Oxidative dehydrogenation of ethane over Ni-W-O mixed metal oxide catalysts

Article abstract of DOI:10.1016/j.jcat.2011.02.010

Ni-W-O mixed oxides were prepared through the evaporation of aqueous solutions of nickel nitrate and ammonium tungstate and calcined in air at 500 °C for 2 h. The catalysts were characterized by several techniques (N ₂ adsorption, X-ray diffraction, temperature-programmed reduction, X-ray photoelectron spectroscopy, Fourier transform infrared spectroscopy of adsorbed CO, and ¹⁸O/¹⁶O isotope exchange) and tested in the oxidative dehydrogenation of ethane. The catalytic activity and catalyst reducibility decrease when the W content increases. Thus, nickel sites seem to be the active centers for ethane activation in these catalysts. However, the selectivity to ethylene strongly changes depending on the Ni/W ratio. In W-rich catalysts, in which NiWO₄ and WO₃ are mainly observed, a strong influence of ethane conversion on the selectivity to ethylene is observed. However, in Ni-rich catalysts, in which NiO crystallites and WO _x nanoparticles are mainly observed, ethane conversion hardly influences the selectivity to ethylene. It has been demonstrated that the nature of the Ni sites and the characteristics and number of the acid sites determine the catalytic behavior of these catalysts. The presence of Lewis acid sites with high acid strength in W-rich catalysts facilitates the decomposition of ethylene during ethane oxidation.

Full text of DOI:10.1016/j.jcat.2011.02.010

Journal of Catalysis 280 (2011) 28–39
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Oxidative dehydrogenation of ethane over Ni–W–O mixed metal oxide catalysts
a
b,
⇑
b
a
b
a
B. Solsona , J.M. López Nieto , P. Concepción , A. Dejoz , F. Ivars , M.I. Vázquez
a
Dep. Ingeniería Química, Univ. Valencia, 46100 Burjassot, Spain
Instituto de Tecnología Química (UPV-CSIC), Campus Universidad Politécnica de Valencia, Avenida de los Naranjos s/n, 46022 Valencia, Spain
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Ni–W–O mixed oxides were prepared through the evaporation of aqueous solutions of nickel nitrate and
Received 28 December 2010
Revised 23 February 2011
Accepted 24 February 2011
Available online 13 April 2011
ammonium tungstate and calcined in air at 500 °C for 2 h. The catalysts were characterized by several
2
techniques (N adsorption, X-ray diffraction, temperature-programmed reduction, X-ray photoelectron
18 16
spectroscopy, Fourier transform infrared spectroscopy of adsorbed CO, and O/ O isotope exchange)
and tested in the oxidative dehydrogenation of ethane. The catalytic activity and catalyst reducibility
decrease when the W content increases. Thus, nickel sites seem to be the active centers for ethane acti-
vation in these catalysts. However, the selectivity to ethylene strongly changes depending on the Ni/W
Keywords:
Nickel tungsten metal oxide catalysts
Oxidative dehydrogenation of ethane
Ethylene oxidation
ratio. In W-rich catalysts, in which NiWO
conversion on the selectivity to ethylene is observed. However, in Ni-rich catalysts, in which NiO crystal-
lites and WO nanoparticles are mainly observed, ethane conversion hardly influences the selectivity to
4 3
and WO are mainly observed, a strong influence of ethane
Catalyst characterization
x
XPS
TPR
XRD
Raman
FTIR of adsorbed CO
Oxygen isotopic exchange
ethylene. It has been demonstrated that the nature of the Ni sites and the characteristics and number of
the acid sites determine the catalytic behavior of these catalysts. The presence of Lewis acid sites with
high acid strength in W-rich catalysts facilitates the decomposition of ethylene during ethane oxidation.
Ó 2011 Elsevier Inc. All rights reserved.
1
. Introduction
Ethylene is the principal petrochemical building block and is a
been proposed in the past two decades, and it is known that the
reaction conditions and the mechanism involved strongly depend
on the catalytic system [2–7]. At the moment, the limiting factor
in the development of most of the proposed catalytic systems is
the high formation of carbon oxides, especially by the consecutive
deep oxidation of ethylene [2–7]. This has been improved by using
MoVTeNbO catalysts, which seems to be the most competitive
alternative nowadays to the current industrial process in ethylene
production [8,9]. However, the ethylene productivity achieved
with these catalysts is probably still low for their industrial
implementation.
NiO-based mixed oxides are also potential catalysts for ODHE
[18–21], especially Ni–Nb–O materials [11–14]. Nickel is a low-
cost metal capable of activating ethane at very low temperatures.
The incorporation of niobium makes this catalyst more selective.
A drastic decrease in surface area has been observed after catalytic
tests, but it only causes a small decrease in catalytic performance.
After 100 h on line, the surface area of a Ni–Nb–O catalyst de-
creased 40%, while the ethane conversion only dropped from 42%
to 38%, with the selectivity to ethylene remaining constant [18].
Recently, it has been reported that alumina-supported Ni–W–O
mixed oxides are also active and relatively selective in ODHE [21].
As with Ni–Nb–O catalysts, one interesting feature of this catalytic
system is that ethylene deep oxidation during the oxidative dehy-
drogenation of ethane can be minimized by controlling the catalyst
composition.
major feedstock for polymers. Global ethylene production capacity
as of January 2009 was 126.7 million tons per year [1], and it is ex-
pected to increase significantly in the near future. However, ethyl-
ene is currently produced in addition to propylene by steam
cracking of hydrocarbons, the most energy-consuming process in
the chemical industry [2]. Oxidative dehydrogenation of lower al-
kanes is an interesting alternative to the current industrial pro-
cesses (steam cracking or catalytic dehydrogenation) for the
production of olefins [3–6]. In fact, ethylene could be produced
by the oxidative dehydrogenation of ethane (ODHE), an exothermic
process operating at about 400 °C in which catalyst deactivation by
coke can be minimized because of the presence of molecular oxy-
gen as an oxidant in the reactor feed. However, the development of
both active and selective catalytic materials proves to be a very
hard task, since the desired products, olefins, are usually more
reactive than the corresponding alkanes, so that the formation of
carbon oxides is the thermodynamically favored route [3–6].
From an industrial point of view, it has been proposed that eth-
ylene yields between 65% and 70% are required to compete with
the steam-cracking process [3,6]. Several catalytic systems have
⇑
Corresponding author.
E-mail address: jmlopez@itq.upv.es (J.M. López Nieto).
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.02.010

B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
29
The goal of this work is the synthesis, characterization, and
investigation of the catalytic behavior of bulk Ni–W–O mixed oxi-
des in the ODHE. It will be shown that the variation of the selectiv-
ity to ethylene with ethane conversion and the physicochemical
properties (using appropriate characterization techniques) depend
strongly on the catalyst composition. In addition, it will be shown
that both primary and consecutive nonselective steps in this reac-
tion can be tuned by changing the Ni/W ratio in the catalysts.
change experiments (TPIE), the catalyst (0.162 g) was subjected
1
8
ꢀ1
to a 10%
2
O /Ar flow (22 ml min ), and the temperature was
raised from 25 to 650 °C at a heating rate of 10 °C/min. The concen-
tration profiles of the exit gas composition were obtained by
1
6
acquiring the mass spectra signals relative to
O
2
(m/e = 32),
1
6
18
18
O
O (m/e = 34) and
O (m/e = 36). Blank run experiments were
2
carried out using an empty reactor in order to check contributions
of the gas-phase reactions and stability of the mass spectrometer.
2.3. Catalytic tests
2
. Experimental
The catalytic experiments were carried out at atmospheric pres-
sure, in the temperature range 200–475 °C, mainly at 400–450 °C,
2.1. Catalyst preparation
using a fixed-bed quartz tubular reactor (i.d. 20 mm, length
Ni–W–O mixed metal oxide catalysts were prepared through
4
00 mm). The feed consisted of a mixture of C
2 6 2
H /O /He with mo-
the evaporation at 90 °C of aqueous solutions of nickel nitrate
and ammonium tungstate. NiO and WO were obtained by evapo-
lar ratio of 30/30/40. Several catalyst weights (from 0.05 to 5 g) and
3
ꢀ1
total flows (from 20 to 150 ml min ) were studied. Catalyst sam-
ples were introduced into the reactor diluted with silicon carbide
in order to keep constant volume in the catalytic bed. Reactants
and products were analyzed by gas chromatography using two
packed columns: (i) molecular sieve 5 Å (2.5 m) and (ii) Porapak
Q (3 m). Blank runs showed no conversion at a reaction tempera-
ture of 475 °C.
ration of the corresponding aqueous solutions at 90 °C. The solids
were dried overnight in a furnace at 120 °C and finally calcined
in static air for 2 h at 500 °C. The catalysts will be named NiWn,
in which n corresponds to the atomic ratio W/(Ni + W).
The NiWO single phase was prepared similarly to the other Ni–
4
W–O catalysts by calcining the samples twice, first at 500 °C and
later at 600 °C. The composition observed was that corresponding
to NiWO
.2. Catalyst characterization
Catalyst surface areas were determined by multipoint N
4
, i.e., presenting a Ni/W atomic ratio of 1.
3
. Results and discussion
2
3.1. Catalyst characterization
2
adsorption at ꢀ196 °C using the BET method.
Table 1 lists the characteristics of catalysts. Chemical analysis of
Powder X-ray diffraction (XRD) was used to identify the crystal-
line phases present in the catalysts. An Enraf Nonius FR590 sealed
W and Ni is in good agreement with the nominal composition. On
the other hand, it can be seen that both pure NiO and WO present
3
2
ꢀ1
tube diffractometer with a monochromatic Cu K
ing at 40 kV and 30 mA was used.
Temperature-programmed reduction (TPR) was carried out in a
Micromeritics Autochem 2910 equipped with a TCD detector, in
a
1
source operat-
low surface areas (4 and 9 m g , respectively). However, mixed
oxides show higher surface areas even in the case of the lowest
W loadings. The sample with a W/(Ni + W) atomic ratio of 0.26 pre-
sents the highest surface area, ca. 57 m g . This enhanced surface
area of mixed Ni–W–O catalysts is likely due to the presence of a W
heteroatom that hinders the crystallization of NiO, preventing the
formation of large particles.
2
ꢀ1
which the reducing gas was 10% H
5
2
in Ar (total flow rate
0 ml min ). The temperature range explored was from room
temperature to 800 °C. The heating rate was maintained at
ꢀ1
ꢀ1
1
0 °C min
X-ray photoelectron spectroscopy (XPS) measurements were
taken on a SPECS spectrometer with an MCD-9 detector and using
a nonmonochromatic Al K (1486.6 eV) X-ray source. Spectra were
recorded using analyzer pass energy of 50 V and an X-ray power of
.
XRD patterns of calcined samples are shown in Fig. 1, while the
description of crystalline phases is summarized in Table 1. For
comparison, the XRD patterns of pure NiWO , with the main dif-
4
fraction peaks at 2h = 31.0°, 54.7°, 65.9°, 36.6°, 41.8°, 19.4°, 24.0°,
and 25.0° [JCPDS: 15-0755] (Fig. 1, pattern h) have also been in-
cluded. A W-free sample presents diffraction peaks at 2h = 37.3°,
43.3°, and 62.9°, characteristic of cubic NiO [JCPDS: 78-0643]
(Fig. 1, pattern a), while an Ni-free sample (i.e., NiW1) presents
main diffraction peaks at 2h = 23.19°, 23.59°, 24.38°, and 34.14°,
which correspond to the (0 0 2), (0 2 0), (2 0 0), and (2 2 0) crystal-
a
ꢀ9
2
00 W and under an operating pressure of 10 mbar. Spectra
treatment was performed using the CASA software. Binding ener-
gies (BE) were referenced to O1s at 361.5 eV.
FTIR spectra were collected with a FTS-40A BioRad spectrome-
ter equipped with a DTGS detector (4 cm–1 resolution, 32 scans).
An IR cell allowing in situ treatment under controlled atmospheres
and temperatures from ꢀ176 to 500 °C was connected to a vacuum
system with a gas dosing facility. Self-supporting pellets (ca.
lographic planes, respectively, of monoclinic WO
3
[JCPDS: 43-
1035] (Fig. 1, pattern g). For mixed metal oxides, the XRD patterns
suggest an important decrease in the crystal sizes of both NiO and
WO and the appearance of new crystalline phases.
3
In the Ni-rich region (i.e., W/(Ni + W) ratio lower than 0.4), NiO
is formed over almost all the catalysts, although the appearance of
new diffraction peaks can also be seen at 2h = 14.4°, 25.1°, 28.9°,
32.7°, 47.1°, 53.6°, 56.0°, 57.7°, and 61.7°, whose intensities in-
crease with the W content up to W/(Ni + W) ratios of 0.36. Similar
diffraction peaks have also been observed when a stoichiometric
–
2
1
0 mg cm ) were prepared from the sample powders and treated
ꢀ1
at 250 °C in oxygen flow of 20 ml min for 1.5 h, followed by evac-
ꢀ4
uation at 10 mbar at the same temperature for 1 h. After activa-
tion, the samples were cooled to ꢀ176 °C under dynamic vacuum
conditions, followed by CO dosing at increasing pressure (0.4–
8
.5 mbar). IR spectra were recorded after each dosing.
Oxygen isotopic-exchange experiments were conducted using a
quartz microreactor coupled to a quadrupole mass spectrometer
mixture of H26
N
6
O
41
W
12ꢁ18H
2
O and Ni(NO
3
)
2
ꢁ6H
2
O (Ni/W molar
(
Omnistar QMG 220 M1). Before each experiment, the catalyst
ratio of 1) at 400 °C was treated in air, and it has been related to
tungsten trioxide of low crystallinity [22].
16
was pretreated in 50%
2
2
O /Ar flow (36 ml/min) at 450 °C for
16
.5 h, followed by cooling to 150 °C in the same 10%
O
2
/Ar flow.
On the other hand, no changes in the NiO lattice constants have
6
+
Once 150 °C was attained, oxygen was replaced by argon
been observed (Table 1), suggesting that W ions are not incorpo-
rated into the lattice of the NiO crystals. However, a broadening of
the diffraction peaks related to NiO is clearly observed in the XRD
ꢀ1
(
20 ml min ) and keep at that temperature for 1.5 h before being
cooled to 25 °C. For the temperature-programmed isotopic-ex-

3
0
B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
Table 1
Characteristics of Ni–W–O mixed metal oxide catalysts.
BET (m2
g
ꢀ1
)
Crystalline phasesb
NiO lattice constantc (Å)
Apparent activation energy
Catalysts
W/(Ni + W)
Atomic ratio
S
a
d
ꢀ1
E
a
ethylene (kJ mol
)
NiW0
0
3.8
NiO
4.180
4.181 (1)
4.183 (1)
4.178 (2)
106
109
96
91
91
e
NiW0.1
NiW0.2
NiW0.3
NiW0.36
NiW0.45
NiW0.5
NiW0.7
NiW1
0.09
0.16
0.26
0.33
0.41
0.47
0.68
1.0
35.7
37.9
57.1
50.7
38.1
42.9
29.0
9.4
NiO + WO
NiO + WO
NiO + WO
NiO + WO
NiWO
NiWO
NiWO
x
x
x
x
e
e
+ NiWO
4
4.178 (3)
4
4
4
+ WO
+ WO
+ WO
3
3
3
–
–
–
–
nd
117
110
nd
WO
3
a
b
c
Ni and W contents were determined by atomic absorption spectroscopy.
Determined by XRD and Raman spectroscopy.
Between brackets is the calculated error for the last significant cipher.
d
e
Apparent activation energy (in kJ/mol) for the ODH of ethane; nd = not determined.
WO could also be nanoparticles of undoped and NiO-doped WO
x
x
.
x2
h
g
x4
h
g
f
x4
x8
e
d
f
x2
x2
x4
x4
e
c
d
c
b
x4
x4
x1
a
b
a
200
400
600
800
Temperature (ºC)
2
Fig. 2. H -TPR curves of Ni–W–O catalysts: (a) NiW0; (b) NiW0.1; (c) NiW0.2; (d)
20
40
60
NiW0.3; (e) NiW0.36; (f) NiW0.45; (g) NiW0.7; (h) NiW1.
2
θ/ degree
quence of the interaction between NiO particles (with small crystal
sizes) and tungsten oxide nanoparticles (surface area significantly
increases with W addition) [24]. In addition to this, new bands of
less reducible species begin to appear. Thus, the intensity of the
band at around 400–460 °C decreases as the tungsten amount is in-
creased, while the intensity of the peaks at higher temperature in-
creases. The reduction peaks appearing at temperatures higher
than 460 °C are clearly visible and mainly related to the reduction
in tungsten oxide species. However, some contribution of NiO par-
ticles with different properties to the reduction bands centered ca.
500 and 600 °C, due to the interaction with surrounding particles
of nonstoichiometric tungsten oxide, cannot be rejected (as ob-
served similarity to reported TPR of supported NiO with no reduc-
tion event apparently associated with the support) [25,26].
Fig. 1. XRD patterns of Ni–W–O catalysts: (a) NiW0; (b) NiW0.2; (c) NiW0.3; (d)
NiW0.36; (e) NiW0.45; (f) NiW0.7; (g) NiW1; (h) NiWO . Symbols: NiO ( ); WO
h); NiWO ); WO monoclinic (s). (For interpretation of the references to colour
4
x
(
4
(
3
in this figure legend, the reader is referred to the web version of this article.)
patterns of W-containing catalysts, which suggests a large de-
crease in the crystal size of NiO crystals. In the W-rich region
(
4 3
i.e., W/(Ni + W) ratios higher than 0.45), NiWO and WO are
mainly observed, although the intensity of the corresponding
phases depends on the Ni/W ratio.
Fig. 2 shows the TPR patterns of the Ni–W–O catalysts. The W-
free sample, NiW0, shows a peak with maximum reduction at ca.
3
35 °C, attributed to the reduction in bulk NiO according to litera-
ture data [23]. A shoulder at 360 °C is also observed, which is prob-
ably due to some large particles present in this sample [24].
In W-doped catalyst, i.e., W/(Ni + W) < 0.45, the catalyst reduc-
ibility decreases as the maximum NiO reduction peak shifts to
higher temperature. The shift of the former peak is likely a conse-
In the case of W-rich samples, i.e., 0.45 < W/(Ni + W) < 0.7, the
bands related to the reduction in NiO practically disappear
(Fig. 2, patterns f and g). Moreover, the TPR profiles become more
complex, since the remaining peaks broaden and strongly overlap,
and bands also appear at temperatures higher than 700 °C. In this

B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
31
way, NiWO
4
is clearly observed by XRD, which shows a character-
with NiO and the others to the presence of NiWO
4
and monoclinic
istic reduction peak at ca. 710 °C, as previously described [23]. The
Ni-free sample (Fig. 3, pattern h) shows, in agreement with previ-
ous work [23,27], a broad band with a shoulder at ca. 600 °C and
one peak at ca. 700 °C that spreads beyond 800 °C. These bands
WO , whose intensities depend on the W/(Ni + W) ratio.
3
The results of XPS analysis of Ni–W–O catalysts are summarized
in Table 2. It can be observed that the compositions of bulk and
surface are similar for all the catalysts, without a clear outer
enrichment of nickel or tungsten.
Fig. 3A shows the Ni2p3/2 spectra of the Ni–W–O catalysts. For
4
comparison, the spectra of NiW0 (i.e., W-free NiO) and NiWO have
5
+
are related to the reduction in tungsten oxide forming first W
4
+
species, then W , and finally tungsten species with an apparent
oxidation number of 2+ (derived from the transformation of the
corner-linked WO
6
octahedra along the shear planes into an
also been included (Fig. 3B). In all Ni–W–O samples, the Ni2p3/2
XPS main peak shows two additional structures at the high BE side,
named satellite peaks: one at ꢀ1.5 eV higher BE (Satellite I) and the
other at ꢀ7 eV higher BE (Satellite II). A similar spectrum is also ob-
served in the NiO sample, as previously reported in the literature
[11]. The origin of Satellite I is a matter of controversy. It has been
arrangement of edge-sharing octahedra) [27]. The relative intensi-
ties, as well as the temperature of the maxima of these bands,
change with the W content, probably due to variations in the
WO
x 4
particle size and their interaction with NiO and/or NiWO
particles.
3
+
2+
Accordingly, the incorporation of tungsten favors a partial de-
crease in the reducibility of Ni sites and the appearance of at least
three types of W sites: one is related to WO
related to the presence of Ni ions [28–30], Ni –OH species [31],
2
+
or Ni vacancies [32] or to a nonlocal screening mechanism [33–
35]. On the other hand, the satellite line at 7 eV higher BE involves
x
species interacting
A
main line
sat I
sat II
B
a
b
NiW0
c
d
e
f
NiWO4
x2
x2
890
880
870
860
850
890
880
870
860
850
BE (eV)
BE(eV)
Fig. 3. X-ray photoelectron spectra of Ni2p3/2 of Ni–W–O catalysts. (A) (a) NiW0.1; (b) NiW0.2; (c) NW0.3; (d) NiW0.36; (e) NiW0.5; (f) NiW0.7. (B) For comparison, the
spectra of NiW0 and NiWO are also presented.
4
Table 2
XPS results for Ni–W–O mixed oxide catalysts.
Sample
W/(Ni + W)
Bulk
W/(Ni + W)
Surface
Ni2p3/2
W4f7/2
Main peaka (eV)
Satellite,
I)
D
BEb (eV)
(II)
Satellite/main-peak intensity ratio
Main peaka (eV)
Secondary peak (eV)
(
(I)
(II)
NiW0
0
0
854.3
856.1
856.0
856.1
856.0
857.1
857.1
857.0
–
2.3
2.0
2.0
1.9
1.7
2.0
2.2
2.0
–
8.0
6.9
6.8
7.0
7.1
6.4
6.4
6.5
–
4.07
0.65
0.41
0.86
1.41
0.29
0.24
0.53
–
3.9
–
–
NiW0.1
NiW0.2
NiW0.3
NiW0.36
NiW0.45
NiW0.5
NiW0.7
NiW1
0.09
0.16
0.26
0.33
0.41
0.47
0.68
1.0
0.10
0.13
0.20
0.25
0.51
0.52
0.61
1.0
1.28
0.93
1.35
1.83
0.92
0.85
1.21
–
36.7
36.9
36.5
36.4
36.4
36.3
36.2
36.8
36.3
37.2
35.7
37.3
–
–
–
37.6
NiWO
4
0.50
0.49
857.0
1.7
6.3
0.32
1.07
36.6
a
Estimated experimental error of ±0.1 eV.
BE shift of the satellite peak versus the main peak.
b

3
2
B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
a ligand–metal charge transfer [33–35]. The intensity and position
of both satellite lines are very sensitive to the structural arrange-
ment and the nature of the surrounding atoms, giving interesting
information about the local environment of the Ni surface atoms.
Table 2 shows the BE of the main line, as well as the separation
and intensity ratio of the satellite peaks versus the main peak. A
shift to higher BE in the Ni2p3/2 XPS main line of Ni–W–O catalysts
versus pure NiO is clearly observed, which should be related to the
presence of tungsten atoms in the structure of the catalysts. More-
over, in the case of samples with W/(Ni + W) ratios higher than 0.5,
the Ni2p3/2 BE shifts ca. 1 eV to higher energy (857.0 eV) from the
rest of the samples (856.0 eV), which could be related to the pres-
ence of NiWO
agree with the presence of NiWO
(
4
. Indeed, Raman (not shown here) and XRD results
in Ni–W–O samples with a W/
Ni + W) ratio higher than 0.50. In the case of Ni-rich samples,
4
3
3
2
2
1
1
.5
.0
.5
.0
.5
.0
0
.10 < W/(Ni + W) < 0.36, the BE of the Ni2p3/2 XPS main line does
not shift significantly, while the intensity and position of both sa-
tellite peaks clearly change with increasing tungsten content in the
sample. In fact, the satellite (I + II)-main peak intensity ratio fol-
lows a Gaussian-type curve vs the tungsten content in the samples,
with a maxima at a W/(Ni + W) ratio of 0.30–0.36 (Fig. 4). This
could account for a different local environment of the Ni atoms
in the bulk structure, influencing the Ni–O interaction.
The W4f XPS spectra of the Ni–W–O catalysts, together with
those of NiW1 (i.e., NiO-free WO
Fig. 5. A progressive shift to lower BE in the W4f7/2 line, compared
with that of bulk WO , with increasing tungsten content in the Ni–
3 4
) and NiWO , are shown in
3
0
.1
0.2
0.3
0.4
0.5
0.6
0.7
W–O catalysts is clearly observed (Table 2), which should be re-
lated to the presence of Ni atoms in the local environment of
W/(Ni+W) atomic ratio
6
+
W
species [36,37].
Fig. 4. Variation of the XPS Ni2p3/2 satellite (I + II)–main peak intensity ratio with
the W/(Ni + W) atomic ratio Ni–W–O mixed oxides.
A
a
b
B
c
NiW1
d
e
NiWO4
4
4
40
BE (eV)
36
32
44
40
36
32
BE (eV)
Fig. 5. X-ray photoelectron spectra of W4f transition of Ni–W–O catalysts. (A) (a) NiW0.1; (b) NiW0.2; (c) NiW0.3; (d) NiW0.36; (e) NiW0.7. (B) For comparison, the spectra of
NiW1 and NiWO are also presented.
4

B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
33
On the other hand, at a high W/(Ni + W) molar ratio, a W4f7/2
BE similar to that observed on NiWO is observed. Moreover, a lit-
tle contribution of W at higher BE (around 37.2 eV) is observed in
almost all samples, which could be due to highly dispersed poly-
meric tungsten species in a different environment, as suggested
from the Raman results (not shown).
Since XPS is not a highly surface-sensitive technique, more spe-
cific characterization techniques have been chosen to analyze the
nature of catalytic surface sites present on the Ni–W–O samples.
For this reason, samples with different catalytic behavior have
been analyzed by IR spectroscopy of CO adsorption. CO has been
widely used for characterization of different types of surface Lewis
acid sites, due to the high sensitivity of this molecule. In our study,
two samples, named NiW0.36 and NiW0.7, have been selected. For
In the case of pure WO
3
, i.e., the NiW1 sample (spectrum not
4
shown), no IR bands associated with carbonyl species are observed
after CO adsorption, indicating a small number of Lewis surface
sites.
CO adsorption on NiWO
4
(Fig. 6B) shows an intense IR band
ꢀ1
with two components at 2188 and 2176 cm , together with an-
ꢀ1
ꢀ1
other band at 2156 cm and a shoulder at 2140 cm . The IR band
ꢀ
1
at 2188 cm is related to CO interacting with Lewis acid sites, the
ꢀ1
2176 and 2156 cm IR bands are associated with CO interacting
with OH groups, according to the respective shift in the OH groups,
ꢀ
1
and the 2140 cm IR band is due to physisorbed CO. The greater
ꢀ1
shift of the CO frequency in the 2176 cm IR band compared with
ꢀ
1
the 2156 cm band points to higher acidity of the OH group in the
ꢀ
1
former. Indeed, two hydroxyl groups at 3666 and 3631 cm are
observed in the OH region, which are shifted due to CO adsorption,
4
comparison, NiW0, NiW1 and NiWO samples have also been ana-
lyzed and used as reference compounds.
ꢀ1
giving rise to the corresponding 2176 and 2156 cm IR bands.
On the NiW0.36 sample, two intense IR bands at 2189 and
The IR spectrum in the carbonyl stretching frequency region
ꢀ
1
ꢀ1
ꢀ1
(
2250–2050 cm ) of the NiW0 sample, i.e., pure NiO, after CO
2156 cm together with a shoulder at 2143 cm are observed
ꢀ1
ꢀ1
adsorption shows the presence of one intense band at 2153 cm
Fig. 6A). The band at 2153 cm is associated with CO interacting
(Fig. 6C). The shoulder at 2143 cm is ascribed to physisorbed
CO. The IR band at 2156 cm can be assigned to CO interacting
ꢀ1
ꢀ1
(
with OH groups, according to the corresponding shift of the OH
groups and the very low stability of the IR band [38]. The absence
of Lewis acid sites observed in the NiW0 sample could be related in
part to their low surface area. On the other hand, fast hydroxyl-
ation of surface sites due to water by cooling of the pellet under
the experimental IR procedure has been observed (inset of
Fig. 6A), which could also account for the absence of Lewis acid
sites detected on this sample.
with OH groups [38], while the higher frequency band at
2189 cm is associated with CO interacting with Lewis acid sites
on the catalysts’ surface [39,40].
ꢀ1
On the NiW0.7 sample, three IR bands are observed at
ꢀ
1
2+
ꢀ1
2200 cm , associated with Ni Lewis acid sites, 2164 cm , asso-
ciated with OH groups according to the corresponding shift in the
ꢀ1
hydroxyl region, and 2143 cm , due to physisorbed CO (Fig. 6D).
A
B
C
D
Fig. 6. FTIR spectra at ꢀ176 °C of NiW0 (A), NiWO
A): IR spectra in the hydroxyl region of (a) NiO sample after activation at 250 °C, (b) after cooling down to 25 °C, (c) after 8.5 mbar CO adsorption. Inset in (B–D): changes in
the IR spectra in the hydroxyl region after 8.5 mbar CO adsorption.
4
(B), NiW0.36 (C), and NiW0.7 catalysts (D) after CO adsorption at increasing coverage (a–g or a–h, 0.4–8.5 mbar). Inset in
(

3
4
B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
According to the IR spectra, the presence of tungsten in the Ni–
A
B
C
W–O samples leads to the formation of Lewis and Brønsted acid
sites. The Lewis acidity of the samples increases in the order Ni-
ꢀ
1
ꢀ1
ꢀ
WO
4
= NiW0.36 (2186 cm ) < NiW0.7 (2200 cm ), while the
1
Brønsted acidity increases as follows: NiO (2153 cm ) < NiW0.36
(2164 cm ) < NiWO
176 cm ). The higher acidity of surface sites could explain the
ꢀ1
ꢀ1
(
2
2156 cm ) < NiW0.7
4
(2156
and
ꢀ
1
appearance of consecutive reactions during the oxidation of ethane
over W-containing samples, as will be shown later.
On the other hand, temperature-programmed 8O
1
isotope-ex-
2
change measurements (TPIE) make it possible to investigate the
nature of oxygen species participating in the reaction mechanism
and the type and density of surface sites available for oxygen acti-
vation. Indeed, the distribution of isotopic-exchange products
1
6
16 18
(
O
2
and
O O) depends on the relative rates of oxygen dissoci-
ation, surface exchange, lattice incorporation, and diffusion into
the bulk [14]. Thus, fast oxygen dissociation followed by rapid sur-
face oxygen exchange compared with the oxygen diffusion rate
1
6 18
into the bulk leads to the formation of
O O, while a faster incor-
poration reaction and subsequent diffusion step into the bulk re-
1
6
sult in
2
O as the dominant product [41]. The TPIE profiles in
the temperature range 25–650 °C for NiW0, NiW0.36, and
NiW0.7 samples are shown in Fig. 7. The oxygen isotopic-exchange
process starts at a lower temperature (320 °C) on the NiW0 sample
than on the Ni–W–O mixed oxides (500 °C), indicating that the
reactivity of exchangeable oxygen species is higher on the W-free
sample (NiW0) than those achieved for W-containing samples.
This agrees with a greater ability of the NiW0 sample for oxygen
activation (i.e., dissociation) and a higher activity of these oxygen
species.
The amount of exchangeable oxygen species (16 and 16O18O)
O
2
normalized to Ni atoms is shown in Fig. 8. On the NiW0 sample, ra-
pid surface oxygen exchange is observed, which can be attributed
to fast oxygen dissociation on surface defects, leading to the forma-
tion of highly active surface oxygen species. In agreement with the
TPIE results, the dissociation of oxygen on surface defects is elim-
inated by the incorporation of tungsten into the catalyst. In addi-
tion, the reactivity of surface oxygen species is decreased,
according to the higher Tonset for oxygen exchange. This agrees
with TPR results in which the addition of W species decreases
the reducibility of Ni surface sites. On the other hand, different dis-
tributions of isotopic oxygen species are observed on both
NiW0.36 and NiW0.7 samples. Larger amounts of doubly ex-
1
6
changed oxygen species (
gen species (
O
2
) compared with cross-labeled oxy-
Fig. 7. Temperature-programmed 18O
A), NiW0.36 (B), and NiW0.7 (C).
isotopic oxygen exchange profiles of NiW0
16 18
2
O O) are observed on the NiW0.36 sample,
(
8
3
A
B
6
4
2
1
6
6
16O2
O2
2
1
1
18O
16
O O
18
O
0
0
2
00
300
400
500
600
700
400
500
600
700
Temperature (ºC)
Temperature (ºC)
Fig. 8. Evolution of the oxygen exchange species per Ni atom during TPIE experiments on NiW0 (
,
), NiW0.36 (
,
), and NiW0.7 (
,
) catalysts. Symbols: 16O
2
(j,
,
);
16 18
O
O (
,
,
). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
35
indicating faster diffusion of oxygen into the bulk and a smaller
amount of oxidizing surface oxygen species. However, the preva-
lent formation of cross-labeled oxygen species for NiW0.7 indi-
cates the presence of a larger amount of surface oxidizing species
than of doubly exchanged oxygen species in the same catalyst. This
type of surface oxygen species has been referred to in the literature
conversion increases from the NiW0 sample at ca. 310 °C to more
than 450 °C for samples with W/(Ni + W) atomic ratios of 0.36
and higher.
Fig. 10 shows the variation of the rate of formation of ethylene
ꢀ
1
ꢀ1
per unit mass of catalyst (STYC2H4, in gC2H4
h
g
), the activity per
cat
ꢀ
1
ꢀ1
gram of nickel (in gC2H6
s
g
), and the catalytic activity per Ni
Ni
as electrophilic oxygen species and as responsible for CO
tion [14].
2
forma-
atom normalized per surface area of catalyst (in
g
C2H6
ꢀ1
ꢀ2 ꢀ1
Ni
m
h ) with the W content in Ni–W–O catalysts. It can be
observed that in all three cases, the activity (per mass of catalyst,
per surface area, or per both mass and surface area) decreases
when the tungsten loading of the catalyst increases. Therefore,
the catalytic activity of these catalysts depends on something apart
from the surface area and the amount of nickel.
3
.2. Catalytic performance in ethane oxidation
Fig. 9 shows the variation of ethane conversion with the reac-
tion temperature achieved over Ni–W–O mixed oxide catalysts. It
can be seen that the temperature necessary to achieve 10% ethane
Fig. 11 shows the variation of the selectivity to ethylene with
the ethane conversion achieved on Ni-rich (Fig. 11a) and W-rich
(
Fig. 11b) catalysts. A completely different pattern is observed for
the two types of catalysts. In the case of Ni-rich catalysts, i.e.,
< W/(Ni + W) < 0.36 (Fig. 11a), the initial selectivity to ethylene
selectivity at very low ethane conversion) increases with increas-
ing W content. Furthermore, in the range of ethane conversion
studied, these catalysts do not show appreciable variation of the
selectivity to ethylene with the ethane conversion, except in the
case of samples NiW0.3 and NiW0.36 (the most selective cata-
lysts), in which a very small decrease in selectivity to ethylene
takes place with ethane conversion (the selectivity to ethylene
drops from ca. 60% to ca. 50% when the ethane conversion is in-
creased from 5% to 35%). It is clear that in this case, the reaction
network can be explained by parallel and consecutive reactions
0
(
4
3
2
1
0
0
0
0
0
(Scheme 1a), in which the incorporation of small amount of W
atoms favors a decrease in k and an increase in the k /k ratio.
2
1
2
In the case of W-rich catalysts, i.e., samples with 0.45 < W/
Ni + W) < 1 (Fig. 11b), the initial selectivity to ethylene (at very
(
low ethane conversion) is very high in all cases. In addition, the
selectivity to ethylene decreases with the ethane conversion (this
loss of selectivity is also greater when the amount of W in the cat-
alyst is increased). On the other hand, in this type of catalyst, a
reaction network based on both parallel and consecutive reactions
240
280
320
360
400
440
480
Temperature,ºC
Fig. 9. Variation of ethane conversion with the reaction temperature during the
oxidation of ethane over Ni–W–O mixed oxides. Symbols: NiW0 (j); NiW0.1 ( );
NiW0.2 ( ); NiW0.3 ( ); NiW0.36 ( ); NiW0.45 ( ); NiW0.5 ( ); NiW0.7 ( );
NiW1 ( ). (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)
2 3 1
can be proposed (Scheme 1b), in which k is very low and the k /k
ratio increases with increasing W loading. The initial selectivity
reaches a maximum for a W/(Ni + W) atomic ratio of about 0.5,
although for this catalyst an important selectivity loss with ethane
conversion is observed. In any case, the best behavior at high
r
1
E-5
E-6
a
b
1E-3
1E-4
1E-5
1E-6
1E-7
1E-3
1E-4
1E-5
1E-6
1E-7
1
1E-7
1
E-8
E-9
1
0
20
40
60
80
100
0
20
40
60
80
100
W/(Ni+W) ratio, at.%
W/(Ni+W) ratio, at. %
Niꢀ
1
ꢀ1
Fig. 10. Variation of the formation rate of ethylene, STY (in gC2H4/gcat/h) (j), catalytic activity for ethane conversion normalized gram of nickel (in gC2H6
g
h
) ( ), and
) ( ) with the W content of catalysts achieved during the
ODHE over Ni–W–O catalysts at 425 °C. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
ꢀ
1
ꢀ2 ꢀ1
catalytic activity for ethane conversion per Ni atom normalized per surface area of catalyst (in gC2H6 Ni
m
h

3
6
B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
8
6
4
2
0
0
0
0
0
80
a
b
60
40
20
0
NiWO4
0
10
20
30
40
0
10
20
30
40
Ethane conversion, %
Ethane conversion, %
Fig. 11. Variation of selectivity to ethylene with the ethane conversion achieved at 400 °C over Ni-rich (a) and W-rich (b) catalysts. Symbols: NiW0 (j); NiW0.1 ( ); NiW0.2
); NiW0.3 ( ); NiW0.36 ( ); NiW0.45 ( ); NiW0.5 ( ); NiW0.7 ( ); NiW1 ( ). For comparison, the results achieved over pure NiWO4 ( ) have also been included. (For
(
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
k3
k1
k2
C H
k1
C H
2
4
2
4
CO + CO2
C H
C H
2 6
2
6
k3
k2
CO2
CO2
Ni-rich catalysts
W-rich catalysts
Scheme 1. Reaction networks for the oxidative dehydrogenation of ethane over Ni-rich and W-rich Ni–W–O catalysts.
ethane conversion is obtained for catalysts with lower tungsten
3.3. Catalytic performance in ethylene oxidation
content, in which the W/(Ni + W) ratio is between 0.30 and 0.36.
Fig. 12 shows the variation of the selectivity to ethylene, CO,
Ethylene oxidation experiments have been conducted on some
representative Ni-rich (NiW0.2 and NiW0.3) and W-rich (NiW0.5
and NiW0.7) catalysts (Table 3). CO and CO were only detected
2
as reaction products, although the relative amount of each carbon
oxide was demonstrated to depend on the catalyst composition.
and CO
loading, NiW0.2, shows no influence of the ethane conversion on
the selectivity to ethylene and CO , and the formation of CO is
2
on four characteristic catalysts. The sample with low W
2
not observed. However, over NiW0.36, a small decrease in the
selectivity to ethylene with ethane conversion can be seen, while
the formation of CO is clearly observed (especially at high ethane
conversion).
Finally, for samples with higher W loading, i.e., NiW0.5 and
NiW0.7, the formation of CO clearly increases with ethane conver-
sion and the W loading of the catalyst, while the selectivity to eth-
ylene presents an opposite trend (Fig. 12). According to these
results, it can be said that CO is mainly formed from ethylene, by
consecutive oxidation, ethylene is formed directly from ethane
Thus, the Ni-rich catalyst (NiW0.2) showed mainly CO
2
with selec-
tivity 98–100%. The selectivity to CO achieved by the NiW0.3 sam-
2
ple was ca. 75%; for samples NiW0.5 and NiW0.7, it was ca. 45%
and ca. 30%, respectively. Thus, it can be observed that the higher
the W content of the catalyst the lower was the selectivity to
CO
ferent carbon oxides obtained remained almost constant regard-
less of the ethylene conversion, suggesting that (i) CO and CO
2
obtained. It must be mentioned that the selectivity to the dif-
2
2
are primary products and (ii) CO is not formed by the further oxi-
2
and CO can be formed either from ethane (main) or from ethylene
dation of CO.
decomposition. However, the extension of parallel and consecutive
reactions strongly depends on the W/(Ni + W) ratio in the catalyst.
The apparent activation energy for the direct transformation of
ethane to ethylene over Ni–W–O catalysts is shown in Table 1. The
apparent energy for ethylene formation varies depending on the
catalyst (from 91 to 117 kJ mol for catalysts with a W/(Ni + W)
ratio between 0.20 and 0.36, those that present the lowest ener-
gies, similar to that observed for NiO).
Table 3 compares the catalytic activity for ethane and ethyl-
ene oxidation. It can be seen that in Ni-rich samples, the relative
activity between ethane and ethylene oxidation is much higher
than that achieved over W-rich catalysts. While the ethylene con-
version is similar on NiW0.2, NiW0.3, and NiW0.5 (and only
twice more reactive than on NiW0.7), the ethane conversion
greatly decreases in all cases when the tungsten loading in-
creases (the reactivity of NiW0.2 is more than 20 times higher
than that of NiW0.7). Therefore, the different selectivity to ethyl-
ene observed over these catalysts in the ODHE can be explained
by the large differences in the catalytic activation of ethane over
Ni-rich samples with respect to those achieved over W-rich
samples.
ꢀ1
Similar apparent activation energy for ethane oxidation to eth-
ꢀ1
ylene has also been reported for Ni–Nb–O catalyst (96.2 kJ mol
14], although the apparent activation energy for ethane oxidation
to CO was lower than that observed in our catalysts. This could ex-
plain the different behavior with reaction temperature.
)
[
2

B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
37
100
100
NiW0.2
NiW0.36
C H
80
60
40
20
0
80
CO2
60
2
4
40
CO2
C H
2
4
20
0
CO
CO
0
10
20
30
0
10
20
30
Ethane conversion, %
Ethane conversion, %
1
00
100
80
60
40
20
0
NiW0.50
NiW0.7
8
6
4
2
0
0
0
0
0
CO
CO2
C H
2
4
C H
2
4
CO
CO2
0
10
20
30
0
10
20
30
Ethane conversion, %
Ethane conversion, %
2
Fig. 12. Variation of selectivity to ethane ( ), CO ( ), and CO (s) with the ethane conversion achieved at 400 °C over Ni–W–O catalysts: NiW0.2, NiW0.36, NiW0.5, and
NiW0.7. Experimental conditions in text. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 3
Catalytic activity for ethane and ethylene oxidation over Ni–W–O mixed oxide catalysts.
Catalyst
Ethylene oxidation a
Selectivityc (%)
Catalytic activity b
Relative activity (ethane/ethylene)
Ethylene (103 molC2H4
g
catꢀ1 ꢀ1
h
)
Ethane (103 molC2H6
g h )
catꢀ1 ꢀ1
CO
2
CO
NiW0.2
NiW0.3
NiW0.5
NiW0.7
99
75
48
29
1
25
52
71
1.54
1.10
1.45
0.60
7.4
3.7
0.87
0.32
4.8
3.7
0.60
0.53
a
b
c
Reaction conditions in text, reaction temperature = 425 °C.
Catalytic activity for ethylene or ethane oxidation at 425 °C.
The selectivity to carbon oxides is almost constant in the conversion range studied (until 15%).
3.4. General remarks
2
species present in NiO [10,42]. In this way, TPD-O experiments
on NiO/Al catalysts suggested the partial elimination of the
2 3
O
The interaction of NiO with other metal oxides can improve the
more reactive (and unselective) oxygen species without modifica-
tion of the less active but more selective ones [11].
catalytic properties of NiO during the oxidative dehydrogenation of
ethane. This is the case for NiO supported on Al [10,11,21] and
O
2 3
A similar mechanism has been proposed for Nb-promoted NiO
catalysts [13,14,19,20], although it leads to higher production of
ethylene. The incorporation of Nb into the NiO lattice (by substitu-
tion of nickel atoms and/or filling of the cationic vacancies in the
defective nonstoichiometric NiO surface) favors a partial reduction
in the nonstoichiometric NiO crystals abundant on pure nickel
oxide and responsible for the total oxidation of ethane to carbon
dioxide [13]. This has been confirmed by SSITKA (steady state iso-
MgO [17], in which it is possible to achieve selectivity to ethylene
higher than that observed over pure NiO. However, as a conse-
quence of the interaction between the support and NiO particles,
lower catalytic activity for ODHE is observed over supported cata-
lysts, especially in NiO/MgO, for which reaction temperatures
higher than 500 °C are required. The higher selectivity achieved
over alumina-supported NiO catalysts, with high ethane reactivity
at reaction temperatures lower than 450 °C, has been related to the
low affinity of this catalyst for ethylene oxidation [11] or to a selec-
tive partial elimination of one of the two types of active oxygen
topic transient kinetic analysis) experiments with isotopic 18
O
2
[14], suggesting that the dissociation of oxygen, leading to a high
concentration of intermediate electrophilic oxygen species on the

3
8
B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
surface, is the fast step of the exchange process in pure NiO. These
electrophilic oxygen species are active for the total oxidation of
ethane. However, doubly exchanged species were mainly observed
on the Ni–Nb–O catalyst, which suggest that Nb cations in NiO fa-
vor the diffusion of oxygen species in the bulk of the catalyst and
increase the presence of nucleophilic oxygen species (also decreas-
ing the presence/formation of the most oxidizing species). The
importance of nucleophilic oxygen species for oxidative dehydro-
genation of lower alkanes has been proposed in recent years [3–
left). In agreement with the characterization results, NiO crystals
are mainly observed in Ni-rich catalysts (although the XRD pat-
terns show lower intensity of the corresponding reflections), while
x
polymeric WO species can also be deduced from the XRD and Ra-
man spectra (not shown here). High surface areas have also been
observed in catalysts with W/(Ni + W) ratios of 0.30 and 0.36.
Accordingly, these catalysts can be described as NiO in which part
of the unselective sites is eliminated by the deposition of poly-
x
meric tungsten species and/or WO crystals. Indeed, oxygen iso-
5
,14–16].
tope experiments have shown a strong decrease in the number
of surface exchange oxygen species. On the other hand, XPS results
show differences in the coordination environment of Ni atoms
according to their satellite structure. Maximum selectivity has
been observed in samples with maximal satellite-main peak inten-
sity ratio. Finally, the number of acids sites increases with the W
content (Fig. 6).
The catalytic results over Ni–W–O mixed oxides show different
catalytic performance depending on the W/(Ni + W) ratio of the
catalysts, which is related to the different characteristics of the ac-
tive sites in these materials. In fact, the presence of tungsten in Ni–
W–O catalysts strongly modifies the nature of active phases and
the characteristics of active and selective sites on the surfaces of
the catalysts compared with pure NiO. Fig. 10 shows the evolution
In the case of W-rich catalysts, WO
3 4
and NiWO are mainly ob-
of the formation rate of ethylene per unit mass of catalyst (STYC2H4
)
served, the first increasing at higher W/(Ni + W) ratios and the sec-
2
+
and the activity per gram of nickel with the W content of catalysts,
achieved during the ethane oxidation. As can be observed, both the
formation of the olefin and the catalytic activity for ethane oxida-
tion normalized per nickel content decrease when the W loading
ond presenting an opposite trend. It is clear that Ni species in
NiWO are active and relatively selective, while WO shows low
4
3
activity and low selectivity. However, as a consequence of the pres-
ence of different active phases, these catalysts are active in both
ethane and ethylene oxidation, and deep oxidation of ethylene to
CO is also observed. Accordingly, a reaction pathway as that shown
in Scheme 1 (right), with both parallel and consecutive reactions,
can be proposed. The presence of polytungstate species on the sur-
face of the catalysts can also contribute to improve the selectivity
to ethylene in catalysts with W/(Ni + W) ratios between 0.5 and
3
increases. Since WO has been shown to present extremely low
activity, Ni–O sites can be considered as the active centers for
the selective activation of ethane over Ni–W–O catalysts, although
the intrinsic catalytic activity of Ni sites decreases in W-containing
samples.
XP spectra have demonstrated that the concentration of Ni
near the surfaces of the catalysts is quite similar to that of the
bulk, and therefore, the number of available sites on the surface
will depend, apart from the surface area, on the amount of nickel
in the catalyst. However, the different characteristics of the Ni
species must be an important factor in the catalytic behavior of
these catalysts. Also, a modification of the Ni species (variations
in the oxidation state and/or coordination) was observed by the
XPS technique when the W content in the catalyst increased
0.7. However, the presence of WO
ings should have a little influence on catalytic behavior, as sug-
gested by the extremely low activity achieved over pure WO
while NiWO is relatively active in both ethane ODH and ethylene
3
in catalysts with higher W load-
3
,
4
deep oxidation (Fig. 9b). We must note that the ratio between eth-
ane oxidation and ethylene oxidation for the NiW0.7 sample (ca.
0.5) is much lower than those observed for catalysts with low W
loading (ca. 4.8).
(
Fig. 5).
On the other hand, it has been proposed for Ni–Nb–O catalysts
that, apart from improving the dispersion of the nickel phase, nio-
bium facilitates C–H bond activation by acting as an electron trans-
fer promoter [13]. This behavior could also be proposed in our case.
In a recent article, the Me-valence in Ni–Me–O mixed oxides
has been demonstrated to be of paramount importance for the
characteristics of the catalysts and consequently for catalytic per-
formance during ethane oxidative dehydrogenation [15]. Without
going into detail, it can be said that high Me valences favor the
formation of ethylene. In our case, tungsten is mainly present
Taking into account that redox is the most likely mechanism
for the ODHE reaction on these catalysts, lower catalyst
reducibility would account for the decrease in the activity per
nickel site. TPR profiles show a modification of the catalyst
reducibility that explains the low catalytic activity of the W-rich
catalysts (a decrease in both catalytic activity and catalyst reduc-
ibility was observed when the W content increased). This is also
confirmed by the oxygen isotope-exchange experiments, in
which the number and activity of exchangeable oxygen species
decrease considerably upon the addition of tungsten oxide to
NiO.
6
+
as W : (i) as WO
x 4
in Ni-rich catalysts or (ii) as NiWO and
6
+
WO in W-rich catalysts. Thus, W is assumed to be an interest-
3
On the other hand, IR spectra of CO adsorption show a different
nature of Lewis acid sites in the Ni–W–O catalysts depending on
the W content. Indeed, higher acidity of both surface Lewis acid
sites and OH groups is observed in samples with high W content
ing promoter of NiO at W/(Ni + W) atomic ratios in the range 0.3–
0.4.
Finally, it has been proposed that an efficient catalyst for the
ODH reaction should achieve a compromise between the ability
to activate the CꢀH bond and the ability to release olefin [43].
Accordingly, one selective catalyst could be prepared by optimiz-
ing the ethane activation and facilitating the ethylene desorption,
although other possibilities must be considered. Thus, one strat-
egy could be the selective elimination of deep oxidation of ethane
without the formation of new actives sites which could favor the
deep oxidation of ethylene and try to optimize the acidity of Le-
wis surface sites, in order to favor activation of the alkane but
avoid readsorption of the olefin. This is the case of Ni–W–O
mixed oxide and maybe in part in Ni–Nb–O, in the Ni-rich region,
(
3
sample NiW0.7), although it is not observed for pure WO sample.
Accordingly, the addition of tungsten to the catalysts changes the
nature of the Ni species, increases both the number of Lewis acid
sites and the diffusion rate of oxygen species into the bulk of the
catalyst, and decreases the presence of highly active (i.e., electro-
philic) surface oxygen species. All these changes can favor a higher
selectivity to ethylene, decreasing the number of deep oxidation
sites in catalysts.
In the Ni-rich Ni–W–O catalysts, the selectivity to ethylene in-
2
creases while that to CO decreases as the tungsten loading in-
creases. In this case, the formation of CO is negligible and the
selectivity to ethylene remains almost constant with the ethane
conversion. This suggests that deep oxidation of the ethylene
formed does not occur or takes place to a low extent (see Scheme 1,
2
+
with Ni sites in W-doped NiO. However, this is not observed in
2
+
Ni–W–O mixed oxide in the W-rich region, with Ni sites in a
4
less active phase (i.e., NiWO ), which is also active in ethylene
oxidation.

B. Solsona et al. / Journal of Catalysis 280 (2011) 28–39
39
4
. Conclusions
References
[
[
[
1] D.N. Nakamura, Oil Gas J. 107 (28) (2009) 43.
2] T. Ren, M.K. Patel, K. Blok, Energy 33 (2008) 817.
3] F. Cavani, N. Ballarini, A. Cericola, Catal. Today 127 (2007) 113.
The results of this study show that Ni–W–O mixed oxides are
active and relatively selective in the oxidative dehydrogenation
of ethane. Ni–O sites seem to be the active species in these cata-
lysts, although the nature of crystalline phases, the environment
of Ni sites (as well as their physicochemical characteristics), and
the catalytic behavior strongly depend on the catalyst composition.
Pure NiO shows a low selectivity to ethylene but remains al-
most constant when the ethane conversion is increased. This is
consistent with the fact that it shows a higher reducibility and a
greater amount of exchangeable oxygen species than observed
[4] J.M. López Nieto, Top. Catal. 41 (2006) 3.
[
[
5] H.X. Dai, C.T. Au, Curr. Top. Catal. 3 (2002) 33.
6] G. Grubert, E. Kondratenko, S. Kolf, M. Baerns, P. Van Greem, R. Parton, Catal.
Today 81 (2003) 337.
[7] M.D. Argyle, K. Chen, A.T. Bell, E. Iglesia, J. Phys. Chem. B 106 (2002) 5421.
[
8] J.M. López Nieto, P. Botella, M.I. Vázquez, A. Dejoz, Chem. Commun. (2002)
906.
1
[
9] P. Botella, E. García-González, A. Dejoz, J.M. López Nieto, M.I. Vázquez, J.
González-Garcia, J. Catal. 225 (2004) 428.
[10] X. Zang, J. Liu, J. Xie, Appl. Catal. A 240 (2003) 143.
11] E. Heracleous, A.F. Lee, K. Wilson, A.A. Lemonidou, J. Catal. 231 (2005) 159.
12] (a) Liu, US Patent 6891,075, 2005;
[
[
for Ni–W–O samples. On the other hand, pure NiWO
cially WO present high initial selectivity to ethylene, which rap-
idly drops when the ethane conversion increases.
For W-rich N–W–O catalysts, the presence of NiWO and WO
3
4
and espe-
3
(
b) Y. Liu, US Patent 7227,049 A2, 2007.
[13] E. Heracleous, A.A. Lemonidou, J. Catal. 237 (2006) 162.
[
[
[
14] E. Heracleous, A.A. Lemonidou, J. Catal. 237 (2006) 175.
15] E. Heracleous, A.A. Lemonidou, J. Catal. 270 (2010) 67.
16] B. Savonova, S. Loridant, D. Filkova, J.M.M. Millet, Appl. Catal. A 390 (2010)
148.
17] K.-I. Nakamura, T. Miyake, T. Konishi, T. Suzuki, J. Mol. Catal. A 260 (2006) 144.
18] M.L. Rodriguez, D.E. Ardissone, A.A. Lemonidou, E. Heracleous, E. López, M.N.
Pedernera, D.O. Borio, Ind. Eng. Chem. Res. 48 (2009) 1090.
[19] E. Heracleous, A. Delimitis, L. Nalbandian, A.A. Lemonidou, Appl. Catal. A 325
2007) 220.
20] A.A. Lemonidou, Abstract 9th Novel Gas Conversion Symposium, Lyon, 2010,
KN11, p. 36.
[21] B. Solsona, A. Dejoz, M.I. Vázquez, F. Ivars, J.M. López Nieto, Top. Catal. 52
2009) 751.
22] J.M. Quintana Melgoza, J. Cruz Reyes, M. Avalos-Borja, Mater. Lett. 47 (2001)
14–318.
[23] B. Scheffer, P. Molhoek, J.A. Moulijn, Appl. Catal. 46 (1989) 11.
[24] Ch. Li, Y.-W. Chen, Thermochim. Acta 256 (1995) 457–465.
4
in different proportions, depending on the W content of the cata-
lyst, determines their catalytic behavior. Moreover, the presence
of Lewis acid sites, with relatively high acid strength, facilitates
the consecutive decomposition of the ethylene formed during the
ethane oxidation. Thus, the decrease in the selectivity to ethylene
with ethane conversion is greater at high W content.
For Ni-rich catalysts, i.e., with 0.1 < W/(Ni + W) < 0.4, interme-
diate catalytic behavior is observed. The catalytic activity for eth-
ane oxidation decreases (and the selectivity to ethylene
increases) with the tungsten loading as a consequence of the lower
reducibility and smaller amount of the Ni–O sites on the catalyst
surface. However, no influence of ethane conversion on the selec-
tivity to ethylene is clearly observed (and a very low selectivity
to CO is observed in the ethane conversion range studied). The
characterization results presented here suggest that small NiO par-
[
[
(
[
(
[
3
[
[
25] R. Fiuza, M.A. Silva, J.S. Boaventura, Int. J. Hydrogen Energy 35 (2010) 11216.
26] A. Spojakina, R. Palcheva, K. Jiratova, G. Tyuliev, L. Petrov, Catal. Lett. 104
(
2005) 45.
[27] Q. Zhao, Sh. Chen, J. Gao, Ch. Xu, Transition Met. Chem. 34 (2009) 621.
28] M.J. Tomellini, J. Chem. Soc. Faraday Trans. 1 (84) (1988) 350.
29] M.W. Roberts, R.J. Smart, J. Chem. Soc. Faraday Trans. 1 (80) (1984) 2957.
30] P. Salagre, J.L.G. Fierro, F. Medina, J.E. Sueiras, J. Mol. Catal. A 106 (1996) 125.
[31] J.C. Vedrine, G. Hollinger, T.M. Duc, J. Phys. Chem. 82 (1978) 1515.
32] M.A. van Veenendaal, G.A. Sawatzky, Phys. Rev. Lett. 70 (1993) 2459.
33] V. Biju, M. Abdu Khadar, J. Nanopart. Res. 4 (2002) 247.
34] M.A. van Veenendaal, D. Alders, G.A. Sawatzky, Phys. Rev. B 51 (1995) 13966.
[
[
[
ticles and WO
meric tungstate species could also be present in samples with
higher W content. The interaction of NiO particles with WO nano-
x
nanoparticles are mainly present, although poly-
x
[
[
[
particles seems to be an important factor in the improvement of
the selectivity to ethylene, probably blocking the active and nonse-
lective sites of pure nickel oxide, similarly to Ni–Nb–O catalysts.
According to these results, it has been concluded that both par-
allel and consecutive reactions (including selective and nonselec-
tive steps) in this reaction can be tuned by changing the Ni/W
ratio in the catalysts.
[35] D. Alders, F.C. Voogt, T. Hibma, G.A. Sawatzky, Phys. Rev. B 54 (1996) 7716.
[
[
[
36] E. Salje, A.F. Carley, M.W. Roberts, J. Solid State Chem. 29 (1979) 237.
37] J. Haber, J. Stoch, L. Ungier, J. Solid State Chem. 19 (1976) 113.
38] F. Hilbrig, H. Schmelz, H. Knözinger, in: L. Guzci, F. Solymosi, P. Tétényi (Eds.),
Proceedings of the 10th International Congress on Catalysis, Elsevier,
Amsterdam, 1993, p. 1351.
39] G. Busca, V. Lorenzelli, V. Sánchez-Escribano, Chem. Mater. 4 (1992) 595.
40] K. Hadjivanov, H. Knözinger, M. Mihaylov, J. Phys. Chem. B 106 (2002) 2618.
[41] J. Zhu, J.G. van Ommen, H.J.M. Bouwmeester, L. Lefferts, J. Catal. 233 (2005)
34.
[
[
Acknowledgments
4
[
[
42] X. Zhang, Y. Gong, G. Yu, G.Y. Xie, J. Mol. Catal. A Chem. 180 (2002) 293.
43] H. Fu, Z.-P. Liu, Z.-H. Li, W.-N. Wang, K.-N. Fan, J. Am. Chem. Soc. 128 (2006)
Financial support from DGICYT in Spain (Project CTQ-2009-
4495) and Generalitat Valenciana (ACOMP/2010/091) is grate-
1
11114.
fully acknowledged.