Unprecedented selectivity to the direct desulfurization (DDS) pathway in a highly active FeNi bimetallic phosphide catalyst

Article abstract of DOI:10.1016/j.jcat.2011.08.006

The hydrodesulfurization (HDS) of the refractory compound 4,6-dimethyldibenzothiophene (DMDBT) normally proceeds through a hydrogenation pathway that removes the planarity of the ring system and makes the hindered sulfur atom more accessible to the desulfurization centers. In this study, a highly active dispersed bimetallic NiFeP catalyst is found to have high selectivity for a direct desulfurization pathway, which does not require prior hydrogenation. The compound has equal numbers of Ni and Fe atoms which extended X-ray absorption fine structure analysis indicates are distributed randomly in the hexagonal Fe₂P structure, with just a slight enrichment of Fe on the surface. This is supported by density functional theory calculations. The remarkable properties of the catalyst are ascribed to a ligand effect of Fe on the active Ni atoms.

Full text of DOI:10.1016/j.jcat.2011.08.006

Journal of Catalysis 285 (2012) 1–5
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Priority Communication
Unprecedented selectivity to the direct desulfurization (DDS) pathway
in a highly active FeNi bimetallic phosphide catalyst
a,b,
⇑
, Haiyan Zhao , Hans-Joachim Freund , Kiyotaka Asakura d,e, Radosław Włodarczyk f,
a
c
S. Ted Oyamaf
Marek Sierka
a
Department of Chemical Systems Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
Department of Chemical Engineering, Virginia Tech, Blacksburg, VA 24061, USA
Department of Chemical Physics, Fritz Haber Institute of the Max Planck Society, Faradayweg 4-6, Berlin 14195, Germany
Catalysis Research Center, Hokkaido University, Kita 21-10, Sapporo 001-0021, Japan
Department of Quantum Science and Technology, Hokkaido University, Kita 21-10, Sapporo 001-0021, Japan
Institut für Chemie, Humboldt-Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Germany
b
c
d
e
f
a r t i c l e i n f o
a b s t r a c t
Article history:
The hydrodesulfurization (HDS) of the refractory compound 4,6-dimethyldibenzothiophene (DMDBT)
normally proceeds through a hydrogenation pathway that removes the planarity of the ring system
and makes the hindered sulfur atom more accessible to the desulfurization centers. In this study, a highly
active dispersed bimetallic NiFeP catalyst is found to have high selectivity for a direct desulfurization
pathway, which does not require prior hydrogenation. The compound has equal numbers of Ni and Fe
atoms which extended X-ray absorption fine structure analysis indicates are distributed randomly in
the hexagonal Fe P structure, with just a slight enrichment of Fe on the surface. This is supported by den-
2
sity functional theory calculations. The remarkable properties of the catalyst are ascribed to a ligand
effect of Fe on the active Ni atoms.
Received 8 May 2011
Revised 29 June 2011
Accepted 8 August 2011
Available online 22 October 2011
Keywords:
NiFeP
Deep HDS
DFT
Ó 2011 Elsevier Inc. All rights reserved.
EXAFS
Nickel iron phosphide
4
,6-Dimethyldibenzothiophene
1
. Introduction
Reduction of the sulfur content in transportation fuels is an area
valuable high-octane aromatics in the product. Analysis of the bime-
tallic catalyst by X-ray absorption fine structure (EXAFS) reveals that
it exposes low-coordination sites that are particularly active for
DDS. Density functional theory (DFT) calculations provide informa-
tion on the site occupancy in the alloy. We anticipate that these find-
ings will lead to the exploration of other phosphide catalysts with
structures that can be modified and controlled by alloying.
of great current activity in the refining industry [1]. Recent reviews
describe conventional hydroprocessing and emerging processes
such as oxidative desulfurization and biodesulfurization [2,3].
Conventional hydroprocessing is challenging because it requires
the hydrodesulfurization (HDS) of refractory compounds such as
Early work [6,7] showed that the HDS activity of common phos-
4
,6-dimethyldibenzothiophene (4,6-DMDBT), in which the sulfur
phides follows the order: Ni
of bimetallic phosphides such as Ni
and Ni Co P [12,13] have also been studied because a synergistic
effect between the components was foreseen as found for pro-
moted metal sulfides. Unexpectedly, however, these bimetallic
phosphide phases did not show enhanced activity or selectivity
over the component Ni, Co, or Mo phosphides, except for the case
2
P > WP > MoP > CoP > Fe
2
P. A number
is sterically protected [4]. Sulfur removal on commercial promoted
catalysts such as CoMo or NiMo sulfides proceeds by a hydrogena-
tion (HYD) route in which the 4,6-DMDBT is partly hydrogenated
to render the molecule non-planar and the sulfur accessible to the
catalyst surface [5]. Here, we report a bimetallic NiFe phosphide cat-
alyst, which has higher activity than commercial catalysts, and
which follows a direct desulfurization (DDS) route that allows re-
moval of the sulfur without prior hydrogenation (Scheme 1a). This
has been long a goal in the field because it permits HDS to be carried
out with less consumption of expensive hydrogen while retaining
x
Mo P [8,9], Co Mo
y
x
y
P [10,11],
x
y
of an unsupported [10] and supported [13] Co0.08Ni
2
P where
respective increases in conversion over Ni P of 67% were found
2
in the HDS of 4,6-DMDBT and 34% in the HDS of thiophene.
Recently, studies with unsupported and silica-supported Fex-
Ni2Àx
P
y
catalysts [14] showed a modest 10% activity increase in
the HDS of dibenzothiophene with a Ni-rich catalyst (x = 0.03). In
⇑
Corresponding author at: Department of Chemical Systems Engineering, The
University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan.
this study, we explore the activity of similar FeNiP/SiO
2
catalysts
E-mail address: oyama@vt.edu (S. Ted Oyama).
for the deep HDS of 4,6-DMDBT.
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.08.006

2
S. Ted Oyama et al. / Journal of Catalysis 285 (2012) 1–5
(
a)
DDS
CH3
CH3
3,3'-Dimethylbiphenyl
S
CH3
CH3
HYD
4,6-Dimethyldibenzothiophene
CH3
CH3
CH
3
CH
3
3-Methylcyclohexyltoluene
3,3'-Dimethylbicyclohexane
(
b)
DEHYD
HYD
Naphthalene
H
H
Tetralin
H
H
t-Decalin
c-Decalin
Scheme 1. (a) Products of 4,6-dimethyldibenzothiophene desulfurization, (b) products of tetralin dehydrogenation and hydrogenation.
2
. Methods
with a model feed liquid containing 500 ppm sulfur as 4,6-DMDBT,
3
000 ppm sulfur as dimethyl disulfide, 200 ppm nitrogen as quin-
The compounds Ni
2
P/SiO
2
, NiFeP(3:1)/SiO
2
, NiFeP(1:1)/SiO
2
, and
oline, 1 wt.% tetralin, 0.5 wt.% n-octane as internal standard, and
balance n-tridecane.
Fe
2
P/SiO were prepared wherethe numbers in parenthesis are molar
2
ratios. The synthesis involved temperature-programmed reduction
TPR), following procedures reported previously [15,16]. Briefly, the
Periodic density functional theory (DFT) calculations were
carried out using the Vienna Ab Initio Simulation package (VASP)
[24] and the Perdew, Burke and Ernzerhof (PBE) exchange–correla-
tion functional [25]. The calculations were performed using the pro-
jector augmented wave method (PAW) [26]. Only the valence
electrons were explicitly considered. Optimizations of cell parame-
ters used a 7 Â 7 Â 13 Monkhorst–Pack k-point mesh [27]for the
Brillouin-zone sampling and an energy cutoff of 1000 eV for a plane
wave basis set. The Pulay stress arising from the incomplete basis set
was minimized by restarting the optimization until self-consistency
of the total energy was reached. Optimizations of atomic positions
were performed with a 4 Â 4 Â 7 Monkhorst–Pack k-point mesh
and a plane wave cutoff of 400 eV. 1 Â 1 Â 2 supercells was used
for calculations of electronic density of states (DOS).
(
synthesis involved impregnation of metal phosphate precursors on
a silica support (Cabosil EH5), followed by reduction to the phos-
À1
phides at ꢀ560 °C. The total metal molar loading was 1.6 mmol g
(
mmol per g of support) in all cases. The proportion of metals in the
precursors were in the stoichiometric ratios indicated above, but
the phosphorus content was quadrupled because earlier work had
shownthatthisproducedactive compositionsinthecaseof Ni
2
P [16].
The activity of the catalysts for 4,6-DMDBT HDS was compared at
3
3
2
40 °C and 30 bar [6,7] with a H flow rate of 150 cm (NTP)/min and
a contact time of 2.4 s. Quantities of catalysts loaded in the reactor
corresponded to the same amount of CO or atomic oxygen uptake
(
240
eP(3:1)/SiO
for the Fe P/SiO
l
mol) and were 2.18 g for the Ni
catalyst, 4.00 g for the NiFeP(1:1)/SiO
2 3
, and 1.26 g for the Ni–Mo–S/Al O
2
P/SiO
2
, 2.50 g for the NiF-
catalyst, 4.62 g
catalyst. Note
2
2
2
2
that the Ni–Mo–S catalysts are an optimized commercial catalyst
with very high dispersion. The use of CO chemisorption to count
the surface metal sites in phosphides is reasonable. Early studies
comparing unsupported and supported materials indicated that
the same turnover frequencies were obtained when CO was used
3. Results and discussion
The high activity of Ni
synthesis, structure and reactivity [28]. The crystal structure
[29]of Ni P is hexagonal (a = 0.5859 nm, c = 0.3382 nm) with
2
P has prompted many studies of its
2
o
o
ꢀ
as a probe [6,17]. The use of O
2
chemisorption to titrate sites in sul-
space group p62m where the metal atoms are of two types, M(1)
at position 3(f) with tetrahedral coordination and M(2) at position
3(g) with square pyramidal coordination (Fig. 1). From the average
of M–P bond distances, the size of the M(1) site can be calculated to
be 0.223 nm and that of the M(2) site 0.243 nm, so the M(1) site is
smaller. A recent study [30] showed that the M(1) site carries out
DDS while the pyramidal M(2) site is particularly active for HYD
fides serves only as an approximation [18], but it has been shown
that pulse chemisorption at dry ice/acetone temperatures avoids
corrosive chemisorption [19,20], and this method was used here.
X-ray absorption spectra at the Ni K-edge (8.333 keV) and Fe
K-edge (7.112 keV) of reference and catalyst samples were re-
corded in the energy range 8.233–9.283 keV at beam line X18B
at the National Synchrotron Light Source at Brookhaven National
Laboratory. Measurements were taken by standard absorption
methods on samples loaded in cells with Kapton windows without
exposure to the atmosphere. The EXAFS data were analyzed by the
program REX [21,22]. Phase shift and amplitude functions were de-
rived from the FEFF8 program [23].
and that the latter is exposed on small Ni
interest to study NiFeP because Fe P is isostructural with Ni P
[31,32] and substitution of Fe for Ni was surmised to change the
electronic properties of the cluster. From the structure [33] of
2 o o
Fe P (a = 0.5865 nm, c = 0.3456 nm), the size of the M(1) site is
2
P crystallites. It was of
2
2
0.225 nm and that of the M(2) site is 0.246 nm.
Hydrotreating activities of the samples were measured in a
three-phase, packed-bed reactor operated at 30 bar and 340 °C
The following series of compounds were tested: Ni
2 2 2 2
NiFeP(3:1)/SiO , NiFeP(1:1)/SiO , and Fe P/SiO2. The Ni P showed
2
P/SiO
2
,

S. Ted Oyama et al. / Journal of Catalysis 285 (2012) 1–5
3
the HDS reaction, except that 10% tetralin was added and the solvent
reduced accordingly. It was found that the conversion of tetralin was
2
6% and the selectivities were 25% naphthalene, 49% trans-decalin,
and 26% cis-decalin. If dehydrogenation were the thermodynami-
cally preferred reaction at the reaction conditions, then tetralin
should produce naphthalene. In fact naphthalene was the minority
product. The experiment demonstrates directly that the selectivity
to the diphenyl is kinetically controlled even at close to 100% HDS
conversion, and this is supported by thermodynamic calculations
[
28]. The DDS products are kinetically favored because the sulfur
compounds are adsorbed strongly on the surface and prevent hydro-
genation of the primary aromatic products. Note that the tetralin
level was substantially higher than the 4,6-DMDBT (10,000 ppm
vs. 500 ppm) in the above measurements.
Density functional theory calculations (DFT) were undertaken
to shed light on the bulk electronic structure of the materials.
The calculated structural parameters for Ni
2 o
P (a = 0.5886 nm,
c
o
= 0.3372 nm) and Fe P (a = 0.5809, c = 0.3428 nm) show less
2
o
o
than 1% deviation from the experimental [29,33] data. Calculations
on the FeNiP structures containing 100% Fe and 100% Ni in M(1)
sites, respectively, indicate only a 1.30 kJ mol of metal sites pref-
2
Fig. 1. Crystal structure of Ni P showing tetrahedral M(1) and square pyramidal
M(2) sites.
À1
erence for Fe occupation of M(1) sites. This minute energetic differ-
ence shows that M(1) and M(2) sites could be occupied by both Ni
and Fe. Indeed, additional calculations performed for different Fe-
NiP structures containing 67%, 50%, as well as 33% Fe in M(1) sites
confirm that configurations with mixed M(1) and M(2) occupations
are the most stable. Electronic density of states of FeNiP (Not
shown) clearly indicates changes in the metallic nature of this alloy
the highest HDS conversion (99%), while the catalysts containing
increasing amounts of Fe showed slightly lower activity while
the Fe P catalyst was essentially inactive, as found earlier for
2
dibenzothiophene HDS [16]. The selectivity results at steady state
showed remarkable dependency on composition (Fig. 2). Whereas
the Ni
Ni/Fe ratio of 3:1 and 1:1 showed increasing DDS selectivity of
0% and 85%. These high values for DDS are unprecedented in
2 2
P/SiO had a DDS selectivity of only 12%, the samples with
compared to the pure Fe
reveal metallic character [37,38].
Recent characterization of Fe
2
P and Ni
2
P compounds, which both also
7
x
Ni2ÀxP catalysts with Mössbauer
y
the hydrotreating literature where typical selectivity for a CoMo
effect spectroscopy [14] indeed confirms that for x = 1, the M(1) sites
are occupied by both Fe and Ni atoms (about 67% Ni and 33% Fe). The
experiments also indirectly confirm the small energy differences be-
tween different occupation patterns since the fraction of Fe atoms
that occupy M(1) site is not constant, but smoothly changes with re-
spect to the composition of catalyst. Thus, for low Ni contents
catalyst is ꢀ30% and for a NiMo catalyst is ꢀ20% in the HDS of
4
,6-DMDBT at similar conditions [34]. The results are particularly
notable as the high DDS selectivity is obtained at high conversion,
which would tend to favor the HYD products. In contrast to these
compounds, the Fe P/SiO had very low activity, as also found pre-
2 2
viously [14]. This is not surprising, among sulfides iron [35] and
iron alloys [36] are among the least active compositions.
To rule out that the observed selectivity to DDS products could be
due to the secondary dehydrogenation of the HYD products, exper-
iments were conducted with a cofeed of tetralin, which can undergo
dehydrogenation to naphthalene or hydrogenation to decalin
(x > 0.60), the Ni atoms preferentially occupy M(1) sites, but for high
Ni contents (x < 0.60), this site preference is reversed, and Ni atoms
preferentially occupy M(2) sites.
2
To obtain further insight on the materials, the NiFeP(1:1)/SiO ,
which showed the highest selectivity for DDS products, was char-
acterized by X-ray absorption fine structure (EXAFS) spectroscopy.
Analysis was carried out with a two-shell model for comparison
between the samples of different composition. Earlier work used
a three-shell analysis, which explicitly accounted for M(1) and
M(2) contributions [30], but this was not considered appropriate
with the current multi-element data. Comparison was made with
(Scheme 1b). The experimental conditions were the same as for
Conversion
100
Ni
the known structure of the compounds. The bond distances in
Ni P are smaller than those of Fe P, in agreement with the crystal-
lographic results. For NiFeP, fits were made with Ni in the M(1)
sites, Fe in the M(1) sites and 50–50 mixtures, and only the latter
gave a good agreement with the experimentally determined spec-
tra (Fig. 3). With this, the bond distances in the NiFeP compound
2 2 2 2
P/SiO and Fe P/SiO . The results (Table 2) are consistent with
8
6
4
2
0
0
0
0
0
2
2
MCHT
HYD
were obtained and found to be in between those of Ni
2
P and
3
,3'-DMDP
Fe P. The coordination numbers in the compounds are lower than
2
DDS
3
,3'-DMBCH
the average bulk M–P and M–M coordinations of 5 and 8, consis-
tent with the small particle size determined by X-ray line broaden-
ing (Table 1).
Examination of the details (Table 2) allows a number of observa-
tions. First, it is noted that the metal–phosphorus (M–P) and
Ni P
NiFeP(3:1) NiFeP(1:1) NiMoS
/SiO2
/SiO2
/Al O
2
2 2
metal–metal (M–M) distances are larger in Fe P than in Ni P, consis-
/SiO2
2
3
tent with the known lattice spacings (Table 1). Second, in the NiFeP
material, the Ni–P and Fe–P distances of 0.218 and 0.216 nm are
close, as are the Ni–M and Fe–M distances of 0.261 and 0.265 nm,
2
Fig. 2. Steady-state reactivity of NiFeP/SiO samples and reference sulfide at 340 °C
and 30 bar.

4
S. Ted Oyama et al. / Journal of Catalysis 285 (2012) 1–5
Table 2
2
X-ray absorption fine structure parameters for NiFeP (1:1)/SiO .
Sample
Bond pair
CNa
Distance (nm)
Debye–Waller factor (nm)
Ni
2
P/SiO
2
2
Ni–P
Ni–Ni
1.8(2)
3.2(5)
0.218(3)
0.260(3)
0.0077(2)
0.0092(3)
Fe
2
P/SiO
Fe–P
Fe–Fe
2.3(2)
1.9(7)
0.221(3)
0.268(3)
0.0092(2)
0.0083(3)
NiFeP (1:1)/SiO
2
Ni–P
Ni–Ni (or Ni–Fe)
Fe–P
1.9(2)
4.3(5)
1.7(2)
1.9(7)
0.218(3)
0.261(4)
0.216(3)
0.265(3)
0.0091(2)
0.0088(3)
0.0089(2)
0.0080(3)
Fe–Fe (or Fe–Ni)
a
Coordination number. The values in parenthesis show the error in the last digit.
(
a) 15
(b)
1
2
6
0
6
Ni P/SiO
2 2
Ni P/SiO
2
2
1
0
5
0
NiFeP/SiO2
NiFeP/SiO2
-
-
1
0
1
2
3
4
5
6
7
8
4
6
8
10
12
14
r / 0.1 nm-1
k / 10 nm
-1
(
c) 10
(d)
1
2
6
0
Fe P/SiO2
2
Fe P/SiO2
2
5
0
NiFeP/SiO2
NiFeP/SiO2
-6
0
1
2
3
4
5
6
4
6
8
10
12
14
-
1
-1
r / 0.1 nm
k / 10 nm
2 2 2
Fig. 3. EXAFS data of Ni P, Fe P, and NiFeP (1:1)/SiO , (a) magnitude of the Fourier transform of the Ni K-edge spectra, (b) Ni K-edge EXAFS spectra (points) and model (line)
from NiFeP/SiO , (c) magnitude of the Fourier transform of the Fe K-edge spectra, (d) Fe K-edge EXAFS spectra (points) and model (line) NiFeP/SiO .
2 2
so that the Fe and Ni atoms likely occupy sites interchangeably in the
structure. This agrees with the DFT results. Third, the Ni–P and Fe–P
distances are short, 0.218 and 0.216 nm, and correspond to dis-
tances expected for a metal in the tetrahedral M(1) site (0.22 nm)
rather than the pyramidal M(2) site (0.24 nm). Fourth, the coordina-
tion numbers of the Fe–P and Fe–M bonds (1.7 and 1.9) are smaller
than those of the Ni–P and Ni–M bonds (1.9 and 4.3).
probing just the M(1) site, with average M–P distance of
0.223 nm, rather than the M(2) site with average M–P distance
of 0.243 nm. Thus, it can say little about the relative numbers
of M(1) and M(2) sites. Interpretation (b): The EXAFS intensity
can have contributions from both M(1) and M(2) sites, and the
short distance indicates an organization of the structure that
makes M(1) sites more numerous.
There are two different interpretations for the third observation
of short Ni–P and Fe–P distances. Interpretation (a): The EXAFS
intensity for a collection of distances around a scattering atom will
be stronger for the shorter distances [39] so in this case, EXAFS is
There are also two interpretations for the fourth observation
about the lower coordination numbers associated with the Fe atoms
in comparison with the Ni atoms. (c) The Fe is preferentially, though
not exclusively, located at the surface, where nearest neighbor num-
bers are smaller. (d) Fe prefers M(1) sites that have lower coordina-
tion (4) than M(2) sites (5). Taking possibilities a and c in
combination gives a first model of the NiFeP catalyst as consisting
Table 1
Characterization of NiFeP/SiO
2
samples.
Particle size,a nm
(±2 nm)
ꢀ
Sample Surface area
CO uptake
of the hexagonal p62m structure (Fig. 1) with mostly random occu-
2
À1
À1
(m
g
)
(
lmol g
)
pation of the M(1) and M(2) sites but with Fe in a slight excess at the
surface in either of these sites. Taking possibilities b, c, and d in com-
bination gives a second model of the catalyst in which the hexagonal
NiFeP nanoparticles are terminated by M(1) sites that are slightly
more occupied by Fe atoms than Ni atoms. This is because in the bulk
material, there should be an equal numbers of M(1) sites and M(2)
sites, and the excess M(1) sites would only be possible if the nano-
particles were terminated by these sites. The DFT calculations are
consistent with both structures.
Ni
NiFeP(3:1)/
SiO
NiFeP(1:1)/
SiO
P/SiO
NiMoS/
Al
2
P/SiO
2
135
138
110
96
10
7
2
153
60
52
9
2
Fe
2
2
148
160
5
N/D
b
95
2 3
O
N/D: Not determined.
a
From Scherrer equation.
O Uptake.
2
The special reactivity of the FeNi bimetallic phosphide can be
understood from either structural model as revealed by EXAFS. In
b

S. Ted Oyama et al. / Journal of Catalysis 285 (2012) 1–5
5
the first model, which has both Ni and Fe on the surface, the cata-
lytic activity is attributed to the Ni atoms, as Fe P is inactive. The
References
2
[
[
[
1] R.G. Leliveld, S.E. Eijsbouts, Catal. Today 130 (2008) 183.
2] A. Stanislaus, A. Marafi, M.S. Rana, Catal. Today 153 (2010) 1.
3] S.T. Oyama, T. Gott, H. Zhao, Y.K. Lee, Catal. Today 143 (2009) 94.
special reactivity of the bimetallic material is probably due to a
ligand effect of Fe on Ni, which must also account for the high
selectivity to DDS. In the second model, the activity is again as-
cribed to the Ni atoms, with the high activity again due to a ligand
effect. However, the presence of the M(1) sites on the surface have
[4] T.C. Ho, Catal. Today 98 (2004) 3.
[
[
[
5] T. Kabe, A. Ishihara, Q. Zhang, Appl. Catal. A: Gen. 97 (1993) L1.
6] X. Wang, Y.K. Lee, W.J. Chun, S.T. Oyama, F. Requejo, J. Catal. 210 (2002) 207.
7] P. Clark, X. Wang, S.T. Oyama, J. Catal. 207 (2002) 256.
been associated with the DDS reaction on Ni
2
P nanoparticles [30],
[8] F. Sun, W. Wu, Z. Wu, J. Guo, Z. Wei, Y. Yang, Z. Jiang, F. Tian, C. Li, J. Catal. 228
(2004) 298.
and their presence on the surface of the bimetallic NiFeP nanopar-
ticles could account for the observed selectivity. The M(1) sites are
only fourfold coordinate and so are more accessible to the hindered
sulfur atom in 4,6-DMDBT. Although DFT calculations suggested
that the M(1) sites are more preferable on the surface of
[
9] J.A. Rodriguez, J.-Y. Kim, J.C. Hanson, S.J. Sawhill, M.E. Bussell, J. Phys. Chem. B
07 (2003) 6276.
10] I.I. Abu, K.J. Smith, J. Catal. 241 (2006) 356.
[11] V. Zuzaniuk, R. Prins, J. Catal. 219 (2003) 85.
1
[
[
[
[
12] I.I. Abu, K.J. Smith, Appl. Catal. A: Gen. 328 (2007) 58.
13] A.W. Burns, A.F. Gaudette, M.E. Bussell, J. Catal. 260 (2008) 262.
14] A.F. Gaudette, A.W. Burns, J.R. Hayes, M.C. Smith, R.H. Bowker, T. Seda, M.E.
Bussell, J. Catal. 272 (2010) 18.
Ni
SiO
ticles [30]. The recent work with Mössbauer effect spectroscopy ci-
ted earlier [14] finds that a Fe0.03–Ni1.97 catalyst had a
thiophene HDS activity 40% higher than an optimized 25 wt.%
catalyst (at 370 °C) and suggested that Ni in M(2)
2
P(0001) [40], it should be noted that previous work on Ni
2
P/
2
proposed that M(2) sites are found on the surface of small par-
[
[
[
15] J.F. Deng, H. Li, W. Wang, Catal. Today 51 (1999) 113.
16] X. Wang, P. Clark, S.T. Oyama, J. Catal. 208 (2002) 321.
17] P.A. Clark, S.T. Oyama, J. Catal. 218 (2003) 78.
P2.00/SiO
2
[18] S.J. Tauster, T.A. Pecoraro, R.R. Chianelli, J. Catal. 63 (1980) 515.
[
[
[
19] W. Zmierczak, G. MuraliDhar, F.E. Massoth, J. Catal. 77 (1982) 432.
20] T.A. Bodrero, C.H. Bartholomew, K.C. Pratt, J. Catal. 78 (1982) 253.
21] T. Taguchi, AIP Conf. Proc. 882 (2007) 162.
Ni2.00P1.60/SiO
2
sites (about 70%) is present on the surface of this low Fe content
sample. For a FeNi(1:1) composition, the entire sample contains
only about 22% Ni in M(2) sites, so at the surface, it may be that
the presence of Fe makes Ni prefer M(1) sites, contrary to what
[22] K. Asakura, in: Y. Iwasawa (Ed.), X-ray Absorption Fine Structure for Catalysts
and Surfaces, World Scientific, Singapore, 1996, pp. 33–58.
[
23] J.J. Rehr, J.J. Kas, M.P. Prange, A.P. Sorini, Y. Takimoto, F. Vila, CR Phys. 10 (2009)
48.
5
is found in pure Ni
2
2
P/SiO . More work is necessary to clarify this.
[24] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169.
[25] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 78 (1997) 1396.
It should also be noted that past studies have indicated that the ac-
tive phase is a phosphosulfide [41–43]. Indeed, EXAFS examination
of the catalyst after reaction, a detailed description of which is be-
yond the scope of this paper, reveals the presence of sulfur in quan-
tities less than a monolayer. So the description of the sites above
must encompass the presence of sulfur at reaction conditions.
Regardless of the interpretation, the activity results are remarkable
and the possibility of ligand effects with metals other than Fe pre-
sents a unique opportunity to design new improved catalysts.
[
[
[
26] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758.
27] H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188.
28] Y. Shu, Y.K. Lee, S.T. Oyama, J. Catal. 236 (2005) 112.
[29] S. Rundqvist, Acta Chem. Scand. 16 (1962) 992.
[
[
[
30] S.T. Oyama, Y.K. Lee, J. Catal. 258 (2008) 393.
31] J.B. Goodenough, J. Appl. Phys. 40 (1969) 1250.
32] J.P. Sénateur, A. Rouault, P. L’Héritier, A. Krumbügel-Nylund, R. Fruchart, D.
Fruchart, P. Convert, E. Roudaut, Mater. Res. Bull. 8 (1973) 229.
33] S. Rundqvist, F. Jellinek, Acta Chem. Scand. 13 (1959) 425.
34] J.H. Kim, X. Ma, C. Song, Y.-K. Lee, S.T. Oyama, Energy Fuels 19 (2005) 353.
35] T.A. Pecoraro, R.R. Chianelli, J. Catal. 67 (1981) 430.
[
[
[
[
[
36] T.C. Ho, Catal. Today 130 (2008) 206.
37] F. Luo, H.-L. Su, W. Song, Z.-M. Wang, Z.-G. Yan, Ch.-H. Yan, J. Mater. Chem. 14
Acknowledgments
(
2004) 111.
[
[
[
38] K. Edamoto, Y. Nakadai, H. Inomata, K. Ozawa, S. Otani, Solid State Commun.
148 (2008) 135.
39] E.D. Crozier, J.J. Rehr, R. Ingalls, in: D.C. Koningsberger, R. Prins (Eds.), EXAFS,
XANES and SEXAFS, Wiley, New York, 1988.
40] P. Liu, J.A. Rodriguez, T. Asakura, J. Gomes, K. Nakamura, J. Phys. Chem. 109
(2005) 4575.
We acknowledge support from the US Department of Energy, Of-
fice of Basic EnergySciences, through Grant DE-FG02-963414669, the
National Renewable Energy Laboratory through Grant DE-
FG3608GO18214, the Humboldt Foundation for a Senior Research
Award to STO, and the Japan Ministry of Agriculture, Forestry, and
Fisheries (Norinsuisansho). RW and MS acknowledge support from
the Deutsche Forschungsgemeinschaft (Cluster of Excellence ‘‘Unify-
ing Concepts in Catalysis’’).
[41] S.T. Oyama, X. Wang, Y.-K. Lee, W.-J. Chun, J. Catal. 221 (2004) 263–273.
[
42] T. Kawai, K.K. Bando, Y.-K. Lee, S.T. Oyama, W.-J. Chun, K. Asakura, J. Catal. 241
2006) 20.
43] A.E. Nelson, M. Sun, A.S.M. Junaid, J. Catal. 241 (2006) 180.
(
[