Distribution of Metal Cations in Ni-Mo-W Sulfide Catalysts

Article abstract of DOI:10.1002/cctc.201500788

The distribution of metal cations and the morphology of unsupported NiMo, NiW, and NiMoW sulfide catalysts were explored qualitatively and quantitatively. In the bi- and trimetallic catalysts, Mo(W)S₂ nanoparticles are deposited on Ni sulfide particles of varying stoichiometry and sizes (crystalline Ni₉S₈, and Ni₃S₄ were identified). These nanoparticles are stacks of Mo(W)S₂ slabs with varying size, degrees of bending and mismatch between the slabs. High resolution electron microscopy and X-ray absorption spectroscopy based on particle modeling revealed a statistical distribution of Mo and W within individual layers in sulfide NiMoW, forming intralayer mixed Mo_1-xW_xS₂. Ni is associated with MoS₂, WS₂, and Mo_1-xW_xS₂ creating Ni-promoted phases. The incorporation of Ni at the edges of the slabs was the highest for sulfide NiMoW. This high concentration of Ni in sulfide NiMoW, as well as its long bent Mo_1-xW_xS₂ slabs, were paralleled by the highest activity for nitrogen and sulfur removal from model hydrocarbons such as o-propylaniline and dibenzothiophene.

Full text of DOI:10.1002/cctc.201500788

DOI: 10.1002/cctc.201500788
Full Papers
Distribution of Metal Cations in Ni-Mo-W Sulfide Catalysts
Jennifer Hein,[a] Oliver Y. GutiØrrez,*[a] Eva Schachtl,[a] Pinghong Xu,[b] Nigel D. Browning,[c]
Andreas Jentys,[a] and Johannes A. Lercher*[a]
The distribution of metal cations and the morphology of un-
supported NiMo, NiW, and NiMoW sulfide catalysts were ex-
plored qualitatively and quantitatively. In the bi- and trimetallic
catalysts, Mo(W)S2 nanoparticles are deposited on Ni sulfide
particles of varying stoichiometry and sizes (crystalline Ni9S8,
and Ni3S4 were identified). These nanoparticles are stacks of
Mo(W)S2 slabs with varying size, degrees of bending and mis-
match between the slabs. High resolution electron microscopy
and X-ray absorption spectroscopy based on particle modeling
revealed a statistical distribution of Mo and W within individual
layers in sulfide NiMoW, forming intralayer mixed Mo1ÀxWxS2. Ni
is associated with MoS2, WS2, and Mo1ÀxWxS2 creating Ni-pro-
moted phases. The incorporation of Ni at the edges of the
slabs was the highest for sulfide NiMoW. This high concentra-
tion of Ni in sulfide NiMoW, as well as its long bent Mo1ÀxWxS2
slabs, were paralleled by the highest activity for nitrogen and
sulfur removal from model hydrocarbons such as o-propylani-
line and dibenzothiophene.
Introduction
MoS2 and WS2 are isostructural and isomorphic semiconductors
with a layered structure. The ability of MoS2 and WS2 to acti-
vate H2 and to catalyze hydrogenation and hydrogenolysis of
CÀN and CÀS bonds makes them interesting catalysts. Thus,
MoS2 and WS2 (bulk and supported) have been widely applied
as catalysts in hydrotreating of oil fractions,[1–4] biomass-de-
rived feedstocks,[5] as well as in photo- and electrochemistry.[6–8]
The wide applicability triggered activities to synthesize well-de-
fined morphologies and in turn tailored band structures.[9,10]
There has been also impressive progress in atom-level charac-
terization of MoS2 and WS2.[11,12] Most of these studies, howev-
er, focused on model catalysts, prepared under conditions facil-
itating the analysis (e.g., in situ monolayer growth). Wet
chemistry prepared sulfide materials similar to those used in-
dustrially are significantly more complex and hardly accessible
for these advanced characterizations. This holds also true for
the promotion of the MoS2 and WS2 by Ni2+ or Co2+ cations,
which are mainly explored via averaging techniques such as
XAS.[13–15]
Although bimetallic catalysts are already challenging, trime-
tallic Ni-Mo-W sulfide materials make the task even more com-
plex.[16,17] Studies on the consequences of combining Mo and
W in a single sulfide were in consequence only performed in
the absence of Ni or Co.[11,18,19]
The target of the current work is, therefore, to investigate
the structure, morphology, and the distribution of catalyst con-
stituents in three wet-chemically prepared sulfides qualitatively
and quantitatively. We address not only the distribution of W
and Mo, but also the location and nature of the incorporated
Ni. A comprehensive EXAFS study of all three metal edges in
combination with HAADF-STEM was used and combined with
TEM, XRD, and Raman spectroscopy. Systematic comparison of
theoretical EXAFS of model clusters was used to analyze in
detail the influence of the backscatter Mo, W, and Ni at differ-
ent distances. The impact of the physicochemical properties
on catalytic activity was explored for nitrogen and sulfur re-
[a] J. Hein, Dr. O. Y. GutiØrrez, E. Schachtl, Prof. Dr. A. Jentys,
Prof. Dr. J. A. Lercher
Department of Chemistry and Catalysis Research Center
Technische Universität München
Lichtenbergstraße 4, 85747 Garching (Germany)
Fax: (+49)89-28913544
moval
from
o-propylaniline
and
dibenzothiophene,
respectively.
E-mail: oliver.gutierrez@mytum.de
johannes.lercher@ch.tum.de
Experimental Section
[b] P. Xu
Department of Chemical Engineering and Materials Science
University of California-Davis
One Shields Avenue, Davis, CA 95616 (USA)
Catalyst preparation: Two bimetallic, NiMoS and NiWS, and one
trimetallic NiMoWS materials were prepared by a pH controlled co-
precipitation in aqueous solution in accordance to Ref. [20]. The
precursors were subsequently sulfided in 10 vol.% H2S in H2 flow
at 4008C and 1.8 MPa for 12 h. Ammonium heptamolybdate
(AHM), ammonium metatungstate (AMT), Ni nitrate, aqueous am-
monia and maleic acid were used as reactants during the
synthesis.
[c] Prof. Dr. N. D. Browning
Fundamental and Computational Sciences Directorate
Pacific Northwest National Laboratory
Richland, WA 99352 (USA)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201500788.
This publication is part of a Special Issue on “Microscopy and Spectros-
copy for Catalysis”. Once the full issue has been assembled, a link to its
Table of Contents will appear here.
Powder X-Ray diffraction (XRD): The crystal structure of the sam-
ples was determined by X-ray diffraction by using an X’Pert Pro
ChemCatChem 2015, 7, 3692 – 3704
3692
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
PW 3040/60 (PANalytical) diffractometer equipped with a copper X-
ray tube, a Ni-Kb filter to obtain monochromatic Cu-Ka1 radiation
(0.154 nm) and a solid state detector (X’Celerator). The measure-
ments were performed with a 10”10À9 m slit mask in a range from
2q=58 to 2q=708 at the operating conditions of 45 kV and
40 mA. The sulfided catalysts were measured for 1 h with step size
of 0.0178 and scan time of 115 s per step and for selected refer-
ence materials 5 min scans with a step size of 0.0178 and 10 s per
step were used. The crystallographic phases were identified by
using the Inorganic Crystal Structure Database (ICSD).[21] The Scher-
rer equation was used to determine the stacking degree of sulfide
slabs in the catalysts, knowing that the diffraction at around
148 2q corresponds to the (002) plane with interplanar distance of
6.1 Š (distance between the metal cation layers in MoS2 or WS2).
The measured full width at the half maximum (FWHM) was correct-
ed by the diffractometer typical line broadening of 0.1 rad estimat-
ed by instrument calibration.
IFEFFIT.[24–25] After background removal and normalization to the
average post-edge height to one, the oscillations were weighted
with k3 and Fourier-transformed within the limits of k=2.5–
14.0 ŠÀ1. The local environments of the Mo, W, and Ni atoms in the
sulfided catalysts were determined in k-space from the EXAFS.
Single and multiple scattering phase shifts and backscattering am-
plitudes were calculated with FEFF[25] based on crystallographic in-
formation files (cif files) of the ICSD[21] and on the structure of
model mixed MoxWyS2 clusters. The EXAFS at the Mo K-edge, W LIII-
edge and at the Ni K-edge were simultaneously fitted. During fit-
ting the Debye–Waller factor s2 and the distance r between two
types of the metals were constrained to be equal (i.e. rAÀB =rBÀA
)
and the coordination numbers NAÀB and NBÀA were constrained by
the molar ratio of A and B in the catalysts (NAÀB/NBÀA =n(B)/n(A)).
The R factor and the absolute errors of all parameters, which are
the estimated standard deviations and the statistical uncertainties
of the starting parameter determined by IFEFFIT,[24] were used to
evaluate a certain fit result.
Raman Spectroscopy: Raman spectra were recorded with a Renish-
aw Raman system (Type 1000, dispersive spectrometer) equipped
with CCD detector and a Leica microscope DM LM. The excitation
wavelength of 514 nm was generated by a multi-line argon-ion
gas laser (Stellar-Pro Select 150 of MODU-Laser) operating at
20 mW power. The wavenumber accuracy was within 1 cmÀ1. Sul-
fide catalysts and reference materials were analyzed under ambient
conditions in the form of self-supported wafers.
Catalytic activity studies: Kinetic studies were performed in a con-
tinuous flow trickle bed reactor system equipped with high pres-
sure mass flow meters and a HPLC pump. The stainless steel, glass-
coated tubular reactor was loaded with catalyst (0.025 g), diluted
in SiC (1 g). The liquid samples were analyzed by off line gas chro-
matography (HP 6890 GC) equipped with a flame ionization detec-
tor and 60 m DB-17 capillary column. The hydrotreating reactions
were performed as temperature dependent experiments at con-
À1
stant space-time of 49 hgcat molOPA and total pressure of 5.0 MPa.
Electron microscopy: The morphology and particle size of the dif-
ferent samples were analyzed by electron microscopic methods.
Standard measurements of the sulfide catalysts were performed in
transmission mode coupled with selective area electron diffraction
(TEM-SAED) with a JEM-2011 (JEOL) with an accelerating voltage of
120 keV. The average length of the sulfide slabs in the catalysts
was estimated by the length measurements of around 500 differ-
ent bundles of metal sulfide slabs from different sample spots.
Moreover, high resolution scanning electron microscopy coupled
with energy dispersive X-ray spectroscopy (HR-SEM-EDX) was per-
formed with a high resolution FE-SEM, JSM 7500 F (JEOL) with EDX
detector (Oxford). The HR-SEM micrographs were taken with the
lower secondary electron imaging detector (LEI) and an accelerat-
ing voltage of 2 keV. Additionally, high-angle annular dark-field
Prior to the activity test reactions, the precursors were activated
in situ in 10 vol.% H2S in H2 flow at 4008C and 1.8 MPa for 12 h.
The reactions were performed in excess of H2 and with a mixture
of hydrocarbons keeping the flow ratio of H2 to liquid constant at
330 Ndm3 dmÀ3. The initial reactant concentration was set to
1000 ppm N as o-propylaniline (OPA) in a mixture of 1000 ppm S
as dimethyl disulfide (DMDS), 4.94 wt% hexadecane and
93.95 wt% tetradecane as solvent. Together with the OPA hydrode-
nitrogenation (HDN) activity, also the hydrodesulfurization (HDS)
activity was studied by co-feeding 500 ppm S as dibenzothiophene
(DBT) to the liquid feed. At the beginning of the experiment,
a feed with OPA and DMDS was introduced at 3708C. These start-
ing conditions were kept for 48 h, although steady state conditions
were usually reached after 24 h time on stream. Afterwards, the
temperature was decreased to 360 and 3508C and the liquid feed
was changed to an OPA-DBT mixture to perform HDN and HDS si-
multaneously. After 30 h at 3708C, the activity of the parallel HDN
and HDS was also tested at 360 and 3508C. At the end of the run,
the initial reaction conditions (3708C and pure OPA feed) were ap-
plied again. For all experiments, the same results were found after
the initial stabilizing time of 48 h and at the end of the run.
imaging with
a scanning transmission electron microscope
(HAADF-STEM) was performed using an aberration-corrected FEI
Titan 80/300S. The device was operating at 80 keV and the HAADF
collection inner angle was 75 mrad.
X-ray absorption spectroscopy (XAS): The structural properties of
the sulfided catalysts were studied by X-ray absorption spectrosco-
py at the X1 beamline at Hasylab, DESY, Hamburg, Germany. The
data set was completed with experiments performed on the BM
26A-DUBBLE, (Dutch-Belgian) beamline at the ESRF, Grenoble,
France. Spectra were recorded in transmission mode at the Mo K-
edge (20000 eV), W LIII-edge (10207 eV) and at the Ni K-edge
(8333 eV). Prior to EXAFS measurements, the sulfide catalysts were
re-sulfided in the stainless steel in situ flow XAS cell.
Results and discussion
Composition and crystallinity
The composition of NiMoS, NiWS and the trimetallic NiMoWS
sulfide are summarized in Table 1. All materials had similar Ni
molar fractions and the molar ratio of Mo to W in the trimetal-
lic catalyst was 1.3.
Prior to the analysis of the experimental XAS data, a systematic
EXAFS modeling was performed using mixed MoxWyS2 clusters cre-
ated with Accelrys Material Studio 7.0 on the basis of the crystallo-
graphic structure of pure MoS2 and WS2. These mixed disulfide
phases were used to calculate the Mo-W and W-Mo phase shifts
and backscattering amplitudes at different distances using FEFF9
and VIPER.[22,23] All XAS spectra were analyzed with the Demeter-
package (ATHENA and ARTEMIS, version 0.9.20) using FEFF6 and
The XRD patterns showed Mo(W)S2 and Ni sulfide phases as
well as an X-ray amorphous material in all catalysts (Figure 1).
The reference materials, MoS2, WS2, and Ni3S2 are shown for
comparison. The reflections of all references are in agreement
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3693
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
angles, are an indication for turbostratic disorder and uncorre-
lated single sulfide layers.[26,27]
Table 1. Composition and stacking degree of the unsupported sulfided
catalysts.
Catalyst
Composition [mmolgÀ1
]
CNi
Av. stacking degree
[a]
Ni
Mo
W
[mol/mol] of MoS2 and WS2
Raman spectroscopic characterization
NiMoS
NiMoWS 4.0
NiWS 2.5
4.6
3.7
1.8
0.55
0.56
0.47
5.7
5.1
4.4
Raman spectra of catalysts and reference materials are shown
in Figure 2. The assignment of the bands is described in the
1.4
2.8
[a] Average number of slabs forming a MoS2 particle as determined by
applying the Scherrer-equation on the (002) reflection at 14.28 2q which,
is associated to the interplanar distance of 6.1 Š.
Figure 2. Raman spectra (lex =514 nm) of the reference Ni3S2 (a; no Raman
bands), b) MoS2, and c) WS2, and of the unsupported sulfided catalysts
d) NiMoS, e) NiMoWS, and f) NiWS. The shift of the most intense Raman
bands of the references are indicated by the dotted vertical lines and the
corresponding atomic displacement of the E12g (left) and A1g (right) mode is
illustrated.
Figure 1. XRD pattern of the reference bulk materials a) MoS2, b) WS2, and
c) Ni3S2, and of the unsupported sulfide catalysts d) NiMoS, e) NiMoWS, and
f) NiWS. The profiles fitted under the (002) reflection at around 14.28 2q
(grey filled) were used to determine the stacking degree of the MoS2 and
WS2 slabs. The most important reflection of Ni9S8 (*) and Ni3S4 (8) are
indicated.
Supporting Information. Direct evidence for Ni sulfides was not
obtained through Raman spectroscopy.
In the sulfide catalysts NiMoS and NiWS, only the A1g and
with the data published in the ICSD database (#MoS2: 644245;
#WS2: 202366; #Ni3S2: 27521).[21] Sharp signals in the patterns
of the sulfide catalysts correspond to different Ni sulfides. In
NiMoS and NiMoWS, orthorhombic Ni9S8 (ICSD#: 63080) was
the dominant phase (as indicated by the intense reflection at
27.48 2q) with traces of trigonal Ni3S2 (21.98 2q). The reflection
at 26.68 2q in NiWS is assigned to the cubic phase of Ni3S4
(ICSD#: 57435). Additionally, Ni9S8 was found in low amounts in
NiWS.
E1 modes were observed (Figure 2(d) and (f)). The corre-
2g
sponding bands were shifted to lower wavenumbers and were
broader than the bands of the reference materials. For in-
stance, the A1g mode appeared at 405 cmÀ1 for NiMoS, and at
413 cmÀ1 for NiWS (in the reference materials this band ap-
peared at 409 cmÀ1 for MoS2 and 421 cmÀ1 for WS2). The
downward shift was attributed to weaker metal sulfur bonds
caused by the low stacking degree in the catalysts and the
concomitant weak van der Waals forces, which allow atom dis-
Broad reflections at 14.2, 33, 40 and 608 2q are assigned to
hexagonal phases of MoS2 and WS2. The (002) reflection
around 14.28 2q of the catalysts appeared at smaller angles
compared to the references. In NiWS this reflection was ob-
served at 14.08 2q, for NiMoWS at 14.18 2q, and for NiMoS at
14.28 2q. This indicates that the lattice parameters d (i.e., the
distance between the metal sulfide layers) for the catalysts are
between 6.32 and 6.23 Š, whereas the distance between the
stacked layers in the bulk reference materials is 6.15 Š in WS2
(14.48 2q) and 6.1 Š in MoS2 (14.58 2q). This difference reflects
disorder of the metal sulfide layers, e.g., bending, which occurs
when the slabs grow significantly longer in the x and y direc-
tion than in the z direction.[16] Additionally, line broadening
and the amorphous background, especially at low diffraction
placement.[28] The asymmetry of the bands of A1g and E1
2g
modes was concluded to be caused by highly bent slabs,
which influenced the symmetry selection rules leading to
a second-order Raman signal, which overlapped with A1g and
E12g bands.[29]
The spectrum of NiMoWS (Figure 2(e)) appears to be a com-
bination of the spectra of the bimetallic catalysts. The band at
410 cmÀ1 is assigned to the A1g mode of Mo-W composite spe-
cies.[30] The bands at 374 cmÀ1 and 349 cmÀ1 are assigned to
E1 modes of MoÀS and WÀS, respectively, in agreement with
2g
observations when W systematically replaced Mo in Mo(W)S2
crystals.[19] The shift of both E1 modes, compared to the refer-
2g
ences MoS2 and WS2, is attributed to a structural disorder of
the sulfide slabs.
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3694
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Electron microscopic characterization
terns (Figure 3(B)), to small particles, which seem to be occlud-
ed inside the Mo(W)S2 agglomerates.
Figure 3(C) shows a representative HR-SEM image of NiMoWS.
The microstructure of the catalysts consists of large NixSy crys-
tals (only Ni and S in a ratio of 1:1 are observed by EDX map-
ping) covered by spherical particles identified as MoS2 or WS2
by EDX mapping. Interestingly, Ni and S were identified across
the whole sample, whereas Mo and W were detected only in
the round pellets. The characterization by scanning He ion mi-
croscopy (SHIM) of all catalysts confirms the observations
made by HR-SEM (Supporting Information).
Aberration corrected HAADF-STEM with atomic resolution
was used to analyze the sulfide slabs with their basal (001)
plane parallel and perpendicular to the electron beam
(Figure 4). Figure 5(A) shows a representative HAADF-STEM
image of the top view of the basal plane of NiMoWS. The fact
that particles do not exhibit sharp edge structures is attributed
to the non-aligned terminations of the stacked sulfide slabs.[31]
The distance between the bright spots assigned to Mo or W
atoms is around 0.6 nm, which matches the distance of the
second metal–metal coordination sphere of the hexagonal lat-
tice of metal disulfides (5.48 Š). Additionally, the ADF profile
shows weak signals in a distance of 0.2 nm from W and Mo,
which are assigned to S atoms.[11,31] The appearance of relative-
ly bright and dark spots with different contrast (within 3 Š)
next to each other in Mo or W
Micrographs obtained by TEM (Figure 3A for NiMoWS and
S1 of the supporting information) show that the sulfide ag-
glomerates contained crystalline domains with varying sizes in
the order of nanometers. The averaged stacking degree of
around five for NiMoWS in the small crystals is in good agree-
positions hints to the presence
of projections of metal atoms
with very different averaged
molar mass in the same sulfide
slab. We propose that this corre-
sponds to the preferred pres-
ence of Mo and W forming bi-
metallic sulfide slabs, as also in-
dicated by Raman spectroscopy.
Such differences in Z-contrast
were also used to deduce the
Figure 3. Representative electron microscopy images study of sulfide NiMoWS; a) TEM image, B) selective area
electron diffractogram, and C) HR-SEM image. The rings in (B) and stacked pellets in (C) are identified as MoS2 and
WS2 phases, whereas the dots in (B) and the large particles in (C) belong to different Ni sulfides.
formation of mixed Mo1ÀxWxS2
layers by ADF studies of model
Mo1ÀxWxS2 particles combined
ment with the values obtained from XRD (Table 1). Further-
more, the structural disorder, i.e., bending (the MoS2 slabs are
not straight along the x and y axis), and random orientation of
the crystal domains observed in the micrographs is consistent
with the increased lattice parameters in z direction derived
from XRD and the shifts and asymmetry of the Raman signals
of the A1g and E1 modes. The longest slabs were found for
2g
NiMoWS with an average length of 20 nm, whereas slabs of
15 nm and 10 nm were present in NiWS and NiMoS, respec-
tively. The presence of NixSy species associated to WS2 and
MoS2 was confirmed by electron diffraction (SAED) of selected
area. For instance, Figure 3(B) shows the SAED of NiMoWS,
Figure 4. Aberration-corrected HAADF-STEM image of the sulfided NiMoWS
catalyst; A) overview image at 300 keV, B) side view in [001] direction at
80 keV. The brighter atoms indicate the heaviest element in NiMoWS, that is,
W atoms (marked by the circles).
where the dots corresponded to single Ni9S8 crystals (interpla-
nar distances of 2.8, 2.6, 2.3, 1.8, and 1.7 Š according to ICSD#:
63080[21]). The presence of MoS2/WS2 phases was indicated by
the broad rings in agreement with the broad reflections in the
XRD patterns.
Hence, all catalysts are concluded to contain mixtures of sul-
fide phases. MoS2 or WS2 slabs form stacks with a relatively
high degree of disorder, i.e., bending and misalignment
among them. In turn, these microcrystalline domains agglom-
erate with random orientations forming spherical particles.
Nickel sulfides exist in a variety of phases and particle sizes
ranging from very large (few microns), which act as support
for Mo(W)S2 agglomerates and may produce diffraction pat-
with atomically resolved electron energy loss spectroscopy
(EELS).[11]
In contrast to microscopy studies of monolayers or cleaved
single crystals reported,[9,12,31] strong Z-contrast of atoms and
perfectly resolved ADF images are not expected for the multi-
layer bent structures studied. On the other hand, Ni atoms, if
incorporated to the Mo(W)S2 structure should be located at
the edges of the slabs, according to the Co(Ni)ÀMoÀS model.
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3695
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
second pronounced contribution at around 3.5 Š (not phase-
shift corrected) was assigned to Ni-Ni contributions in Ni3S2
(ICSD #27521[21]). In the catalysts, this contribution is very
weak. The FT of the EXAFS at the Mo K-edge and at the W LIII-
edge are shown in Figure 6(B) and (C), respectively. The contri-
butions of Mo-S[14] and W-S[32] were observed around 2 Š (not
phase-shift corrected). The same distances (and similar intensi-
ties at the Mo K-edge) were found for the references and the
catalysts. The second shell contributions at around 3 Š (not
phase-shift corrected) at both edges were assigned to metal
backscatter within the Mo(W)S2 structure.[14,32] At both edges
(Mo K, and W LIII), the metal–metal contributions were weaker
in the catalysts than in the references. In NiMoWS the intensi-
ties of the second shell contributions were even lower than
the FT-EXAFS of the bimetallic one.
Figure 5. Representative aberration-corrected HAADF STEM image at 80 kV
of the sulfided NiMoWS catalyst; A) top view, MS2 [001] direction along the
electron beam, B) distribution of the ADF counts along the indicated white
line in (A). The circles indicate the heavier and therefore brighter W atoms.
The weak metal–metal contributions at around 3 Š in
NiMoWS might be explained by the rather poor alignment of
the atoms in the short and bent MoS2 or WS2 slabs.[32,33] How-
ever, this apparently contradicts the conclusions from XRD,
Raman spectroscopy, and electron microscopy that the slabs in
NiMoWS are ordered and are the largest among the studied
materials. The reason for the weak metal–metal contributions
in the FT is attributed to the specifics of the k3 weighted
EXAFS of the catalysts at the Mo K-edge and W LIII-edge and
the corresponding bulk reference materials in Figure 7 (the k3
weighted EXAFS functions of the samples at the Ni K-edge are
presented in Figure S7 of the Supporting Information). Note
that the amplitudes of the oscillations were weaker for the cat-
alysts and the fine structure was less pronounced. Moreover, at
the Mo K-edge and the W LIII-edge, the EXAFS of the trimetallic
NiMoWS catalyst was different from the reference and the bi-
metallic catalysts between 9 and 16 ŠÀ1. These differences in
the EXAFS could be caused by neighboring atoms with oppo-
site backscattering phases like Mo and W,[34,35] resulting in de-
structive interference. This is in line with the intralayer
Mo1ÀxWxS2 mixed sulfides suggested by HAADF-STEM.[31,32] To
assign the amplitudes and phase shifts of the MÀM absorber-
backscatter pairs, a series of mixed Mo1ÀxWxS2 model clusters
was generated based on the structure of MS2 and the corre-
sponding EXAFS were calculated. Special emphasis was given
to the different backscattering of Mo and W atoms at the Mo
However, identification of a third element by differences in Z-
contrast was not possible due to the less defined edges of the
slabs and the qualitative nature of the HAADF-STEM image
analysis. In order to unequivocally stablish the formation of
mixed Mo-W sulfides within one layer level and to elucidate
the interaction of Mo(W)S2 with Ni, detailed analysis of the X-
ray absorption spectra is required.
X-ray absorption spectroscopy
The formation of intralayer Mo1ÀxWxS2 slabs in NiMoWS has to
be reflected by Mo-W and W-Mo contributions at 3.16 Š,
whereas the scattering between Ni and the other metals
should be observed in higher coordination spheres. The X-ray
absorption near edge structure (XANES) at the Ni and Mo K-
edge as well as at the W LIII-edge are summarized in the Sup-
porting Information (Figure S3). The corresponding Fourier
transforms (FT) of the k3 weighted extended X-ray absorption
fine structure (EXAFS) are presented in Figure 6. Detailed de-
scriptions of the XANES and of the corresponding linear com-
bination fittings are presented in the Supporting Information.
The first contributions in the Fourier transforms of the
EXAFS at the Ni K-edge of the catalysts (Figure 6(A)), at around
2 Š (not phase-shift corrected) is assigned to Ni-S[14] contribu-
tions, which shifted to lower distances compared to Ni3S2. A
Figure 6. Fourier transforms of k3 weighted EXAFS at the A) Ni K-edge, B) Mo K-edge, and C) W LIII-edge of the references; Ni3S2 (A,a), MoS2 (B,a), and WS2
(C,a), as well as of the catalysts NiMoS (A,b and B,b), NiMoWS (c), and NiWS (A,d and C,b).
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3696
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
could reduce the coordination number at the specific
distance. This is the focus of an ongoing theoretical
EXAFS investigation. For the present case, it suffices
to understand the intermetallic interactions using the
small 5”5 particles with two layers and the central
metal atoms as absorber.
Let us analyze next the incorporation of W into the
MoS2 lattice. For the first possibility of model clusters,
interlayer mixed Mo1ÀxWxS2 particles, Mo and W disul-
fide layers were stacked in different sequences to
reach a stacking degree of 6 (similar to that found by
TEM and XRD for NiMoWS). The chosen stacking se-
quences were abbaab and ababab, where a stands
for the [MoS2] layer and b for the [WS2] layer. The
Mo-W coordination number at 6.42 Š was varied
from 0 for a monometallic disulfide slab over 3 for
the abbaab stacking sequence to the maximum
Figure 7. EXAFS at the A) Mo K-edge and at the B) W LIII-edge of MoS2 (A, a), WS2 (B, a)
and of the catalysts NiMoS (A, b), NiWS (B, b), and NiMoWS (c). In the grey highlighted
region, the EXAFS differ strongly from each other.
K- and at the W LIII-edges, respectively. Intralayer and interlayer
mixtures were included as model clusters as well as different
intermetallic arrangements and coordination numbers.
number possible of 6 in the ababab sequence. Additionally,
clusters of different sizes and shapes were created.
Selected clusters are presented in the Supporting Informa-
tion as Particle 2, and the corresponding EXAFS and FT at the
Mo K-edge are shown in Figure 8. In comparison to the pure
MoS2 cluster, the Mo EXAFS has new weak features at 12.5 and
15 ŠÀ1, especially for NMoÀW =6 (Figure 8(A, c)). However, the
presence of W in the next sulfide layers is just slightly noticea-
ble in the Mo EXAFS and FT. Increasing size or a hexagonal
shape for the sulfide layer does not change the EXAFS oscilla-
tions. These observations demonstrate that the distance to the
next layer at 6.42 Š is too far away to influence the EXAFS of
the absorber Mo atom significantly.
EXAFS of interlayer and intralayer mixed Mo1ÀxWxS2 model
particles
A MoS2 particle (16”16 Š) was created by using 5 hexagonal
unit cells in a- and b-direction, which corresponds to two
stacked sulfide layers containing 5Mo atoms in a and b-direc-
tion (presented in the Supporting Information as Particle 1).
The EXAFS function and its Fourier transformation are present-
ed in Figure S8. The oscillations of the Mo-S scattering domi-
nate in the lower k region up to 8 ŠÀ1, whereas the
metallic backscatter (i.e., the element with the higher
molecule weight) determines the backscattering in
the higher k region. As predicted by the EXAFS func-
tion,[36] the most intense oscillations were found for
the two next neighbors at rMoÀS =2.41 Š and rMoÀMo
=
3.16 Š. However, the backscattering of the more dis-
tant neighbors is needed to describe the structure as
visualized by Figure S8(C) and (D). Note that the ap-
proach using the crystallographic structure describes
perfectly ordered, and large particles (where most of
the atoms have full coordination) leading to the
highest possible intensity of the oscillations. The co-
ordination number of the next neighbors and at a dis-
tance of 6.42 Š, i.e., NMoÀS and NMoÀMo is, therefore, 6
for the model Particle 1 (also NMo-Mo is 6 at a distance
of 6.42 Š).
Figure 8. EXAFS at the Mo K-edge and the corresponding Fourier transforms of MoS2 (a),
MoWS2 with Mo-W=3 at 6.42 Š, stacking sequence abbaab (b), MoWS2 with Mo-W=6 at
6.42 Š, stacking sequence ababab (c), MoWS2 with Mo-W=2 at 5.48 Š (d), MoWS2 with
Mo-W=2 at 3.16 Š (e). The used particles are also shown in Supporting Information
(Color code: grey=Mo, black=W, light grey=S).
In distorted particles, the structural disorder leads
to a decrease in the intensity of the oscillations at
higher k values, which are typically accounted for in
the analysis of the EXAFS by including higher Debye–Waller
terms. However, the metal–metal coordination number is re-
duced for MoS2 particles with high distortion and disorder, es-
pecially at the particle edges.[33] Bending or distortion, as ob-
served for the investigated sulfides, causes also differences in
the bond distances of the metal–metal neighbors for a few
atoms such as in edge distorted particles. Therefore, bending
On the other hand, the particle size or shape does not influ-
ence the calculated phase shifts and amplitudes for the central
Mo atom, whereas the coordination number averaged over the
whole particle is influenced by size and particle shape. The
squared MoS2 particle with 5”5 Mo atoms (Particle 1) has aver-
age NMoÀS =5.2 and NMoÀMo =4.5. These values increase with
particle size and change with the shape. For the 9”9 abbaab
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3697
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
MoWS2 particle (Particle 2(b)) the coordination numbers were
MoÀS =NWÀS =5.6 and NMoÀMo =NWÀW =5.1, whereas for the hex-
N
agonal ababab MoWS2 particle with a diagonal of 27 Mo
atoms (a good representation of the investigated catalysts)
N
MoÀS =NWÀS =5.9 and NMoÀMo =NWÀW =5.7 were calculated (Par-
ticle 2(c)).
To determine the influence of Mo-W interactions on the
EXAFS of model clusters with Mo and W in the same slabs, W
atoms were incorporated within a layer of MoS2 (Particle 2(d)
and (e) in the Supporting Information). These clusters contain
4 W atoms and 21 Mo atoms per sulfide layer and mixed met-
allic coordination numbers of 2. In Particle 2(d), two Mo atoms
were replaced by W at 6.33 Š and 5.48 Š with regard to the
central Mo atom. In Particle 2(e), a coordination number of
Figure 9. Calculated k3 weighted Mo-Mo paths (black lines) and Mo-W paths
(grey lines) at 3.16 Š at the Mo K-edge visualize the p shift between both
scattering paths (solid lines correspond to N(MÀM)=6 and dotted lines to
N(MÀM)=2).
N
MoÀW =2 was calculated at distances of rMoÀW =5.48 Š and
3.16 Š. The EXAFS and the corresponding FT calculated for
these particles are shown in Figure 8. The EXAFS of the Parti-
cle 2(d) with long Mo-W distances were very similar to those
of pure MoS2, i.e., the coordination number of NMoÀW =2 is too
low and the distance between Mo and W atoms is too large to
influence the overall EXAFS. Interestingly, the presence of W at
were calculated with different coordination numbers (Figure 9).
The p shift between the two scattering paths was observed in
À1
a wide range between 9 and 14 Š and was consistent for dif-
r
MoÀW =3.16 Š in Particle 2(e) has a large influence on the
ferent coordination numbers and distances. The presence of
both scattering paths in one sample leads to a destructive in-
terference. Therefore, the EXAFS of an Mo/W=1:1 solid solu-
tion in a Mo1ÀxWxS2 system with a homogeneous dispersion of
Mo and W is only determined by the metal–sulfur atom-back-
scatter pairs in the k range from 9 to 14 ŠÀ1.
EXAFS despite the low coordination number of NMoÀW =2. The
EXAFS oscillations change between k=10–16 Š and the
À1
second contribution in the FT at around 3 Š is strongly re-
duced compared to pure MoS2, similar to the EXAFS observed
for the unsupported NiMoWS sulfide catalyst.
The analysis of the EXAFS of model clusters demonstrates
that the formation of structures containing MoS2 next to WS2
phases cannot be ruled out by XAS. On the other hand, the
theoretical EXAFS shows that the presence of the intralayer
mixed Mo1ÀxWxS2 EXAFS of NiMoWS is clearly established.
After concluding that Mo and W form intralayer MoWS2 mix-
tures in NiMoWS, several intralayer Mo1ÀxWxS2 clusters consist-
Figure S9 shows all involved single scattering paths and the
resulting overall EXAFS and FT of the 5”5 mixed sulfide cluster
with NMoÀW =3 at rMoÀW =3.16 Š (Particle 3(f)) at the Mo K-edge.
The metal–sulfur paths are dominant, although complete ex-
tinction of the metal–metal contributions does not occur. All
EXAFS and the corresponding FT for the clusters created (Parti-
cles 3 and 4) are presented in Figure S10 for the Mo K-edge
and in Figure S11 for the W LIII-edge. The EXAFS oscillations
and their FT at both edges strongly changed in the region k=
9–16 Š for clusters with varying intermetallic coordination
number at r=3.16 Š. The second metal–metal contribution
steadily decreases with increasing mixed metallic coordination
number starting from a value for NMoÀW and NWÀMo of 2. Further-
more, a splitting of the signal was observed and the lowest in-
tensity was obtained for NMoÀW =3 and NWÀMo =2.
The experimental data of NiMoWS catalyst and the EXAFS of
model clusters are compared at the Mo K-edge and the W LIII-
edge in Figures 10 and 11. For the experimental data, the in-
tensity was lower and less features are visible in the EXAFS as
well as in FT compared to the model clusters. This indicates
that the NiMoWS catalyst does not have a long-range crystal-
line structure and consists of different phases.
ing of two layers with 5”5 atoms and varying NMoÀW at rMoÀW
=
3.16 Š were created and the corresponding EXAFS were calcu-
lated (Particles 2(d) and 2(e) were already discussed). Other
clusters in this series with Mo and W as central atom are pre-
sented as Particle 3 and Particle 4 in the Supporting Informa-
tion. Clusters with NMoÀW of 0, 1, 2, 3, 4, and 6 were generated
by this approach. As the first metal–sulfur contribution at 2.4 Š
was not influenced by the replacement of Mo by W (see
Figure 8) only metal–metal scattering paths of the model clus-
ters are discussed in the following.
The EXAFS of an absorber-backscatter pair is the result of
the interference between the outgoing spherical wave of the
photoelectron generated by the absorption process and the
spherical wave backscattered from the neighboring atoms. The
phase differences of both waves depend on the type of the
atoms involved and distance between them. As an example,
the phase functions (shown in the Supporting Information)
Figures 10 and 11 also allow a qualitative comparison of the
EXAFS of NiMoWS with those of the model clusters, the pat-
terns of NiMoWS (shown in the lines labelled with (d)) fit well
between the model patterns with N=2 and N=3. Therefore,
mixed metallic coordination numbers between 2 and 3 exist in
NiMoWS. Thus, for the final EXAFS fitting (multi-edge fitting),
the theoretical FEFF-paths of the clusters with N(Mo-W) and
N(W-Mo) of 3 and 2, respectively, were used.
were calculated at k=11.34 AÀ1 as fMoÀMo =1.42*p, fMoÀW
2.41*p at the Mo K-edge, and fWÀW =4.38*p, and fWÀMo
=
=
3.39*p at the W LIII-edge by using tabled phase shifts.[37] These
values indicate that the phase functions of these particular ab-
sorber-backscatter pairs are shifted by p at both metal edges.
Subsequently, the EXAFS of the metal–metal scattering paths
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3698
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 2–4 summarize the best-fit results for the k3
weighted EXAFS data of the sulfided catalysts at the
Mo K-edge, W LIII-edge, and Ni K-edge. For the sulfide
NiMoS catalyst (Table 2), full Mo-S coordination envi-
ronment was found at the Mo K-edge, and the bond
distances match those corresponding to MoS2 (Fig-
ure S15 and S16 show the experimental EXAFS data
compared to the fit). The presence of Mo-O neigh-
bors at short distances was not required during the
fit procedure. The NMoÀMo in NiMoS was 5.1, which is
below the coordination number in the reference
MoS2 (NMoÀMo =6). This implies that, although the
metal was in a trigonal-prismatic environment, not all
atoms have full coordination environment (i.e., metal
atoms at the edges of the slabs). The value for
Figure 10. Experimental EXAFS of NiMoWS compared to EXAFS at the Mo K-edge (A) and
the corresponding Fourier transforms (B) of model MoWS2 with varying Mo-W coordina-
tion number N at 3.16 Š, namely MoS2 bulk reference (a), MoWS2 with NMoÀW =2 and NWÀ
Mo =6 (b), MoWS2 with NMoÀW =2 and NWÀMo =2 (c), NiMoWS catalyst (d), MoWS2 with
N=3 (e), and MoWS2 with N=4 (f).
N
NiÀS =4.5 (Table 4) indicates that Ni is either tetrahe-
drally or pentagonally coordinated by sulfur, which
fits to the observed pre-edge feature of the XANES
(Supporting Information). The second contribution at
r
NiÀNi =2.6 Š with NNiÀNi =1.2, as well as the third con-
tribution at rNiÀNi =3.9 Š with NNiÀNi =4.4 were smaller
than those in Ni3S2. The metal–metal coordination
numbers (Mo-Mo) and (Ni-Ni), smaller in NiMoS than
in the references, reflect lower crystallinity (smaller
size of the MoS2 slabs and Ni sulfide particles and
large disorder). Moreover, other Ni sulfide phases
than Ni3S2 are probably present as well. Additionally,
a Ni-Mo contribution was observed at 2.7 Š with the
coordination number of 0.2, which indicates that Ni
is indeed associated with MoS2. The quality of the fit
was improved by 6.5% after adding this contribution.
The assumption of a Mo-Ni contribution at the Mo K-
edge and the associated constraints improved the fit
further by 2%.
Figure 11. Experimental EXAFS of NiMoWS compared to the EXAFS at the W LIII-edge (A)
and the corresponding Fourier transforms (B) of model MoWS2 clusters with varying W-
Mo coordination number N at 3.16 Š, namely WS2 bulk reference (a), MoWS2 with NWÀ
Mo =2 and NMoÀW =6 (b), MoWS2 with NWÀMo =2 and NMoÀW =2 (c), NiMoWS catalyst (d),
MoWS2 with N=3 (e), and MoWS2 with N=4 (f).
Table 2. Best-fit results for k3 weighted EXAFS data of the sulfided
catalysts at the Mo K-edge.[a]
EXAFS analysis
In the discussion above, we have demonstrated that the X-ray
absorption spectra suggest the presence of intralayer
Mo1ÀxWxS2 clusters in NiMoWS in accordance with electron mi-
croscopy. As a final step of the XAS data analysis, a multi-edge,
multi-scattering fit procedure was applied to analyze the
EXAFS of the sulfides at all metal edges simultaneously. The
three references, MoS2, WS2, and Ni3S2 were fitted by using
FEFF paths calculated from the crystallographic structure. The
fit of the EXAFS and the corresponding Fourier transforms of
the references are presented in Figure S12–S14 and Tables S3–
S5. MoS2 and WS2 showed the expected trigonal-prismatic co-
ordination environment with the maximum metal–sulfur and
metal–metal coordination of six, respectively. Ni3S2 was difficult
to fit, since the distances to the 4S and the 4Ni neighbors
were very close. Moreover, the two distances for Ni-S neigh-
bors at 2.26 and 2.27 Š could not be differentiated and were,
therefore, fitted together resulting in an overall coordination
number of 4.4. The most intense single scattering paths up to
4.1 Š were added to the analysis to obtain an appropriate fit.
Catalyst
Shell
r
[Š]
N
s2
[Š2]
E0
[eV]
NiMoS
Mo-S
2.40 (0.01) 6.2 (0.2) 0.0025 (0.0002) 0.74 (0.67)
R=0.0013 Mo-Mo 3.17 (0.01) 5.1 (0.1) 0.0033 (0.0001)
Mo-Ni
2.68 (0.02) 0.3 (0.1) 0.0024 (0.0022)
NiMoWS
Mo-S
2.40 (0.01) 5.2 (0.1) 0.0027 (0.0002) 1.52 (0.60)
R=0.0055 Mo-Mo 3.16 (0.01) 3.1 (0.1) 0.0043 (0.0010)
Mo-W
Mo-Ni
3.17 (0.01) 1.6 (0.2) 0.0042 (0.0004)
2.68 (0.03) 0.6 (0.1) 0.0093 (0.0029)
[a] Abbreviations: r: distance, N: coordination number, s2: Debye–Waller
like factor, E0: inner potential; in parenthesis the absolute errors.
In NiWS, W-S and W-W contributions were found at the
same distances as in the reference WS2 structure. The experi-
mental and fitted EXAFS and FT at the W and Ni edge, are
shown in Figure S17 and S18, results of the fits are summarized
in Table 3 and Table S5. The coordination numbers NWÀS and
NWÀW were 4.5 and 3.1, respectively. Both are smaller than for
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3699
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the model clusters, Mo-W and W-Mo contributions were found
at 3.16 Š and 3.17 Š. The quality of the overall EXAFS fit im-
proved by 25% by using the FEFF-paths of the model
Mo1ÀxWxS2 cluster with N(Mo-W) of 3 (Particles 3(f) and 4(f)) as
a model to fit the experimental EXAFS. At the Mo K-edge,
Table 3. Best-fit results for k3 weighted EXAFS data of the sulfided
catalysts at the W LIII-edge in k space.[a]
Catalyst
Shell
r
N
s2
[Š2]
E0
[Š]
[eV]
NiWS
R=0.0033 W-W
W-S
2.40 (0.01) 4.5 (0.2) 0.0037 (0.0002) 7.36 (0.44)
3.15 (0.01) 3.1 (0.4) 0.0045 (0.0005)
N
MoÀMo =3.1 and NMoÀW =1.6 were found, which leads to an
average Mo-metal coordination number of 4.7 at around 3.2 Š.
The average NMoÀNi in NiMoWS is 0.6, i.e., twice as high as in
NiMoS. The coordination numbers at the W LIII-edge for
NiMoWS were 2.0 for NWÀMo, 2.1 for NWÀW and 0.6 for NWÀNi. The
latter coordination number was higher than in the bimetallic
NiWS catalyst. The finding of higher Mo-Ni and W-Ni coordina-
tion numbers in NiMoWS than in NiMoS and NiWS is impor-
tant, because it indicates better interaction of Ni on the sulfide
slabs in the trimetallic than in the bimetallic sulfides. For this
NiMoWS catalyst, NMoÀMo is higher than NWÀW, whereas NMoÀW is
lower than NWÀMo. These observations suggest that the content
of Mo and W within a sulfide slab is similar to that of the bulk
(slightly more Mo than W, Table 1). Thus, most of the sulfide
slabs must be Mo1ÀxWxS2 particles containing both metals in
one layer. NNiÀS of 3.8 at rNiÀS =2.27 Š was found in NiMoWS,
which is the lowest NNiÀS value among all measured samples.
W-Ni
2.75 (0.05) 0.1 (0.1) 0.0033 (0.0030)
NiMoWS
W-S
2.41 (0.01) 4.9 (0.2) 0.0044 (0.0006) 8.02 (1.23)
3.17 (0.02) 2.1 (0.3) 0.0042 (0.0006)
R=0.0055 W-W
W-Mo 3.17 (0.01) 2.0 (0.2) 0.0042 (0.0004)
W-Ni 2.82 (0.04) 0.6 (0.2) 0.0030 (0.0021)
[a] Abbreviations: r: distance, N: coordination number, s2: Debye–Waller
like factor, E0: inner potential; in parenthesis the absolute errors.
Table 4. Best fit results for k3 weighted EXAFS data of the sulfided
catalysts at the Ni K-edge in k space.[a]
Catalyst
Shell
r
[Š]
N
s2
E0
[eV]
2
[Š ]
NiMoS
Ni-S
2.27 (0.02) 4.5 (0.5) 0.0075 (0.0010) 3.00 (2.66)
2.60 (0.03) 1.2 (0.5) 0.0071 (0.0031)
R=0.0013 Ni-Ni
Ni-Mo 2.68 (0.02) 0.2 (0.1) 0.0024 (0.0022)
N
NiÀNi =0.7 at 2.59 Š lies between the values compared to the
Ni-Ni
Ni-S
3.88 (0.06) 4.4 (3.5) 0.0169 (0.0076)
bimetallic catalysts. The Ni-Mo contribution was found at a dis-
tance of 2.68 Š, which is the same as in in NiMoS. A slightly
longer distance was found for the Ni-W contribution in
NiMoWS compared to that in NiWS (both are longer than the
Ni-Mo distance). These mixed Ni-Mo and Ni-W contributions
have higher coordination numbers than in the bimetallic cata-
lyst, especially for Ni-W. The observations confirm that the
Mo(W)-Ni incorporation was higher in the trimetallic than in
both bimetallic catalysts.
NiWS
2.26 (0.01) 4.4 (0.6) 0.0089 (0.0013) 2.35 (1.51)
2.57 (0.02) 0.5 (0.2) 0.0052 (0.0020)
2.75 (0.05) 0.1 (0.1) 0.0033 (0.0030)
3.97 (0.08) 3.1 (2.7) 0.0239 (0.0119)
R=0.0033 Ni-Ni
Ni-W
Ni-Ni
NiMoWS
Ni-S
2.27 (0.03) 3.8 (0.7) 0.0076 (0.0034) 2.85 (4.75)
2.59 (0.05) 0.7 (0.5) 0.0055 (0.0045)
R=0.0055 Ni-Ni
Ni-Mo 2.68 (0.03) 0.3 (0.1) 0.0093 (0.0029)
Ni-W 2.82 (0.04) 0.2 (0.2) 0.0030 (0.0021)
[a] Abbreviations: r: distance, N: coordination number, s2: Debye–Waller
like factor, E0: inner potential; in parenthesis the absolute errors.
Catalytic activity
The kinetic data are compiled in the Supporting information.
Figure 12 summarizes the dependence of o-propylaniline (OPA)
and dibenzothiophene (DBT) conversion rates on temperature.
OPA is converted to propylbenzene (PB) via a Csp2-N cleav-
age (direct denitrogenation route, DDN); and to propylcyclo-
hexylamine (PCHA) via hydrogenation of the benzoic ring. This
hydrogenation route (HYDN) continues with fast nitrogen re-
moval to propylcyclohexene (PCHE) via Hoffman-elimination
and hydrogenation to propylcyclohexane (PCH).[38] The HDS of
DBT follows two pathways, direct desulfurization (DDS) and hy-
drogenation (HYDS). The former pathway has biphenyl (BiPh)
as only product, whereas HYDS leads to tetrahydrodibenzo-
thiophene (H-DBT), which is further hydrogenated to phenylcy-
clohexane (PhCH) or biphenyl (BiCH) via dodecahydrodibenzo-
thiophene (DH-DBT).[39] The reaction networks for HDN and
HDS are illustrated in Figures S22 and S23. All mentioned prod-
ucts were found for all three catalysts.
the reference WS2, which indicates smaller particle size and dis-
tortion in the catalyst. The addition of a W-O contribution at
around 2 Š resulted in the degradation of the fit quality. Thus,
its presence was excluded. The distances and coordination
numbers of the first Ni-S contribution match those of Ni3S2
(Table 4). The second and third Ni-Ni contributions appear at
2.57 Š (NNiÀNi =0.5) and at 3.97 Š (NNiÀNi =3.1), i.e., the coordina-
tion numbers are much smaller and a shift of the Ni-Ni contri-
bution is observed compared to Ni3S2 and NiMoS. These obser-
vations also suggest smaller particle sizes of different Ni sulfide
phases in NiWS. The fit was slightly improved by adding W-Ni
and Ni-W contributions at 2.75 Š, which indicates that also in
NiWS, Ni associates to the WS2 slabs although probably to
a minor extent compared to NiMoS.
The results of the EXAFS analysis and the corresponding FT
of the trimetallic NiMoWS catalyst are summarized in Tables 2
to 4 and Figure S19, S20, and S21. The Mo-S and W-S coordina-
tion numbers were both around 5.0, lower than in NiMoS, but
higher than in NiWS. This reflects an intermediate sulfidation
(less complete sulfur coverage at the edges) state for the tri-
metallic catalyst. Furthermore, as predicted by the EXAFS of
The rates normalized to metal content were identical for
both HDN and HDS with the HDN rates being 4–5 times higher
than rates of HDS (Figure 12). The nearly identical energy of ac-
tivation for related reactions suggests that the differences are
caused by differences in the concentrations of active sites.
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3700
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
values of the HYDS route increased in the order
NiWS<NiMoWS<NiMoS (the very low yields of the
products of this route hinder quantitative discussion).
Structure activity correlations
Control experiment have shown that species of Ni
sulfides are relatively unreactive for HDN and HDS,
compared to Mo(W)S2 phases. Therefore, in the fol-
lowing only Mo(W)S2 species are discussed as catalyt-
ically active. All estimated rates in the HDS and HDN
reaction of the trimetallic NiMoWS sulfide catalyst
were higher than those of bimetallic catalysts. The
properties determining the activity of sulfide catalysts
are the dispersion of the active phase and type of
active sites, as both together control the concentra-
tion, availability, and its intrinsic activity. Especially,
Lewis acidic sulfur vacancies (CUS) acting as adsorption sites
and Brønsted acidic -SH groups providing hydrogen are essen-
tial for hydrotreating.[4,40] The -SH groups are not only needed
for the hydrogenation steps, but also to provide hydrogen
during the cleavage of the carbon-heteroatom bonds and
could act as weak adsorption site.[4,38] The intrinsic activity is in-
fluenced by the nature of the active phase (WS2 or MoS2), the
sulfidation degree, and the concentration of Ni. The dispersion
is determined by the sulfide morphology and particle size.
Therefore, the higher activity of NiMoWS suggests a higher
concentration and/or availability of CUS and -SH groups com-
pared to the bimetallic sulfides.
Figure 12. Hydrotreating activity normalized per metal content in function of tempera-
ture for the determination of the apparent activation energy for the HDN of OPA (A) and
!
&
*
HDS of DBT (B) of NiMoS ( ), NiMoWS ( ) and NiWS ( ). The HDN rates in the absence
!
&
*
of DBT are as well shown (open symbols
).
NiMoWS exhibited the highest conversion rates followed by
the bimetallic NiWS and NiMoS, which had similar HDN rates.
The presence of DBT decreased the HDN rates, especially on
NiMoS, without showing impact on the selectivity (Figure 13
for NiMoWS and in the Supporting Information for the bimetal-
lic catalysts). The HDN rates in absence of DBT were higher for
NiMoS than for NiWS, whereas NiWS had the higher HDN rates
in the presence of DBT. In this regard, the HYDN route was
faster than DDN by one order of magnitude. In contrast, the
HYDS route was slower than DDS also by one order of magni-
tude. The marked preference for DDS is illustrated by the se-
lectivity towards BiPh, Figure 13(B). The apparent activation
energies (Ea) for the DDN route were larger than those of the
HYDN route. For instance, 191–230 kJmolÀ1 (DDN) and 119–
129 kJmolÀ1 (HYDN), in the absence of DBT (Table S6). In gen-
eral, the presence of DBT decreased the Ea values of both
routes slightly, which is hypothesized to indicate that the ad-
sorption of OPA becomes stronger in the presence of DBT
(more negative adsorption enthalpies would decrease the ap-
parent activation energies). The apparent activation energies
for the DDS route of HDS were 137 kJmolÀ1 for NiMoS,
139 kJmolÀ1 for NiMoWS and 148 kJmolÀ1 for NiWS. The Ea
The trimetallic sulfide has intralayer mixed Mo1ÀxWxS2 slabs
and an average metal–sulfur coordination number of 5, where-
as the NWÀS was only 4.5 for WS2 in NiWS. For NiMoS the NMoÀS
of 6 suggests fully coordinated Mo atoms in MoS2. An average
metal–sulfur coordination number lower than 6 for the W-con-
taining sulfides suggest the presence of CUS (reduced edges)
or of distorted sulfide structures. The metal–sulfur coordination
is especially interesting to analyze for WS2, since the sulfidation
of W oxides is slower than that of Mo oxides (WÀO bond
strength of 7.0 eV compared to 5.8 eV of Mo-O).[41] In turn, the
activity of Ni-WS2 has been correlated to its sulfida-
tion degree.[42,43] The results demonstrate, however,
that the contribution of oxide species is minor for
both W containing catalysts as deduced from LCF
and EXAFS, both pointing to a nearly complete sulfi-
dation. Thus, the relatively low NWÀS values are attrib-
uted to the bent and less aligned morphology.
Differences in morphology and particle size were
observed between the three unsupported catalysts.
The highest stacking degree was observed for NiMoS,
the largest sulfide slabs for NiMoWS, and the most
disordered slabs for NiWS. Thus, we conclude that
the low NWÀS value in NiWS is caused by distortion
and bending of WS2 slabs, which leads to variations
Figure 13. Selectivities along with conversion over NiMoWS for (A) OPA HDN and (B) DBT
in bond length and angles. Therefore, in the EXAFS
not all sulfur neighbors were visible at the normal
distance, leading to an underestimation of the sulfur
coordination number.[33] The same is concluded to
~
!
&
*
HDS. The products in (A) are PCHE ( ), PCH ( ), PB ( ), and PCHA ( ).The HDN selectivi-
~ !
&
*
ties are presented in the presence of DBT (black symbols
) and in the absence of
~ !
!
&
&
*
*
DBT (open symbols
). The products in (B) are BiPh ( ),BiCH ( ), H-DBT ( ), and
~
PhCH ( ).
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3701
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
occur in NiMoWS as suggested by the strong sulfide slab
bending. Such exposed sulfur atoms in a distorted environ-
ment are more labile and, hence, easier to remove to create
CUS, which is connected to a formal reduction to W or Mo.
Note that for the pure MoS2 and WS2 phases, the metalÀS
bond strength is 2.6 eV, and 2.9 eV, respectively,[44] suggesting
that only small differences in the reduction degree of MoS2
and WS2 are probable. However, the substitution of W by Ni
might be more effective for WS2 than for MoS2, leading to
higher CUS concentration than for promoted MoS2.[32] In con-
trast, the well-ordered highly stacked slabs, i.e., high crystallini-
ty, as observed for NiMoS influences the active site concentra-
tion negatively and, in turn, lowers the activity. Note in passing
that increasing particle size and stacking degree increase are
among the causes of deactivation. Thus, we conclude that
both W containing catalysts have a higher CUS concentration,
in line with their higher activity compared to NiMoS.
to NiMoWS with NMoÀNi and NWÀNi of 0.3 and 0.1 in NiMoS and
NiWS, respectively. Therefore, the interaction between the in-
tralayer mixed Mo1ÀxWxS2 phase in NiMoWS and the promoter
Ni is concluded to be more effective than with the pure MoS2
and WS2 phase.
The formation of promoted sulfide phases has been demon-
strated for Ni-W and bulk sulfides[13,32,34,49] and is indicated by
the significant mixed metallic coordination numbers (Mo-Ni or
W-Ni). It is concluded that the concentration of Ni-promoted
sulfide CUS was higher for NiMoWS than for NiMoS, which
translates to an increased -SH concentration. The interactions
of Ni with WS2 in NiWS are in contrast very weak.
It is hypothesized that the faster sulfidation rates for Mo
species than for W increases the probability of incorporation of
Ni. In addition to the Ni promoting species, the formation of
a variety of Ni sulfide species is observed. The relatively low
NNiÀNi values deduced from EXAFS compared to the reference
Ni3S2 are attributed to the existence of several Ni-Ni distances
and NNiÀNi values broadening the corresponding EXAFS. In line
with this hypothesis, different Ni sulfide phases were observed
by XRD and electron microscopy. Note, that in NiMoS, NNiÀNi at
2.6 Š is 1.2, whereas in NiWS and NiMoWS, NNiÀNi at around the
same distance is 0.5 and 0.7, respectively. This indicates that
the NiSx particles are, in average, smaller when interacting with
WS2 or Mo1ÀxWxS2.[32]
Thus, the performance of unsupported NiMoS is dominated
by the Ni promoted MoS2 phase. Ni- and Mo-associated CUS
are present, which act as adsorption and reaction sites for
OPA, DBT, and H2. However, the overall concentration of active
sites is lower compared to NiMoWS due to the morphology of
NiMoS. Therefore, the reactants compete for fewer Ni-promot-
ed sites, which is in line with the very low conversion rate for
the HYDS route of DBT and the decrease of the conversion
rate for the HYDN route in the presence of DBT. The DDN rates
were hardly affected by DBT because the active sites for DDN
are Mo associated sites instead of Ni-CUS.[38]
CUS and -SH groups are located at the perimeter, i.e., at ex-
posed edges of the Mo and W sulfide slabs. Considering hex-
agonal geometry, the number of Mo and W atoms at the
edges of a sulfide particle derived from the average slab
lengths, is 185 atoms for NiMoWS, 134 atoms for NiWS and
91 atoms for NiMoS per slab.[45] The trend for the edge atoms
of the catalysts matches the found activity trend in HDS, i.e.,
the larger the slab the higher the HDS activity. However, the
average fraction of Mo and W at the sulfide edge in relation to
the total number of Mo and W atoms per sulfide slab[45] results
in very low values for large particles such as in NiMoWS
(fMo,W =0.06) and NiWS (fW =0.08). This fraction is used as a mea-
sure for the dispersion of supported sulfide systems, low dis-
persion accounts for low hydrotreating activity.[39,46] This appar-
ent contradiction is resolved by considering that the slabs of
the catalysts are neither rigid nor straight. We speculate that
the distortion of large sulfide slabs strongly increases the
active site concentration.
Following the hypothesis that small, poorly crystalline sulfide
particles are required for high catalytic activity, NiWS would be
expected to be more active than NiMoWS based on its shorter
sulfide slabs and the lower metal–sulfur coordination numbers.
The question also arises as to why NiMoWS is more active than
the strongly disordered NiWS, whereas at the same time the
HDN rate on NiMoS was comparable to the HDN rate of NiWS.
As discussed below the answer is related to the nature of the
reactive perimeter and the way Ni influences it.
The morphology of WS2 appears to be better suited to stabi-
lize a high active site concentration. The fraction of Ni promot-
ed WS2 is relatively low compared to NiMoS, however, the high
hydrogenation rates of DBT and OPA in presence of DBT (Figur-
es S25–S26 and Table S6) might be due to higher intrinsic ac-
tivity of the Ni-promoted W sites. This is consistent with the re-
ported higher hydrogenation rates for Ni-W sulfides compared
to Ni-Mo,[50] being more active for, e.g., HDS of substituted di-
benzothiophenes (for which hydrogenation is critical in the re-
action pathway).[51,52] These sites are less active to convert OPA
via the HYDN route in the absence of DBT compared to NiMoS.
However, in the presence of DBT the HYDN rates decrease at
most by 18%, whereas in NiMoS a decrease of up to 43% was
observed. This suggests that the W associated sites are less af-
fected by the presence of DBT.
Hydrogenation has been found to scale with the concentra-
tion of -SH groups, which are created by dissociative adsorp-
tion of H2S and H2 at CUS. The incorporation of Ni in MoS2/
Al2O3 increased the concentration of -SH groups,[40] leading in
turn to a correlation of the hydrogenation rates to the Ni con-
centration.[38,47] The incorporation of Ni occurs on the edges of
mixed sulfide phases, i.e., Mo(W)S2 slabs.[48] All three materials
contained large concentrations of Ni and Ni-promoted MoS2
and WS2 cations were identified by probe molecules (here not
presented). In NiMoWS, the highest coordination numbers for
Mo-Ni (Ni-Mo) and W-Ni (Ni-W) were observed as well as NNiÀS
of 3.8 (the lowest value in the series). The intermetallic coordi-
nation numbers in the bimetallic catalysts are low compared
The unsupported trimetallic NiMoWS catalyst exhibit a mix-
ture of the sites and structural features found in NiMoS and
NiWS. This is also indicated by the activation energies of the
catalytic routes being in between the values observed on
NiMoS and NiWS. Moreover, this mixture results in a higher
density of active sites. It is concluded that the intralayer mix-
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3702
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ture (Mo and W present in the same slab) is synergistic for the
hydrotreating activity, stabilizing a concentration of Ni-promot-
ed sites and a high concentration of -SH groups. The intralayer
Mo1ÀxWxS2 phase enabled the growth of long slabs with a mod-
erate stacking degree and a distortion providing a high edge
area and, therefore, a higher CUS concentration than in NiWS.
The higher concentration of Ni in the Mo1ÀxWxS2 mixed phase
compared to NiMoS is concluded to lead in turn to a higher
-SH group concentration.
clusters[47] remains unresolved as evidence of Ni atoms at the
edges of Mo(W)S2 slabs and/or defined Mo(W)S2-NiSx phases[54]
were not observed by microscopy. The small Ni-Ni coordination
numbers suggest that a substantial fraction of NiSx consists of
very small clusters or atomically dispersed Ni.
The catalytic properties (hydrodesulfurization of dibenzo-
thiophene and hydrodenitrogenation of o-propylaniline) were
governed by the concentration of accessible cations and the
concentration of -SH groups. The trimetallic catalyst had the
highest concentration of active sites leading to the highest
HDN and HDS rates. The trimetallic sulfide was, thus, conclud-
ed to have the largest specific perimeter. We conclude that the
simultaneous presence of Mo and W in the same slab in
NiMoWS retards the growth and favors nucleation of Ni pro-
moting species, allowing so the largest fraction of Ni to be in-
corporated. The additional slow growth in z direction leads to
a maximizing of the active sites at the perimeter of the
particles.
Conclusions
All three mixed sulfide phases, i.e., NiMoS, NiWS, and NiMoWS
offer interesting possibilities for hydrogenation, hydrodesulfuri-
zation as well as hydrodenitrogenation. The characterization
suggests that all catalysts consist of mixtures of Ni containing
Mo(W)S2 and Ni sulfides (Ni9S8, Ni3S2, and Ni3S4). The
(Ni)Mo(W)S2 phase is formed by stacks of 4–6 sulfide slabs with
some degree of bending and mismatch between the layers.
Stacks of the sulfide particles agglomerate in random direc-
tions forming spheres with sizes in the submicron range on
a mesoscopic level. The Ni sulfides show in contrast a broad
distribution of particle sizes ranging from few microns, on
which the Mo(W)S2 agglomerates deposit, to small particles
completely covered by the Mo(W)S2 domains.
Acknowledgements
This work was supported by the Chevron Energy Technology
Company. The authors would like to specially thank Dr.
Alexander Kuperman, and Dr. Jinyi Han for fruitful discussions.
We thank Prof. Roel Prins for the critical review of the results and
Prof. Matthias Bauer and Prof. Moniek Tromp for their assistance
throughout the whole presented work. The authors acknowledge
the light source facility DORIS III at DESY (member of the Helm-
holtz Association, Germany) and ESRF (Grenoble, France) for the
provision of beam time as well as the HASYLAB staff at DESY
(beamline X1) and the DUBBLE staff at ESRF. We are also grateful
to Dr. Marianne Hanzlik for TEM measurements, Dipl.-Min. Katia
Rodewald for HR-SEM measurements (Institute of Silicon Chemis-
try, TU Munich) and Dipl.-Ing. Xaver Hecht for technical support.
Further, Robert Colby and Bernd C. Kabius from PNNL are
acknowledged for the SHIM measurements.
Microscopy images with atomic resolution showed metal
(Mo or W) rich stacks of atoms with remarkable differences in
Z-contrast in NiMoWS, which suggested the formation of intra-
layer Mo1ÀxWxS2 particles (Mo and W in the same sulfide slabs),
as well as the preference of alignment of homotopic cations in
the projection direction of the STEM measurements. To confirm
this hypothesis series of model clusters were constructed and
the corresponding EXAFS were calculated. Models with inter-
layer Mo1ÀxWxS2 particles (slabs of MoS2 and WS2 stacked in dif-
ferent sequences) and intralayer Mo1ÀxWxS2 particles were con-
sidered. The analysis of the theoretical EXAFS showed that the
presence of mixed Mo and W in different slabs did not influ-
ence the EXAFS and Fourier transforms at the Mo (K-edge). In
contrast, the presence of W in close vicinity of Mo within a sul-
fide slab decreased the metal–metal (Mo-W or W-Mo) backscat-
tering. This was caused by destructive interference between
the Mo-W and Mo-Mo scattering pairs with opposite phases.
The same effect was observed in the EXAFS and Fourier trans-
forms of the NiMoWS catalyst, which suggested the formation
of sulfides slabs with Mo and W.
Keywords: electron microscopy
· hydrodenitrogenation ·
hydrodesulfurization · hydrotreating · multimetallic sulfides · X-
ray absorption spectroscopy
[1] W. Bensch in Comprehensive Inorganic Chemistry II, 2nd ed. (Eds.: J.
Reedijk, K. Poeppelmeier), Elsevier, Amsterdam, 2013, pp. 287–321.
[2] R. Prins, Adv. Catal. 2001, 46, 399–464.
[3] R. R. Chianelli, G. Berhault, B. Torres, Catal. Today 2009, 147, 275–286.
[4] O. Y. GutiØrrez, A. Hrabar, J. Hein, Y. Yu, J. Han, J. A. Lercher, J. Catal.
2012, 295, 155–168.
[5] P. M. Mortensen, J. D. Grunwaldt, P. A. Jensen, K. G. Knudsen, A. D.
Jensen, Appl. Catal. A 2011, 407, 1–19.
[6] U. Maitra, U. Gupta, M. De, R. Datta, A. Govindaraj, C. N. R. Rao, Angew.
Chem. Int. Ed. 2013, 52, 13057–13061; Angew. Chem. 2013, 125, 13295–
13299.
[7] J. P. Wilcoxon, T. R. Thurston, J. E. Martin, Nanostruct. Mater. 1999, 12,
993–997.
[8] T. F. Jaramillo, K. P. Jørgensen, J. Bonde, J. H. Nielsen, S. Horch, I. Chor-
kendorff, Science 2007, 317, 100–102.
[9] 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, I. V. Shvets, S. K. Arora, G. Stanton,
H. Y. Kim, K. Lee, G. T. Kim, G. S. Duesberg, T. Hallam, J. J. Boland, J. J.
Wang, J. F. Donegan, J. C. Grunlan, G. Moriarty, A. Shmeliov, R. J. Nich-
The EXAFS fittings of bimetallic and the trimetallic sulfides
confirm the presence of MoS2, WS2, in NiMoS and NiWS, re-
spectively showing smaller coherent domains than the refer-
ence materials. The EXAFS fitting of NiMoWS confirms the
presence of intralayer Mo1ÀxWxS2. The presence of Ni-Mo (Mo-
Ni), and Ni-W (W-Ni) paths in all materials suggests that Ni is ef-
fectively interacting with MoS2, WS2, or Mo1ÀxWxS2, allowing us
to conclude that mixed Ni-Mo(W) sulfide phases were present
in all catalysts. The values of the coordination numbers sug-
gest that the most pronounced Mo(W)-Ni interaction occurs in
the trimetallic NiMoWS. The nature of the promoting Ni spe-
cies, i.e., single atoms (like in a classic Ni-Mo-S model[53]) or
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3703
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
olls, J. M. Perkins, E. M. Grieveson, K. Theuwissen, D. W. McComb, P. D.
Nellist, V. Nicolosi, Science 2011, 331, 568–571.
[10] B. T. Yonemoto, G. S. Hutchings, F. Jiao, J. Am. Chem. Soc. 2014, 136,
8895–8898.
[11] D. O. Dumcenco, H. Kobayashi, Z. Liu, Y.-S. Huang, K. Suenaga, Nat.
Commun. 2013, 4, 1351.
[12] 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.
[35] C. Thomazeau, C. Geantet, M. Lacroix, V. HarlØ, S. Benazeth, C. Marhic,
M. Danot, J. Solid State Chem. 2001, 160, 147–155.
[36] M. Bauer, H. Bertagnolli in Methods in Physical Chemistry (Eds.: P. R.
Schäfer, P. P. C. Schmid), Wiley-VCH, 2012, pp. 231–269.
[37] B. K. Teo, P. A. Lee, J. Am. Chem. Soc. 1979, 101, 2815–2832.
[38] A. Hrabar, J. Hein, O. Y. GutiØrrez, J. A. Lercher, J. Catal. 2011, 281, 325–
338.
[39] O. Y. GutiØrrez, S. Singh, E. Schachtl, J. Kim, E. Kondratieva, J. Hein, J. A.
Lercher, ACS Catal. 2014, 4, 1487–1499.
[13] S. D. Kelly, N. Yang, G. E. Mickelson, N. Greenlay, E. Karapetrova, W. Sin-
kler, S. R. Bare, J. Catal. 2009, 263, 16–33.
[40] E. Schachtl, E. Kondratieva, O. Y. GutiØrrez Tinoco, J. A. Lercher, J. Phys.
Chem. Lett. 2015, 6, 2929–2932.
[14] J. Hein, A. Hrabar, A. Jentys, O. Y. GutiØrrez, J. A. Lercher, ChemCatChem
2014, 6, 485–499.
[41] CRC Handbook of Chemistry and Physics, 90th ed., Internet Version, CRC
Press, Boca Raton, FL, 2010.
[15] S. P. A. Louwers, R. Prins, J. Catal. 1992, 133, 94–111.
[16] S. Eijsbouts, S. W. Mayo, K. Fujita, Appl. Catal. A 2007, 322, 58–66.
[17] J. Bocarando, R. Huirache-AcuÇa, W. Bensch, Z. D. Huang, V. Petranov-
skii, S. Fuentes, G. Alonso-NfflÇez, Appl. Catal. A 2009, 363, 45–51.
[18] S. Srivastava, D. Palit, Solid State Ionics 2005, 176, 513–521.
[19] D. O. Dumcenco, Y.-C. Su, Y.-P. Wang, K.-Y. Chen, Y.-S. Huang, C.-H. Ho,
K.-K. Tiong, Solid State Phenom. 2011, 170, 55–59.
[42] H. R. Reinhoudt, R. Troost, A. D. van Langeveld, J. A. R. van Veen, S. T.
Sie, J. A. Moulijn, J. Catal. 2001, 203, 509–515.
[43] M. J. Vissenberg, Y. van der Meer, E. J. M. Hensen, V. H. J. de Beer, A. M.
van der Kraan, R. A. van Santen, J. A. R. van Veen, J. Catal. 2001, 198,
151–163.
[44] H. Toulhoat, P. Raybaud, S. Kasztelan, G. Kresse, J. Hafner, Catal. Today
1999, 50, 629–636.
[20] T. Maesen, A. E. Kuperman, US 2009/007883, United States, 2008.
[21] A. Belsky, M. Hellenbrandt, V. L. Karen, P. Luksch, Acta Crystallogr. Sect. B
2002, 58, 364–369.
[22] J. J. Rehr, J. J. Kas, F. D. Vila, M. P. Prange, K. Jorissen, Phys. Chem. Chem.
Phys. 2010, 12, 5503–5513.
[45] S. Kasztelan, H. Toulhoat, J. Grimblot, J. P. Bonnelle, Appl. Catal. 1984,
13, 127–159.
[46] 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,
224–235.
[23] K. V. Klementiev, VIPERfreeware, K. V. Klementev, J. Phys. D 2001, 34,
209.
[24] B. Ravel, M. Newville, J. Synchrotron Radiat. 2005, 12, 537–541.
[25] S. I. Zabinsky, J. J. Rehr, A. Ankudinov, R. C. Albers, M. J. Eller, Phys. Rev. B
1995, 52, 2995–3009.
[47] E. Schachtl, E. Kondratieva, L. Zhong, J. Hein, O. Y. GutiØrrez, A. Jentys,
J. A. Lercher, ChemCatChem 2015, DOI: 10.1002/cctc.201500706.
[48] H. Topsøe, B. S. Clausen, Catal. Rev. 1984, 26, 395–420.
[49] A. Olivas, D. H. Galvµn, G. Alonso, S. Fuentes, Appl. Catal. A 2009, 352,
10–16.
[26] D. Yang, R. F. Frindt, Mol. Cryst. Liq. Cryst. A 1994, 244, 355–360.
[27] J. Moser, F. LØvy, Thin Solid Films 1994, 240, 56–59.
[28] C. Lee, H. Yan, L. E. Brus, T. F. Heinz, J. Hone, S. Ryu, ACS Nano 2010, 4,
2695–2700.
[50] J. A. R. van Veen in Zeolites for Cleaner Technologies (Eds.: M. Guisnet,
J.-P. Gilson), Imperial College Press, 2002, pp. 131–152.
[51] W. R. A. M. Robinson, J. A. R. v. Veen, V. H. J. d. Beer, R. A. v. Santen, Fuel
Energy Abstr. 2000, 41, 79–80.
[29] J. W. Chung, A. Adib, Z. R. Dai, K. Adib, F. S. Ohuchi, Thin Solid Films
1998, 335, 106–111.
[52] T. Kabe, Y. Aoyama, D. Wang, A. Ishihara, W. Qian, M. Hosoya, Q. Zhang,
Appl. Catal. A 2001, 209, 237–247.
[30] L. Wang, Y. Zhang, Y. Zhang, Z. Jiang, C. Li, Chem. Eur. J. 2009, 15,
12571–12575.
[53] M. W. J. Craje, S. P. A. Louwers, V. H. J. De Beer, R. Prins, A. M. Van der K-
raan, J. Phys. Chem. 1992, 96, 5445–5452.
[31] L. P. Hansen, Q. M. Ramasse, C. Kisielowski, M. Brorson, E. Johnson, H.
Topsoe, S. Helveg, Angew. Chem. Int. Ed. 2011, 50, 10153–10156;
Angew. Chem. 2011, 123, 10335–10338.
[54] M. Ramos, G. Berhault, D. A. Ferrer, B. Torres, R. R. Chianelli, Catal. Sci.
Technol. 2012, 2, 164–178.
[32] E. J. M. Hensen, Y. van der Meer, J. A. R. van Veen, J. W. Niemantsverdriet,
Appl. Catal. A 2007, 322, 16–32.
[33] T. Shido, R. Prins, J. Phys. Chem. B 1998, 102, 8426–8435.
[34] C. Thomazeau, C. Geantet, M. Lacroix, M. Danot, V. Harle, Oil Gas Sci.
Technol. 2005, 60, 781–790.
Received: July 17, 2015
Published online on September 25, 2015
ChemCatChem 2015, 7, 3692 – 3704
www.chemcatchem.org
3704
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim