Understanding Ni Promotion of MoS2/γ-Al2O3 and its Implications for the Hydrogenation of Phenanthrene

Article abstract of DOI:10.1002/cctc.201500706

The chemical composition and structure of NiMo sulfides supported on γ-Al₂O₃ and its properties for hydrogenation of polyaromatic compounds is explored. The presence of Ni favors the formation of disperse octahedrally coordinated Mo in the oxide precursors and facilitates its reduction during sulfidation. This decreases the particle size of MoS₂ (measured by transmission electron microscopy) and increases the concentration of active sites up to a Ni/(Mo+Ni) atomic ratio of 0.33. At higher Ni loadings, the size of the MoS₂ did not decrease further, although the concentration of adsorption sites and accessible Ni atoms decreased. This is attributed to the formation of NiS_x clusters at the edges of MoS₂. Nickel also interacts with the support, forming separated NiS_x clusters, and is partially incorporated into the γ-Al₂O₃, forming a Ni-spinel. The hydrogenation of phenanthrene follows two pathways; by adding one or two H₂ molecules, 9,10-dihydrophenanthrene or 1,2,3,4-tetrahydrophenanthrene are formed as primary products. Only symmetric hydrogenation, leading to 9,10-dihydrophenanthrene, was observed on unpromoted MoS₂/γ-Al₂O₃. In contrast, symmetric and deep hydrogenation (leading to 9,10-dihydrophenanthrene and 1,2,3,4-tetrahydrophenanthrene, respectively) occur with similar selectivity on Ni-promoted MoS₂/γ-Al₂O₃. The rates of both pathways increase linearly with the concentration of Ni atoms in the catalyst. The higher rates for symmetric hydrogenation are attributed to increasing concentrations of reactive species at the surface, and deep hydrogenation is concluded to be catalyzed by Ni at the edge of MoS₂ slabs. Well, everybody knows that Ni is the word: In promoted MoS₂/γ-Al₂O₃, Ni substitutes Mo at the perimeter of the MoS₂ slabs, forming particles of Ni sulfides with varying sizes at the edges of MoS₂ or on the support. The proportions of these species depend on the Ni content. Ni-substituted sites perform faster and deeper hydrogenation of phenanthrene than non-promoted sites.

Full text of DOI:10.1002/cctc.201500706

DOI: 10.1002/cctc.201500706
Full Papers
Understanding Ni Promotion of MoS /g-Al O and its
2
2
3
Implications for the Hydrogenation of Phenanthrene
Eva Schachtl, Lei Zhong, Elena Kondratieva, Jennifer Hein, Oliver Y. GutiØrrez,*
[
a]
Andreas Jentys, and Johannes A. Lercher*
The chemical composition and structure of NiMo sulfides sup-
tion of phenanthrene follows two pathways; by adding one or
ported on g-Al O and its properties for hydrogenation of poly-
two H molecules, 9,10-dihydrophenanthrene or 1,2,3,4-tetra-
2
3
2
aromatic compounds is explored. The presence of Ni favors
the formation of disperse octahedrally coordinated Mo in the
oxide precursors and facilitates its reduction during sulfidation.
hydrophenanthrene are formed as primary products. Only sym-
metric hydrogenation, leading to 9,10-dihydrophenanthrene,
was observed on unpromoted MoS /g-Al O . In contrast, sym-
2
2
3
This decreases the particle size of MoS (measured by transmis-
metric and deep hydrogenation (leading to 9,10-dihydrophe-
nanthrene and 1,2,3,4-tetrahydrophenanthrene, respectively)
occur with similar selectivity on Ni-promoted MoS /g-Al O .
2
sion electron microscopy) and increases the concentration of
active sites up to a Ni/(Mo+Ni) atomic ratio of 0.33. At higher
2
2
3
Ni loadings, the size of the MoS did not decrease further, al-
The rates of both pathways increase linearly with the concen-
tration of Ni atoms in the catalyst. The higher rates for sym-
metric hydrogenation are attributed to increasing concentra-
tions of reactive species at the surface, and deep hydrogena-
tion is concluded to be catalyzed by Ni at the edge of MoS2
slabs.
2
though the concentration of adsorption sites and accessible Ni
atoms decreased. This is attributed to the formation of NiSx
clusters at the edges of MoS . Nickel also interacts with the
2
support, forming separated NiS clusters, and is partially incor-
x
porated into the g-Al O , forming a Ni-spinel. The hydrogena-
2
3
Introduction
MoS - and WS -based materials are well-established catalysts in
sites and reactant molecules is not fully understood qualita-
tively or quantitatively on an atomistic level. Thus, substantial
efforts have been made towards this goal over the last several
2
2
major industrial processes, ranging from coal liquefaction and
the production of clean fuels in petroleum refineries (hydro-
[
1–3]
treating) to the synthesis of alcohols and thiols.
Addition of
years by imaging MoS -based systems, as well as by theoretical
2
[
4]
Ni or Co helps to increase the rate of hydrotreating reactions.
chemistry describing selected aspects of these catalytic materi-
In consequence, a wide body of information exists with respect
to catalyst formulations and their catalytic properties. It has
been established, for instance, that the maximum promoter
effect is obtained at well-defined Ni or Co concentrations. Vari-
ous structural models have been developed to explain this
als, for example, promoter effect, H activation, desulfurization
2
[
15–17]
mechanisms,
well-controlled morphologies.
Most of the research described above has been stimulated
and in developing synthesis procedures for
[
18,19]
[
20]
by the mandate to reduce the sulfur levels in fuels. Thus, the
effects of promoters have been rationalized for hydrodesulfuri-
zation (HDS) by the presence of (Co)Ni-S-Mo sites with opti-
mum metal–sulfur bond strength, which optimizes the rates of
[5]
[6]
promotion on MoS /g-Al O . The Co-Mo-S model, nowadays
2
2
3
widely accepted and adapted to Ni promotion, proposes sub-
stitution of individual Mo atoms by Co (or Ni) atoms at the
[
21–25]
edges of the MoS slabs. Abundant evidence for this substitu-
adsorption as well as of removal of sulfur.
Similar interpre-
2
[
7]
tion has been provided by diverse methods, including STM,
tations are put forward for direct nitrogen removal in hydrode-
[
8,9]
[10]
[11,12]
[26–28]
STEM,
ed X-ray absorption fine structure (EXAFS) studies.
this consensus, the nature of the interactions between active
DFT calculations, IR spectroscopy,
and extend-
nitrogenation (HDN).
In addition to heteroatom removal,
[13,14]
Despite
the catalysts enable hydrogenation, but the role of the pro-
[
10,24]
moter has not been unequivocally elucidated.
The conven-
tional interpretation invokes adjacent vacancies as sites for ad-
sorption of aromatic rings, but recent results indicate sites
having metal-like character on the plane close to the edge of
[
a] E. Schachtl, Dr. L. Zhong, Dr. E. Kondratieva, J. Hein, Dr. O. Y. GutiØrrez,
Prof. Dr. A. Jentys, Prof. Dr. J. A. Lercher
Department of Chemistry and Catalysis Research Center
Technische Universität München
Lichtenbergstraße 4, 85748 Garching (Germany)
Fax: (+49)8928913544
[
10]
the sulfide particles.
Most of the evidence for the Co-Mo-S model and recent ad-
vances in the understanding of sulfide catalysts has been ob-
tained from model Co-promoted catalysts. This suggests that
the nature of Ni promotion and its concomitant effect on hy-
drogenation must be explored in more detail. One of the ques-
tions that we address is related to the structural and local
E-mail: oliver.gutierrez@mytum.de
johannes.lercher@ch.tum.de
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201500706.
ChemCatChem 2015, 7, 4118 – 4130
4118
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
chemical impact of Ni on the MoS structure. The second ques-
The specific surface area of the g-Al O support was
2
2
3
À1
2
À1
3
tion addresses the correlation of the formation of the Ni-Mo-S
phase with hydrogenation in the absence of defunctionaliza-
tion pathways. Our target reaction, therefore, is the hydroge-
nation of polyaromatics.
248 m g with a pore volume of 0.67 cm g . Impregnation
2
À1
with Mo decreased the specific surface area to 216 m g ,
which corresponded to the value expected considering the
density increase. The pore volume decreased by almost 20%,
suggesting effective deposition of Mo species in the meso-
pores of g-Al O . Further addition of Ni into the Mo-containing
To address these questions, we have synthesized a series of
(
Ni)MoS /g-Al O materials with varying Ni loading at a fixed
2 2 3
2
3
Mo content. These materials were characterized by IR and X-
ray absorption spectroscopy and kinetically explored in the hy-
drogenation of phenanthrene as a model compound for poly-
aromatic hydrocarbons. We provide models for the structure of
materials decreased the surface area and pore volume insignifi-
cantly. Thus, we conclude that deposition of Mo and Ni oc-
curred inside the pores without substantial pore blocking.
Considering the surface area of the support and the metal
loadings, the concentration of Mo in all materials was around
the Ni-MoS phase evolving with increasing concentrations of
2
2
2
Ni and describe the consequences of different levels of promo-
tion on the rates and depth of hydrogenation.
2.3 Mo atoms per nm (2.4 Ni atoms per nm in NiO/g-Al O ).
2 3
The total metal surface density (Mo+Ni) increased from 2.3 in
2
MoO /g-Al O to 2.8, 3.4, 4.2, and 5.8 metal atoms per nm in
3
2
3
Ni(1.5)MoO , Ni(3)MoO , Ni(6)MoO , and Ni(10)MoO , respec-
3
3
3
3
Results
tively. Thus, in all catalysts the metal content was below the
[
29]
value of a monolayer on alumina.
The profiles of the H and H S consumption and evolution
Catalyst characterization
2
2
The materials containing only Mo or Ni are denoted as MoO3/
recorded during selected temperature-programmed sulfidation
(TPS) experiments are compiled in Figure 1 (all profiles are pre-
sented in Figure S4 of the Supporting Information). Four main
g-Al O and NiO/g-Al O in their oxide form and as MoS /g-
2
3
2
3
2
Al O and NiS /g-Al O when sulfided. The bimetallic g-Al O -
2
3
x
2
3
2
3
supported materials are denoted as Ni(x)MoO or Ni(x)MoS in
sections were distinguished in the H
below 373 K ascribed to the desorption of H
lower temperatures, 2) an initial region of H
below 520–545 K, 3) a sharp maximum of H S evolution with
2
S profiles: 1) a small peak
S physisorbed at
S consumption
3
2
the oxide and sulfide forms, respectively, where x is the nomi-
2
nal weight percentage of Ni in the sample. An overview of the
2
elemental composition and textural properties of the g-Al O3
2
2
a temperature characteristic for
the material, and 4) a second
consumption region.
2 3
Table 1. Elemental composition and textural properties of the g-Al O support and all investigated catalyst
oxide precursors.
The processes occurring in
each of these steps are analyzed
in accordance with an earlier
Catalyst
Metal content
wt%] (mmolg
Ni
Ni/(Mo+Ni)
BET surface area
Pore volume
À1
À1
2
À1
3
À1
[
)
[molmol
]
[m g
]
[cm g
]
[
31,32]
Mo
–
report.
During the low tem-
g-Al
2
O
3
–
–
–
248
216
207
198
194
187
210
0.67
0.52
0.54
0.53
0.50
0.47
0.58
perature H S consumption, sec-
2
MoO
3
/g-Al
2
O
3
9.1 (0.95)
8.8 (0.92)
9.1 (0.95)
8.5 (0.89)
8.7 (0.91)
–
–
tion 2, O is gradually exchanged
Ni(1.5)MoO
Ni(3)MoO
Ni(6)MoO
Ni(10)MoO
NiO/g-Al
3
1.3 (0.22)
2.7 (0.46)
5.0 (0.85)
8.8 (1.50)
5.8 (0.97)
0.19
0.33
0.49
0.62
–
6+
by S at Mo , producing H O.
2
3
The following section 3 is char-
3
acterized by rapid H S evolution
3
2
2
O
3
accompanying H2 consumption
and corresponds to the reduc-
support and the oxide catalyst
precursors is given in Table 1.
The Mo content was very similar
among all the Mo-containing
samples and was slightly lower
than the nominal one. The four
different bimetallic Ni(x)MoO3
samples had Ni concentrations
close to the nominal ones, and
the Ni/(Mo+Ni) molar fractions
increased from 0.19 to 0.62. The
Ni content of NiO/g-Al O was
2
3
À1
1
mmolg , that is, identical to
the Mo content in the other
samples.
Figure 1. Profiles of a) H
2 2 3 2 3
and b) H S for the temperature-programmed sulfidation of A) MoO /g-Al O ,
B) Ni(3)MoO , and C) Ni(10)MoO .
3
3
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4119
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
6
+
4+
[30]
tion of Mo to Mo coupled with formation of H S (reduc-
not significant.
The reflections at 2q =36 and 668 that
2
tion temperature T ). In section 4, the rest of O is exchanged
appear in some diffractograms (Figure 3a and 3e) result from
the incomplete removal of SiC used for dilution in the sulfida-
tion procedure.
red
for S.
Although the H S and H consumption and evolution were
2
2
similar, increasing concentrations of Ni lowered Tred (see
The sulfided NiO/g-Al O sample did not exhibit any reflec-
2
3
Figure 2) and accelerated the H S consumption after the reduc-
tion that could be assigned to a crystalline Ni sulfide phase;
2
therefore, we name it NiS /g-Al O . Note that in other samples
x
2
3
specifying the Ni and S stoichiometry implies that the corre-
sponding phase has been identified by XRD (e.g., Ni S ). Note
3
2
that the value of x is between 0.7 and 1.3 as the Ni phases
found in Ni–Mo sulfide catalysts are typically Ni S , Ni S , or
3
2
9 8
[
26]
Ni S . Reflections ascribed to Ni S (2q = 22, 32, 50, 558, Sup-
3
4
3 2
porting Information) appeared only for Ni(10)MoS . For the
2
sake of clarity, a comparison between the X-ray diffractograms
of Ni(10)MoS and g-Al O is presented in Figure S3 (in the Sup-
2
2
3
porting Information).
Typical fringes of MoS were observed in the transmission
2
electron microscopy (TEM) micrographs of all Mo-containing
[
33]
sulfide catalysts. Most of the particles consist of one or two
Figure 2. Reduction temperature, Tred, determined by temperature-pro-
MoS layers with a length range of 2–8 nm. Owing to the flexi-
2
grammed sulfidation of MoO
3
2 3 3 3
/g-Al O ( ), Ni(3)MoO
(^), Ni(1.5)MoO
&
(~),
(^).
ble structure of MoS , the crystals bend (i.e., the slabs are not
2
Ni(6)MoO
3
(*), and Ni(10)MoO
3
straight along the x and y axes) when the slabs grow longer in
the x and y directions than in the z direction. However, signifi-
cant differences in the degree of bending (other than those re-
tion was complete. Both observations suggest an easier reduc-
tion of Mo with increasing Ni loading. The TPS profile of refer-
ence NiO/g-Al O (see the Supporting Information) exhibited
lated to the length) were not observed among the MoS parti-
2
cles on the different catalysts. Representative micrographs as
well as distributions of slab length and stacking, resulting from
the statistical analysis, are shown in Figure S5 and S6 of the
Supporting Information. Table 2 presents the average slab
2
3
a strong H S consumption signal at approximately 500 K. The
2
total H S consumption by Ni(x)MoS increased with the content
2
2
of Ni, which is in line with the higher overall metal loading.
Further characterization of the oxide precursors is given in the
Supporting Information. Results corresponding to sulfide mate-
rials are described in the following.
Table 2. Average slab length, stacking degree, and dispersion fMo of the
supported MoS phase, determined by statistical analysis of TEM micro-
2
graphs (standard deviation given in brackets). NO adsorption measured
by pulse experiments at RT.
X-ray diffractograms of the sulfide catalysts are depicted in
Figure 3. All samples showed the diffraction peaks characteris-
tic for MoS at 2q=33 and 598. The absence of a peak at 2q
Catalyst
Slab length
nm]
Stacking
f
Mo
NO
[mmolg ]
2
À1
[
=
148, corresponding to the (002) plane of MoS with an inter-
2
planar distance of 6.1 Š, indicated that stacking of MoS was
MoS
Ni(1.5)MoS
Ni(3)MoS
Ni(6)MoS
Ni(10)MoS
2
/g-Al
2
O
3
5.5 (2.2)
4.7 (1.9)
4.5 (1.9)
4.5 (1.9)
4.6 (1.9)
1.5 (0.7)
1.6 (0.8)
1.9 (0.9)
1.7 (0.9)
1.8 (1.1)
0.22
0.26
0.26
0.26
0.26
151
231
298
314
407
2
2
2
2
2
length and stacking degree of the supported MoS phase on
2
all catalysts. The average length of the MoS slabs decreased
2
upon the addition of Ni from 5.5 nm for MoS /g-Al O to
2
2
3
4
.7 nm for Ni(1.5)MoS . At the same time, the distribution of
2
the slab length became narrower in the presence of Ni, giving
rise to a lower standard deviation (graphical display in the Sup-
porting Information, Figure S6). Interestingly, increasing the Ni/
(
Mo+Ni) ratio did not have a significant effect on the average
slab length. The stacking degree, on the other hand, increased
slightly with increasing Ni loading from 1.5 to 1.9 (MoS /g-
2
Al O and Ni(3)MoS , respectively) and remained constant at
2
3
2
Figure 3. X-ray diffractograms of the sulfide catalysts: a) MoS
b) Ni(1.5)MoS , c) Ni(3)MoS , d) Ni(6)MoS , e) Ni(10)MoS , f) NiS
marked reflections correspond to MoS
(*) and Ni
(!). Unmarked reflec-
tions are assigned to the g-Al support.
2
/g-Al
2 3
O ,
higher loadings.
2
2
2
2
x
/g-Al
2 3
O . The
Using the structural information obtained from the analysis
of the TEM micrographs, we determined the fraction of Mo
2
3 2
S
2
O
3
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4120
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
atoms at the edges of the sulfide slab, f , assuming a perfect
fore, NO adsorption can be considered as a method to probe
the adsorption sites (and concomitantly dispersion) of all
phases present in the catalysts (e.g., Ni-MoS and NiS ), even
Mo
[34,35]
hexagonal geometry for the MoS crystals.
Recent investi-
2
gations suggest that the morphology of MoS might deviate
2
2
x
[10,36]
from a perfectly hexagonal one.
However, the assumption
though not all of them are catalytically active for hydrogena-
tion.
of hexagonal shape allows us to determine the dispersion by
using the thermodynamically most stable structure of MoS as
The IR spectra of CO adsorbed on g-Al O , MoS /g-Al O , and
2
2
3
2
2
3
a model. The fMo value reflects the dispersion of MoS and, con-
NiS /g-Al O are shown in Figure S7 of the Supporting Informa-
x 2 3
2
À1
comitantly, is a semi-quantitative measure of the active sites.
tion. For alumina, two bands appeared at 2192 and 2156 cm ,
3
+
The fMo values for all catalysts are reported in Table 2. The pres-
corresponding to CO adsorbed on Lewis (n(CO/Al )) and
[
44]
ence of Ni in MoS increased fMo from 0.22 to 0.26.
Brønsted acid sites (n(CO/OH)), respectively. The intensity of
these bands decreased in the spectra of MoS /g-Al O and
2
The concentration of NO adsorbed on the sulfide catalysts at
2
2
3
room temperature is compiled in Table 2. The reference NiS /g-
NiS /g-Al O , most likely owing to sulfide phases covering
x 2 3
x
Al O adsorbed 351 mmol of NO per gram of material. Compar-
these sites on alumina. Regarding CO adsorbed on the sulfide
2
3
ing MoS /g-Al O and Ni(1.5)MoS , it was evident that the addi-
phase, for MoS /g-Al O (Figure S7 or Figure 5a) two bands, at
2
2
3
2
2
2
3
tion of small concentrations of Ni (1.5 wt%) had a dramatic
impact on the adsorption capacity of MoS , increasing the
2
uptake of NO by 50%, in agreement with literature exam-
[
37–40]
ples.
Further Ni addition steadily increased the NO uptake
of the samples.
The concentration of adsorbed NO reflects the concentration
of coordinatively unsaturated sites (CUS) in the MoS2
[
22,40–43]
phase.
However, the NiS /g-Al O reference material also
x 2 3
adsorbed a significant amount of NO. Thus, the concentration
of adsorbed NO on Ni(x)MoS requires some caution in inter-
2
pretation, as increasing concentrations of NiS were observed
x
with increasing Ni loading.
To account for the effect of increasing concentrations of NiSx
species, the overall NO uptake was normalized per mol of
metal in the samples. Figure 4 shows the dependence of the
Figure 5. Spectra of CO adsorbed on Ni(x)MoS
2
/g-Al
2
O
3
: a) MoS
2
/g-Al
2
O
3
,
b) Ni(1.5)MoS
2
, c) Ni(3)MoS
2
, d) Ni(6)MoS
2
, e) Ni(10)MoS
2
, and f) NiS
x
/g-Al
2
O
3
.
À1
2
106 (high intensity) and 2094 cm (appearing as a shoulder)
were observed. The former corresponded to CO adsorbed on
Mo located at the edge of MoS terminated by metal cations
2
and the latter was attributed to CO adsorbed on Mo at the
[
44]
sulfur edge of MoS2. Both were identified as active sites for
Figure 4. Concentration of NO adsorbed on the catalysts normalized to the
[
45,46]
hydrodesulfurization.
In addition to the bands of CO ad-
metal content of each sample: MoS
~), Ni(6)MoS
(*), and Ni(10)MoS
(^).
2 2 3 2 2
/g-Al O ( ), Ni(3)MoS
(^), Ni(1.5)MoS
&
(
2
2
sorbed on the support, the spectra of NiS /g-Al O (Figure S7
x
2
3
and Figure 5 f) exhibited an intense asymmetric band at
À1
À1
2
080 cm with shoulders at 2056 and 2025 cm . The main
À1
normalized concentration of adsorbed NO on the Ni/(Mo+Ni)
fraction. The NO uptake increased with the Ni content, peaking
at 3 wt% Ni. This indicated that Ni promotion increased the
concentration of CUS on the catalysts up to a certain Ni con-
tent. The decrease in the concentration of adsorbed NO at
higher loadings (instead of leveling off as expected from the
small changes in dispersion observed by TEM) is attributed to
band (2080 cm ) is attributed to CO adsorbed on NiS , the
x
[
12,47,48]
shoulders to polycarbonyl species (Ni(CO)x).
The spectra of CO adsorbed on Ni(x)MoS2 (Figure 5b–e)
were complex and cannot be understood by combinations of
the spectra of the components (Figure S7 or Figure 5a/f). The
spectra of Ni(x)MoS had a complex group of bands between
2
À1
2140 and 1950 cm . Based on separate spectroscopic and DFT
[
12]
the formation of large NiS particles, which reduce the fraction
studies, as well as considering the spectra of reference mate-
rials, adsorption of CO on five different sites has been hypothe-
x
of surface available to adsorb NO per gram of catalyst. There-
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4121
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
sized to contribute to these bands. The adsorbed species iden-
tified were (i) CO adsorbed on Ni atoms decorating MoS2
À1
(
2118 cm ), (ii) CO adsorbed on non-promoted Mo at the M-
À1
À1
edge of MoS2 (2106 cm ), (iii) between 2080–2020 cm the
group of CO bands was concluded to be from CO adsorbed on
NiS species, on Mo at the S-edge of MoS , and on promoted
x
2
Mo sites. The spectra of Ni(x)MoS (Figure 5) showed clearly
2
how the increasing concentration of Ni shifted the maximum
À1
of CO adsorption from 2106 cm (non-promoted MoS ) to
2
À1
2
080 cm (NiS and promoted MoS ). This indicated that the
x
2
accessible surface of the catalysts at high Ni loading was domi-
nated by NiS . To quantify this, the areas of bands were fitted
x
by deconvolution and the results of the deconvolution are
summarized in detail in the Supporting Information (Figure S8
and Table S1).
Figure 7. A) Mo K-edge XANES and corresponding B) Fourier transforms of
Figure 6 shows the concentrations of CO calculated from the
3
k -weighted EXAFS of: a) reference MoS
2
, b) MoS
2
/g-Al
2
O
3
, c) Ni(1.5)MoS
2
,
À1
intensities of the bands at 2118 cm (associated with CO on
2 2 2
d) Ni(3)MoS , e) Ni(6)MoS , and f) Ni(10)MoS .
catalyst samples and reference MoS . According to the well-
2
known 2H-MoS2 structure, the first contribution at approxi-
mately 1.9 Š (not phase shift corrected) belonged to Mo–S and
the second at approximately 2.8 Š (not phase shift corrected)
[
13]
to Mo–Mo distances. The Mo–Mo paths of the catalysts are
less intense than those of bulk MoS , because the reference
2
material has a considerably larger crystal size and much less
structural disorder (bending and low-coordinated Mo atoms)
than the supported (Ni)MoS particles. Thus, a large proportion
2
of Mo atoms are positioned within well-defined bond lengths
and angles, which enhances the characteristic Mo–Mo scatter-
ing.
Figure 6. Relative concentrations of CO adsorbed on A) Ni atoms and
B) non-promoted Mo sites: MoS
Ni(6)MoS
(*), and Ni(10)MoS
(^).
2 2 3 2 2
/g-Al O ( ), Ni(3)MoS
(^), Ni(1.5)MoS
&
(~),
2
2
Figure 8 depicts the XANES and FT-EXAFS data at the Ni K-
edge. The sample Ni(1.5)MoS was omitted owing to the inade-
2
À1
Ni atoms decorating MoS ), and at 2106 cm (assigned to
quate quality of the spectra caused by the low concentration
of Ni (see Figure S9 in the Supporting Information). A small
pre-edge at around 8333 eV was visible for Ni S and the sul-
2
non-promoted Mo at the M-edge of MoS ). The bands in the
2
À1
2
080–2020 cm
region were not quantitatively analyzed
3
2
owing to strong overlapping of bands from different species.
fide catalysts (shown in the inset (i) of Figure 8), which is typi-
cal for tetrahedrally or pentagonally coordinated Ni. The refer-
Figure 6 shows that the concentration of CO on Ni atoms
À1
(
2118 cm ) was the highest for Ni(3)MoS , indicating an opti-
2
mum decoration of MoS at a molar fraction of Ni of 0.33. The
2
À1
concentration of CO on non-promoted Mo (2106 cm ), how-
ever, decreased steadily with increasing Ni loading and
reached values close to zero for Ni(10)MoS . This indicates that
2
at Ni/(Mo+Ni) ratios above 0.33, both the decoration of MoS2
with Ni and the fraction of MoS without contact to the Ni pro-
2
moter decreased. These seemingly contradicting observations
were related to changes in structure and diverse promotion
mechanisms that were not related to Ni decoration (see
below). Not surprisingly, the steady increase in the intensity of
À1
the bands at 2080–2020 cm indicated that the concentration
of CO adsorbed on promoted Mo and/or NiS species increased
x
with increasing Ni loading.
The X-ray absorption near-edge structure (XANES) and the
corresponding Fourier transformed (FT) extended X-ray absorp-
tion fine structure (EXAFS) at the Mo K-edge are presented in
Figure 7. The absorption edge energy and the local environ-
ment around Mo were almost identical between the sulfide
3
Figure 8. A) Ni K-edge XANES and B) FT of k -weighted EXAFS of: a) refer-
ence Ni
3
S
2
, b) reference NiAl
2
O
4
, c) NiS
x
/g-Al
2
O
3
, d) Ni(3)MoS
2
, e) Ni(6)MoS
2
,
and f) Ni(10)MoS
samples (c–f).
2
. The inset (i) displays the pre-edge found in Ni S
3 2
(a) and
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4122
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ence NiAl O4 spinel displayed a quite sharp feature at the
2
3
Table 4. Best-fit results for k -weighted EXAFS data of sulfided catalysts
edge, which was observed to some extent also in NiS /g-Al O
3
2
x
2
at the Ni K-edge in k-space. N: coordination number; R: distance; s :
and in the Ni(x)MoS materials with a lower intensity. This im-
Debye–Waller factor; E : inner potential.
2
0
plied that a small fraction of Ni was always present as NiAl O4
2
2
2
Catalyst
NiS /g-Al
Path
N
R [Š]
s [Š ]
spinel, in line with the UV/Vis spectra (see the Supporting In-
[
49]
formation). The absorption edge energy and the local envi-
ronment around Ni for the sulfide catalysts were similar to the
x
2
O
3
Ni–S
1.6Æ0.1
1.0Æ0.1
4.4Æ0.2
0.7Æ0.3
2.22Æ0.02
2.61Æ0.02
2.11Æ0.02
3.14Æ0.04
0.0038Æ0.0022
0.0058Æ0.0021
0.0071Æ0.0049
0.0054Æ0.0030
(
(
R=0.030)
=8.0Æ2.5)
Ni–Ni
Ni–O
Ni–Ni
E
0
Ni3S reference, in agreement with the XRD results of the
2
sample with the highest Ni loading. The structure of Ni S con-
3
2
Ni(3)MoS
2
Ni–S
2.4Æ0.1
1.5Æ0.2
1.0Æ0.2
0.6Æ0.1
0.3Æ0.2
2.21Æ0.02
2.58Æ0.04
2.00Æ0.04
3.07Æ0.04
2.81Æ0.06
0.0042Æ0.0005
0.0067Æ0.0046
0.0079Æ0.0052
0.0070Æ0.0028
0.0073Æ0.0068
tains four tetrahedrally coordinated S atoms in the first shell at
(
(
R=0.007)
=1.0Æ2.9)
Ni–Ni
Ni–O
Ni–Ni
Ni–Mo
2
.3 Š, and four Ni atoms at 2.5 Š in a second coordination
E
0
[
50]
shell.
Owing to the closeness of these contributions, one
broad and intense signal from backscattering of Ni-Ni and Ni-S
pathwas appeared in EXAFS. The contributions observed for
the sulfide catalysts had lower intensities and were shifted to
Ni(6)MoS
R=0.031)
2
Ni–S
2.7Æ0.4
0.3Æ0.2
1.6Æ0.3
0.5Æ0.3
0.3Æ0.2
2.27Æ0.02
2.61Æ0.02
2.06Æ0.01
3.07Æ0.07
2.82Æ0.02
0.0091Æ0.0011
0.0072Æ0.0029
0.0021Æ0.0008
0.0059Æ0.0023
0.0078Æ0.0021
(
Ni–Ni
Ni–O
Ni–Ni
Ni–Mo
shorter distances. Moreover, the NiÀNi contribution of Ni S at
3
2
(E =2.7Æ2.3)
0
3
.7 Š (not phase shift corrected) was absent.
Linear combination fitting (LCF) was performed to obtain in-
formation on the chemical environment of the metal species
in the catalysts, which is required during the EXAFS fitting.
This analysis, shown in the Supporting Information, revealed
a mixture of phases in the catalysts. Thus, the XANES was mod-
eled with small fractions of oxide species and a new, bimetallic
phase in addition to the sulfide phases MoS or Ni S . The un-
Ni(10)MoS
R=0.016)
2
Ni–S
2.4Æ0.2
0.6Æ0.1
1.5Æ0.1
0.5Æ0.2
0.4Æ0.2
2.28Æ0.01
2.57Æ0.02
2.08Æ0.01
3.14Æ0.02
2.84Æ0.02
0.0091Æ0.0011
0.0072Æ0.0021
0.0021Æ0.0008
0.0056Æ0.0023
0.0078Æ0.0021
(
Ni–Ni
Ni–O
Ni–Ni
Ni–Mo
(E =9.1Æ1.3)
0
2
3 2
known phase has been identified as the Ni-Mo-S phase with
neighboring Mo and Ni atoms, which also contributed to the
The average coordination numbers N
=4.8 and N
=
MoÀS
MoÀMo
[26]
XANES of the sulfide catalysts.
3.9 were found for MoS /g-Al O , which are low compared to
2
2
3
Therefore, MoÀNi and NiÀMo scattering contributions were
the reference MoS2 with N
=5.8 and N
=6.0 (see
MoÀS
MoÀMo
included in the multi-edge fitting of the EXAFS data. Ni S was
Table S4 in the Supporting Information). The addition of Ni led
to an increase in the coordination numbers of Mo. A maximum
3
2
described by four different contributions, whereas for the sul-
fide catalysts only one NiÀS path at ~2.3 Š and one NiÀNi path
was observed for Ni(3)MoS with N
=6.0 and N
=4.8.
MoÀMo
2
MoÀS
[26]
at ~2.6 Š were distinguishable. Moreover, a NiÀO path at
For all Ni-containing catalysts, a Mo–Ni contribution was found
~
2.0 Š and a NiÀNi path at ~3.1 Š were introduced to account
with N =0.2 at ~2.8 Š.
MoÀNi
3
for NiÀO interactions. The best-fit results for the k -weighted
EXAFS data at the Mo K-edge and the Ni K-edge are presented
in Table 3 and Table 4, respectively. The corresponding fitted
EXAFS spectra are depicted in the Supporting Information.
The first coordination shell of Ni in the sulfide catalysts con-
sisted of oxygen, with N =1.0 for Ni(3)MoS , and N =4.4
NiÀO
2
NiÀO
for NiS /g-Al O . The Ni–Ni path corresponding to NiAl O
x
2
3
2
4
[
26]
spinel was found to be 0.6 (N ) for Ni(3)MoS2. The second
NiÀNi
coordination shell of Ni consisted of S atoms at ~2.25 Š, which
were attributed to Ni S . The N
value for Ni(x)MoS is be-
3
2
NiÀS
2
tween 2.4–2.7. The Ni–Ni distance of Ni S followed in the third
3
2
3
Table 3. Best-fit results for k -weighted EXAFS data of sulfided catalysts
coordination shell at ~2.6 Š with coordination numbers rang-
ing from 0.3 to 1.5. Hence, the coordination numbers for Ni–S
and Ni–Ni are remarkably low for the sulfide catalyst materials
2
at the Mo K-edge in k-space. N: coordination number; R: distance; s :
Debye–Waller factor; E
0
: inner potential.
2
2
Catalyst
MoS /g-Al
Path
N
R [Š]
s [Š ]
compared with reference Ni S (N =N
3
2
NiÀS
=4, see Table S4 in
NiÀNi
the Supporting Information). A Ni–Mo contribution was found
2
2
O
3
Mo–S
Mo–Mo
4.8Æ0.2
3.9Æ0.2
2.41Æ0.01
3.16Æ0.01
0.0026Æ0.0002
0.0033Æ0.0002
(
(
R=0.019)
=2.2Æ1.0)
for all Ni(x)MoS catalysts at ~2.8 Š, being 0.3 for Ni(3)MoS2
2
E
0
and Ni(6)MoS and 0.4 for Ni(10)MoS .
2
2
Ni(3)MoS
2
Mo–S
Mo–Mo
Mo–Ni
6.0Æ0.2
4.8Æ0.2
0.2Æ0.1
2.40Æ0.01
3.17Æ0.01
2.81Æ0.05
0.0031Æ0.0002
0.0036Æ0.0002
0.0075Æ0.0050
(
(
R=0.038)
=1.8Æ1.1)
Hydrogenation of phenanthrene
E
0
The hydrogenation activity of the series of (Ni)MoS catalysts
2
Ni(6)MoS
2
Mo–S
Mo–Mo
Mo–Ni
5.5Æ0.2
4.4Æ0.2
0.2Æ0.1
2.42Æ0.01
3.18Æ0.01
2.82Æ0.06
0.0020Æ0.0003
0.0029Æ0.0003
0.0089Æ0.0078
was evaluated through the conversion of phenanthrene as
a model compound that has three condensed aromatic rings.
The conversion of phenanthrene on all sulfide catalysts as
a function of space-time is presented in Figure S13 of the Sup-
porting Information (points represent experimental values,
lines display the fitting model described below). The addition
(
(
R=0.01)
=2.2Æ1.4)
E
0
Ni(10)MoS
2
Mo–S
Mo–Mo
Mo–Ni
5.4Æ0.2
4.3Æ0.2
0.2Æ0.1
2.41Æ0.01
3.17Æ0.01
2.84Æ0.06
0.0022Æ0.0003
0.0027Æ0.0003
0.0073Æ0.0070
(
(
R=0.01)
=2.8Æ1.5)
E
0
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4123
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
conversion plots in Figure 9(A–C) and the selectivity–conver-
sion plots in Figure 9(D–F) show that DiHPhe and TetHPhe are
Table 5. Reaction rate (r) and rate constants (k
drogenation of phenanthrene. The unit of r and k
i
) determined for the hy-
À1
i
is molPhe (h·gcat) .
[
52,53]
the primary products, in agreement with previous studies.
With MoS /g-Al O , the selectivity to DiHPhe and TetHPhe re-
3
3
3
3
3
Catalyst
MoS /g-Al
r”10
k
1
”10
k
2
”10
k
3
”10
k
4
”10
1 2
k /k
2
2
3
2
2
O
3
0.58
1.13
1.54
1.26
1.08
0.87
1.10
1.35
1.07
0.90
0.10
0.81
1.29
1.03
0.87
0
0
8.6
1.4
1.0
1.0
1.0
mained constant at 90 and 10%, respectively, in the studied
range of conversion. With Ni(3)MoS2, Ni(6)MoS2, and
Ni(1.5)MoS
2
1.14
1.58
1.35
1.31
3.63
5.18
4.42
4.02
Ni(3)MoS
Ni(6)MoS
2
Ni(10)MoS , the initial selectivity towards DiHPhe and TetHPhe
2
2
was almost 50% (Figure 9(F)). The product distribution of
Ni(10)MoS
2
Ni(1.5)MoS2 was in between those of MoS /g-Al O and the
2
2
3
other Ni(x)MoS catalysts. The products sym- and asymOHPhe
2
(
detected with Ni(x)MoS catalysts) were regarded as secondary
2
of Ni significantly increased the rate of phenanthrene hydroge-
nation (Table 5). For instance, at low conversions the rate of
products based on the yield- and selectivity–conversion plots
in Figure 9 (their initial selectivities equal zero when extrapolat-
ed to zero conversion).
phenanthrene
hydrogenation
reached
0.58”
À3
À1
1
0
molPhe (h·g ) for MoS /g-Al O , and about twice that
To determine the origins of sym- and asymOHPhe, let us
cat
2
2
3
À3
À1
value with Ni(1.5)MoS2 (1.13”10 molPhe (h·g ) ). The most
comment on the selectivity–conversion plot of Ni(6)MoS (Fig-
cat
2
active catalyst was Ni(3)MoS , which led to a reaction rate of
ure S15 in the Supporting Information). The selectivity to
DiHPhe and TetHPhe (primary products) decreased steadily
with increasing phenanthrene conversion. Note that at, for ex-
ample, 17% of phenanthrene conversion, the decrease in the
selectivity to DiHPhe (compared with the initial selectivity) was
3%, which perfectly corresponded to the increase in selectivity
for asymOHPhe. On the other hand, the decrease of TetHPhe
selectivity equaled the increase of symOHPhe selectivity, the
differences (compared with initial selectivities) being 9% each.
These observations allowed us to conclude that phenanthrene
was hydrogenated in two parallel reaction pathways. One, sub-
sequently referred to as “symmetric hydrogenation”, comprised
consecutive hydrogenation to DiHPhe and asymOHPhe. The
second pathway, “deep hydrogenation”, proceeded from phen-
anthrene to symOHPhe via TetHPhe. Further hydrogenation
2
À3
À1
1
.54”10 molPhe (h·g ) . Increasing the amount of Ni further
cat
À3
À1
decreased the rates to 1.26”10 molPhe (h·g ) and 1.08”
1
cat
À3
À1
0
molPhe (h·g ) for Ni(6)MoS and Ni(10)MoS , respectively.
cat 2 2
It is important to mention that NiS /g-Al O did not show any
x
2
3
activity for phenanthrene hydrogenation under the applied re-
action conditions.
Under the present experimental conditions, only hydrogena-
tion was observed; ring opening and hydrogenolysis were not
detected. The products observed during the activity tests were
9
,10-dihydrophenanthrene (DiHPhe), 1,2,3,4-tetrahydrophenan-
threne (TetHPhe), 1,2,3,4,5,6,7,8-octahydrophenanthrene (sy-
mOHPhe), and 1,2,3,4,4a,9,10,10a-octahydrophenanthrene (asy-
mOHPhe). The fully hydrogenated product, perhydrophenan-
threne, was not observed. The formation rates and hydrogena-
tion depth of the products dra-
matically depended on the
presence of Ni. In Figure S14 of
the Supporting Information, the
yields for MoS /g-Al O and
2
2
3
Ni(x)MoS are shown. In the case
2
of MoS /g-Al O , only two prod-
2
2
3
ucts appeared; DiHPhe and
TetHPhe. With Ni(x)MoS2, the
deeper hydrogenated products,
symOHPhe and asymOHPhe,
were also produced.
To develop the reaction net-
work, the experimental data was
analyzed by using the delplot
[
51]
technique.
The product yield
and selectivity observed with
MoS /g-Al O , Ni(1.5)MoS2, and
2
2
3
Ni(6)MoS are presented as func-
2
tions of phenanthrene conver-
sion in Figure 9. The depend-
ence of yield and selectivity on
conversion for Ni(3)MoS2 and
Ni(10)MoS were almost identical
2
Figure 9. Yield and selectivity of the products of phenanthrene hydrogenation as a function of conversion. Yield:
A) MoS /g-Al O , B) Ni(1.5)MoS , and C) Ni(6)MoS . Selectivity: D) MoS /g-Al O , E) Ni(1.5)MoS , and F) Ni(6)MoS .
to those for Ni(6)MoS2 and,
2
2
3
2
2
2
2
3
2
2
therefore, not shown. The yield– Phenanthrene hydrogenation products: DiHPhe (^), TetHPhe (~), symOHPhe (~), asymOHPhe (^).
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4124
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
The presence of Ni decreased the size of MoS slabs (TEM),
2
but slightly increased stacking. The concentration of Ni did not
influence this effect. The fraction of Mo at the MoS -edges (f )
2
Mo
followed this variation perfectly (Table 2). Thus, the structural
impact of Ni was confined to a slight reduction in size and an
equally slight increase in the stacking degree. In contrast, the
concentration of CUS increased linearly with increasing con-
centration of metals.
Figure 10. Reaction network for the hydrogenation of phenanthrene under
the present reaction conditions. Phe: phenanthrene, DiHPhe: 9,10-dihydro-
phenanthrene, TetHPhe: 1,2,3,4-tetrahydrophenanthrene, symOHPhe:
The IR spectra of adsorbed CO allowed us to conclude that
Ni decorated MoS (Ni was incorporated at the edge of MoS ,
2
2
forming the so-called Ni-Mo-S phase), leading to a variety of
adsorption sites, depending on the concentration of Ni. The
1,2,3,4,5,6,7,8-octahydrophenanthrene, asymOHPhe: 1,2,3,4,4a,9,10,10a-octa-
hydrophenanthrene.
presence of Ni incorporated into the MoS edge was further
2
confirmed by the presence of NiÀMo contributions in EXAFS.
The distance and coordination number of this contribution
was in good agreement with a model of direct substitution of
was not observed, for example, formation of perhydrophenan-
threne, probably owing to the low phenanthrene conversions
accessed in this work. The low conversions also allowed rever-
sible reactions to be neglected, as the concentrations of the
products were well below equilibrium compositions (Support-
ing Information). The reaction network corresponding to this
description is shown in Figure 10.
[
14,26,49]
Ni for Mo in the whole edge structure.
The concentration of incorporated Ni atoms reached a maxi-
mum at 3 wt% Ni, whereas the concentration of parent MoS2
sites decreased with Ni concentration (see Figure 6). The quan-
titative variations did not complement each other, as one
would expect that the concentrations of promoted Ni atoms
and non-promoted Mo show opposite trends. Therefore, we
conclude that further surface species must exist at high Ni
loadings, and that at least a small fraction of the incorporated
Ni atoms is negatively affected by high concentrations. The for-
mation of this additional phase is concluded to begin at rela-
tively low concentrations of Ni because of the presence of
a significant fraction of non-promoted Mo atoms. A simple
mass balance, therefore, points to the existence of separate
The apparent rate constants for each step of the reaction
network were calculated by assuming first order in the hydro-
carbons and zero order in H in all steps (the hydrogen was
2
[4]
present in great excess). The network in Figure 10 is repre-
sented by the set of differential equations shown in the Sup-
porting Information [Equations (3)–(7)]. The resulting rate con-
stants are reported in Table 5. Over MoS /g-Al O , the reaction
2
2
3
toward DiHPhe (k ) was considerably faster than that to TetH-
1
Phe (k ), giving k /k =8.6. Upon addition of Ni, k and k in-
2
1
2
1
2
creased. The former increased by 26, 55, 23, and 4% for
Ni(1.5)MoS , Ni(3)MoS , Ni(6)MoS , and Ni(10)MoS , respectively.
NiS clusters. The negative impact of this phase on the concen-
x
tration of accessible Ni in the edge, suggests that at least
2
2
2
2
The increase of k was much higher, that is, 7, 12, 9, and 8
a part of it grows from the edge of the MoS crystals, not
2
2
[
55]
times for Ni(1.5)MoS , Ni(3)MoS , Ni(6)MoS , and Ni(10)MoS , re-
unlike early postulates by Delmon et al. and van der Kraan
2
2
2
2
[
56,57]
spectively. This marked difference in the increase of k led to
et al.
for Co-promoted catalysts. We hypothesize that Ni
2
the k /k ratio of 1.4 for Ni(1.5)MoS and 1.0 for the catalysts
cations incorporated in the MoS edge act as nucleation sites
1
2
2
2
with higher Ni content. The effect of Ni addition on the forma-
for NiS . This process would rapidly turn the Ni-decorated sites
x
tion of OHPhe (k and k ) was even more dramatic, as forma-
into NiSx clusters, but would not hinder the promotion of
nearby Mo sites. Both sorption of CO and the EXAFS analysis
3
4
tion of these products did not occur with MoS /g-Al O .
2
2
3
confirm their presence. During the adsorption of CO, the inten-
À1
sity of the 2084 cm band (carbonyls on NiS ) strongly in-
x
Discussion
creased with the addition of Ni above 3 wt%. On the other
hand, the rather low NiÀS and NiÀNi coordination numbers in
EXAFS analysis compared with reference Ni S pointed to an
State of Ni and its effect on the structure of MoS /g-Al O
3
2
2
3
2
[
26,49]
The chemical composition strongly influences the physico-
chemical and catalytic properties of the mixed sulfides. The in-
teractions between Ni and Mo were observed even in the
important contribution of very small Ni S particles.
The
3
2
coordination number N
decreased for high Ni concentra-
NiÀNi
tions. This was explained by the formation of NiS clusters with
x
oxide precursors. Ni disperses agglomerated MoO species (in
a broad distribution of particle sizes and NiÀNi distances
(many kinds of Ni sulfides are stable under the experimental
conditions). It should also be emphasized at this point that an
important fraction of Ni forms a NiAl O spinel (evidence from
3
the XRD patterns, reflections from MoO are present in MoO /
3
3
g-Al O but not in Ni(x)MoO /g-Al O , see the Supporting Infor-
2
3
3
2
3
mation), and decreases its interaction with the support, favor-
2
4
[
40,54]
ing the formation of O Mo species (UV/Vis spectroscopy).
UV/Vis spectroscopy of the oxides and XAS measurements in
the sulfide catalysts). Thus, Ni was present in three phases in
the sulfided catalysts, that is, Ni atoms incorporated into the
edges of MoS , NiS (segregated on the support or as small
h
Weaker interactions with the support facilitated the reduction
of Mo oxide as shown by the reduction at lower temperature
during TPS. This suggests that proximity between Ni and Mo
exists in the oxide precursors. This proximity is maintained in
the sulfide form as all characterizations suggest.
2
x
clusters at the MoS edges), and NiAl O spinel.
2
2
4
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4125
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 11. Visualization of the Ni-Mo-S and NiS
x
phases in a) Ni(3)MoS
2
, b) Ni(6)MoS
2
, and c) Ni(10)MoS
2
. Yellow: S, light blue: Mo, dark blue: Ni.
Visualizations constructed by Accelrys Material Studio 7.0
software (see Figure 11 and the Supporting Information) exem-
Table 6. Comparison of the coordination numbers determined by
a) EXAFS fitting and b) the coordination numbers calculated from the cor-
responding models in Figures S16 and 11.
plify such a situation. The Ni-Mo-S phase in Ni(3)MoS is con-
2
sidered to exhibit the highest proportion of decorating Ni
atoms without full decoration or full replacement of Mo at the
edges. The excess of Ni, which was not present in the spinel,
Catalyst
MoS /Al
N
MoÀS
N
MoÀMo
N
NiÀS
N
NiÀNi
N
MoÀNi
N
NiÀMo
2
2
O
3
a) 4.8Æ0.2 3.9Æ0.2 –
b) 5.5 5.8
–
–
–
–
–
–
formed mainly NiS clusters on the support. In the Ni(6)MoS2
–
x
Ni(3)MoS
Ni(6)MoS
2
2
a) 6.0Æ0.2 4.8Æ0.2 2.4Æ0.1 1.5Æ0.2 0.2Æ0.2 0.3Æ0.2
b) 5.8 5.3 2.9 2.2 0.3 0.6
a) 5.5Æ0.2 4.4Æ0.2 2.7Æ0.4 0.3Æ0.2 0.2Æ0.2 0.3Æ0.2
b) 5.9 5.3 2.9 2.4 0.3 0.4
a) 5.4Æ0.2 4.3Æ0.2 2.4Æ0.2 0.6Æ0.1 0.2Æ0.1 0.4Æ0.2
b) 5.9 5.3 3.0 2.7 0.4 0.6
model, more Ni atoms were incorporated at the edges, which
increased the concentration of Mo atoms adjacent to Ni. How-
ever, clusters of NiS species also formed at the edge, which re-
x
duced the net concentration of exposed Ni atoms. The Ni sul-
Ni(10)MoS
2
fide species not associated with the MoS edges also grew in
2
size and abundance. In the Ni(10)MoS model, full decoration
2
was virtually reached, as there was a minimum of Mo atoms
exposed at the edges. However, it was far from an ideal deco-
ration, as many of the decorating species were not single Ni
models were in good agreement with the ones resulting from
the EXASF fitting, although the numbers for the models were
atoms, but NiS clusters. This drastically decreased the concen-
slightly higher. The N
values of the models exceed the
NiÀNi
x
tration of exposed Ni atoms. In addition, the concentration of
values from EXAFS analysis, especially in case of Ni(6)MoS and
2
Ni was so high that NiS species agglomerate to form crystal-
Ni(10)MoS . This was attributed to the large variety of NiS spe-
x
2
x
line species detectable by XRD.
cies that formed at the edges or on the support. Contributions
of all those species (with different NiÀNi distances) broadened
the XAS spectra, driving the fitting to underestimate the NiÀNi
The quantification of Mo and Ni atoms in the models is pre-
sented in Table S7 of the Supporting Information. Table S7 also
shows a comparison of the Ni/(Mo+Ni) ratios determined by
elemental analysis and those corresponding to the models in
Figures 11 and S16 (in the Supporting Information). The ratios
of the models were lower than determined by elemental analy-
sis, because a fraction of Ni is present in the NiAl O spinel,
coordination numbers strongly. The N
and N
values of
NiÀMo
MoÀNi
the models are generally in accordance with the EXAFS fitting.
The coordination numbers N were 5.7 and 5.8 for the
MoÀMo
MoS -14”14 and MoS -17”17 models, respectively (Table S6 in
2
2
the Supporting Information). These values were higher than
those determined in the EXAFS fitting, which were 3.9 (MoS2/
g-Al O ) and 4.3–4.8 (Ni(x)MoS /g-Al O ). This constant discrep-
2
4
which was not considered for the construction of the models.
The deviation of the Ni/(Ni+Mo) ratios in the models and
those from elemental content increased with Ni loading, be-
cause more Ni was engaged with the support. The atomic con-
tents of the models (Table S7), in combination with experimen-
tal elemental analysis, allowed an estimation of the concentra-
2
3
2
2
3
ancy was attributed to structural disorder of MoS in the sam-
2
ples, which led to underestimation of N
values (correlated
MoÀMo
[
58]
with the particle size). The N
value of MoS /g-Al O was
MoÀS
2
2
3
lower than that of the MoS -17”17 model (N
=5.5), likely
2
MoÀS
tion of decorating Ni atoms per gram of catalyst as 87, 46, and
owing to incomplete sulfidation of the MoS2 phase in the
À1
[26]
1
5 mmol g
, for Ni(3)MoS , Ni(6)MoS , and Ni(10)MoS , re-
sample. The introduction of Ni eased reduction (TPS), lead-
Ni cat
2
2
2
spectively. In contrast, the concentration of decorating Ni
ing to better sulfided catalysts and, therefore, the N
values
MoÀS
atoms observed by the IR spectra of adsorbed CO was 10.6,
of Ni(x)MoS (5.4–6.0) were in better agreement with the value
2
À1
5
.1, and 1.5 mmol g
(Figure 6), which indicates that by CO
of 5.9 for the models.
Ni cat
titration we observed about one tenth of the total decorating
Ni atoms present under the probing conditions.
Structure–activity correlations for (Ni)MoS /g-Al O
3
2
2
Comparison of the coordination numbers of the catalysts
determined by EXAFS analysis and those of the models in Fig-
The hydrogenation rate of phenanthrene on catalysts with
varying Ni concentrations showed a maximum in activity with
ures 11 and S16 are given in Table 6. The N
values of the
NiÀS
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4126
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Ni(3)MoS , which had a Ni/(Mo+Ni) ratio of 0.33. This optimum
Deep hydrogenation (k ) showed a more complex depend-
2
2
coincided with the maximum HDS and HDN activity on sup-
ence on the concentration of Ni atoms incorporated in the
ported Ni-promoted MoS catalysts reported at Ni/(Mo+Ni)
edge of the MoS slab (Figure 12B). The values for MoS /g-
2
2
2
[
5,14,24,59,60]
ratios of 0.3–0.4.
Al O , Ni(1.5)MoS , and Ni(3)MoS correlate well, indicating that
2 3 2 2
However, the presence of Ni also influenced the selectivity
of phenanthrene hydrogenation. Deep hydrogenation was fa-
vored, leading to a DiHPhe/TetHPhe (k /k ) selectivity ratio of
Ni(1.5)MoS consisted of a mixture of non-promoted and Ni-
2
promoted MoS as also suggested by the selectivity (Figure 9).
2
The values for Ni(3)MoS , Ni(6)MoS , and Ni(10)MoS were cor-
1
2
2
2
2
1
(versus k /k of 8.6 for MoS /g-Al O ). The formation rates of
related with a different function, one with a less pronounced
1
2
2
2
3
TetHPhe and OHPhe also increased dramatically. Interestingly,
increasing Ni loading above 3 wt% did not influence the selec-
tivity, but only the activity of the catalyst. This indicates that
adding Ni to MoS /g-Al O increased the concentration of
slope and with an intercept way above the level of MoS /g-
2
Al O . This indicated that the presence of Ni in the active sites
2
3
was mandatory for deep hydrogenation and that variations in
the magnitude of k depend mostly on the concentration of
2
2
3
2
active sites and changed their intrinsic nature, whereas further
variation of Ni only modified the concentration of sites.
these sites. The same trend is observed for k and k (see Fig-
3
4
ure S17 in the Supporting Information).
The linear correlation between the rate constant k (Phe to
The mechanistic picture that arises from this work is present-
ed in Figure 13. The adsorption of phenanthrene, driven by
1
DiHPhe, symmetric hydrogenation) and the concentration of
Ni atoms in the edge of MoS slabs (Figure 12A) indicates simi-
van der Waals forces, occurs on the basal plane of the (Ni)MoS
2
2
[
62]
slab. Hydrogenation is likely to occur, however, at regions
close to the edge, where SH groups provide hydrogen. The
chemisorption through p-interactions is considered decisive
for the selectivity owing to the properties of phenanthrene.
The angular arrangement of the three rings creates two elec-
tron sextets at the corners of the molecule, leaving a single
[
63]
double bond with high reactivity in the 9,10-position. There-
fore, adsorption of phenanthrene through the middle ring
leads to fast hydrogenation towards DiHPhe (symmetric hydro-
genation) even on non-promoted Mo sites. On the other hand,
adsorption of phenanthrene through a lateral ring does not
lead to a stable product (TeHPhe) unless four hydrogen atoms
have been added to the adsorbed molecule. This seems to be
difficult on non-promoted sites. We hypothesize that the deep
hydrogenation is promoted by the stronger interaction of Ni
cations with these sites, shifting from the more planar interac-
tion in the parent MoS2.
Figure 12. Correlation of the rate constants A) k
tration of CO adsorbed on Ni atoms decorating MoS
Ni(1.5)MoS ), Ni(3)MoS
(~), Ni(6)MoS
(*), and Ni(10)MoS
2
1
and B) k
2
with the concen-
/g-Al
(^),
(^).
2
. MoS
2
2 3
O
2
(
&
2
2
lar k values for MoS /g-Al O and Ni(10)MoS , although both
The adsorption of phenanthrene through the middle ring on
promoted catalysts leads to faster symmetric hydrogenation
than on MoS /g-Al O owing to the higher edge fraction and
1
2
2
3
2
had very different sites, that is, only non-promoted sites and
only promoted sites (and NiS clusters), respectively. This dem-
x
2
2
3
onstrates that hydrogenation of the middle ring in phenan-
threne does not depend directly on the presence of Ni in the
active site, but mainly on the
concentration of SH groups (available hydrogen). On the other
hand, deep hydrogenation would occur if phenanthrene ad-
total concentration of adsorption
sites. A recent study demonstrat-
ed that the variation of Ni on
(
Ni)MoS2 increased the concen-
tration of active hydrogen by
almost 30% in Ni(3)MoS com-
2
[61]
pared with MoS /g-Al O . This
2
2
3
increase was comparable to the
increase in k with the optimum
1
promotion (55%). Indeed, the in-
crease of k1 is concluded to
result from a combination of in-
creased active hydrogen (30%)
and increase in dispersion (fMo in-
creases about 20% by adding
3
wt% Ni).
2 2 3
Figure 13. Illustration of the hydrogenation routes of phenanthrene on (Ni)MoS /g-Al O .
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4127
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
sorbs through a lateral ring on a promoted site, where a Ni
entity (atom or cluster according to correlations shown above)
is indeed present.
ing the described thermal treatments. Prior to activity tests and ex-
situ characterization, the active sulfide form of the catalysts was
obtained by sulfidation of the oxide precursors in a plug flow reac-
À1
tor under a flow of H S (10 vol%) in H2 (20 mLmin total) at
2
2
0 bar and 673 K for 8 h.
Conclusion
Using a series of (Ni)MoS /g-Al O catalysts, it is shown that in-
Catalyst characterization
2
2
3
corporation of Ni influences the state of Mo precursors and
sulfides alike. Ni is concluded to weaken the interaction of Mo
with the support. This increases the oligomerization of molyb-
date species at the expense of tetrahedral Mo species and con-
comitantly decreases the reduction temperature of Mo species
N physisorption isotherms were measured at liquid N tempera-
2
2
ture by using a Thermo Finnigan Sorptomatic 1990 series instru-
ment. Prior to the measurements, the samples were outgassed at
523 K for 2 h. The texture of the oxide precursors, including the
specific surface area (BET analysis) and the specific pore volume
(
Gurvich analysis), were determined from the N adsorption data.
2
during sulfidation. The dispersion of the MoS phase is modest-
2
The elemental composition of the catalysts was determined by the
Microanalytical Laboratory at TU München.
ly increased by 20%, whereas the concentration of coordina-
tively unsaturated sites substantially increases with the Ni load-
ing. Ni is incorporated in three forms, as spinel within the sup-
port, as Ni atoms decorating MoS (Ni-Mo-S phase), and as NiS
UV/Vis diffuse-reflectance (DR) spectra of the oxide materials were
obtained in an Avantes Avaspec 2048 spectrometer. The samples
were used as powders, placed in a PTFE sample holder and excited
by UV and visible light using an optical fiber (Avantes FCR-7V400-
2
x
sulfide species ranging from NiSx clusters to agglomerated
Ni S . A fraction of the small clusters grows at the edge of
3
2
2
ME-HTX UV/Vis reflection probe) at RT. The UV/Vis DR spectra
MoS decreasing the concentration of single Ni atoms. These
2
were collected with an integration time of 33 ms, and 30 scans
were accumulated. All obtained spectra are presented in the Sup-
porting Information in the form of the Kubelka–Munk function de-
changes can be modeled perfectly, supporting feasibility of the
conclusions derived from the physicochemical measurements.
Hydrogenation of phenanthrene occurs by two routes, “sym-
metric hydrogenation” (phenanthrene!9,10-dihydrophenan-
threne) and “deep hydrogenation” (Phe!1,2,3,4-tetrahydro-
phenanthrene!1,2,3,4,5,6,7,8-octahydrophenanthrene). Only
2
fined as F(R)=(1ÀR) /(2R), where R is the reflectance of the
sample.
The crystallinity of the oxide and sulfide materials was determined
by powder X-ray diffraction (XRD) with a Phillips/PANalytical’s
X’Pert PRO system (CuKa radiation, 0.154 nm) operating at 45 kV
and 40 mA. The XRD patterns were recorded by using a scan
symmetric hydrogenation is observed on non-promoted MoS /
2
g-Al O . The addition of Ni increases the rate of symmetric hy-
2
3
À1
speed of 0.0178s . The samples were deposited on a silicon single
drogenation and opens the route towards deep hydrogenation
the selectivity towards both routes is almost the same on Ni-
crystal with a (111) surface for the analysis.
(
containing catalysts). The activity for the hydrogenation of
phenanthrene correlates with the concentration of exposed Ni
atoms. The enhancement of symmetric hydrogenation (on ad-
sorbing the middle ring with the reactive bond in the 9,10-po-
sition) by promotion is attributed to higher concentrations of
available hydrogen (SH groups) at the surface, the presence of
Ni at the active site is not needed. In contrast, Ni is part of the
active site that performs deep hydrogenation after the adsorp-
tion of a lateral ring of phenanthrene.
The transition of the oxides to the sulfide forms of the materials
was studied by temperature-programmed sulfidation (TPS). In a typ-
ical experiment, the oxide catalyst precursor (0.1 g, 250–355 mm)
was placed in a quartz flow reactor inside a ceramic oven. The
À1
samples were treated for 2 h at 393 K in He (10 mLmin ). Subse-
quently, the samples were sulfided in H2S (2 vol.%) in H2
À1
À1
(
10 mLmin total), heating with 5 Kmin up to 673 K, where
a final dwell time of 2 h was applied. The evolution and consump-
tion of H and H S was monitored by a mass spectrometer (Pfeiffer
2
2
Vacuum QME 200) recording the signals of the masses (m/e) 2 (H2)
and 34 (H S).
2
The concentration of NO adsorbed on the sulfide materials at RT
was determined by volumetric NO pulse experiments. The oxide
precursors were sulfided in situ during a TPS experiment (see
above) and, after cooling the reactor in a sulfiding atmosphere to
RT, purged thoroughly with He. Subsequently, pulses of NO in He
Experimental Section
Material synthesis
The oxide catalyst precursors were prepared by incipient wetness
impregnation of g-Al O , provided by the Chevron Energy Technol-
2
3
(
1.68 mL, 10 vol.%) were periodically introduced into the reactor
until further NO uptake (followed by a mass spectrometer, m/e=
0) was not observed. The concentration of NO adsorbed in
ogy Company. The support was treated in flowing synthetic air at
33 K for 2 h before impregnation. The two precursor salts, that is,
ammonium heptamolybdate ((NH ) Mo O ·4H O, Aldrich, ꢀ99.0%)
8
3
4
6
7
24
2
a given pulse was determined as the integral difference between
that NO signal and the NO signal at saturation. The total concen-
tration of NO adsorbed was calculated as the sum of the NO up-
takes per pulse.
and nickel nitrate (Ni(NO ) ·6H O, Aldrich, ꢀ98.5%) were deposited
3
2
2
from aqueous solutions in two consecutive impregnation steps
first Mo and then Ni). After each impregnation step, the materials
(
were dried at 393 K in static air and subsequently treated at 773 K
in flowing synthetic air for 4 h. The nominal Mo content was kept
constant (1 mmol Mo per gram of g-Al O ) in all the prepared ma-
terials, while increasing the concentration of Ni to obtain NiMo ma-
terials with varying Ni/(Mo+Ni) ratios. Additionally, an oxide pre-
cursor containing 1 mmol of Ni per gram of g-Al O was prepared
Transmission electron microscopy of the sulfide catalysts was per-
formed with a JEOL JEM-2011 instrument using an acceleration
voltage of 120 keV. Prior to the analysis, suspensions of the materi-
als in ethanol were prepared and deposited on copper grids with
supporting carbon films. The statistical analysis of the properties of
2
3
2
3
as a reference material by impregnation with the Ni salt and apply-
the supported MoS phase (length and stacking degree) was per-
2
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4128
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
formed by measuring at least 300 slabs distributed in micrographs
taken from different regions of the sample.
Kinetic measurements
The kinetic measurements were carried out in a continuous trickle
bed reactor with a glass-coated stainless steel tube (ø = inches).
1
The adsorption of CO on the sulfide catalysts was followed by IR
spectroscopy by using a Nicolet 6700 FTIR spectrometer equipped
4
High-pressure mass flow controllers (Bronkhorst EL-FLOW) and
a HPLC pump (Gilson 307, pump head 5SC) were used to introduce
gas and liquid feeds, respectively. The oxide catalyst precursor
(250–355 mm) was diluted with SiC (1:20 , 63–90 mm) and posi-
tioned in the middle of the reactor tube by filling up with SiC of
the same size. Prior to each experiment, the oxide precursors were
sulfided in situ (see above). The hydrogenation reactions were per-
formed as space–time-dependent experiments at 573 K and 6 MPa
total pressure after 20 h on stream (in order to reach a steady
state). Space–time was defined as m /F where m is the mass of
À1
with a MCT detector using a resolution of 4 cm . A dedicated
flow/vacuum IR cell allowed the in-situ sulfidation of the samples
at high pressure and temperature (up to 20 bar and 673 K) and the
subsequent adsorption of CO at ꢁ123 K in vacuum without any
exposure of the sample to air. The samples were diluted with g-
Al O (ratio 1:3) to improve the transmission of catalysts with high
2
3
Ni content. The powders were pressed into self-supported wafers
À2
(
10 mgcm ), and placed inside the IR cell. Sulfidation was per-
À1
formed in situ at 673 K (heating rate 3 Kmin ) and in H S in H
20 bar, 20 mLmin , 10 vol.%). After sulfidation for two hours, the
2
2
cat
cat
À1
(
the catalyst precursor (40 mg) and F is the molar flow of phenan-
cell was flushed with He, evacuated to a residual pressure of
threne. The reactions were performed by keeping a H -to-hydrocar-
2
À6
3
À3
1
0
mbar and after 1 h cooled to 323 K. To perform the adsorption
bon ratio of 300 Ndm dm . The reactant solution had a concentra-
tion of 1 wt% of phenanthrene (Alfa Aesar, 98%) and 1000 ppm S
(present as dimethyl disulfide, Aldrich ꢀ99.0%) in decalin (Merck,
ꢀ99.0%) and 2 wt% n-tetradecane (Alfa Aesar, 99+%) as internal
standard. The absence of transport artifacts or incomplete wetting
was verified by performing experiments using different amounts of
catalyst, varying flow rates, and changing the catalyst and SiC par-
ticle sizes. The reactions were monitored by periodically collecting
samples by using a multiport sampling valve. The liquid samples
were analyzed off-line by gas chromatography with a Shimadzu
GC-2010 gas chromatograph equipped with a 50 m HP-1 column
and a flame ionization detector.
experiments, the IR cell was cooled with liquid nitrogen to 123 K in
the presence of He (3 mbar). After taking a reference spectrum, the
cell was evacuated and small doses of CO were admitted to the
cell up to an equilibrium pressure of 1 mbar. The spectra reported
here have had the background subtracted and were recorded in
the presence of CO (1 mbar) and He (2 mbar) in the IR cell to reach
a stable temperature. The intensities of the bands were quantita-
tively determined by deconvolution using the software GRAMS.
Concentrations of adsorbed CO species were calculated using
molar extinction coefficients reported in Refs. [64,65].
X-ray absorption spectroscopy (XAS) was employed on sulfide cata-
lysts for a complete study of their structural properties. The data
were recorded at two beamlines, the BM 26A-DUBBLE at ESRF (Gre-
noble, France), and the BL 22-CLÆSS at ALBA (Barcelona, Spain).
Prior to the experiments, samples of the catalysts were sulfided at
Acknowledgements
2
0 bar (see above), pressed into self-supporting wafers and re-sul-
This work was supported by the Chevron Energy Technology
Company. The authors would like to specially thank Dr. Alexand-
er Kuperman, Dr. Axel Brait, and Dr. Jinyi Han for fruitful discus-
sions. We thank Prof. Roel Prins for the critical discussion of the
results. We are also grateful to Dr. Marianne Hanzlik for TEM
measurements and Dipl.-Ing. Xaver Hecht for technical support.
Parts of this research were undertaken at the light source facili-
ties ESRF (Grenoble, France) and ALBA (Barcelona, Spain). Both
facilities are acknowledged for the provision of beamtime and
their kind assistance during experiments.
fided in situ in the stainless-steel flow cell used for the XAS meas-
urements. All spectra were recorded in transmission mode in flow-
ing He, and at liquid N temperature to minimize thermal vibra-
tions. For energy calibration, Mo and Ni foils were measured simul-
taneously with each sample. Two spectra per sample were aver-
aged to enhance the signal-to-noise ratio. The X-ray absorption
near-edge structure (XANES) and extended X-ray absorption fine
structure (EXAFS) data was analyzed by using IFEFFIT together
with the Demeter package (Athena and Artemis, version
2
[66–68]
0
.9.20).
After removing the background absorption, the spec-
tra were normalized to an average post-edge height of one. The
XANES were analyzed by linear combination fitting (LCF) in the
energy range of À20 eV to 50 eV relative to the absorption edge.
Keywords: aromatic hydrogenation · hydrotreating · MoS2
·
3
For EXAFS analysis, the oscillations were weighted with k and
Ni-Mo-S · promoter effect
Fourier transformed (limits k=3–12 Š for Ni-edge and k=3–13 Š
for Mo-edge). The local environment of Mo and Ni atoms was de-
termined in k-space from the EXAFS. Single and multiple scattering
contributions for MoÀS, MoÀMo, NiÀS, and NiÀNi (phase shifts and
[
[
1] A. Stanislaus, B. Cooper, Catal. Rev. 1994, 36, 75–123.
2] J. M. Christensen, P. M. Mortensen, R. Trane, P. A. Jensen, A. D. Jensen,
Appl. Catal. A 2009, 366, 29–43.
[68]
backscattering amplitudes) were calculated with FEFF based on
crystallographic information files (cif) of the Inorganic Crystal Struc-
[3] O. Y. GutiØrrez, L. Zhong, Y. Zhu, J. A. Lercher, ChemCatChem 2013, 5,
[50]
2
ture Database (ICSD). The amplitude reduction factor So was de-
termined from reference compounds and was 0.85 for Mo at ESRF
and 0.98 at ALBA and 1.0 for Ni at both beamlines. For the
Ni(x)MoS /g-Al O catalysts the Mo K-edge and Ni K-edge were
fitted simultaneously to identify Mo–Ni and Ni–Mo contributions.
The distances, R, for Mo–Ni and Ni–Mo were constrained to be
equal and the coordination numbers, N, were defined by the molar
3249–3259.
[4] M. J. Girgis, B. C. Gates, Ind. Eng. Chem. Res. 1991, 30, 2021–2058.
[5] H. Topsøe, B. S. Clausen, F. E. Massoth, Hydrotreating Catalysis, Springer,
Berlin, Germany, 1996.
2
2
3
[
[
6] H. Topsøe, B. S. Clausen, Catal. Rev. 1984, 26, 395–420.
7] J. V. Lauritsen, S. Helveg, E. Lægsgaard, I. Stensgaard, B. S. Clausen, H.
Topsøe, F. Besenbacher, J. Catal. 2001, 197, 1–5.
[
[
8] Y. Zhu, Q. M. Ramasse, M. Brorson, P. G. Moses, L. P. Hansen, C. F. Kisie-
lowski, S. Helveg, Angew. Chem. Int. Ed. 2014, 53, 10723–10727; Angew.
Chem. 2014, 126, 10899–10903.
9] F. L. Deepak, R. Esparza, B. Borges, X. Lopez-Lozano, M. Jose-Yacaman,
ACS Catal. 2011, 1, 537–543.
ratio of Mo and Ni in the catalysts (N
=n(Ni)/n(Mo)”N
).
MoÀNi
NiÀMo
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4129
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
10] F. Besenbacher, M. Brorson, B. S. Clausen, S. Helveg, B. Hinnemann, J.
Kibsgaard, J. V. Lauritsen, P. G. Moses, J. K. Nørskov, H. Topsøe, Catal.
Today 2008, 130, 86–96.
[41] O. Y. GutiØrrez, T. Klimova, J. Catal. 2011, 281, 50–62.
[42] N. Koizumi, S. Jung, Y. Hamabe, H. Suzuki, M. Yamada, Catal. Lett. 2010,
135, 175–181.
[
[
11] N.-Y. Topsøe, H. Topsøe, J. Catal. 1982, 75, 354–374.
12] A. Travert, C. Dujardin, F. MaugØ, E. Veilly, S. Cristol, J.-F. Paul, E. Payen, J.
Phys. Chem. B 2006, 110, 1261–1270.
13] S. M. A. M. Bouwens, R. Prins, V. H. J. de Beer, D. C. Koningsberger, J.
Phys. Chem. 1990, 94, 3711–3718.
14] S. P. A. Louwers, R. Prins, J. Catal. 1992, 133, 94–111.
15] D. O. Dumcenco, H. Kobayashi, Z. Liu, Y.-S. Huang, K. Suenaga, Nat.
Commun. 2013, 4, 1351.
16] M. Sun, A. Nelson, J. Adjaye, J. Catal. 2005, 233, 411 –421.
17] L. S. Byskov, J. K. Nørskov, B. S. Clausen, H. Topsøe, J. Catal. 1999, 187,
[43] N. Topsøe, A. Tuxen, B. Hinnemann, J. V. Lauritsen, K. G. Knudsen, F. Be-
senbacher, H. Topsøe, J. Catal. 2011, 279, 337–351.
[44] A. Travert, C. Dujardin, F. MaugØ, S. Cristol, J.-F. Paul, E. Payen, D. Bou-
geard, Catal. Today 2001, 70, 255–269.
[45] C. Dujardin, M. LØlias, J. van Gestel, A. Travert, J. Duchet, F. MaugØ, Appl.
Catal. A 2007, 322, 46–57.
[46] B. M. Vogelaar, N. Kagami, T. F. van der Zijden, A. D. van Langeveld, S.
Eijsbouts, J. A. Moulijn, J. Mol. Catal. A 2009, 309, 79–88.
[47] E. Marceau, A. Lçfberg, J.-M. Giraudon, F. NØgrier, M. Che, L. Leclercq,
Appl. Catal. A 2009, 362, 34–39.
[
[
[
[
[
1
09–122.
[48] G. Poncelet, M. A. Centeno, R. Molina, Appl. Catal. A 2005, 288, 232–
242.
[
[
18] J. N. Coleman, M. Lotya, A. O’Neill, S. D. Bergin, P. J. King, U. Khan, K.
Young, A. Gaucher, S. De, R. J. Smith, et al., Science 2011, 331, 568–
[49] N. Koizumi, Y. Hamabe, S. Jung, Y. Suzuki, S. Yoshida, M. Yamada, J. Syn-
chrotron Radiat. 2010, 17, 414–424.
[50] A. Belsky, M. Hellenbrandt, V. L. Karen, P. Luksch, Acta Crystallogr. Sect. B
2002, 58, 364–369.
5
71.
19] B. T. Yonemoto, G. S. Hutchings, F. Jiao, J. Am. Chem. Soc. 2014, 136,
895–8898.
8
[
[
20] R. G. Leliveld, S. Eijsbouts, Catal. Today 2008, 130, 183–189.
21] W. Bensch in Comprehensive Inorganic Chemistry II, 2nd ed. (Eds.: J.
Reedijk, K. Poeppelmeier) Elsevier, Amsterdam, The Netherlands, 2013.
22] O. Y. GutiØrrez, S. Singh, E. Schachtl, J. Kim, E. Kondratieva, J. Hein, J. A.
Lercher, ACS Catal. 2014, 4, 1487–1499.
[51] N. A. Bhore, M. T. Klein, K. B. Bischoff, Ind. Eng. Chem. Res. 1990, 29,
313–316.
[52] M. J. Girgis, B. C. Gates, Ind. Eng. Chem. Res. 1994, 33, 2301–2313.
[53] A. R. Beltramone, D. E. Resasco, W. E. Alvarez, T. V. Choudhary, Ind. Eng.
Chem. Res. 2008, 47, 7161–7166.
[
[
23] R. Prins, V. H. J. De Beer, G. A. Somorjai, Catal. Rev. Sci. Eng. 1989, 31, 1–
[54] C. C. Williams, J. G. Ekerdt, J. M. Jehng, F. D. Hardcastle, I. E. Wachs, J.
Phys. Chem. 1991, 95, 8791–8797.
4
1.
[
[
24] A. N. Startsev, Catal. Rev. 1995, 37, 353–423.
25] L. S. Byskov, B. Hammer, J. K. Nørskov, B. S. Clausen, H. Topsøe, Catal.
Lett. 1997, 47, 177–182.
[55] B. Delmon, Stud. Surf. Sci. Catal. 1989, 53, 1–40.
[56] A. M. van der Kraan, M. W. J. CrajØ, E. Gerkema, W. L. T. M. Ramselaar,
V. H. J. de Beer, Appl. Catal. 1988, 39, L7–L10.
[
[
26] J. Hein, A. Hrabar, A. Jentys, O. Y. GutiØrrez, J. a. Lercher, ChemCatChem
[57] M. W. J. CrajØ, E. Gerkema, V. H. J. de Beer, A. M. van der Kraan, Stud.
Surf. Sci. Catal. 1989, 50, 165–179.
2014, 6, 485–499.
27] R. Prins, M. Egorova, A. Rçthlisberger, Y. Zhao, N. Sivasankar, P. Kukula,
Catal. Today 2006, 111, 84–93.
28] R. Prins, M. Jian, M. Flechsenhar, Polyhedron 1997, 16, 3235–3246.
29] X. Wang, B. Zhao, D. Jiang, Y. Xie, Appl. Catal. A 1999, 188, 201–209.
30] R. W. G. Wyckoff, Crystal Structures, Vol. 1, Interscience, New York, USA,
[58] T. Shido, R. Prins, J. Phys. Chem. B 1998, 102, 8426–8435.
[59] Y. C. Park, H.-K. Rhee, Korean J. Chem. Eng. 1998, 15, 411 –416.
[60] V. H. J. de Beer, T. H. M. van Sint Fiet, J. F. Engelen, A. C. van Haandel,
M. W. J. Wolfs, C. H. Amberg, G. C. A. Schuit, J. Catal. 1972, 27, 357–368.
[61] E. Schachtl, E. Kondratieva, O. Y. GutiØrrez, J. A. Lercher, J. Phys. Chem.
Lett. 2015, 6, 2929–2932.
[
[
[
1
963.
31] P. Arnoldy, J. A. M. van den Heijkant, G. D. de Bok, J. A. Moulijn, J. Catal.
985, 92, 35–55.
[
[62] P. G. Moses, J. J. Mortensen, B. I. Lundqvist, J. K. Nørskov, J. Chem. Phys.
2009, 130, 104709.
1
[
[
32] B. Scheffer, P. J. Mangnus, J. A. Moulijn, J. Catal. 1990, 121, 18–30.
33] S. Eijsbouts, L. C. A. van den Oetelaar, R. R. van Puijenbroek, J. Catal.
[63] E. Clar, Polycyclic Hydrocarbons, Band I, Academic Press, London, New
York, 1964.
2
005, 229, 352–364.
[64] F. MaugØ, J.-C. Lavalley, J. Catal. 1992, 137, 69–76.
[65] E. V. Kondratieva, O. V. Manoilova, A. A. Tsyganenko, Kinet. Catal. 2008,
49, 451–456.
[
[
34] E. J. M. Hensen, P. J. Kooyman, Y. van der Meer, A. M. van der Kraan,
V. H. J. de Beer, J. A. R. van Veen, R. A. van Santen, J. Catal. 2001, 199,
2
24–235.
35] S. Kasztelan, H. Toulhoat, J. Grimblot, J. Bonnelle, Appl. Catal. 1984, 13,
27–159.
[66] B. Ravel, M. Newville, J. Synchrotron Radiat. 2005, 12, 537–541.
[67] M. Newville, J. Synchrotron Radiat. 2001, 8, 322–324.
[68] S. I. Zabinsky, J. J. Rehr, A. Ankudinov, R. C. Albers, M. J. Eller, Phys. Rev. B
1995, 52, 2995–3009.
1
[
[
[
36] J. Chen, F. MaugØ, J. El Fallah, L. Oliviero, J. Catal. 2014, 320, 170–179.
37] N.-Y. Topsøe, H. Topsøe, J. Catal. 1983, 84, 386–401.
38] O. Y. GutiØrrez, A. Hrabar, J. Hein, Y. Yu, J. Han, J. A. Lercher, J. Catal.
2
012, 295, 155–168.
39] O. Y. GutiØrrez, C. Kaufmann, J. A. Lercher, ACS Catal. 2011, 1, 1595–
603.
40] A. Hrabar, J. Hein, O. Y. GutiØrrez, J. A. Lercher, J. Catal. 2011, 281, 325–
38.
[
[
Received: June 19, 2015
1
Revised: July 31, 2015
3
Published online on September 25, 2015
ChemCatChem 2015, 7, 4118 – 4130
www.chemcatchem.org
4130
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim