294
H. Zhu et al. / Journal of Catalysis 285 (2012) 292–303
occupation factor of nickel and the isotropic atomic displacement
parameter of nickel and oxygen. Strong correlation between atom-
ic displacement parameters and site occupation factor prevented
the structure refinement to be conclusive about the occurrence of
bulk niobium doping.
Additional peaks localized at 2h = 30.43, 35.64, 41.22, 43.80, and
60.82°, characteristic of NiNb2O6 [ICSD card: 37212] [36,39], were
fitted with a pseudo-Voigt function and used to improve the over-
all quality of the refinement. Indeed, the latter peaks are in vicinity
of bunsenite reflections and tend to influence the calculation of
bunsenite crystallite size when not taken into account. Those reflec-
tions are considered as an amorphous NixNbyO phase which might
constitute a precrystalline network of the crystalline mixed oxide
NiNb2O6, that was identified by HRTEM-FFT studies described be-
low. Evolution of the amorphous content as a function of Nb con-
tent was roughly computed by calculating a pseudo degree of
ethylene are calculated on a carbon basis. To obtain different con-
version levels at constant temperature (330 °C), the W/F ratio is
varied from 0.05 to 0.6 g s/mL. For stability tests, gas mixtures of
10% C2H6/5% O2 in He are introduced through the catalytic bed held
at 330 °C, at a total flow rate of 10 mL/min (W/F = 0.6 g s/mL), and
the reaction mixture is sampled every 30 min for about 70 h.
4. Results and discussion
4.1. NiO catalysts
4.1.1. Morphology
Three methods were considered for the preparation of NiO cat-
alysts: precipitation of nickel nitrate in oxalic acid (Protocol 1),
precipitation of nickel chloride in ammonia (Protocol 2) and a
sol–gel method starting from nickel nitrate and citric acid (Protocol
3). The specific surface areas and porosities, as well as the corre-
sponding scanning electron micrographs, are given in Table 1 and
Fig. 1.
crystallinity q.
Total peak areas of crystalline phases
Total peak areas of crystalline phases þ Total peak areas of NixNbyO amorphous phase
q
¼
The morphologies of the materials prepared by precipitation are
consistent with the morphology obtained in [37] by precipitation
of nickel nitrate in ammonia, with similar scanning electron micro-
graphs (Fig. 1) and surface areas of 60–65 m2/g (Table 1). The
nickel oxide prepared by sol–gel exhibits a lower surface area of
31 m2/g. It is attributed to the presence of the more dense Ni
metallic phase (14 wt.% present in NiO_SG, as determined by XRD,
Table 2), which appeared after calcination at 450 °C in the form
of a shiny black micro-sheet on top of the black NiO powder. The
use of citric acid in the synthesis, which remains in the dry gel as
evidenced by TGA, indeed results in incomplete oxidation of NiO
after calcination at 450 °C, as observed in [40]. After removal of
the shiny nickel micro-sheet (<1 wt.% Ni left in NiO_SG (Ni re-
moved)), the surface area reaches a value of 48 m2/g. This is lower
than the 88 m2/g obtained in [37], despite the fact that the NiO
crystallite size (13 nm, Table 2) is similar (10 nm). It is also lower
than the theoretical surface (69 m2/g) expected for spheres of
NiO (density 6.67 g/cm3) of 13 nm in diameter. This indicates a cer-
tain degree of aggregation of the NiO crystallites, in agreement
with SEM observations (Fig. 1).
Rwp, RBragg, GOF were, respectively, in the range of 0.9, 0.7, and 1.8
for all refinements.
Imaging of samples was performed on a Titan G2 80–300 kV
transmission electron microscope (TEM) from FEI Company (FEI
Company, Hillsboro, OR) equipped with a 4 k ꢃ 4 k changed couple
device (CCD) camera model US4000 and an energy filter model GIF
Tridiem from Gatan, Inc. (Gatan Inc., Pleasanton, CA). Multiple
locations of the specimens are investigated using high-resolution
transmission electron microscopy (HRTEM) technique. The ac-
quired HRTEM micrographs were processed to obtain their diffrac-
tograms using Fast-Fourier transforms (FFT) technique to measure
the spatial frequencies present in the images. The measurement of
the spatial frequencies allowed obtaining the d-spacings of the
phases present in the sample with an error less than 5%. The
TEM was also set to scanning TEM (STEM) mode for carrying out
the electron energy loss spectroscopy (EELS) characterization of
sample with about ꢄ1 nm spatial resolution. Energy resolution of
GIF in EELS mode was nearly 1 eV, and furthermore, it was set to
an energy dispersion of 0.2 eV/channel. EELS signal from Oxygen
K edge (O-K edge) of 532 eV is acquired with convergence angle
of 9 mrad and collection angle of ꢄ20 mrad. GIF was also utilized
in energy-filtered TEM (EFTEM) mode for the distribution of Ni-
and Nb-based phases in the samples. Ni M-edge of energy 68 eV
and Nb M-edge of energy 205 eV were selected for their EFTEM
maps. Furthermore, each elemental map is created by using a 3-
window method of generating such maps in GIF.
4.1.2. Catalytic properties in ethane ODH
These nickel oxides prepared by three different methods have
been tested in low-temperature ODH of ethane (Fig. 2). Despite
lower surface area (48 m2/g) and presence of <1 wt.% Ni metal,
NiO_SG exhibits a higher efficiency than NiO_NH4OH, which is simi-
lar to that displayed by NiO_oxalic acid. The intrinsic activity of
NiO_SG, in terms of number of mole of ethane converted per surface
unit per time unit (Table 3, Entry 3), is similar to that reported for
nickel oxides prepared by the evaporation method [34,36]. But, gi-
ven the high selectivity of our catalyst (Table 3, Entry 3), the effi-
ciency of NiO_SG is 2–3 times higher than those reported so far,
3. Catalyst evaluation in low-temperature ethane oxidative
dehydrogenation (ODH)
Catalytic properties are evaluated at atmospheric pressure in
the oxidative dehydrogenation of ethane using a P and ID micro-
pilot equipped with a stainless steel reactor (internal diameter
4 mm). Gas mixtures of 10% C2H6/5% O2 and 10% C2H6/10% O2 in
He are introduced through the catalytic bed at a total flow rate
of 10 mL/min. Gas compositions are controlled by calibrated
mass-flow controllers. 100 mg catalyst was placed in the reactor
with glass wool as support (W/F = 0.6 g s/mL). It is ramped at
1°/min from 200 to 400 °C. The reaction mixture is sampled at
the outlet of the reactor at regular intervals, typically every
5 min, and analyzed with on-line Varian 490 micro-GC equipped
with a TCD detector and two columns: a MolSieve 5 Å column
(Ar as carrier gas) to quantify O2 (no CO is detected under our con-
ditions), and a poraPLOT Q column (He as carrier gas) to quantify
CO2, C2H4 as well as C2H6. Ethane conversion and selectivity to
with a surface-specific ethene production rate at 350 °C of
3.9 ꢃ 10ꢀ8 molethene/m2/s. Finally, given the low surface area of
the published nickel oxides (17 m2/g [34] and 12 m2/g [36]), the
mass-specific ethene production rate (Table 3, Entry 3) is one order
of magnitude higher on our 48 m2/g NiO_SG and ethene yields of 9%
and 24% are obtained at 300 °C and 350 °C, respectively (Table 3,
Entry 3).
These preliminary results clearly show the impact of the prep-
aration method on the structural and catalytic properties of nickel
oxides catalysts for low-temperature ethane ODH. The selectivity
of about 60% displayed by NiO_SG (held up to 25% conversion) is
significantly higher than selectivities normally reported for pure
nickel oxides (20% [34] – 30% [36]). It is normally only achieved
with composite materials, such as the recently reported nickel
oxide supported on high surface area zirconia, which managed to