Simultaneous hydrodenitrogenation and hydrodesulfurization on unsupported Ni-Mo-W sulfides

Article abstract of DOI:10.1016/j.cattod.2017.05.083

The catalytic properties of unsupported Ni-Mo-W sulfides (composites of Ni-Mo(W)S₂ mixed sulfides and Ni₃S₂) obtained from precursors synthesized via co-precipitation, hydrothermal, and thiosalt decomposition were explored

Full text of DOI:10.1016/j.cattod.2017.05.083

CatalysisTodayxxx(xxxx)xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Simultaneous hydrodenitrogenation and hydrodesulfurization on
unsupported Ni-Mo-W sulfides
Sylvia Albersbergera, Jennifer Heina, Moritz W. Schreibera, Santiago Guerraa, Jinyi Hanb,
Oliver Y. Gutiérreza,⁎, Johannes A. Lerchera,⁎
a
Technical University Munich, Department of Chemistry, Catalysis Research Center, 85747 Garching, Germany
Chevron Energy Technology Company, Richmond, 94802 CA, USA
b
A R T I C L E I N F O
A B S T R A C T
Keywords:
The catalytic properties of unsupported Ni-Mo-W sulfides (composites of Ni-Mo(W)S2 mixed sulfides and Ni3S2)
obtained from precursors synthesized via co-precipitation, hydrothermal, and thiosalt decomposition were ex-
plored for hydrodenitrogenation (HDN) of o-propylaniline or quinoline in presence and absence of di-
benzothiophene undergoing hydrodesulfurization (HDS). The conversion rates varied with the reacting substrate
and were related to specific catalyst properties such as morphology, texture, surface and composition. For the
HDN of o-propylaniline and the HDS of dibenzothiophene in presence of o-propylaniline, high concentrations of
coordinatively unsaturated cationic sites (as characterized by NO adsorption) and the specific surface areas
determined the rates of reaction. For the HDN of quinoline and the HDS of dibenzothiophene in the presence of
quinoline, the high hydrogenation activity of tungsten sulfide and length of the slabs was found to be more
important than in the cases with o-propylaniline. Overall, the activity of unsupported catalysts relates to the size
of sulfide slabs provided that Ni is present at the perimeter of the slabs.
Hydrodenitrogenation
Hydrodesulfurization
Unsupported catalysts
Sulfides
Trimetallic catalysts
1. Introduction
that such properties are indicators of the parameters that directly de-
termine activity, i.e., concentration and intrinsic activity of active sites.
Hydrotreating of sulfur and nitrogen rich feeds requires highly ac-
tive MoS2 and WS2 catalysts that retain higher rates in hydrogenation
and hydrodefunctionalization in presence of significant concentrations
of heterocompounds [1]. Most catalyst formulations include Co or Ni as
promoters and Al2O3 as support [2–4]. However, the exceptional ac-
tivity of unsupported trimetallic Ni-Mo-W sulfides [5] has triggered
substantial efforts to improve catalytic activity beyond present levels
and to understand the role of the catalyst components [6–10].
Unsupported multimetallic sulfide catalysts consist of complex
mixtures of mono-, bi-, and trimetallic phases. To understand these
local structures, it is important to note that MoS2 and WS2 are aniso-
tropic materials and their activity strongly depends on their mor-
phology and crystallinity [5]. Thus, several parameters might influence
the intrinsic activity and availability of active sites. As the catalytic
pathway strongly depends on the nature of the reacting substrates,
catalyst formulation have to be adjusted to the characteristics of the
feed [11].
In order to verify the hypothesis, the diverse materials were tested in
the conversions of model molecules that require well-defined hydro-
genation and hydrodefunctionalization steps. In the final step, the ac-
tivity trends were compared with the accessed physicochemical prop-
erties and structure − activity correlations were discussed.
Common routes towards synthesis of unsupported sulfide catalysts
comprise sulfidation of oxide precursors, which are obtained by co-
precipitation [12]. This synthesis approach is usually practiced in in-
dustry because it produces materials with Mo/W/Ni ratios optimized
for hydrodesulfurization and hydrodenitrogenation [13,14]. Co-pre-
cipitation also allows varying the Mo:W ratio in the resulting solids,
leading to different oxide precursor structures. The structure and
composition of the precursor influence the sulfidation kinetics and the
cation distribution in the final mixture of sulfide phases [15]. As an
alternative to the “oxide route”, direct precipitation of sulfides from the
synthesis solution has been reported as advantageous for the activity of
the materials, which is often attributed to improved Mo(W)-Ni inter-
actions [6,10,16,17].
In the context of drawing activity-structure correlations for complex
unsupported multimetallic sulfides, this work aims to study series of
materials with diverse physicochemical properties. It was hypothesized
Thus, in this work, several Ni-Mo-W catalyst precursors were syn-
thesized following co-precipitation, and decomposition of thiosalts.
⁎
Corresponding authors.
E-mail addresses: oliver.gutierrez@mytum.de (O.Y. Gutiérrez), johannes.lercher@ch.tum.de (J.A. Lercher).
http://dx.doi.org/10.1016/j.cattod.2017.05.083
Received 23 November 2016; Received in revised form 14 February 2017; Accepted 29 May 2017
0920-5861/PublishedbyElsevierB.V.
Pleasecitethisarticleas:Albersberger,S.,CatalysisToday(2017),http://dx.doi.org/10.1016/j.cattod.2017.05.083

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
Some procedures have been adapted to hydrothermal conditions in
order to obtain precursor phases different to those resulting of low
temperature precipitation from identical precursor solutions [15,18].
Precursors and the corresponding sulfides were characterized by means
of N2 physisorption, X-ray diffraction, Raman spectroscopy, pulse NO
titration, and scanning and transmission electron microscopy. The sul-
fide catalysts were studied with respect to activity and selectivity for
hydrodenitrogenation of o-propylaniline, and quinoline, in presence
and absence of dibenzothiophene undergoing hydrodesulfurization.
Structure-activity correlations were rationalized in terms of properties
and availability of active sites.
(2.7 M, Sigma-Aldrich, ≥99.0%) in 15 mL bidistilled water, was
dropped to solution A and stirred for 20 min. Solution C, containing
nickel(II) nitrate hexahydrate (20 mmol Ni, Sigma-Aldrich, ≥98.5%) in
30 mL bidistilled water, was added to the reaction mixture. The red
precipitate was filtrated and dried at 120 °C for 12 h in synthetic air.
2.1.4. Hydrothermal decomposition (WMoNi-sHTOx
)
The WMoNi-sHTOx precursor was prepared by a novel hydrothermal
method using a Parr autoclave (Series 4843). The salt precursors, i.e.,
ammonium tetrathiomolybdate (7.2 mmol, Aldrich, 99.97%), ammo-
nium tetrathiotungstate (7.2 mmol W, Aldrich, ≥99.9%), and nickel(II)
hydroxide (14.4 mmol Ni, Aldrich) were suspended in 75 mL bidistilled
water in the autoclave. The mixture was heated to 250 °C reaching a H2
pressure of 4 MPa. This solution was isothermally kept at 250 °C and
4 MPa for 16 h. A dark blue precipitate was recovered by hot filtration,
dried under vacuum overnight, and at 120 °C for 12 h in synthetic air.
2. Experimental
2.1. Synthesis of oxide precursors
2.1.1. Salt precursor route (WMoNi-aOx and WMoNi-aHTOx
)
The synthesis was performed according to Maesen et al. [12]. So-
lution A was prepared by consecutively dissolving ammonium hepta-
molybdate (25 mmol Mo, Sigma-Aldrich, ≥99%) and ammonium me-
tatungstate (25 mmol W, Sigma-Aldrich, ≥99%) in 200 mL bidistilled
water. Under continuous stirring, 7.5 mL of aqueous ammonia
(28–30 wt%, Sigma-Aldrich) were added before heating up the mixture
to 90 °C. For solution B, nickel(II) nitrate hexahydrate (50 mmol Ni,
Sigma-Aldrich, ≥98.5%) was dissolved in 12.5 mL bidistilled water and
heated to 70 °C. The warm solution B was added dropwise to solution A,
which forced the precipitation of the catalyst precursor. After 30 min at
90 °C, and hot filtration, the obtained solid was suspended in an aqu-
eous solution of maleic acid (0.05 M, Fluka, ≥98.0%) and kept at 70 °C
(solution C). The precipitate was filtrated and dried in vacuum over-
night, and in synthetic air at 120 °C for 12 h.
2.2. Synthesis of sulfide catalysts
The precursors, denoted as WMoNi-aOx, WMoNi-aHTOx, WMoNi-
bOx, WMoNi-bHTOx, WMoNi-sOx, and WMoNi-sHTOx (according to the
experimental descriptions above) were sulfided to obtain the active
catalysts. The sulfidation was carried out in a trickle bed reactor using a
mixture of H2S in H2 (10 vol.% H2S, 40 mL/min) at 2 MPa and 400 °C
for 12 h. The resulting sulfide materials are denoted according to the
name of the corresponding precursor, i.e., WMoNi-a, WMoNi-aHT,
WMoNi-b, WMoNi-bHT, WMoNi-s, and WMoNi-sHT.
2.3. Characterization
The elemental analysis of the oxidic precursors and sulfide catalysts
were conducted by the micro analytic laboratory of the Technische
Universität München. The concentrations of Ni, Mo, and W were pho-
tometrically determined whereas the contents of H, C, N, and S were
analyzed using an automated element analyzer instrument (vario EL
CHN analyser, ELEMENTAR). The analysis of several batches of se-
lected samples showed identical elemental contents. The concentration
of oxygen was determined as the difference between the total mass and
the masses of the elements quantified.
The synthesis of the WMoNi-aHTOx precursor was performed under
hydrothermal conditions using a 300 mL Parr autoclave (Series 4843).
Solution A was heated in the autoclave to 250 °C for 30 min reaching a
H
2 pressure of 4 MPa. Solution B was heated to 90 °C before injecting it
via an injection pipette to the autoclave. After 90 min at 250 °C, the
reaction mixture was cooled to 80 °C and filtrated. The resulting yellow
precipitate was slurred in solution C at 70 °C for 30 min. The obtained
solid was dried as described for WMoNi-aOx
.
Isotherms of adsorption and desorption of N2 at 77 K were per-
formed with an automated nitrogen adsorption analyzer Sorptomatic
1990 Series (Thermo Finnigan). Prior to the measurements, the samples
were evacuated at 120 °C for 4 h. BET analysis was used to determine
the surface area of the oxide and sulfide materials. The error de-
termined for the N2 physisorption was up to 20% of the reported
value.
The crystal structures of the materials were analyzed by powder X-
ray diffraction. The measurements were carried out in a X’Pert Pro PW
3040/60 instrument by PANalytical equipped with a copper X-ray tube
(Cu-Kα radiation, 0.1542 nm), a nickel Kβ-filter, and solid-state de-
tector (X'Celerator) operated at 45 kV/40 mA with step size of 0.017°
and scan time of 115 s per step.
2.1.2. Oxide precursor route (WMoNi-bOx and WMoNi-bHTOx
)
The synthesis of WMoNi-bOx, derived from that reported by Soled
et al. [13], was performed as follows. Suspension A was prepared by
slurring molybdenum(VI) oxide (10 mmol Mo, Aldrich, 99.98%) and
tungstic acid (10 mmol W, Sigma-Aldrich, ≥99.0%) in 160 mL of bi-
distilled water. Subsequently, this mixture was heated to 90 °C. Sus-
pension B, consisting of nickel(II) carbonate hydroxide tetrahydrate
(20 mmol Ni, Sigma-Aldrich) and 40 mL bidistilled water, was also
heated to 90 °C before adding dropwise to solution A. The resulting
green precipitate was separated and dried under vacuum overnight and
later in synthetic air at 120 °C for 12 h.
Hydrothermal conditions were applied for the synthesis of WMoNi-
bHTOx. Suspensions A and B were mixed in an autoclave (Series 4843)
and heated to 250 °C under a H2 pressure of 4 MPa for 6 h. The drying
procedure of WMoNi-bOx was also applied for this material.
SEM images of the materials were recorded using a REM JEOL 5900
LV microscope.
A secondary electron detector and an Everhart-
Thornley detector for backscattered electrons were employed. The SEM
images of the secondary electron detector were taken with an accel-
eration voltage of 25 kV. Before the measurements, the samples were
outgassed for two days, transferred on a graphite foil and coated with
gold by sputtering.
Transmission electron microscopy was performed in an instrument
(JEOL JEM-2011) with an accelerating voltage of 120 keV. SAED
measurements were taken with the same instrument. The samples were
prepared grinding a small amount of material and dispersing it ultra-
sonically in ethanol. Subsequently, drops of this dispersion were ap-
plied on a copper carbon grid and the ethanol was evaporated at room
temperature. Statistical analysis of the length and stacking height was
2.1.3. H2S precipitation route (WMoNi-sOx
)
This synthesis was adapted from Yi et al. and Nava et al. [16,19].
Ammonium heptamolybdate (10 mmol Mo, Sigma-Aldrich, ≥99%),
and ammonium metatungstate (10 mmol W, Sigma-Aldrich, ≥99%)
were dissolved in 30 mL of aqueous ammonia (28–30 wt%, Sigma-Al-
drich). A gaseous mixture of H2S in H2 (10 vol.% H2S, 40 mL/min) was
bubbled through solution A for 6 h under vigorous stirring. At 70 °C a
deep red solution was obtained (solution A). After cooling to room
temperature,
a solution B, consisting of tetramethyl ammonium
chloride (2.7 M, Sigma-Aldrich, ≥99.0%) and sodium hydroxide
2

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
H2S precipitation
route
Hydrothermal
decomposition
Salt precursor route
Oxide precursor route
(NH4)6Mo7O24·4H2O
(NH4)6H2W12O40
Ni(NO3)2·6 H2O
(NH4)6Mo7O24·4H2O
(NH4)6(W12O40)H2
Ni(NO3)2·6H2O
(NH4)2MoS4
(NH4)2WS4
and Ni(OH)2
MoO3
WO3·H2O
2NiCO3·3Ni(OH)2·4H2O
Low
temperature
Low
temperature
Hydrothermal
conditions
Hydrothermal
conditions
Hydrothermal
conditions
Low temperature
WMoNi-aOx
WMoNi-aHTOx
WMoNi-bOx
WMoNi-bHTOx
WMoNi-sOx
WMoNi-sHTOx
Fig. 1. Schematic representation of the synthesis routes applied in this work to prepare precursors of Ni-Mo-W sulfide catalysts.
achieved by counting at least 300 slabs distributed in different regions
of the sample.
measureds the temperature in the isothermal zone of the reactor. The
pressure was fixed by a homemade back-pressure controller connected
to a high pressure N2 line. The schematic outline of the reactor setup
can be found in Ref. [20]. Prior to the activity test, pellets of the pre-
cursors (typically 50 mg, 250–355 μm) were mixed with γ-Al2O3 (1 part
of precursor and 1 part of Al2O3 in weight) and with SiC (355–500 μm,
1 part catalyst 20 parts SiC in weight). The precursor packed in the
reactor was sulfided with a mixture of H2S in H2 (10 vol.% H2S, 40 mL/
min) at 2 MPa and 400 °C for 12 h. Tetradecane (Alfa Aesar, ≥99%)
was used as solvent and hexadecane (Sigma-Aldrich) as internal stan-
dard. The mixtures of the liquid feed applied for the activity tests are
listed in Table S11 (supporting information). The ratio of liquid to H2
was 1:330 Ndm3/dm3. Typical liquid flows of around 0.1 mL/min were
applied, which is equivalent to a LHSV of 109 h−1 based on the volume
of catalyst. Alternatively, the space time was typically 153 mol/
(gcat∙min) for o-propylaniline, 154 mol/(gcat∙min) for quinoline, and
705 mol/(gcat∙min) for dibenzothiophene. Here, space time is defined as
Raman measurements of the catalyst precursors and their corre-
sponding sulfides were carried out in a Renishaw Raman Spectrometer
(Type 1000). The instrument was equipped with a CCD detector and a
microscope (Leica microscope DM LM). As excitation source, an argon
laser (514 nm) was used.
The NO adsorption on sulfided samples was carried out at ambient
conditions using a pulse technique. Prior to the experiment, the pre-
cursors were treated under a mixture of 2 mL/min of H2S in H2 (10 vol.
% H2S) and 8 mL/min of helium at 400 °C (5 °C/min) for 2 h. Once the
reactor reached room temperature and was flushed with helium
(10 mL/min), pulses of NO were introduced periodically until the NO
adsorption-desorption equilibrium was established. The adsorption of
NO was monitored by a mass spectrometer. The concentration of ad-
sorbed NO was calculated as the sum of the individual NO uptakes per
pulse. The error in the determination of adsorbed NO was around 10%.
X-ray photoelectron spectroscopy (XPS) measurements were per-
formed using a Phi Quantera Scanning X-ray Microprobe instrument.
This instrument was equipped with a hemispherical energy analyzer
with multichannel detection at an energy resolution of 1.1 eV. As en-
ergy source, monochromatic Al Kα X-rays (h∙ν=1486.7 eV) were used.
The powdered catalyst was mounted in an area of approximately
0.8 cm × 0.8 cm of double-sticky tape. The surface of the tape was
completely covered by the catalyst powder. Five analysis areas
(1.2 mm × 100 micron) were selected for detailed spectral character-
ization for each catalyst. At each area, detailed spectra were collected
for W 4f, S 2p, Mo 3d, C 1s, O 1s, and Ni 2p3 photoelectron peaks. Total
spectral accumulation times were 100 min per analysis area while ir-
radiating with 100 W of X-radiation. The binding energies were refer-
enced to the C 1s peak (284.8 eV) to account for charging effects.
Spectral envelopes were deconvoluted using an iterative least square
algorithm provided in Phi Multipak software.
m
cat/F, where mcat is the mass of the catalyst and F is the molar flow of
the particular reactant. The temperatures applied were 310 °C, 330 °C,
350 °C, and 370 °C at constant pressure of 5 MPa H2. These conditions
allowed to control the reaction rates in a wide range below complete
reactant conversion in order to obtain reliable kinetic data. Prior to
sampling, the reaction conditions applied were kept for at least 12 h to
ensure steady state. The product stream was analyzed off-line using a
gas chromatograph (HP 6890) equipped with a flame ionization de-
tector (FID) and an Agilent DB-17 capillary column. Liquid samples
were collected by a 16-port sampling valve after liquid-gas separation.
The presented data were acquired after 24 h on stream to obtain results
representative of stable catalysts. The reproducibility of the measure-
ments was 5%.
3. Results and discussion
Six precursors and the corresponding sulfide catalysts were syn-
thesized as described in the experimental section and schematically
shown in Fig. 1. The figure also shows the names of the precursors. The
characterization of the precursors is shown in the supporting informa-
tion. In the following, the physicochemical and kinetic characterization
of the materials after sulfidation are described.
2.4. Catalysis
The hydrotreating activity of the catalysts was evaluated with re-
spect to hydrodenitrogenation (HDN) using o-propylaniline (OPA,
Sigma-Aldrich, 97%), and quinoline (Q, Sigma-Aldrich, 98%) as model
compounds. In addition, dimethyl disulfide (DMDS, Sigma-Aldrich,
≥99%) was added to the liquid feed. In a separate series of experi-
ments, hydrodesulfurization (HDS) of dibenzothiophene (DBT, Sigma-
Aldrich, ≥99%) was performed simultaneously with the HDN of OPA
and quinoline. The experiments were performed in a flow trickle fixed
bed reactor, where liquid and gas feeds were introduced using a HPLC
pump (Shimadzu LC-20AD) and high pressure mass flow controllers
(Bronkhorst). Gas and liquid streams were introduced in concurrent
downflow mode in the glass-coated tubular reactor (ø ¼ inches) loaded
with the catalytic material as described below. The reactor was sur-
rounded by a brass heating jacket equipped with a thermocouple that
3.1. Chemical and physicochemical properties of sulfide catalysts
The sulfide catalysts are denoted according to the name of the
corresponding precursor, i.e., WMoNi-a, WMoNi-aHT, WMoNi-b,
WMoNi-bHT, WMoNi-s, and WMoNi-sHT. The metal contents in all
materials led to Ni metals fractions from 0.47 to 0.59 and varying Mo to
W molar ratios (Table 1). The complete elemental compositions of the
catalysts differed from those of the corresponding precursors (Tables S1
and S8) due to elimination of carbon and nitrogen, and the substitution
of oxygen by sulfur during sulfidation. The concentration of Ni
3

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
the same for spent and freshly sulfided catalysts.
Table 1
Metal molar fractions (xW, xMo, xNi) derived from the bulk elemental analysis and the XPS
characterization. BET surface area (SBET), concentration of NO adsorbed on the sulfide
catalysts (NO), and slab length (L) determined from TEM.
In TEM micrographs of all materials (Fig. 3), typical lattice fringes
with interplanar distance of 0.62 nm, corresponding to the layered
structure of MoS2 and WS2, were observed. The trends in average
stacking degrees of the Mo(W)S2 slabs, obtained by inspection of TEM
images, were in qualitative agreement with those determined from the
Scherrer equation (Table S9). The average slabs lengths varied in a
relatively wide range (6–48 nm) and exhibited broad distributions (Fig.
S5). A second phase was identified by means of selected area diffraction
(SAED) as Ni9S8. For instance, Fig. 3 shows for WMoNi-b that the SAED
contained reflections of Ni9S8 (bright dots) besides those of Mo(W)S2
(rings). It was hypothesized that these nickel sulfide crystals indicated
by SAED are not those that produced X-ray diffraction peaks, which
correspond to crystals sizes in the range of microns. The large crystals of
Ni sulfide are directly observed by SEM (Fig. S6). Thus, Ni sulfide
particles are present in sizes ranging from nanometers (TEM) to mi-
crometers (SEM) and are surrounded by Mo(W)S2 slabs.
Bulk composition
XPS analysis
SBET
NO
L nm
m2/g μmol/
g
xW
xMo
xNi
xW
xMo
xNi
WMoNi-a
0.31
0.16
0.33
0.25
0.27
0.29
0.30
0.53
0.53
0.48
0.47
0.49
0.59
0.04 0.15 0.81 5.3
0.06 0.42 0.51 9.8
0.19 0.23 0.57 7.9
0.28 0.31 0.40 15.2
0.08 0.19 0.73 8.0
0.08 0.15 0.77 19.8
72
57
91
38
109
111
6.0
6.0
10.1
48
9.3
WMoNi-aHT 0.14
WMoNi-b 0.27
WMoNi-bHT 0.27
WMoNi-s 0.22
WMoNi-sHT 0.11
11.6
determined by XPS was higher than the bulk values for many catalysts,
whereas the bulk content of W was higher than the W content de-
termined by XPS. This indicates that the distribution of the metals is not
heterogeneous along the volume of the materials. We proposed, in line
with the conclusions reported in Ref. [15] that this heterogeneous
elemental composition reflects the presence of several phases in the
catalysts. Thus, the high Ni segregation towards the surface (XPS Ni
content higher than the bulk Ni content) was related with the NiSx
species [15].
The BET surface areas of the sulfides did not resemble the trends of
the precursors, because each material followed a particular sulfidation
and reconstruction mechanism, as the precursors were prepared by
different methods and contained different phases. The concentration of
adsorbed NO did not correlate with the specific surface area (see
Table 1), which is attributed to the anisotropy of MoS2 and WS2 (where
adsorption sites for NO are available only at the edges), and to the
varying stoichiometry of NO adsorbed on Mo(W)S2 and Ni sulfides
[21].
The Mo(W)S2 phases in similar sulfide Ni-Mo-W catalysts consisted
of intralayer mixed MoxWx-1S2 slabs as deduced from XAS and Raman
spectroscopy [9,18,22]. These studies also showed that atomically
dispersed Ni is present at the edges of the Mo(W)S2 crystals [18]. In
agreement with these reports, the Raman spectra of the catalysts in this
work (Fig. 4) showed bands in the range of 372–378 cm−1, attributed
to the E1 mode of Mo-S, and bands around 346 cm−1 ascribed to the
2g
E1 mode of W-S (specific position of the bands is described in Table
2g
S10 of the supporting information). The A1g modes of Mo-S and W-S in
the reference materials were observed at 410 cm-1 and 424 cm−1, re-
spectively. The bands shifting to around 408 cm−1 for WMoNi-a, and
WMoNi-bHT and around 402 cm−1 for the other four materials were
assigned to the A1g mode of Mo-W composite species (Mo1-xWxS2)
[23,24]. Thus, the catalytic activity is likely dominated by Ni-promoted
MoxWx-1S2, as Ni sulfides do not have a relevant hydrotreating activity
[25].
The X-ray diffractograms of all materials (Fig. 2) exhibited diffrac-
tion peaks characteristic of MoS2, WS2, Ni9S8, and Ni3S2 (ICSD#:
644245, 202366, 164879, and 27521, respectively). WMoNi-s and
WMoNi-sHT showed additional reflections assigned to α-NiS (ICSD#:
29313). However, after catalysis, the spent catalysts (Fig. S4) exhibited
reflections of MoS2, WS2, and Ni9S8 (ICSD#: 644245, 202366, 27521,
respectively) as the only Ni sulfide phase. Sharper reflections and flatter
baselines indicated higher crystallinity of the spent catalysts than the
freshly sulfided ones. The Scherrer equation, applied on the reflection
at 14 °2θ corresponding to the (002) lattice plane of Mo(W)S2, showed
that the crystal sizes of WMoNi-a, WMoNi-b, and WMoNi-s increased
during the reaction, whereas those of the hydrothermal treated mate-
rials hardly changed (Table S9). However, the trend of crystal sizes was
3.2. Catalysis
3.2.1. Hydrodenitrogenation of o-propylaniline
The apparent rate constants of the sulfide catalysts in the HDN of
OPA as a function of the inverse temperature are shown in Fig. 5
(further graphical representations of the catalytic results are presented
in Figs. S7, and S8 of the supporting information). The rate constants
increased in the sequence WMoNi-s < WMoNi-a < WMoNi-bHT <
WMoNi-aHT < WMoNi-b < WMoNi-sHT. The conversion rates of
OPA (Fig. S9) increased in presence of DBT without changing the order
of activity. The rates of OPA HDN have been concluded to increase in
presence of DBT only on Ni-promoted MoS2/Al2O3 but not on MoS2/
Fig. 2. X-ray diffraction patterns of the sulfided catalysts WMoNi-a (a), WMoNi-b (b), WMoNi-s (c), WMoNi-aHT (d), WMoNi-bHT (e), WMoNi-sHT (f). The marked reflections correspond
to MoS2 or WS2 (♦), Ni9S8 (◊), Ni3S2 (○), and α-NiS (▲).
4

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
Fig. 3. Representative TEM micrographs of
the sulfided catalysts WMoNi-a, WMoNi-b,
WMoNi-s, WMoNi-aHT, WMoNi-bHT, and
WMoNi-sHT.
Al2O3 [26,27]. Hence, the positive influence of DBT on the catalytic
activity suggests that NiMo(W)S2 dominates the catalytic performance.
The products observed during the HDN of OPA were propylcyclo-
hexylamine (PCHA), three isomers of propylcyclohexene (PCHE) (i.e.,
1-propylcyclohexene, 3-propylcyclohexene, and propylidene cyclo-
hexane), and propylcyclohexane (PCH). Thus, hydrogenation of OPA to
PCHA is followed by N removal towards PCHE and consecutive sa-
turation to PCH. Propylbenzene (PB) was the product of the direct
denitrogenation of OPA [27,28]. The corresponding reaction network is
presented in Scheme 1.
Fig. 6 shows the product yields as a function of temperature at si-
milar reactant conversions. PCHA had very low yields, which indicates
that the hydrogenation of OPA is slower than the denitrogenation of
PCHA. Declining PCHA yields with increasing temperature point to
increasing differences in the rates of OPA hydrogenation and PCHA
denitrogenation as the temperature increased. The yields of PB were
higher than those of PCHA, but much lower than those of PCHE and
PCH. This indicates that direct denitrogenation of OPA (C(sp2)-N
5

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
quinoline and OPA conversions differed. For the conversion of quino-
line, the activity increased as follows: WMoNi-s < WMoNi-bHT <
WMoNi-a < WMoNi-sHT < WMoNi-aHT < WMoNi-b.
E1
A1g
2g
The presence of DBT increased the conversion rates of quinoline, as
shown in Fig. S14, and modified the activity ranking making WMoNi-a
more active than WMoNi-sHT. This increase of conversion rates of
quinoline and 14THQ in the presence of DBT is attributed to enhanced
concentration of active sites for the hydrogenation reactions and the
ring opening of 14THQ, and DHQ. The rationale is that direct de-
sulfurization of DBT creates basic S2− groups, which in turn lead to a
higher concentration of activated H (presumably in the form of SH
groups) at the surface [26].
a
b
c
d
e
f
The products observed during the HDN of quinoline were 14THQ,
5,6,7,8-tetrahydroquinoline (58THQ), both isomers of decahy-
droquinoline (DHQ) (cis- and trans-decahydroquinoline), as well as OPA
and the products observed during the HDN of OPA (PCHA, PCHE, PCH,
and PB). The overall reaction network of the HDN of quinoline, which
comprises OPA HDN, is presented in Scheme 1 [26]. Quinoline was
fully hydrogenated to DHQ via 14THQ or 58THQ. Under our reaction
conditions only the (de)hydrogenation equilibrium between quinoline
and 14THQ was reached. Ring opening reactions are key steps in the
HDN network producing monocyclic compounds that react more
readily than bicyclic compounds, i.e., ring opening from 14THQ to
OPA, and ring opening from DHQ to PCHA. DHQ reacts faster than
14THQ while a dehydrogenation step from PCHA to OPA is possible,
especially at high temperatures.
Fig. 8 shows the product yields for the HDN of quinoline over
temperature. The yields of 58THQ are not shown, because they were
negligible. The yields of DHQ increased with temperature up to max-
imum values at approximately 350 °C on most catalysts. The increase of
PB yields paralleled the increase of OPA, although PB yields remained
in very low concentrations. PCHA yields increased with temperature
passing through a maximum at 350 °C. The yields of the denitrogenated
products, PCHE, and PCH increased above 350 °C. Hence, 350 °C seems
to be a turning point, at which the rates of ring opening, and subsequent
denitrogenation become comparable to those of the (de)hydrogenation
steps of bicyclic compounds. Note that, the PCH yields were lower than
those of PCHE. This indicates, in contrast to the observations with OPA
HDN, that in the presence of quinoline (and bicyclic hydrogenated in-
termediates) the hydrogenation of PCHE is suppressed to some extent
probably due to competitive adsorption of bicyclic compounds on hy-
drogenation sites. The product yields in the presence of DBT (Fig. S15)
changed only slightly compared to that observed in the absence of DBT.
WS2
MoS2
Ni3S2
250 275 300 325 350 375 400 425 450 475 500
Raman shift, cm-1
Fig. 4. Raman spectra of the spent catalysts WMoNi-a (a), WMoNi-b (b), WMoNi-s (c),
WMoNi-aHT (d), WMoNi-bHT (e), and WMoNi-sHT (f) as well as Raman spectra of the
references WS2, MoS2, and Ni3S2.
cleavage) is less favored than its hydrogenation and further C(sp3)eN
bond hydrogenolysis. The marked increase of PB yield above 350 °C is
speculated to be caused by dehydrogenation of PCHE. The latter pro-
duct increased with temperature, but passed through a maximum on
the most active catalysts as a function of temperature. The yield of the
final product PCH increased steadily with temperature and followed the
same trend than the OPA conversion. In all cases, the PCH yield was
higher than the PCHE yield, which indicates faster hydrogenation of
PCHE than hydrogenation of OPA followed by denitrogenation of
PCHA.
The differences among the yield profiles observed for the HDN of
OPA were caused by varying conversion degrees on the different cat-
alysts. Thus, the chemical composition of the catalysts marginally in-
fluences the selectivity of OPA HDN. The product yields in the presence
of DBT (Fig. S10 of the supporting information) were very similar to
those observed in the absence of DBT.
3.2.2. Hydrodenitrogenation of quinoline
The rate constants of the sulfide catalysts in the HDN of quinoline at
different temperatures are compiled in Fig. 7 (see also Figs. S11, and
S12). The conversion of quinoline to 1,2,3,4-tetrahydroquinoline
(14THQ) was equilibrated (Fig. S13), while all other steps were kine-
tically controlled. Hence, the sum of the concentrations of quinoline
and 14THQ was used to describe the rates. The conversions rates were
lower than those observed for OPA and the activity rankings for
3.2.3. Hydrodesulfurization of dibenzothiophene
The HDS of dibenzothiophene proceeded at lower rates than the
HDN of OPA and quinoline. Fig. 9 shows Arrhenius plots for the hy-
drodesulfurization of DBT during the HDN of OPA at varying tem-
perature (see also Figs. S16, and S17). The apparent rate constants
4
4
2
2
0
0
-2
-2
1.55
1.65
1/T ·10-3, K-1
1.75
1.55
1.65
1/T ·10-3, K-1
1.75
Fig. 5. Reaction rate constants for the HDN of o-propylaniline at varying temperatures on the sulfided catalysts WMoNi-a (circle, unfilled), WMoNi-b (square, unfilled), WMoNi-s
(triangle, unfilled), WMoNi-aHT (circle, filled), WMoNi-bHT (square, filled), and WMoNi-sHT (triangle, filled) assuming first order reaction.
6

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
DDN
NH2
N
H
N
PB
OPA
14THQ
DHQ
Q
PCH
PCHE
PCHA
58THQ
HYDN
NH2
N
H
N
Scheme 1. Reaction network for the hydrodenitrogenation of quinoline and the intermediate o-propylaniline (within the doted square); the compounds depicted in the figure are
quinoline (Q), 1,2,3,4-tetrahydroquinoline (14THQ), 5,6,7,8-tetrahydroquinoline (58THQ), decahydroquinoline (DHQ), o-propylaniline (OPA), propylcyclohexylamine (PCHA), pro-
pylcyclohexene (PCHE), propylcyclohexane (PCH), and propylbenzene (PB). DDN and HYDN stand for the routes direct denitrogenation and hydrogenation, respectively.
Fig. 6. Product yields of the HDN of o-pro-
pylaniline over temperature on the sulfided
1.5
catalysts WMoNi-a (circle, unfilled), WMoNi-b
(square, unfilled), WMoNi-s (triangle, un-
20
filled), WMoNi-aHT (circle, filled), WMoNi-
bHT (square, filled), and WMoNi-sHT (tri-
angle, filled). The compounds depicted in the
figure are propylcyclohexylamine (PCHA),
1.0
0.5
0.0
10
0
propylbenzene
(PB),
propylcyclohexene
(PCHE), and propylcyclohexane (PCH).
300
320
340
360
380
300
320
340
360
380
Temperature, ºC
Temperature, ºC
30
20
10
0
60
40
20
0
300
320
340
360
380
300
320
340
360
380
Temperature, ºC
Temperature, ºC
increased in the sequence WMoNi-bHT ≅ WMoNi-a < WMoNi-
s ≅ WMoNi-aHT < WMoNi-b < WMoNi-sHT.
In the presence of quinoline, the HDS rates (Fig. 10) were lower
than in the presence of OPA. However, the differences varied randomly.
As consequence, the HDS activity ranking in the presence of quinoline
(WMoNi-s < WMoNi-a ≅ WMoNi-sHT < WMoNi-aHT < WMoNi-
bHT < WMoNi-b) differed from the rates observed for HDS in the
presence of OPA. Hence, OPA, quinoline, and the corresponding HDN
products differently affect the HDS of DBT. In the HDN of quinoline the
strong adsorption of this reactant and the hydrogenated products,
14THQ, 58THQ, and DHQ, hinders HDS to a larger extent than in the
HDN of OPA [29].
In the HDS of DBT (Scheme 2), biphenyl (BiPh), tetra-
hydrodibenzothiophene (HDBT) and phenylcyclohexane (PhCH) were
Fig. 7. Reaction rates of the merged con-
version of quinoline and 1,2,3,4-tetra-
hydroquinoline at varying temperatures on
the sulfided catalysts WMoNi-a (circle, un-
filled), WMoNi-b (square, unfilled), WMoNi-
2
0
2
0
s
(triangle, unfilled), WMoNi-aHT (circle,
filled), WMoNi-bHT (square, filled), and
WMoNi-sHT (triangle, filled).
-2
-2
-4
-4
1.55
1.65
1/T ·10-3, K-1
1.75
1.55
1.65
1.75
1/T ·10-3, K-1
7

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
Fig. 8. Evolution of the yields of selected
products of the HDN of quinoline over
temperature on the sulfide catalysts WMoNi-
a (circle, unfilled), WMoNi-b (square, un-
filled), WMoNi-s (triangle, unfilled),
WMoNi-aHT (circle, filled), WMoNi-bHT
(square, filled) and, WMoNi-sHT (triangle,
filled) assuming first order reaction. The
compounds depicted in the figure are dec-
ahydroquinoline (DHQ), o-propylaniline
(OPA), propylbenzene (PB), propylcyclo-
hexylamine (PCHA), propylcyclohexene
(PCHE), and propylcyclohexane (PCH).
15
12
9
20
15
10
5
6
3
0
0
300
320
340
360
380
300
320
340
360
380
Temperature, ºC
Temperature, ºC
2.0
1.5
1.0
0.5
3
2
1
0
0.0
300
320
340
360
380
300
320
340
360
380
Temperature, ºC
Temperature, ºC
15
12
9
6
4
2
0
6
3
0
300
320
340
360
380
300
320
340
360
380
Temperature, ºC
Temperature, ºC
formed. BiPh was the only product of the direct desulfurization route,
whereas the hydrogenation route of DBT produced PhCH via HDBT
[28]. The yields of the HDS products, observed during the HDS of OPA,
are presented in Fig. 11. The yields of HDBT are not presented because
they were below 1%. Biphenyl was the main product of the reaction and
increased along temperature reaching a maximum between 350 °C and
370 °C. PhCH was formed in negligible concentrations below 340 °C,
but its concentration increased exponentially at higher temperatures.
Thus, BiPh was hydrogenated to PhCH at higher temperature. The HDS
of DBT was concluded, therefore, to occur predominantly via direct
desulfurization followed by hydrogenation. The product distribution of
the HDS of DBT was identical in the presence of quinoline to that ob-
served in the presence of OPA (Fig. S18).
3.3. Apparent activation energies
The activation energies (Ea) for the conversion of OPA (138 kJ/mol
in average) were higher than the Ea of quinoline (94 kJ/mol in average)
regardless of the presence of DBT (Table 2). As the HDN of OPA pro-
ceeded faster than the HDN of quinoline, the higher activation energy
of the former must be overcompensated by the pre-exponential factors,
which merge concentration of active sites and entropic factors. Prob-
ably the concentration of active sites accessible for OPA is higher than
for quinoline. In the presence of DBT, all Ea for OPA HDN increased to
154 kJ/mol in average, but the reaction rates for HDN also increased.
Increasing rates paralleling increasing activation energies suggest that
the presence of DBT enhances the concentration of sites for OPA con-
version at the sulfide edges. The average Ea of DBT HDS (79 kJ/mol)
was lower in the presence of quinoline than in the presence of OPA
Fig. 9. Reaction rate constants for the HDS
of dibenzothiophene at varying tempera-
tures during HDN of o-propylaniline on the
sulfide catalysts WMoNi-a (circle, unfilled),
WMoNi-b (square, unfilled), WMoNi-s (tri-
angle, unfilled), WMoNi-aHT (circle, filled),
WMoNi-bHT (square, filled), and WMoNi-
sHT (triangle, filled).
2.0
1.0
3.0
2.0
1.0
0.0
0.0
-1.0
-1.0
1.55
1.65
1.75
1.55
1.65
1.75
1/T ·10-3, K-1
1/T ·10-3, K-1
8

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
Fig. 10. Reaction rates for the HDS of di-
benzothiophene at varying temperatures on
the sulfided catalysts in the presence of
quinoline: WMoNi-a (sphere, unfilled),
WMoNi-b (square, unfilled), WMoNi-s (tri-
angle, unfilled), WMoNi-aHT (sphere,
filled), WMoNi-bHT (square, filled), and
WMoNi-sHT (triangle, filled) assuming first
order reaction.
2
1
0
0
-2
-1
-4
-2
1.55
1.65
1/T ·10-3, K-1
1.75
1.55
1.65
1/T ·10-3, K-1
1.75
edges of Mo(W)S2 [4,11,33,34]. This substitution leads to optimum
metal-sulfur bond strength resulting in higher conversion rates of S-
containing molecules due to easier CeS bond cleavage [35,36]. More-
over, Ni promotion increases the concentration SH groups at the surface
active for hydrogenation [34]. Based on previous information of the
relation between the presence of Ni and the degree of unsaturation and
concentration of SH groups, the concentration of Ni at the edges is
considered to be the most important parameter governing the activity
of sulfide catalysts [25,34,37–40]. However, the agglomeration of Mo
(W)S2 crystals (associated to the absence of a support) and the presence
of segregated Ni sulfides lead to cases, where the promoter effect of Ni
is overcompensated by low concentrations of available Mo(W)S2 edges
[15]. As there is not a simple and direct correlation between one
physicochemical property and the performances of the catalysts, it is
concluded that in the series of catalysts in this study, there are complex
dependencies of activity on features of the catalyst structure and mo-
lecular structure of the reactant.
Scheme 2. Reaction network for the hydrodesulfurization of dibenzothiophene; the
compounds depicted in the figure are dibenzothiophene (DBT), biphenyl (BiPh), tetra-
hydrodibenzothiophene (HDBT), and phenylcyclohexane (PhCH). DDS and HYDS stand
for the routes direct desulfurization and hydrogenation, respectively.
At this point two levels of structure for unsupported sulfides have to
be distinguished. The primary structure is defined by the morphology of
the Mo(W)S2 particles, that is, slab length, stacking degree and con-
centration of promoting Ni at the edges. The secondary structure is
given by the arrangement of those primary Mo(W)S2 crystals and is
reflected by surface area and NO uptake (as it does not differentiate
among Mo(W)S2 and NiSx). The secondary structure determines the
total amount of active edges exposed during reaction, whereas the
primary structure (and the composition of the slabs) determines the
intrinsic properties of such edges. In the following, it is discussed how
the measured properties reflect the differences of the two levels of
structure of the catalysts and their activity in HDN and HDS. An em-
pirical linear model that quantitatively supports the following discus-
sion is presented in the supporting information.
(92 kJ/mol). Given that faster HDS is observed in the presence of OPA,
it is concluded, as in previous cases, that quinoline decreased the rates
of DBT HDS mainly due to blocking of HDS active sites.
3.4. Structure activity correlations
The active sites in Mo(W)S2 are coordinatively unsaturated sites
(CUS) and acidic SH groups located on the (1010) and (1010) surface of
the crystals, i.e., the edges of the slab terminated formally by metal
cations or sulfur. CUS and SH groups catalyze hydrogenolytic SeC bond
cleavage after coordination of the hydrocarbon via S on the CUS [4,11].
SH groups are involved in hydrogenolytic and hydrogenation pathways
as H-delivering sites. CUS, however, are not directly related with the
hydrogenation of aromatic compounds, which have led to conclude that
two kinds of active sites exist for hydrogenation and defunctionaliza-
tion [30,31]. The hydrogenation sites have been associated to the top
most layer of a Mo(W)S2 particle [3,32], which leads to even stronger
effects of morphology of Mo(W)S2 on hydrogenation than on de-
functionalization.
3.4.1. Hydrotreating activity in the presence of o-propylaniline
Note that WMoNi-sHT has an outstanding activity for the HDN of
OPA (with and without DBT present) and the HDS of DBT in the pre-
sence of OPA. This is attributed to high concentration of accessible
active edges as reflected by the fact that this material combines the
Promoted sites are created by substitution of Mo or W by Ni at the
Fig. 11. Yield of the products of the hydro-
desulfurization of dibenzothiophene on the
sulfided catalysts in the presence of OPA:
WMoNi-a (circle, unfilled), WMoNi-b
(square, unfilled), WMoNi-s (triangle, un-
filled), WMoNi-aHT (circle, filled), WMoNi-
bHT (square, filled), and WMoNi-sHT (tri-
angle, filled).
30
20
10
0
100
80
60
40
20
0
300
320
340
360
380
300
320
340
360
380
Temperature, ºC
Temperature, ºC
9

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
Table 2
Apparent activation energies (kJ/mol) determined from the HDN of o-propylaniline and quinoline and the HDS of dibenzothiophene.
HDN
HDS
DBT-OPAe
OPAa
Qb
OPA-DBTc
Q-DBTd
DBT-Qf
WMoNi-a
WMoNi-aHT
WMoNi-b
WMoNi-bHT
WMoNi-s
WMoNi-sHT
130
159
155
143
103
138
6
7
7
7
5
6
90
98
96
146
45
89
4
4
4
7
2
8
141
160
160
148
139
175
7
8
8
7
6
8
100
107
108
134
43
74
5
5
5
6
88
108
86
78
92
4
5
4
3
4
71
88
72
85
71
89
3
4
3
4
3
4
2
3
102
5
a
HDN of o-propylaniline in the absence of dibenzothiophene.
HDN of quinoline in the absence of dibenzothiophene.
HDN of o-propylaniline in the presence of dibenzothiophene.
HDN of quinoline in the presence of dibenzothiophene.
HDS of dibenzothiophene in the presence of o-propylaniline.
HDS of dibenzothiophene in the presence of quinoline.
b
c
d
e
f
highest specific surface area and concentration of adsorbed NO among
the studied catalysts. Both parameters do not have a direct correlation
with the concentration of the active sites, because of the anisotropy of
Mo(W)S2 and of the adsorption of NO on sites active and inactive for
hydrotreating (Mo(W)S2 and Ni sulfides, respectively) [40]. However,
large surface areas and high concentration of sites adsorbing NO reflect
secondary structures that increase the possibility of having the active
surface of the crystals exposed to the reactants. Thus, both parameters
are indicators of high hydrotreating activity in the presence of OPA.
Increasing the possibility of exposing active edges to the reactant is
an important parameter if, as concluded above, those edges have suf-
ficient Ni (which is a feature of the primary structure). However, as the
morphology of the catalyst varies, the promotion degree varies as well
but without an apparent correlation. While the results of this work do
not allow quantification of the concentration of Ni-decorated sites, it
has been previously noted that an increasing proportion of Ni in seg-
regated Ni sulfides leads to increasing differences between the bulk Ni
concentration and that estimated at the surface by XPS [15,18]. Thus,
in accordance with the Ni molar fractions reported in Table 1, the Ni
decoration follows the qualitative trend: WMoNi-a < WMoNi-sHT <
WMoNi-s < WMoNi-b < WMoNi-aHT < WMoNi-bHT. In the latter
three materials the concentration of Ni at the surface is outstandingly
low. This suggests that segregated Ni sulfides, inactive for hydrotreating
[40], are not present at the surface to a large extent, leading to rela-
tively large concentrations of exposed active sites. Thus, the higher
HDN and HDS activity of WMoNi-b, WMoNi-aHT, and WMoNi-bHT was
caused by lower concentration of segregated Ni sulfides at the surface,
which compensated the disadvantageous structural features of these
materials. In contrast, WMoNi-a and WMoNi-s showed the lowest ac-
tivity for the HDN of OPA because they had low surface area and high
proportions of segregated Ni sulfides at the surface.
higher hydrogenation activity than Ni-MoS2 [42–44]. The role of high
W content on the activity of the catalysts being evident in the presence
of quinoline implies that the strong hydrogenation functionality of Ni-
WS2 becomes more important with increasing complexity of the aro-
matic reactant. The catalyst WMoNi-b has high content of W and
average Ni content at the surface and slab length. Thus, the primary
structure of this catalyst combines the right features to achieve the
highest activity for the conversion of quinoline.
3.4.3. Hydrodesulfurization activity in the presence of o-propylaniline and
quinoline
WMoNi-sHT and WMoNi-b are the most active catalyst for DBT HDS
in the presence of OPA (and the most active for the HDN of OPA as
well). This indicates that secondary structure (surface area, density of
sites for NO adsorption) and low Ni content at the surface are indicators
of high accessibility of Ni-Mo(W)-S sites active for HDS of DBT in the
presence of OPA. WMoNi-bHT has the lowest HDS activity (which
makes the difference between OPA HDN and DBT HDS) due to its low
concentration of active sites reflected in its low NO uptake.
In the presence of quinoline, the HDS of DBT proceeds faster on
WMoNi-b, WMoNi-bHT, and WMoNi-aHT, which is correlated with the
low concentrations of Ni at the surface. This, in turn, implies that, in the
presence of bulky aromatic N-compounds, the strong hydrogenolytic
functionality provided by Ni sites compensates the low overall site
availability.
3.4.4. Comparison of supported and unsupported bimetallic and trimetallic
sulfide catalysts
Rate constants determined over the series of trimetallic sulfide
catalysts studied in this work are compared in Table 3 with those
Table 3
3.4.2. Hydrotreating activity in the presence of quinoline
First order rate constants k (mol/(h·gcat)) for the HDN of o-propylaniline (OPA) and
quinoline (Q) at 370 °C on the unsupported catalysts reported in this work and on other
unsupported and supported catalysts previously reported.
The ranking of rate constants in the presence of quinoline differed
from that in the presence of OPA. Thus, the HDN of polyaromatic
compounds, like quinoline, poses different requirements to the active
phase than anilines. For instance, WMoNi-bHT, which has the largest
Mo(W)S2 slabs, is one of the less active catalysts for HDN of quinoline,
although it is one of the most active catalysts for HDN of OPA. Thus,
small length of Mo(W)S2 slabs becomes very important for the con-
version of quinoline. This is probably caused by short slabs having a
high proportion of corners, which have been hypothesized as preferred
hydrogenation sites for bulky aromatic compounds [41]. Thus, for HDN
of quinoline, the accessible edges must have a favorable primary
structure, i.e., a defined slab length and edge composition.
OPA k [mol/(h∙gcat)]
Q k [mol/(h∙gcat)]
WMoNi-a
WMoNi-aHT
WMoNi-b
WMoNi-bHT
WMoNi-s
5.68·10−3
1.21·10−2
2.65·10−2
1.26·10−2
2.76·10−3
3.91·10−2
1.99·10−3
2.15·10−3
1.16·10−3
5.10·10−3
3.17·10−3
3.42·10−3
5.58·103
2.13·10−3
3.04·10−3
2.07·10−3
–
WMoNi-sHT
a
Ni-MoS2
a
Ni-MoxW1-xS2
–
–
a
Ni-WS2
b
Ni-MoS2/Al2O3
5.28·10−3
In line with this conclusion, note that WMoNi-sHT is less active for
quinoline HDN than for OPA, despite its high specific surface area and
NO uptake. This is attributed to its low W content because Ni-WS2 has
a
Rate constants calculated from the data reported in Ref. [18].
Rate constants calculated from the data reported in Refs. [25] and [27].
b
10

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
previously obtained over other unsupported and Al2O3-supported cat-
alysts. The rate constants for OPA HDN over the trimetallic catalysts
Acknowledgements
studied in this work varied between 2.7·10−3 mol/(h·gcat
)
and
This work was supported by the Chevron Energy Technology
Company. The authors would like to specially thank Alexander
Kuperman, Axel Brait, and Roel Prins for fruitful discussions. We are
also grateful to the members of TU München, Marianne Hanzlik (TEM),
Martin Neukamm (SEM) and Xaver Hecht (BET and technical support).
3.9·10−2 mol/(h·gcat) (WMoNi-s and WMoNi-sHT, respectively). How-
ever, all these catalysts were more active than unsupported catalysts
(Ni-Mo, Ni-W, and Ni-Mo-W) previously reported [18]. Interestingly,
most of the unsupported catalysts of this work were also more active
than an Al2O3-supported Ni-MoS2 catalyst (5.1·10−3 mol/(h·gcat)) re-
ported in Ref. [27]. Most of the rate constants for the HDN of quinoline
on the unsupported catalysts of this work (from 3·10−4 mol/(h·gcat) to
3.4·10−3 mol/(h·gcat)) were lower than the rate on Ni-MoS2/Al2O3
(5.3·10−3 mol/(h·gcat)) [25]. The exception therein is WMoNi-b, which
exhibited activity (5.6·10−3 mol/(h·gcat)) comparable to the supported
catalyst. The wide range of activity observed on unsupported catalysts
shows the strong impact of the precursor (varied by synthesis methods)
on the activity and properties of the sulfides. The rate constants listed in
Table 3 also show that the activity of unsupported Ni-Mo-W catalysts is
higher than that of bimetallic (Ni-Mo, Ni-W) catalysts. This observation,
however, is not common because the activity of the sulfides strongly
depends on their structure and cationic distribution [15,45]. Differ-
ences in intrinsic activity between bimetallic sulfides (NiMo, NiW) and
trimetallic sulfides are still to be elucidated. Finally, the comparison
shown in Table 3 indicates that the activity of the unsupported trime-
tallic sulfides is comparable (HDN of quinoline) or much higher (HDN
of OPA) than that of the archetypical Al2O3-supported Ni-MoS2. This
highlights the importance of studying unsupported multimetallic sul-
fides more in depth in order to optimize their preparation procedures
further.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.cattod.2017.05.083.
References
[1] R. Prins, Hydrotreating, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of
Heterogenous Catalysis, Wiley-VCH Verlag GmbH, Weinheim, 1997, pp.
1908–1928.
[2] R. Prins, Adv. Catal. 46 (2001) 399–464.
[3] R.R. Chianelli, G. Berhault, B. Torres, Catal. Today. 147 (2009) 275–286.
[4] W. Bensch, Hydrotreating: Removal of Sulfur from Crude Oil Fractions with Sulfide
Catalysts, Elsevier Ltd., 2013.
[5] S. Eijsbouts, S.W. Mayo, K. Fujita, Appl. Catal. A Gen. 322 (2007) 58–66.
[6] M. Poisot, W. Bensch, S. Fuentes, G. Alonso, Thermochim. Acta 444 (2006) 35–45.
[7] M. Ramos, G. Berhault, D.A. Ferrer, B. Torres, R.R. Chianelli, Catal. Sci. Technol. 2
(2012) 164.
[8] J. Bocarando, R. Huirache-Acuña, W. Bensch, Z.D. Huang, V. Petranovskii,
S. Fuentes, G. Alonso-Núñez, Appl. Catal. A Gen. 363 (2009) 45–51.
[9] C. Thomazeau, C. Geantet, M. Lacroix, V. Harlé, S. Benazeth, C. Marhic, M. Danot,
J. Solid State Chem. 160 (2001) 147–155.
[10] A. Olivas, D.H. Galván, G. Alonso, S. Fuentes, Appl. Catal. A Gen. 352 (2009) 10–16.
[11] H. Topsøe, B.S. Clausen, F.E. Massoth, Hydrotreating Catalysis, Springer Verlag,
Berlin, 1996.
[12] T. Maesen, A.E. Kuperman, US 2009/0107883 A1 (2009), to Chevron Corporation.
[13] S.L. Soled, S., Miseo, S., Eisjsbouts, F.L. Plantenga, US 2007/0090023 A1 (2007), to
Exxon Mobile Research and Engineering Company.
4. Conclusions
[14] J. Han, A. Kuperman, T.LM. Maesen, H. Treviño, WO 2014/039735 A2 (2014), to
Chevron Corporation.
[15] J. Hein, O.Y. Gutiérrez, S. Albersberger, J. Han, A. Jentys, J.A. Lercher,
ChemCatChem 9 (2017) 629–641, http://dx.doi.org/10.1002/cctc.201601281.
[16] Y. Yi, B. Zhang, X. Jin, L. Wang, C.T. Williams, G. Xiong, D. Su, C. Liang, J. Mol.
Catal. A Chem. 351 (2011) 120–127.
Different catalyst precursors lead to sulfide catalysts with varying
physicochemical properties whereby the concentration of Ni-promoted
sites, their primary structure, and the accessibility are the main para-
meters determining hydrotreating activity. However, the activity of the
Ni-Mo-W sulfides for the conversion of o-propylaniline (OPA), quino-
line, and dibenzothiophene (DBT) showed a complex dependence on
these parameters. Overall, high surface area, short Mo(W)S2 slabs, high
W content, and low Ni content at the surface are concluded to lead to
high hydrotreating activity. Surface area and NO uptake are related to
the concentration of exposed active edges, whereas the size of Mo(W)S2
crystals and the content of W and Ni reflect the intrinsic activity of
those edges. That is, homogeneous distribution of Ni across the catalyst
volume is associated to Ni decoration, whereas the strong hydrogena-
tion functionality of WS2 is required to convert polyaromatic com-
pounds.
[17] Y. Yi, X. Jin, L. Wang, Q. Zhang, G. Xiong, C. Liang, Catal. Today 175 (2011)
460–466.
[18] J. Hein, O.Y. Gutierrez, E. Schachtl, P. Xu, N. Browning, A. Jentys, J. Lercher,
ChemCatChem 7 (2015) 3692–3704.
[19] H. Nava, F. Pedraza, G. Alonso, Catal. Lett. 99 (2005) 65–71.
[20] Y. Yu, B. Fonfé, A. Jentys, G.L. Haller, J.A. Rob Van Veen, O.Y. Gutiérrez,
J.A. Lercher, J. Catal. 292 (2012) 13–25.
[21] N. Topsøe, A. Tuxen, B. Hinnemann, J.V. Lauritsen, K.G. Knudsen, F. Besenbacher,
H. Topsøe, J. Catal. 279 (2011) 337–351.
[22] C. Thomazeau, C. Geantet, M. Lacroix, M. Danot, V. Harle, Oil Gas Sci.
Technol.–Rev. IFP 60 (2005) 781–790.
[23] L. Wang, Y. Zhang, Y. Zhang, Z. Jiang, C. Li, Chem. – A Eur. J. 15 (2009)
12571–12575.
[24] D.O. Dumcenco, Y.C. Su, Y.P. Wang, K.Y. Chen, Y.S. Huang, C.H. Ho, K.K. Tiong,
Solid State Phenom. 170 (2011) 55–59.
In general, if enough Ni decoration of Mo(W)S2 edges exists, the
availability of exposed edges dominates the activity for the conversion
of o-propylaniline and dibenzothiophene in presence of OPA. However,
for the conversion of quinoline and dibenzothiophene in the presence of
quinoline, the most active edges are those, which are short and contain
W and high concentrations of Ni at the edges.
[25] J. Hein, A. Hrabar, A. Jentys, O.Y. Gutiérrez, J.A. Lercher, ChemCatChem 6 (2014)
485–499.
[26] O.Y. Gutiérrez, A. Hrabar, J. Hein, Y. Yu, J. Han, J.A. Lercher, J. Catal. 295 (2012)
155–168.
[27] A. Hrabar, J. Hein, O.Y. Gutiérrez, J.A. Lercher, J. Catal. 281 (2011) 325–338.
[28] O.Y. Gutiérrez, S. Singh, E. Schachtl, J. Kim, E. Kondratieva, J. Hein, J.A. Lercher,
ACS Catal. 4 (2014) 1487–1499.
[29] N. Satterfield, J. Catal. 387 (1988) 375–387.
In this work, WMoNi-b (mainly in the presence of quinoline) and
WMoNi-sHT (mainly in the presence of OPA) showed the highest ac-
tivities for hydrodenitrogenation and hydrodesulfurization. Whereas
WMoNi-sHT is concluded to have high density of active edges exposed
(large surface area and NO uptake), WMoNi-b is concluded to have the
most active edges for hydrogenation (high W content, and best com-
promise among incorporated Ni and slab length). In contrast, WMoNi-a
and WMoNi-s ranked low in most of the reactions due to the dis-
advantageous combination of low concentration of edges and low in-
trinsic activity indicated by low surface area, low W content and large
content of Ni near the surface.
[30] R. Prins, M. Egorova, A. Röthlisberger, Y. Zhao, N. Sivasankar, P. Kukula, Catal.
Today 111 (2006) 84–93.
[31] T.C. Ho, L. Qiao, J. Catal. 269 (2010) 291–301.
[32] J.V. Lauritsen, M. Nyberg, J.K. Nørskov, B.S. Clausen, H. Topsøe, E. Lægsgaard,
F. Besenbacher, J. Catal. 224 (2004) 94–106.
[33] N.-Y. Topsøe, H. Topsøe, J. Catal. 139 (1993) 641–651.
[34] E. Schachtl, E. Kondratieva, O.Y. Gutiérrez, J.A. Lercher, J. Phys. Chem. Lett. 6
(2015) 2929–2932.
[35] L.S. Byskov, J.K. Nørskov, B.S. Clausen, H. Topsøe, J. Catal. 122 (1999) 109–122.
[36] R. Prins, V.H.J. De Beer, G.A. Somorjai, Catal. Rev. Sci. Eng. 31 (1989) 1–41.
[37] Q. Gao, T.N.K. Ofosu, S.G. Ma, V.G. Komvokis, C.T. Williams, K. Segawa, Catal.
Today 164 (2011) 538–543.
[38] L. Fischer, V. Harl, S. Kasztelan, Stud. Surf. Sci. Catal. 127 (1999) 261–268.
[39] P. Tétényi, P. Schnörch, O. Tellinger, React. Kinet. Catal. Lett. 97 (2009) 141–150.
[40] E. Schachtl, L. Zhong, E. Kondratieva, J. Hein, O.Y. Gutiérrez, A. Jentys,
J.A. Lercher, ChemCatChem 7 (2015) 4118–4130.
11

S. Albersberger et al.
CatalysisTodayxxx(xxxx)xxx–xxx
[41] 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. 199 (2001) 224–235.
Technol. 61 (1999) 103–116.
[44] T. Kabe, Y. Aoyama, D. Wang, A. Ishihara, W. Qian, M. Hosoya, Q. Zhang, Appl.
[42] J.A.R. van Veen, Zeolites Clean. Technol. in: M. Guisnet, J.-P. Gilson (Eds.),
Imperial College Press, Singapore, 2002, pp. 131–152.
[43] W.R.A.M. Robinson, J.A.R. van Veen, V.H.J. de Beer, R.A. van Santen, Fuel Process.
Catal. A Gen. 209 (2001) 237–247.
[45] Y. Licea, R. Grau-Crespo, L. Palacio, A. Faro, Catal. Today (2016), http://dx.doi.
org/10.1016/j.cattod.2016.11.031.
12