Ni-Nb-O mixed oxides as highly active and selective catalysts for ethene production via ethane oxidative dehydrogenation. Part I: Characterization and catalytic performance

Article abstract of DOI:10.1016/j.jcat.2005.11.002

This work demonstrates the high potential of a new class of catalytic materials based on nickel for the oxidative dehydrogenation of ethane to ethylene. The developed bulk Ni-Nb-O mixed oxides exhibit high activity in ethane ODH and very high selectivity (～90% ethene selectivity) at low reaction temperature, resulting in an overall ethene yield of 46% at 400 °C. Varying the Nb/Ni atomic ratio led to an optimum catalytic performance for catalysts with Nb/Ni ratio in the range 0.11-0.18. Detailed characterization of the materials with several techniques (XRD, SEM, TPR, TPD-NH₃, TPD-O₂, Raman, XPS, electrical conductivity) showed that the key component for the excellent catalytic behavior is the Ni-Nb solid solution formed upon the introduction of niobium in NiO, evidenced by the contraction of the NiO lattice constant, since even small amounts of Nb effectively converted NiO from a total oxidation catalyst (80% selectivity to CO₂) to a very efficient ethane ODH material. An upper maximum dissolution of Nb ⁵⁺ cations in the NiO lattice was attained for Nb/Ni ratios ≤0.18, with higher Nb contents leading to inhomogeneity and segregation of the NiO and Nb₂O₅ phases. A correlation between the specific surface activity of the catalysts and the surface exposed nickel content led to the conclusion that nickel sites constitute the active centers for the alkane activation, with niobium affecting mainly the selectivity to the olefin. The incorporation of Nb in the NiO lattice by either substitution of nickel atoms and/or filling of the cationic vacancies in the defective nonstoichiometric NiO surface led to a reduction of the materials nonstoichiometry, as indicated by TPD-O₂ and electrical conductivity measurements, and, consequently, of the electrophilic oxygen species (O^-), which are abundant on NiO and are responsible for the total oxidation of ethane to carbon dioxide.

Full text of DOI:10.1016/j.jcat.2005.11.002

Journal of Catalysis 237 (2006) 162–174
www.elsevier.com/locate/jcat
Ni–Nb–O mixed oxides as highly active and selective catalysts
for ethene production via ethane oxidative dehydrogenation.
Part I: Characterization and catalytic performance
∗
E. Heracleous, A.A. Lemonidou
Department of Chemical Engineering, Aristotle University of Thessaloniki and Chemical Process Engineering Research Institute (CERTH/CPERI),
PO Box 1517, University Campus, GR-54006 Thessaloniki, Greece
Received 21 September 2005; revised 2 November 2005; accepted 3 November 2005
Abstract
This work demonstrates the high potential of a new class of catalytic materials based on nickel for the oxidative dehydrogenation of ethane to
ethylene. The developed bulk Ni–Nb–O mixed oxides exhibit high activity in ethane ODH and very high selectivity (∼90% ethene selectivity)
◦
at low reaction temperature, resulting in an overall ethene yield of 46% at 400 C. Varying the Nb/Ni atomic ratio led to an optimum catalytic
performance for catalysts with Nb/Ni ratio in the range 0.11–0.18. Detailed characterization of the materials with several techniques (XRD, SEM,
TPR, TPD-NH , TPD-O , Raman, XPS, electrical conductivity) showed that the key component for the excellent catalytic behavior is the Ni–Nb
3
2
solid solution formed upon the introduction of niobium in NiO, evidenced by the contraction of the NiO lattice constant, since even small amounts
of Nb effectively converted NiO from a total oxidation catalyst (80% selectivity to CO ) to a very efficient ethane ODH material. An upper
2
5+
maximum dissolution of Nb cations in the NiO lattice was attained for Nb/Ni ratios ꢀ0.18, with higher Nb contents leading to inhomogeneity
and segregation of the NiO and Nb O phases. A correlation between the specific surface activity of the catalysts and the surface exposed nickel
2
5
content led to the conclusion that nickel sites constitute the active centers for the alkane activation, with niobium affecting mainly the selectivity
to the olefin. The incorporation of Nb in the NiO lattice by either substitution of nickel atoms and/or filling of the cationic vacancies in the
defective nonstoichiometric NiO surface led to a reduction of the materials nonstoichiometry, as indicated by TPD-O and electrical conductivity
2
−
measurements, and, consequently, of the electrophilic oxygen species (O ), which are abundant on NiO and are responsible for the total oxidation
of ethane to carbon dioxide.

2005 Elsevier Inc. All rights reserved.
Keywords: Ethane; Oxidative dehydrogenation; Nickel oxide; Mixed nickel-niobium oxides; Nonstoichiometric oxygen
1
. Introduction
The most attractive alternative to steam cracking is the cat-
alytic dehydrogenation of light alkanes in the presence of an
oxidizing agent [1]. The energy requirements are minimal be-
cause of the low operation temperature and mild exothermicity
of the reaction, while alkanes constitute a low-cost and readily
available feedstock, conforming to the general need for sustain-
able development. However, the development of both active and
selective catalytic materials proves to be a very hard task, since
the formation of carbon oxides is the thermodynamically fa-
vored route and the resulting product, alkene, is generally more
reactive than the alkane. Therefore, a catalyst should be able to
activate and selectively convert the alkane to the correspond-
ing alkene at relatively low temperature and also preserve the
alkene from secondary oxidation.
Light olefins are currently produced by steam cracking of
various petroleum fractions, mainly naphtha. The reactions that
take place in a steam cracker are highly endothermic, and thus
the process energy demands are extremely high. With the olefin
market growing at a rate of 3% per year and fuel costs ris-
ing constantly, the development of a substitute method for the
production of light olefins has become an increasingly urgent
task.
*
Corresponding author.
E-mail address: alemonidou@cheng.auth.gr (A.A. Lemonidou).
0021-9517/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.11.002

E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
163
Ethane oxidative dehydrogenation has been studied over a
wide range of catalytic materials. Supported early-transition
metal oxides (V, Mo, Cr) operate at low temperature (<550 C)
characterization results and the catalytic performance data shed
light on the nature and functionality of the catalytic active sites.
◦
and exhibit high initial ethene selectivity [2–6]. However,
high activity can only be achieved at the expense of ethene
production, because these catalysts are as a rule more reac-
tive toward olefins. High ethylene yields have been reported
over chlorine-promoted nonreducible oxides (e.g., LiCl/MgO),
2. Experimental
2.1. Catalyst preparation
A series of mixed Ni–Nb–O oxides with varying Nb/Ni
atomic ratio ranging from 0 to 0.67 was prepared by the
evaporation method. Aqueous solutions containing the precur-
sor salts, nickel nitrate hexahydrate (>99%, Merck) and am-
monium niobium oxalate (99.99%, Aldrich), in appropriate
◦
where the reaction proceeds at high temperature (>600 C)
via a homogeneous–heterogeneous reaction scheme [7–10].
Nonetheless, serious problems, such as low catalyst stability
and the release of chlorine, are associated with the use of these
catalysts.
◦
amounts were heated at 70 C under continuous stirring for
Multicomponent mixed oxides seem to be the most promis-
ing class of materials for the ethane oxidative dehydrogenation
reaction. One of the best-performing catalysts, reported in 1978
by Thorsteinson et al. [11], consists of a mixture of Mo–V–Nb
1 h to ensure complete dissolution and good mixing of the
starting compounds. The solvent was then removed by evap-
oration under reduced pressure, and the resulting solids were
◦
◦
dried overnight at 120 C and calcined in synthetic air at 450 C
for 5 h. Pure NiO and Nb2O5 used as reference compounds
were obtained from the decomposition of the corresponding
◦
oxides operating at low temperature (<400 C) with relatively
high efficiency. More recently, Lopez Nieto and co-workers
have developed a catalytic formulation based on mixed Mo–
V–Te–Nb oxides exhibiting about 75% ethylene yield at low
◦
◦
precursors at 450 C for NiO and 550 C for Nb2O5 for 5 h
in synthetic air.
◦
reaction temperature (350–400 C). The enhanced catalytic ac-
The catalysts are referred to as NixNby, where x and y in-
dicate the atomic content of Ni and Nb, respectively, relative to
the total metal content of the materials.
tivity of the proposed mixed oxides was related to the presence
of a multifunctional Te2M20O57 (M = Mo, V, Nb) orthorhom-
bic phase [12].
In a recent investigation we showed that nickel-based mate-
rials exhibit very good potential as catalysts for ethylene pro-
duction via ethane oxidative dehydrogenation [13]. The good
performance of nickel catalysts has been reported previously
2.2. Catalyst characterization
Surface areas of the samples were determined by N2 adsorp-
tion at 77 K, using the multipoint BET analysis method, with
an Autosorb-1 Quantachrome flow apparatus. Prior to the mea-
surements, the samples were dehydrated in vacuum at 250 C
overnight.
X-Ray diffraction (XRD) patterns were obtained using a
Siemens D500 diffractometer, with Cu-Kα radiation. The mor-
phology of the synthesized materials was examined by scanning
electron microscopy (SEM) on a JEOL 6300 microscope, cou-
pled with energy-dispersive X-ray analysis (EDX; Oxford Link
ISIS-2000) for local elemental composition determination.
The reduction characteristics of the catalysts were studied
by temperature-programmed reduction (TPR) experiments per-
formed in a gas flow system equipped with a quadrupole mass
analyzer (Omnistar, Balzers). Typically, the catalyst sample
(50 mg) was placed in a U-shaped quartz reactor and pre-
[
14–18], but with no extensive characterization and/or identifi-
◦
cation of active sites. In our earlier study we demonstrated that
spreading of NiO on alumina causes chemical modification of
the NiO particles, which, even though possess a bulk-like crys-
tallography, exhibit modified electronic properties (monitored
by X-ray photoelectron spectroscopy [XPS]), rendering them
active and selective in ethane ODH; in contrast, unsupported
pure NiO was very unselective toward ethane [13]. Promotion
of the NiO/Al2O3 catalysts with various metals (Mo, V, Co,
Nb, and Ta) significantly modified both structural and catalytic
properties. The introduction of niobium was the most benefi-
cial for ethane ODH, increasing the reactivity toward ethane by
>
50% while maintaining the high ethene selectivity, resulting
◦
in a 30% ethene yield at 425 C. Niobium was found to improve
the dispersion of the nickel phase and facilitate the C–H bond
activation by acting as an electron transfer promoter.
◦
treated in flowing He for 0.5 h at 450 C, followed by cooling at
room temperature. The temperature was then raised from room
◦
◦
In the case of unpromoted NiO/Al2O3 catalysts, the presence
of the alumina support was essential for achieving high selectiv-
ity in the ethane ODH reaction [13]. However, further investi-
gation of the Nb-promoted catalyst revealed that supporting the
nickel–niobium phase was not necessary and that actually the
unsupported Ni–Nb–O mixed oxide yielded a much-improved
catalytic material. We therefore report in this study the devel-
opment of highly active and selective Ni–Nb–O mixed-oxide
catalysts for the ethane oxidative dehydrogenation reaction. The
effect of the Nb/Ni ratio on the performance of the materi-
als is discussed in view of a detailed investigation of the bulk
and surface properties of the oxides. Correlations between the
temperature to 800 C at a rate of 10 C/min in a 5% H2/He
3
flow (50 cm /min). The main (m/z) fragments registered were
H2 = 2, H2O = 18, and He = 4.
NH3 temperature-programmed desorption (TPD) was used
to determine the acid properties. The experiments were con-
ducted on the same apparatus as the TPR experiments described
◦
above. The catalysts (100 mg) were pretreated at 450 C for
◦
0.5 h and then cooled to 100 C under He flow. The pretreated
◦
samples were saturated with 5% NH3/He for 1 h at 100 C, with
subsequent flushing with helium at 100 C for 1 h to remove the
◦
physisorbed ammonia. TPD analysis was carried out from 100
◦
◦
to 700 C at a heating rate of 10 C/min. The following (m/z)

164
E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
fragments were registered: NH3 = 15, H2O = 18; N2 = 28,
NO = 30, N2O = 44, and NO2 = 46. Quantitative analysis of
the desorbed ammonia was based on (m/z) 15.
The stability of the catalyst with the optimum composition was
studied by continuously recording the catalytic performance
◦
for 24 h on line at constant reaction conditions (T = 350 C;
3
The oxygen desorption properties of selected catalysts
W/F = 0.54 gs/cm ).
were studied by O2-TPD measurements. The catalyst sample
The reaction products were analyzed on line by a Perkin
Elmer gas chromatograph equipped with a thermal conductiv-
ity detector. Two columns in a series-bypass configuration were
used in the analysis, a Porapak Q and an MS 5A. The main re-
action products were C H , CO , and H O. Negligible amounts
◦
(
200 mg) was pretreated in a flow of pure oxygen at 450 C
for 1 h and cooled to room temperature under the oxygen flow.
The system was subsequently flushed with helium for 1 h, and
◦
◦
the temperature was raised to 850 C at a rate of 15 C/min in
2
4
2
2
3
He (50 cm /min). The reactor exit was monitored on-line by
of oxygenates were observed at the reactor exit. Ethane conver-
sion and selectivity to the reaction products were calculated on
a carbon basis. Closure of the carbon mass balance was bet-
ter than ± 1%. The contribution of gas-phase-initiated reactions
was tested by conducting experiments using an empty-volume
reactor. The conversion of ethane at these experiments was
<1%, confirming that gas-phase reactions are negligible at the
experimental conditions used for the activity tests.
a quadrupole mass analyzer (Omnistar, Balzers), and the des-
orbed oxygen was detected by following the 32 (m/z) fragment.
Raman spectra were recorded at ambient temperature with a
7
85-nm diode laser excitation (laser power level, 200 mW) on
a Renishaw Raman Spectrometer (type 1000) equipped with a
CCD detector. The catalysts were pressed into self-supporting
wafers and placed in a sample holder, consisting of a gold
plate attached to a heating wire. Before the measurements, the
◦
samples were pretreated in flowing He at 120 C for 1 h. The
3
. Results
−
1
wavenumber accuracy was within 1 cm
.
The surface composition of the catalysts was determined
by XPS. The XPS measurements were performed on a Kratos
AXIS HSi instrument equipped with a charge neutralizer and
Mg-Kα X-ray source. Spectra were recorded at normal emis-
sion, using an analyzer pass energy of 20 eV and X-ray power
of 225 W.
3
.1. Catalyst characterization
The composition and the main physicochemical characteris-
tics of the Ni–Nb–O mixed oxides are presented in Table 1. The
surface area recorded for NiO was relatively low, but increased
considerably upon introduction of Nb. This increase can be at-
tributed to the use of niobium oxalate as starting compound
for Nb. The decomposition (during calcination) of precursors
with organic nature yields a more porous structure than for the
inorganic nitrate salts used for nickel oxide preparation. A max-
imum surface area is recorded for the Ni0.85Nb0.15 catalyst,
whereas a significant reduction is recorded for Ni0.6Nb0.4, due
to the extended phase segregation and formation of large Nb2O5
crystallites (see the XRD results).
XRD was used to investigate the crystalline phases formed
in the calcined catalysts. Fig. 1 illustrates the diffractograms ob-
tained for all of the samples, plus that of NiO and Nb2O5 used
as references. Catalysts with Nb content up to 15% relative to
the total metal content exhibit only the diffraction lines corre-
sponding to a crystalline “NiO-like” phase, with a broad band
Electrical conductivity measurements were performed on
an Agilent 4284 A 20-kHz–1-MHz precision LCR meter. The
samples were compressed under mild pressure to form pellets
of approximately 10 mm diameter and 5 mm thickness, which
were then coated with silver paste, as an electrode, on both sur-
face sides. The electrical resistance (R) of the catalysts was
measured at room temperature in atmospheric air in a frequency
range of 100 kHz–1 MHz. The electrical conductivity (σ) was
−
1
−1
then computed using the formula σ = R tS , where t is the
thickness and S the cross-sectional area of the pellet. The DC
conductivity was calculated by fitting the AC conductivity dis-
persion with frequency using Jonscher’s power law (JPL) [19],
s
σAC = σDC +Aω , by nonlinear least squares fit analysis, where
A is a constant and s is a power law exponent.
◦
centered at 2θ 27 attributed to an amorphous niobium phase.
By only slightly increasing the Nb content (catalyst Ni0.8Nb0.2)
only slightly, sharp diffractions lines at 2θ 35.9 , 41 , and 53.8
indicate the formation of the mixed NiNb2O6 phase. In the
2
.3. Reactivity studies
◦
◦
◦
The catalytic performance of the samples was measured
in a fixed-bed quartz reactor. The catalyst particles were di-
luted with equal amount of quartz particles of the same size
to achieve isothermal operation. The temperature in the middle
of the catalytic bed was measured with a coaxial thermocou-
Table 1
Nomenclature and physicochemical characteristics of the catalysts
◦
Sample
Nb/Ni NiO
Surface NiO aver- NiO lattice Acidity
ple. The samples were activated in flowing oxygen at 450 C
atomic loading area
age crystal constant
(wt%) (m /g) size (nm) (Å)
(µmolNH /
3
for 30 min. The composition of the reaction mixture was 9.1%
C2H6/9.1% O2/81.8% He.
2
2
ratio
m )
NiO
Ni Nb
Ni
Ni0.8Nb0.2
Ni Nb
Ni Nb
0
100
16.7
31.8
21.4
16.7
14.3
14.9
21.6
–
4.1762
4.1751
4.1725
4.1722
4.1723
4.1721
–
0.92
3.30
3.44
3.18
3.13
2.95
5.04
The oxidative dehydrogenation of ethane was investigated
0.111
0.176
0.250
0.428
0.666
–
83.49 55.9
76.10 85.1
69.21 58.6
56.74 57.1
45.74 31.4
0
.9
0.1
◦
in the temperature range 300–425 C. For the determination of
Nb
0.15
0.85
the activity of the catalysts as a function of temperature, the
3
W/F ratio was kept constant at 0.54 g s/cm . To obtain dif-
0.7
0.3
0.4
0.6
ferent ethane conversion levels at constant reaction temperature
Nb O
2 5
0
32.5
◦
3
(
350 C), the W/F ratio was varied from 0.02 to 1.73 g s/cm .

E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
165
segregation phenomena occur between Nb2O5 and NiO phases
for Nb/Ni atomic ratios >0.176. Additional information de-
duced from the XRD analysis is that the Ni–Nb–O mixed ox-
ides consist of crystallites in the nanorange and thus can be
classified as nanomaterials. Clearly, crystallite size is always
smaller for the mixed oxides than for the pure oxides. This phe-
nomenon can be explained by the increased crystallographic
disorder in the Ni–Nb–O catalysts, which could also be respon-
sible for the high surface area recorded for the catalyst with the
highest proportion of the Ni–Nb solid solution (Ni0.85Nb0.15).
The crystal size reaches a minimum for an intermediate Nb/Ni
ratio equal to 0.25 and increases with further increasing Nb con-
tent. The incorporation of niobium in the nickel oxide lattice in-
hibits crystallization, and thus smaller grain sizes are recorded.
However, at high Nb loading and as the two phases (NiO and
Nb2O5) begin to segregate, crystallization of NiO occurs more
freely, leading to the detection of larger crystals. The same ef-
fect has been reported for CeNixOy mixed oxides [21].
The surface morphology of the synthesized materials was
examined by SEM. Micrographs of the Ni–Nb–O mixed ox-
ides of all compositions, together with those of the NiO and
Nb2O5 reference compounds, are presented in Figs. 2A–2G.
Pure nickel oxide consists of well-defined cubic microcrystal-
lites in the nanosize range, in full agreement with the XRD
results. The addition of niobium induces significant changes
in morphology. The Ni0.9Nb0.1 and Ni0.85Nb0.15 samples show
homogeneous particles with a sponge-like appearance, reflect-
ing the lower crystallinity and higher surface area detected by
the previous characterization techniques and consistent with the
formation of a Ni–Nb solid solution. The EDS analysis revealed
a highly homogeneous elemental distribution of Ni and Nb on
these catalysts, as shown in Fig. 3A for Ni0.85Nb0.15, with the
Nb/Ni atomic ratio centered on the nominal value. Further in-
crease in Nb loading leads to the formation of few larger plate-
like crystallites (Ni0.8Nb0.2 catalyst; Fig. 2D), whereas on the
high-Nb content samples (Ni0.7Nb0.3 and Ni0.6Nb0.4), a clear
change of the surface morphology is apparent. The particles of
sponge-like appearance disappear, and the catalysts consist of
well-defined plate-like crystallites, corresponding to the Ni–Nb
phase, and smaller round white particles, corresponding to pure
Nb2O5, as confirmed by the elemental EDS analysis presented
in Fig. 3B for the Ni0.6Nb0.4 oxide.
Fig. 1. X-Ray diffraction patterns of the Ni–Nb–O catalysts and reference ma-
terials NiO and Nb O .
2
5
high-Nb content catalysts (Ni0.7Nb0.3 and Ni0.6Nb0.4), the in-
tensity of the NiNb2O6 lines drops and crystalline Nb2O5 is
detected.
Analysis of the XRD patterns allowed the calculation of par-
ticle size and lattice parameters of the nickel oxide-like phase
in the as synthesized mixed oxides, presented in Table 1. The
XRD analysis shows that pure NiO crystallizes in the cubic rock
salt structure with a lattice constant of a = 4.1762 Å, in agree-
ment with reported data (JCPDS 4-835). The incorporation of
Nb causes a contraction of the NiO lattice, as evidenced by the
monotonous decrease of the lattice constant with increasing Nb
content (Table 1) up to a Nb/Ni ratio of 0.176. Further increase
of the niobium amount does not cause other alterations, and the
lattice constant remains invariable. This observed decrease in
the size of the NiO unit cell could be discussed in terms of a
2+
possible substitution of some of the host cations (Ni ) by the
foreign species (Nb5 ). The substitution process requires that
+
the ionic radii of host and foreign cations be similar. The ionic
radii of Ni2 and Nb are 0.69 Å [20] and 0.64 Å, respec-
tively; therefore, a substitution between the two cations is pos-
sible and would lead to a reduction of the lattice constant, since
+
5+
The reducibility of the catalysts was studied with TPR in
H2. The TPR profiles, shown in Fig. 4, were acquired by con-
tinuously following the H2 signal with a quadrupole mass spec-
trometer while linearly increasing temperature. Pure NiO ex-
hibits a relatively narrow reduction peak with a maximum at
the radii of Nb5 is smaller than that of Ni . The results of
the XRD analysis support this notion. This substitution process
also leads to charge imbalance due to the different valency of
the nickel and niobium cations, which significantly modifies the
properties of the mixed Ni–Nb–O oxides, as discussed in de-
tail later in the paper. The stable NiO lattice constant measured
for the high-Nb loading samples (Ni0.8Nb0.2, Ni0.7Nb0.3, and
Ni0.6Nb0.4) indicates that we have reached an upper limit to the
+
2+
◦
415 C. Calculation of the consumed hydrogen indicates to-
2+
tal reduction of Ni ions to metallic Ni. The introduction
of Nb significantly broadens the reduction profiles, indicat-
ing modification of the catalysts and contributions from differ-
ent species. The main reduction peak is shifted to 390 C in
Ni0.9Nb0.1, and a broad shoulder centered at 500 C appears.
5+
◦
maximum possible dissolution of Nb cations into the NiO
lattice.
◦
◦
Based on the foregoing, the “NiO-like” phase detected in the
diffractograms of the Ni–Nb–O mixed oxides can be identified
as a Ni–Nb solid solution. The highest proportion of this solid
solution is obtained for the Ni0.85Nb0.15 compound, whereas
The same profile, shifted about 30 C lower, is also recorded
for the Ni0.85Nb0.15 catalyst. In Ni0.8Nb0.2 we observe two dis-
◦
◦
tinct peaks with maxima at 340 C and 460 C, ascribed to two
well-defined reducible species. In view of the XRD results,

166
E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
Fig. 2. SEM micrographs of the Ni–Nb–O catalysts and reference materials NiO and Nb O : (a) NiO; (b) Ni Nb ; (c) Ni
Nb
0.15
; (d) Ni Nb
;
2
5
0.9
0.1
0.85
0.8
0.2
(
e) Ni Nb ; (f) Ni Nb ; (g) Nb O .
0.7 0.3 0.6 0.4 2 5
this second peak can be attributed to the crystalline NiNb2O6
phase, detected only in this sample. This observation is also
in line with literature data reporting the reduction temperature
amounts of NiNb2O6 still present in the high-Nb loading sam-
ples.
Quantification of the TPR curves shows that hydrogen con-
sumption is linearly dependent on the wt% amount of NiO in
the catalysts, as graphically illustrated in Fig. 5, and corre-
sponds to the total reduction of Ni ions from +2 to the metallic
form. Given that Nb2O5 is irreducible under the conditions used
◦
of bulk nickel niobate to be around 480 C [22]. The catalysts
with 70 and 60% Ni present similar reduction characteristics,
◦
with a single rather narrow peak shifted to ∼360 C. A very
◦
small shoulder evident near 500 C could be due to minor

E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
167
Fig. 3. SEM micrographs and corresponding EDS elemental analysis of niobium and nickel for (a) Ni
Nb
0.15
; (b) Ni Nb
.
0.85
0.6
0.4
in the TPR measurements (evidenced by testing of the pure nio-
bium oxide), reduction of the catalysts thus occurs by removal
of oxygen from Ni–O–Ni and Ni–O–Nb bonds. Nb incorpo-
ration weakens the Ni–O–Ni bonds, because the oxygen can
be removed at lower temperatures than in pure NiO. Based on
the electronegativity concept, the presence of niobium in the
coordination sphere of nickel should decrease the electronic
density of nickel (Pauling electronegativity of 1.9 for Ni and 1.6

168
E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
Fig. 4. Temperature-programmed reduction profiles of the Ni–Nb–O catalysts
and reference material NiO.
Fig. 6. Ammonia desorption curves of the Ni–Nb–O catalysts and reference
materials NiO and Nb2O5.
◦
ture (700 C) in the TPD profile. The introduction of niobium
causes a significant increase in the total acidity and generates
new acid sites of weak and moderate strength. The Ni–Nb–O
mixed oxides are characterized by a narrow desorption peak
◦
in the range 130–450 C, with maximum temperature of des-
◦
orption at 220–230 C. Although mild differences between the
different catalyst compositions are seen, acidity increases with
increasing Nb content, reaches a maximum for a Nb/Ni ratio of
0
.176, and decreases slightly at higher Nb concentrations. Pure
Nb2O5 has the highest total surface acidity, exhibiting a simi-
lar TPD profile as the catalysts. It is well known that niobium
oxide has high acidic properties, with the Lewis acid sites cor-
responding to highly distorted NbO6 octahedra [23].
In summary, the acidic measurements show that the Ni–Nb–
O mixed oxides exhibit high moderate acidity, much higher than
the sum of the acidity of the individual components in each cat-
alytic composition. The effect is much more pronounced if we
compare, for example, pure NiO and the Ni0.9Nb0.1 catalyst,
where the introduction of only 10% of Nb leads to an almost
Fig. 5. Hydrogen consumption versus wt% NiO loading.
for Nb), and thus facilitate its reduction. The high-temperature
shoulders in the TPR patterns can be ascribed to the reduc-
tion of formed Ni–O–Nb bonds. The absence of such extended
high temperature reduction in the high-Nb content samples
3
00% increase in acidity. According to a model described by
(
30 and 40% Nb) is consistent with the preferential forma-
n+
Tanabe [24], the substitution of cations M of a host oxide by
cations N of another metal oxide can generate acidic centers
tion of niobium oxide, Nb–O–Nb bonds that are irreducible, at
the expense of the formation of Ni–O–Nb heterolinkages. The
TPR results are analogous to reducibility results obtained with
NiO/Al2O3 and Ni–Nb/Al2O3 catalysts presented in our pre-
vious work [13], where the promotion with Nb caused the ap-
pearance of two reduction peaks, one at lower temperature and
one at higher temperature than the Nb-free sample attributed to
reduction of Ni–O–Ni and Ni–O–Nb bonds, respectively.
Acidity was determined by TPD after saturation of the cat-
m+
due to the charge imbalance along M–O–N linkages. This acid-
ity is defined as Brønsted or Lewis based on whether this excess
charge is negative or positive, respectively. The model structure
postulates that each cation retains its parent coordination num-
ber, whereas all oxygen anions in the binary oxide maintain
the coordination of the major component oxide. The applica-
tion of these assumptions to the NiO–Nb2O5 system explains
the remarkable increase in acidity exhibited by the Ni–Nb–O
catalysts compared to pure nickel oxide. Bulk NiO is a well-
known nonstoichiometric oxide with cationic vacancies [25].
Therefore, the filling of the cationic vacancies and/or the sub-
◦
alytic surface with ammonia at 100 C. In all cases the main
desorption product was NH3, with traces of N2, NO, and N2O
also detected. The ammonia desorption curves are compiled in
Fig. 6, and the acidity of the catalysts (expressed as µmol of
2
2+
5+
desorbed NH3/m ) is tabulated in Table 1. Pure NiO exhibits
stitution of Ni ions by Nb in the NiO lattice creates an
excess positive charge along the Ni–O–Nb linkages, and thus
very low acidity, with a small peak appearing at high tempera-

E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
169
Table 2
Nominal and surface atomic composition (derived from XPS) of the catalysts
Sample
Nominal composition (%)
Surface composition (%)
Ni
Nb
O
Ni
Nb
O
NiO
50
42
38
35
29
23
–
50
53
55
57
59
62
46.06
39.90
39.61
38.08
40.01
25.44
–
53.94
54.64
54.50
55.03
54.53
62.14
Ni Nb
5
7
9
6.05
6.51
7.63
0.9
0.1
Ni
0
Nb
0.15
.85
Ni Nb
0.8
0.2
0.3
0.4
Ni Nb
12
15
6.04
0.7
Ni Nb
13.96
0.6
The introduction of Nb in NiO with a Nb/Ni atomic ratio
of 0.111 results in the appearance of a weak broad band cen-
−
1
tered at ∼850 cm , in addition to the features of NiO. With
increasing Nb content, this band becomes more intense and
−
1
shifts to 840 cm for the Ni0.85Nb0.15 catalyst. In the spec-
trum of the Ni0.8Nb0.2 catalyst, this feature develops into a
−
1
well-defined band, red-shifted to 800 cm . The nature of sur-
face NbOx species supported on several carriers (SiO2, Al2O3,
ZrO2, TiO2) has been investigated in detail by Raman and in-
frared spectroscopy by Wachs and co-workers [28]. Accord-
ing to this study, vibrations of Nb=O bonds of monomeric
and polymeric niobia species generally occur at wavenumbers
Fig. 7. Raman spectra of the Ni–Nb–O catalysts and reference materials NiO
and Nb O .
2
5
−
1
above 950 cm ; therefore, the broad band detected in the 950–
−
1
the mixed oxide should exhibit high Lewis acidity. As already
discussed, the acidity reaches a maximum for the Ni0.85Nb0.15
catalyst and then decreases slightly for higher Nb content. The
XRD data indicate, based on the calculated NiO lattice con-
stant, that maximum solubility of niobium cations in NiO is
attained for the same catalyst (Ni0.85Nb0.15), confirming that
the observed acidity is related to the formation of Ni–O–Nb
bonds. A further increase in the Nb/Ni ratio results in extended
segregation of the NiO and Nb2O5 phases and, consequently, in
fewer Ni–O–Nb bonds and relatively reduced acidity.
750 cm range in our samples cannot be assigned to any pos-
sible surface niobia structures on nickel oxide having niobium–
oxygen double bonds. Lower frequency vibrations in the 850–
−
1
950 cm range have been attributed to symmetric stretchings
of [–O–Nb–O–]n bonds [28]. It is thus possible that the broad
bands observed on the Raman spectra of the Ni–Nb–O mixed
oxides result from vibrations of bridging Nb–O–Nb bonds of
surface species or/and vibrations of bridging Ni–O–Nb bonds
located on the surface of the NiO structure.
Additional bands originating from well-ordered crystalline
−
1
Raman spectra of the Ni–Nb–O mixed oxides collected at
ambient conditions are presented in Fig. 7, including the spec-
tra of reference NiO and Nb2O5 compounds. NiO exhibits a
[NbO6] polyhedra appear at 695 and 235 cm in the spectra of
the Ni0.7Ni0.3 and Ni0.6Nb0.4 catalysts, indicating the crystal-
lization of niobium in the pure oxide form in agreement with
the XRD results. Moreover, it should be mentioned that the
Ni0.6Nb0.4 sample is highly inhomogeneous and Raman spectra
acquired from white particles in the sample are identical to that
of pure Nb2O5 (spectrum shown in Fig. 7), confirming the sig-
nificant formation of pure niobium oxide crystals detected by
XRD and EDX analysis.
−
1
strong band at 500 cm due to the Ni–O stretching mode, with
−
1
a shoulder at 410 cm indicative of the nonstoichiometry of
the material [20] and the high nickel vacancy concentration.
The Raman spectrum of Nb2O5 exhibits two strong features at
− −1
1
6
95 and 235 cm , with a shoulder appearing at 310 cm and
−
1
a broad band in the 800–1000 cm range. Niobium pentoxide
generally has an octahedrally coordinated NbO6 structure that
is distorted to varying extents depending on whether its poly-
hedra are corner- or edge-shared. Highly distorted NbO6 octa-
hedra have Nb=O bonds, whereas the slightly distorted ones
The surface composition of the mixed nickel–niobium ox-
ides was determined by XPS, because this technique has a
surface-specific character and allows quantitative determina-
tion of the elemental content in the upper layers of the sample.
The nominal and surface atomic composition of the synthesized
catalysts is presented in Table 2. The values of the surface com-
position were derived from the integration of the Ni 2p, Nb 3d,
and O 1s core-level photoemission spectra, using the appropri-
ate sensitivity correction factors. Concerning pure NiO, a nickel
atom deficiency and an excess concentration of oxygen on the
surface is apparent, consistent with the Raman indications and
the general consensus in literature about the nonstoichiometric
nature of NiO [20,25]. Furthermore, the color of the sample
(gray) is also proof of the nonstoichiometry of the material,
−
1
have only Nb–O bonds [26]. The band at 695 cm is assigned
to the symmetric stretching mode of the niobia polyhedra, and
its position and sharpness reveals an ordered structure and an
increasing bond order of the [NbO6] units, crystallized in the
pseudohexagonal phase (TT-Nb2O5). The Raman bands in the
−
1
low-wavenumber region, 200–300 cm , are characteristic of
the bending modes of the Nb–O–Nb linkages, whereas the ad-
−
1
ditional weak band centered at ∼900 cm is due to a small
concentration of the Nb=O surface sites of highly distorted oc-
tahedra [27].

170
E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
because stoichiometric NiO is green. In the series of the Ni–
Nb–O catalysts with varying Nb/Ni ratio, the surface composi-
tions for Nb/Ni ratios from 0.11 to 0.43 (catalysts Ni0.9Nb0.1,
Ni0.85Nb0.15, Ni0.8Nb0.2, and Ni0.7Nb0.3) appear similar irre-
spective of the different nominal content, leading to an average
Ni surface coverage of 40% and an average Nb surface cover-
age of 6.5%. This observation is in agreement with our previous
results on promoted-Ni/Al2O3 catalysts [13], where the prefer-
ential exposure of the promoter or the nickel on the surface was
found to depend largely on the surface energy, not the compo-
sition, of the respective oxides. Therefore, in multicomponent
systems the surface composition is the result of a thermody-
namic driving force, according to which surface acquires the
composition that leads to the lowest possible surface energy.
Thus, the similar surface compositions detected for four of the
five catalysts indicates that this preferential surface exposure of
nickel and niobium in a fixed analogy, independent of the nom-
inal content, is energetically the most favorable arrangement of
the surface. Only the catalyst with the highest Nb content, the
Ni0.6Nb0.4 catalyst, constitutes an exception to the above, with
the surface composition approaching the nominal values. This
could be due to the severe segregation and formation of crystals
of the independent oxides that was detected on this sample, as
well as the sample’s high inhomogeneity (Figs. 2F and 3B).
Electrical conductivity measurements carried out on oxide-
based catalysts can provide valuable information on the nature
of structural defects, the existence of oxidizing species, and
so on [29]. Furthermore, investigating the electrical properties
provides important insights, particularly in the case of redox
reactions, because these properties are closely related to the na-
ture and reactivity of the oxygen species formed on activation
of the oxidant (O2) on the catalytic surface. Numerous previous
studies have correlated the catalytic performance of metal ox-
ide catalysts in partial oxidation reactions with the conductivity
of the oxide materials, for example, ethane to ethene [30,31],
propane to propene [32–34], n-butane to butenes [35,36], and
butane to maleic anhydride [37].
Fig. 8. DC electrical conductivity measured at ambient conditions versus Nb/Ni
atomic ratio.
Fig. 9. Temperature-programmed oxygen desorption curves of NiO and
Ni
0
Nb
0.15
catalyst.
.85
ments. From an electrical standpoint, the dissolution of het-
erovalent ions in the lattice of a host oxide creates free charge
carriers (electrons or holes) to maintain electrical neutrality.
5+
The dissolution of higher valence cations, such as Nb , in the
p-type NiO lattice suggests an n-doping character of Nb. Nio-
bium oxide is an n-type semiconductor with nonstoichiometry
arising from a lack of oxygen ions and free electrons as main
charge carriers. Therefore, niobium acts as a donor of electrons
and fills the electron (hole) states on Ni, thereby reducing the
The DC electrical conductivity of the Ni–Nb–O mixed ox-
ides and of pure nickel and niobium oxide at room temperature
is plotted as a function of the Nb/Ni atomic ratio in Fig. 8. NiO
exhibits high conductivity, in accordance with data reported
previously [38]. Doping of NiO with even the smallest amount
of Nb (Ni0.9Nb0.1 catalyst) causes a decrease in conductivity by
more than an order of magnitude. Further increases in Nb load-
ing do not cause significant alterations, with a slight decreasing
trend with increasing Nb content leading to the very small value
exhibited by pure Nb2O5 (two orders of magnitude lower than
NiO).
+
positive hole p concentration and, consequently, the conduc-
tivity compared with pure NiO, consistent with our results.
Based on the nonstoichiometry and consequent high con-
ductivity of NiO, the presence of electrophilic oxygen radicals
−
(
O ) on the NiO surface is expected. Indeed, NiO has been
reported to accommodate a large amount of surface excess oxy-
gen [38]. As shown earlier, the incorporation of Nb ions in the
+
NiO lattice reduces the p concentration and thus should reduce
As mentioned earlier, bulk NiO is a well-known p-type
semiconductor with conductivity attributed to cation deficiency
the amount of excess oxygen that the surface can accommo-
date. To verify this hypothesis, we performed O2-TPD mea-
surements on pure NiO and the Ni0.85Nb0.15 catalyst, which,
according to the characterization results, has the highest pro-
portion of the Ni–Nb solid solution. The profiles of oxygen
desorption with increasing temperature are compiled in Fig. 9.
Indeed, NiO exhibits a large oxygen desorption peak in the
[
39,40], justifying the high value recorded for pure NiO. The
+
cationic vacancies induce the formation of positive holes p ,
the main charge carriers, which are introduced in the form
3+
−
of Ni or O ions to maintain charge neutrality conditions
41,42]. Concerning the Ni–Nb–O catalysts, the formation of
[
a Ni–Nb solid solution, suggested by previous characterization
techniques, is directly confirmed by the conductivity measure-
◦
450–850 C range, with maximum temperature of desorption

E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
171
◦
at 695 C, in agreement with O2-TPD measurements on NiO
under the same reaction conditions exhibited negligible activity
(<1% ethane conversion) at 300–450 C, with ethene selectiv-
◦
reported previously [15]. The amount of oxygen desorbed cor-
responds to a 1.14% excess of oxygen (mol) per mole of NiO. It
has been reported previously that oxygen desorption from NiO
occurring at temperatures below 700 C corresponds to the non-
stoichiometric oxygen [43]. For the Nb/NiO catalyst, a much
ity not surpassing 65% even at this low conversion levels.
To account for the different surface areas of the samples,
the activity of the catalysts in terms of specific surface activ-
◦
2
ity (SSAc)—expressed as µmol C H consumed/(m s)—was
2
6
smaller desorption peak, asymmetric on the high-temperature
side, is recorded, with a Tmax of desorption at 625 C. The de-
also considered. Based on this definition, the reactivity of the
tested catalytic materials at 300 C decreased in the follow-
◦
◦
sorbed oxygen amounts to only a 0.2% excess of oxygen per
mole of catalyst, clearly confirming that the introduction of
Nb minimizes the nonstoichiometry of the material, fills the
NiO cationic vacancies, and reduces the chemisorbed oxygen
species on the catalytic surface. It should be noted that repe-
tition of the O2-TPD experiments without prior O2 adsorption
yielded the same desorption patterns. This, in combination with
the absence of low-temperature oxygen desorption peaks, indi-
cates that NiO and the Nb-doped NiO materials do not adsorb
oxygen at the experimental conditions used in this study (am-
bient conditions), and that the desorbed oxygen originates from
the bulk structure of the materials under study.
ing order: NiO > Ni0.85Nb0.15 > Ni0.9Nb0.1 > Ni0.8Nb0.2 >
Ni0.7Nb0.3 > Ni0.6Nb0.4. Pure NiO now exhibits a higher sur-
face reactivity for ethane activation compared with the Ni–
Nb–O oxides, due to the large increase of the surface area on
incorporation of Nb (which thus lowers the surface activity of
the catalysts). However, as shown in the next paragraph, ethane
activation on NiO is unselective, leading to combustion (see
Section 4). The activity trend is maintained for the Ni–Nb–O
catalysts, with optimum reactivity exhibited by Nb/Ni ratios of
0.111–0.176.
The most important requirement for a good ODH cata-
lyst is the ability to effectively convert ethane to ethylene at
low temperature with high selectivity. To study the selectiv-
ity of the mixed Ni–Nb–O oxides, and since selectivity is
generally strongly related to conversion, we conducted a sec-
3
.2. Catalytic performance in ethane ODH
◦
The activity of the Ni–Nb–O mixed oxide catalysts in the
ond series of experiments at constant temperature (350 C),
ethane oxidative dehydrogenation reaction was explored un-
constant ethane/oxygen ratio (1/1), and varying W/F (0.02–
1.73 g s/cm ), to attain different conversion levels. Ethyl-
◦
−3
der steady-state conditions at 300–425 C with a constant W/F
3
(
0.54 g s/cm ) and ethane-to-oxygen ratio (1/1). Ethane con-
ene selectivity as a function of ethane conversion is presented
in Fig. 11. The most prominent feature is the enormous in-
crease of ethene selectivity upon incorporation of Nb, climbing
from 20% for NiO to 80% for the lowest Nb content cata-
lyst (Ni0.9Nb0.1). Varying the Nb/Ni ratio leads to a maximum
in selectivity for the Ni0.85Nb0.15 catalyst (90%), with further
Nb increases decreasing selectivity slightly, down to 80% for
Ni0.6Nb0.4. Overall, all Ni–Nb–O catalysts demonstrate excel-
lent ability to selectively convert ethane to ethylene. It is also of
great importance that selectivity remains virtually constant with
version is plotted as a function of temperature in Fig. 10.
Note that in all cases, oxygen conversion did not exceed 90%.
Generally, all tested catalysts prove to be very reactive, ef-
fectively converting ethane at low reaction temperatures. The
incorporation of even small amounts of Nb (Nb/Ni = 0.111) in
NiO causes a 100% increase in the per weight reactivity com-
pared with the pure nickel oxide. Increasing the Nb content in
the mixed oxides leads to increased ethane conversion, which
reaches a maximum for the Ni0.85Nb0.15 catalyst (66% C2H6
◦
◦
conversion at 400 C). Further increases result in a decline in
increasing ethane conversion at 350 C for the conversion range
reactivity, which still, however, remains higher than that of pure
NiO, except for the highest-Nb loading catalyst (Ni0.6Nb0.4), in
which a sharp decrease is observed. Pure Nb2O5 oxide tested
studied (0–20%). The main problem with ODH catalysts is the
secondary overoxidation of the alkene produced, which causes
the drop in selectivity with increasing conversion. These mate-
Fig. 10. Ethane conversion as a function of temperature (reaction conditions:
Fig. 11. Ethene selectivity as a function of ethane conversion (reaction condi-
3
◦
W/F = 0.54 gs/cm , C H /O = 1/1).
tions: T = 350 C, C H /O = 1/1).
2
6
2
2
6
2

172
E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
Fig. 13. Specific surface activity (SSAc) in ethane ODH and surface Ni content
versus nominal atomic Nb/Ni ratio.
Fig. 12. Ethane conversion and ethene selectivity as a function of TOS (reaction
◦
3
conditions: T = 350 C, W/F = 0.54 gs/cm , C H /O = 1/1).
2
6
2
catalytic performance. A further advantage of these catalysts is
that the only byproduct detected was CO2, making these ma-
terials very attractive from an engineering standpoint because
of the greatly reduced separation costs downstream from the
reactor in any potential industrial application of an oxidative
dehydrogenation process.
rials exhibit very stable ethene selectivity, indicating that they
are very reactive toward ethane but have a very low affinity to
ethylene, thus preserving the olefin from further oxidation. An-
other important advantage of the Ni–Nb–O catalysts is that the
only byproduct detected is CO2, resulting from primary unse-
lective oxidation of ethane.
The key component for this excellent catalytic behavior
seems to be the Ni–Nb solid solution formed upon doping NiO
with Nb. Even small amounts of niobium effectively convert
NiO from a total oxidation catalyst to a very efficient ODH
material. The characterization results show that up to a Nb/Ni
Stability is a very important parameter that often defines the
fate” of a specific catalytic material for potential industrial ap-
“
plication. In a final step, the stability of the catalyst with the
optimum formulation, the Ni0.85Nb0.15 oxide, was investigated
by continuously recording the catalytic performance for 24 h
atomic ratio of 0.176, Nb5 ions efficiently substitute Ni
+
2+
◦
on-line at constant reaction conditions (T = 350 C) and high
ions and/or fill the cationic vacancies in the NiO lattice, with
higher Nb content leading to saturation of the bulk lattices sites
and formation of the mixed NiNb2O6 phase and pure Nb2O5.
The surface composition of the catalysts, determined by XPS,
indicates that the relative exposure of nickel and niobium on
the surface is constant irrespective of the nominal Nb/Ni ratio,
with the exception of the highest-Nb content catalyst, proba-
bly due to the high inhomogeneity and extended segregation
of the two phases detected on this sample. To obtain infor-
mation on the nature of the active sites for ethane activation,
ethane conversion levels. The results of the stability test, pre-
sented in Fig. 12, clearly show the high stability of the catalytic
materials, with both ethane conversion and ethylene selectiv-
ity maintaining their initial values for the entire time span of
the experiment, which, although shorter than what is typically
required for commercial application, demonstrates the high po-
tential of the mixed Ni–Nb–O catalysts.
4
. Discussion
−
2
2
Ni–Nb–O mixed oxides prove to be highly active and selec-
the SSAc, expressed as 10 µmol C2H6 consumed/(m s) at
◦
tive catalytic materials for the production of ethene via ethane
oxidative dehydrogenation, exhibiting an overall ethylene yield
300 C and differential conditions (conversion < 10%), and the
surface concentration of nickel sites are plotted as a function
of the nominal Nb/Ni atomic ratio in Fig. 13. A clear correla-
tion between the ability of the catalytic surface to activate the
ethane substrate and the number of exposed surface Ni sites is
apparent, with these two values following the same trend with
increasing Nb/Ni ratio. Based on this observation, we postulate
that nickel centers constitute the reactive sites for the activation
of alkane, whereas niobium seems to mainly affect the selec-
tivity pattern modifying the route of the reaction from total to
selective oxidation to olefin.
◦
of 46% at 400 C. In comparison with the alumina-supported
Ni–Nb mixed oxide catalyst recently presented in one of our
previous work [13], the binary Ni–Nb–O mixed oxides exhibit
an impressive improvement (160%) in ethene yield. Further-
more, the Ni–Nb–O materials compare favorably with most
ODH catalysts reported in the literature. A summary of per-
formance data for a wide variety of catalytic systems in terms
of selectivity for ethene as a function of ethane conversion and
reaction temperature is available elsewhere [1]. Most catalysts
either exhibit high initial selectivity, which quickly drops with
increasing conversion, or demand high reaction temperatures
for achieving acceptable yields. The developed Ni–Nb–O cata-
lysts for ethane ODH presented in this study appear to be clearly
superior, combining high activity at low temperature and high
ethene selectivity at high conversion levels (66% C2H6 con-
What then is the functionality of niobium, and what are the
modified properties that lead to the enhanced ODH behavior of
the catalysts? XPS, electrical conductivity, and O2-TPD studies
demonstrate that NiO is a highly conductive, nonstoichiometric
oxide with cationic vacancies (excess surface oxygen). Based
on data in the literature, its conductivity results from the pos-
◦
+
version and 70% C2H4 selectivity at 400 C), with very stable
itive holes p , which are induced by the cationic vacancies

E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
173
in the form of Ni3 or O ions, to maintain charge neutral-
+
−
occurring on the NiO surface. Due to the favorable ionic radii of
niobium, Nb cations fill the cationic vacancies and/or substitute
nickel atoms in the NiO lattice, forming a Ni–Nb solid solution.
This substitution process is most likely responsible for the re-
duction of the nonstoichiometry and the cationic defects on the
surface and, consequently, of the unselective oxygen species,
leading to enhanced ethane ODH activity.
ity conditions [41,42]. Iwamoto et al. [44] extensively studied
the chemisorption of O2 on NiO and suggested that O2 and
−
−
O
species are the predominant species formed on the nickel
oxide surface. Therefore, nonstoichiometric oxygen desorbed
from NiO is believed to be in the form of electrophilic oxygen
radicals. The nature and reactivity of the oxygen species on the
catalytic surface is believed to be one of the key factors govern-
ing the performance of metal oxide catalysts in selective oxi-
dation reactions. It is widely accepted that electrophilic species
O2 , O2 , O ) are mostly responsible for the deep oxidation
reactions, whereas nucleophilic ones (O ) are involved in both
the selective and unselective steps of the reaction [45].
Consequently, the low selectivity of NiO in the ethane oxida-
tive dehydrogenation reaction and the preferential total oxida-
tion of ethane to CO2 likely result from the highly reactive elec-
trophilic single-charged oxygen species present on the nickel
oxide surface. As the conductivity measurements point out, Nb
incorporation causes a one-fold decrease in electrical conduc-
tivity, consistent with the principle of controlled valence [25]
for dissolution of higher-valence ions in the lattice of p-type
host oxide. Niobium, therefore, acts as an electron donor and
The effect of niobium on the surface oxygen species partic-
ipating in the reaction and its enhancing action on the ethylene
selectivity is further investigated and confirmed by transient ex-
−
2−
−
18
(
periments with isotopic O2, presented in part II of the present
2−
series. In addition, aspects of the mechanism and the ethane
ODH reaction pathways occurring over this new class of cat-
alytic materials are discussed, and a macroscopic kinetic model
able to predict the catalyst performance for a wide range of op-
erating conditions is developed.
Acknowledgments
The authors thank Drs. L. Nalbandian and V. Zaspalis from
CERTH/CPERI for providing the facilities for XRD, SEM, and
conductivity measurements and Drs. A.F. Lee and K. Wilson
of York University for helping with the XPS measurements.
Financial support was provided by the General Secretariat of
Research and Technology Hellas (grant PENED01).
+
reduces the positive p hole concentration of the NiO accep-
tor. In agreement with this finding, the O2-TPD experiments
clearly indicate that Nb consumes or eliminates the incom-
−
−
pletely reduced electrophilic oxygen species (e.g., O , O2 ,
or O22 ), thus significantly suppressing the total oxidation of
ethane to carbon dioxide and enhancing the selective conversion
to ethylene. The excess oxygen accommodated on the surface
of Ni-based catalysts reportedly correlates inversely with the
selectivity to ethylene [15]. Conclusively, the elimination of
nonstoichiometric oxygen caused by Nb doping is very likely
responsible for the high ethene selectivity exhibited by the Ni–
Nb–O mixed oxides.
−
References
[
1] G. Centi, F. Cavani, F. Trifiro, Selective Oxidation by Heterogeneous
Catalysis, Kluwer Academic Publishers/Plenum Press, New York, 2001.
2] M.A. Banares, Catal. Today 5 (1999) 319.
[
[
3] H.X. Dai, C.T. Au, Curr. Top. Catal. 3 (2002) 33.
[4] H.H. Kung, Adv. Catal. 40 (1994) 1.
[5] E. Heracleous, A.F. Lee, I.A. Vasalos, A.A. Lemonidou, Catal. Lett. 88
(2003) 47.
[6] E. Heracleous, M. Machli, A.A. Lemonidou, I.A. Vasalos, J. Mol. Catal.
A 232 (2005) 29.
5
. Conclusion
[
[
7] S.J. Conway, J.H. Lunsford, J. Catal. 131 (1991) 513.
8] E. Morales, J.H. Lunsford, J. Catal. 118 (1989) 255.
The present work demonstrates the high potential of a new
[9] L. Leveles, S. Fuchs, K. Seshan, J.A. Lercher, L. Lefferts, Appl. Catal.
A 227 (2002) 287.
10] S. Gaab, M. Machli, J. Find, R.K. Grasselli, J.A. Lercher, Top. Catal. 23
class of catalytic materials based on nickel as active component
for the low-temperature oxidative dehydrogenation of ethane to
ethylene. Studying the catalytic performance of a series of Ni–
Nb–O mixed oxides (Nb/Ni atomic ratios of 0–0.67) clearly
indicates that introducing even small amounts of Nb in NiO
converts nickel oxide from an effective total oxidation catalyst
to an efficient low-temperature ODH catalyst, producing ethyl-
[
(2003) 151.
[
11] E.M. Thorsteinson, T.P. Wilson, F.G. Young, P.H. Kasai, J. Catal. 52
(1978) 116.
[12] P. Botella, E. Garcia-Gonzalez, A. Dejoz, J.M. Lopez Nieto, M.I. Vazquez,
J. Gonzalez-Calbet, J. Catal. 225 (2004) 428.
[
13] E. Heracleous, A.F. Lee, K. Wilson, A.A. Lemonidou, J. Catal. 231 (2005)
59.
1
◦
ene with a yield of 46% at 400 C. An optimum in the catalytic
[
14] Y. Schuurman, V. Ducarme, T. Chen, W. Li, C. Mirodatos, G.A. Martin,
Appl. Catal. A 163 (1997) 227.
behavior is achieved for the catalyst containing Nb/Ni in an
intermediate ratio of 0.176 (Ni0.85Nb0.15), which exhibits the
highest per weight activity and selectivity to ethylene (90%).
A correlation between the ethane specific surface activity of
the catalysts and the surface-exposed nickel content indicates
that nickel sites constitute the active centers for activation of the
paraffinic substrate, whereas niobium affects mainly the selec-
tivity to the desired product, olefin. NiO is a nonstoichiometric
oxide with cationic vacancies, which can accommodate large
amounts of excess electrophilic oxygen radicals, strong oxidiz-
ing species responsible for the total oxidation of ethane to CO2
[
[
[
[
[
15] X. Zhang, Y. Gong, G. Yu, Y. Xie, J. Mol. Catal. A 180 (2002) 293.
16] X. Zhang, J. Liu, Y. Jing, Y. Xie, Appl. Catal. A 240 (2003) 143.
17] Y. Liu, US Patent 6 436 871 (2002), to Symyx Technologies.
18] Y. Liu, US Patent 6 417 422 (2002), to Symyx Technologies.
19] A.K. Jonscher, Nature 267 (1977) 673.
[20] S. Lee, H.M. Cheong, N. Park, C.E. Tracy, A. Mascarenhas, D.K. Benson,
S.K. Deb, Solid State Ionics 140 (2001) 135.
21] A. Ponchel, A. D’Huysser, C. Lamonier, L. Jalowiecki-Duhamel, Phys.
Chem. Chem. Phys. 2 (2000) 303.
[
[
22] A.C. Faro, P. Grange, A.C.B. dos Santos, Phys. Chem. Chem. Phys. 4
(2002) 3997.
[23] J. Jehng, I.E. Wachs, Catal. Today 8 (1990) 37.

174
E. Heracleous, A.A. Lemonidou / Journal of Catalysis 237 (2006) 162–174
[
24] K. Tanabe, in: M. Boudart, J.R. Anderson (Eds.), Catalysis Science and
Technology, vol. 2, Springer, Berlin, 1983.
[34] E.V. Kondratenko, M. Baerns, Appl. Catal. A 222 (2001) 133.
[35] I.C. Marcu, J.M.M. Millet, J.M. Herrmann, Catal. Lett. 78 (2002) 273.
[36] L.M. Madeira, J.M. Herrmann, F.G. Freire, M.F. Portela, F.J. Maldonado,
Appl. Catal. A 158 (1997) 243.
[25] P.A. Cox, Transition Metal Oxides: An Introduction to Their Electronic
Structure and Properties, Oxford Univ. Inc., New York, 1992, pp. 170–
1
78.
[37] A.M.D. Farias, W. Gonzalez, P.G. Pries, J.G. Eon, J.M. Herrmann,
M. Aouine, S. Loridant, J.C. Volta, J. Catal. 208 (2002) 238.
[38] A.M. Turky, Appl. Catal. A 247 (2003) 83.
[39] A. Bielanski, R. Dzeinbaj, J. Slozynski, Bull. Acad. Pol. Sci. 14 (1966)
569.
[40] M.E. Dry, F. Stone, Discuss. Faraday Soc. 58 (1959) 192.
[41] M. Atanasov, D. Reinen, J. Electron. Spectrosc. Relat. Phenom. 86 (1997)
185.
[42] E. Antolini, Mater. Chem. Phys. 82 (2003) 937.
[43] T. Chen, W. Li, C. Yu, R. Jin, H. Xu, Stud. Surf. Sci. Catal. 130 (2000)
1847.
[
[
[
[
26] I. Nowak, M. Ziolek, Chem. Rev. 99 (1999) 3603.
27] J.M. Jehng, I.E. Wachs, Chem. Mater. 3 (1991) 100.
28] L.J. Burcham, J. Datka, I.E. Wachs, J. Phys. Chem. B 103 (1999) 6015.
29] J.M. Herrmann, in: B. Imelik, J.C. Védrine (Eds.), Catalyst Characteriza-
tion Physical Techniques for Solid Materials, Plenum Press, New York,
1994, p. 559.
[
[
30] G.H. Yi, T. Hayakawa, A.G. Andersen, K. Suzuki, S. Hamakawa, A.P.E.
York, M. Shimizu, K. Takehira, Catal. Lett. 38 (1996) 189.
31] Q.J. Ge, B. Zhaorigetu, C.Y. Yu, W.Z. Li, H.Y. Xu, Catal. Lett. 68 (2000)
59.
[
[
32] B. Zhaorigetu, W. Li, H. Xu, R. Kieffer, Catal. Lett. 94 (2004) 125.
33] P. Viparelli, P. Ciambelli, J.C. Volta, J.M. Herrmann, Appl. Catal. A 182
[44] M. Iwamoto, Y. Yoda, M. Egashira, T. Seiyama, J. Phys. Chem. 80 (1976)
1989.
(1999) 165.
[45] A. Bielanski, J. Haber, Oxygen in Catalysis, Dekker, New York, 1991.