Oxidative dehydrogenation of n-octane using vanadium-magnesium oxide catalysts with different vanadium loadings

Article abstract of DOI:10.1016/j.apcata.2009.11.004

In this study, vanadium-magnesium oxide (VMgO) catalysts with different vanadium loadings were synthesized by the impregnation method and characterized by BET, ICP-OES, IR, powder XRD, TEM, SEM, TPR-H₂, and TPD of ammonia. The catalysts were tested for the oxidative dehydrogenation (ODH) of n-octane in a continuous flow fixed bed reactor at GHSV of 6000 h^-1 and a temperature range of 350-550 °C using air as an oxidant and nitrogen as a make-up gas to give 9% n-octane in the gaseous mixture. The results showed that the vanadium concentration affects both textural and chemical properties of the catalysts and consequently their catalytic activity and selectivity towards the desired products (octenes and aromatics). The catalyst with a V₂O₅ concentration of 15% showed the best catalytic activity and productivity for the ODH products; especially for 1-octene and styrene.

Full text of DOI:10.1016/j.apcata.2009.11.004

Applied Catalysis A: General 373 (2010) 122–131
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Oxidative dehydrogenation of n-octane using vanadium-magnesium oxide
catalysts with different vanadium loadings
*
Elwathig A. Elkhalifa, Holger B. Friedrich
School of Chemistry, University of KwaZulu-Natal, Westville Campus, Private Bag X54001, Durban 4000, South Africa
A R T I C L E I N F O
A B S T R A C T
Article history:
In this study, vanadium-magnesium oxide (VMgO) catalysts with different vanadium loadings were
synthesized by the impregnation method and characterized by BET, ICP-OES, IR, powder XRD, TEM, SEM,
TPR-H2, and TPD of ammonia. The catalysts were tested for the oxidative dehydrogenation (ODH) of n-
octane in a continuous flow fixed bed reactor at GHSV of 6000 hÀ1 and a temperature range of 350–
550 8C using air as an oxidant and nitrogen as a make-up gas to give 9% n-octane in the gaseous mixture.
The results showed that the vanadium concentration affects both textural and chemical properties of the
catalysts and consequently their catalytic activity and selectivity towards the desired products (octenes
and aromatics). The catalyst with a V2O5 concentration of 15% showed the best catalytic activity and
productivity for the ODH products; especially for 1-octene and styrene.
Received 11 September 2009
Received in revised form 29 October 2009
Accepted 3 November 2009
Available online 10 November 2009
Keywords:
Oxidative dehydrogenation
n-Octane
Styrene
ß 2009 Elsevier B.V. All rights reserved.
Octenes
VMgO catalysts
1. Introduction
the system. A noticeable feature of the VMgO system is its
reluctance to form oxygenates and cracking products [3,12,13].
Medium and long chain linear paraffins are a major byproduct
of coal- and gas-to-liquids technology. With the increasing number
of these plants world-wide, the supply of these low value
chemicals is expected to rise rapidly and a need exists to convert
these to value-added products.
The oxidative dehydrogenation (ODH) over VMgO catalysts of
shorter alkanes to produce the corresponding olefins and dienes
has been an active area of research [1–4]. The phase that is
responsible for the ODH selectivity in the VMgO system, however,
is still a topic of debate. Some attribute this selectivity to the
orthovanadate [1,5–7], while others refer it to the pyrovanadate
Thus VMgO was considered a good candidate for the ODH of higher
alkanes, in this case n-octane, to produce the corresponding olefins
and aromatics.
Dehydrogenation of n-octane has previously been conducted
non-oxidatively [14–16], and when oxygen was employed the aim
was to produce hydrogen or syngas [17,18]. Recently, we reported
an oxidative dehydrogenation of n-octane to styrene using
catalysts derived from hydrotalcite-like precursors that contain
magnesium and vanadium [19]. To the best of our knowledge, no
work concerning the usage of VMgO in the ODH of n-octane has
been reported. The advantage of ODH over non-oxidative
dehydrogenation is that it is thermodynamically favourable [20].
However, because of the stability of the final products, viz. H2O and
CO2, controlling the reaction to intermediate stages (e.g. olefins
and aromatics) is a key challenge. In this approach, influencing the
residence time of the formed ODH products on the catalyst surface
is likely to affect the selectivity pattern of the products. This
residence time is, among other factors, a function of the phasic
composition of the catalyst, as well as its surface acid–base
character [1,3]. Thus vanadium concentrations may be used to
influence these two properties (phasic composition and the acid-
base character) without compromising the merits of the VMgO. In
this work, catalysts with different vanadium loadings have been
prepared and tested for the ODH of n-octane.
[8,9]. Also it has been suggested that the active phase is a
3À
monolayer of amorphous VO4
units scattered over MgO as
isolated and polymeric species [10]. The degree of interaction
between vanadium and magnesium oxide was found to affect the
vanadate phases formed [11]. Therefore changing factors like
vanadium loadings and/or the preparation procedure, e.g. the pH,
will result in different degrees of interaction between vanadium
and MgO. Consequently different proportions of the vanadate
phases will be formed and ultimately affect the ODH selectivity of
* Corresponding author. Tel.: +27 31 2603107; fax: +27 31 2603091.
E-mail address: friedric@ukzn.ac.za (H.B. Friedrich).
0926-860X/$ – see front matter ß 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2009.11.004

E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
123
2. Experimental
ture-programmed reduction (TPR) and NH3-temperature-pro-
grammed desorption (TPD) experiments were carried out in a
Micromeritics 2900 AutoChem II Chemisorption Analyzer. Prior to
the reduction in the TPR, the catalyst was pretreated by being
heated under a stream of argon (30 ml/min) at 400 8C for 30 min
and then cooled down to 80 8C under the same stream of argon. In
the reduction experiment, 5% H2 in argon was used as a reducing
agent at a flow rate of 50 ml/min. To account for the difference in
vanadium concentrations across the catalysts, different weights
were used in the analysis (18–94 mg). Under these reducing
conditions, the temperature was ramped up to 950 8C at a rate of
108/min. To investigate the resultant phase(s) of this reduction,
powder XRD was carried out on catalyst samples after being
reduced and the small sample size was overcome by using a blank
holder. In the TPD experiments, the catalyst (ca. 20 mg) was first
flushed at 350 8C with helium flowing at 20 ml/min for 30 min and
the temperature was thereafter brought down to 80 8C. A 4.1%
ammonia in helium gas mixture was then passed (20 ml/min) over
the catalyst for 30 min. The excess ammonia was removed by
flushing the system with helium (30 ml/min) for 30 min and the
adsorbed ammonia was then stripped off by the same stream of
helium (30 ml/min) and a temperature ramp of up to 900 8C (at
108/min) and the desorption profiles were recorded.
2.1. Synthesis
VMgO catalysts with different vanadium loadings of 5, 15, 25
and 50%, designated 5-, 15-, 25-, 50-VMgO, have been prepared by
impregnation of MgO with aqueous solutions of ammonium
vanadate. The vanadium loading expresses the weight percentage
of V2O5 in the total weight of both V2O5 and MgO. With regard to
the impregnation, the procedure has some similarities with that
reported in [1]. Thus, for the preparation of MgO, an aqueous
solution of magnesium acetate (53.3 g in one liter) was heated to
50 8C. A hot aqueous solution of oxalic acid, which contains the
exact stoichiometric amount needed to form MgC2O4 (31.5 g in
800 ml), was added to the acetate solution and the mixture stirred
for half an hour to allow for complete precipitation of the
magnesium oxalate. The resultant magnesium oxalate was filtered
off and the precipitate was washed thoroughly with cold and hot
water and placed overnight in an oven operated at 110 8C. MgO
was obtained by heating the magnesium oxalate at 600 8C for 6 h.
MgO so-obtained was impregnated with a clear hot aqueous
solution of NH4VO3 (greenish yellow in colour) that contains the
amount of vanadium needed for the required V2O5 loading. As an
example, for the catalyst 15-VMgO, 1.625 g of ammonium
vanadate was dissolved in 750 ml of water and heated at 60 8C
till the vanadate was completely dissolved. This solution was then
added (while hot) to a thick paste of MgO (7.133 g) and the
resultant mixture was heated, and magnetically stirred, till a paste
was formed. The resultant paste was then placed overnight in an
oven at 110 8C. The catalyst so-obtained was ground and calcined
at 550 8C for 5 h. Temperature ramping during the calcinations was
about 1.58/min until 550 8C was reached.
2.3. Catalytic testing
Catalyst testing was carried out in a continuous flow fixed bed
reactor (down flow mode). The reactor tube was stainless steel
with an internal diameter of 10 mm and a length of 220 mm. The
catalyst volume was 1 ml (ca. 0.45 g) sandwiched between two
layers of glass wool and all void spaces were packed with
carborandum. The catalyst pellets’ sizes were between 600 and
1000
mm. An electrically heated block, equipped with a tempera-
2.2. Characterization
ture controller and a thermocouple, was used to heat the reactor
tube. The temperature of the catalyst bed was monitored with a
coaxially centered thermocouple placed at the middle of the
catalyst bed. The concentration of n-octane in the gaseous mixture
was 9% (v/v) and its molar ratio to oxygen was 0.8. Air was used as
an oxidant and nitrogen as a make-up gas to give a total flow of
100 ml/min. n-Octane was delivered to the system by an HPLC
pump. The liquid products and the unreacted octane were
collected in a catchpot cooled to about 2 8C and the volume of
the gaseous components was measured by a Ritter Drum-Type Gas
The elemental composition of the catalysts was determined by
inductively coupled plasma optical emission spectroscopy (ICP-
OES) using an Optima 5300 DV PerkinElmer Optical Emission
Spectrometer. The BET surface areas were measured by nitrogen
physisorption isotherms at 77 K using the standard multipoint
method (eleven points) on a micromeritics Gemini instrument.
Prior to the analysis samples were degassed in a stream of nitrogen
at 473 K for 24 h. The infrared spectra were recorded at room
temperature using a PerkinElmer Spectrum 100 FT-IR Spectro-
meter fitted with a Universal ATR Sampling Accessory. A small
amount of the powder catalyst was placed on top of the ATR crystal
(composite of zinc selenide and diamond) and a pressure of about
120 Gauge was applied to allow for better contact between the
sample and the crystal. Powder X-ray diffraction patterns were
recorded on a Philips PW1050 diffractometer equipped with a
graphite monochromator and operated at 40 kV and 40 mA. The
Meter. The products were analyzed, off line, using
a gas
chromatograph (PerkinElmer Clarus500) equipped with both an
FID and a TCD. The TCD was used for the analysis of carbon oxides
and the FID for the rest of the products in both gaseous and liquid
phase. An Agilent 6890 Series GC System equipped with an Agilent
5973 Network Mass Selective Detector was used to identify the
catalytic test products, and further confirmation was done by
direct injection of the identified products in the GC. Blank tests
under experimental conditions indicate that the non-catalytic
contribution was negligible. Carbon balances ranged from 95 to
101% and all data points were obtained in duplicate.
source of radiation was Co Ka. The 2u covers the range between 10
and 708 at a speed of one degree per minute with a step size of 0.028
and all data were captured by a Sietronics 122D automated
microprocessor. Employing the peak at 508, the average crystallite
size was calculated using the Scherrer equation. The instrumental
broadening on this peak was accounted for by running a standard
3. Results and discussion
(quartz powder) at about the same
2
u.
Scanning electron
3.1. Catalyst synthesis
microscopy (SEM) images were obtained using a LEO 1450
Scanning Electron Microscope. Samples for SEM images were
coated with gold using a Polaron SC Sputter Coater. Transmission
electron microscopy (TEM) images were taken on a Jeol JEM 1010
Transmission Electron Microscope operated at a voltage of 100 kV.
Samples were prepared by deposition of a small amount of the
catalyst between two formvar coated copper grids and the images
were captured with MegaView III Soft Imaging System. Tempera-
An aqueous environment was chosen for the synthesis, though
some studies have opted for organic media [21,22]. Magnesium
oxide, with its isoelectric point of 11 to 12.7, forms a basic solution
that will interact with the ammonium vanadate solution.
Magnesium oxalate was first decomposed to magnesium oxide
prior to the impregnation as it was suspected that the oxalate
stabilizes the V4+ [23]. Before calcination all the catalysts were

124
E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
Table 1
the weakness of these bands implies that both phases are present
in small quantities. Similarly, the band at 816 cmÀ1 and the
shoulder at around 877 cmÀ1 in the spectrum of the 15-VMgO
Chemical composition and textural properties of VMgO catalysts.
Catalyst
Nominal
wt% of
V2O5
Actual wt%
of V2O5
(ICP)
Surface
area
(m2/g)
Crystallite
size from
XRD (nm)
Particle
size from
TEM (nm)
(Fig.
1) indicate the presence of both pyrovanadate and
orthovanadate, respectively. A noticeable feature is that the
presence of pyrovandate is more manifested compared to that
of the 5-VMgO catalyst. The spectrum of the 25-VMgO (Fig. 1) also
shows the existence of both phases but with different vibration
modes. This activation and deactivation of vibrations may probably
be due to different interactions and coexistence modes between
different phases as the vanadium concentration increases. With its
5-VMgO
15-VMgO
25-VMgO
50-VMgO
MgO
5
15
25
50
0
6.46
15.96
26.87
52.24
0
124
173
79
7.0
7.0
7.5
6.6
6.4
7.0
41
8.2
9.0
119
13.3
14.0
C2 point group of symmetry, the pyrovanadate could vibrate
y
white and after calcination they turned yellow, with the colour
intensifying as the vanadium concentration increases.
either through the terminal VO3 or the skeletal VOV bridge [25].
According to Kung and co-workers [1] and Sam et al. [8], the bands
at 667 and 831 cmÀ1 are due to the asymmetric stretching of the
VOV group in the pyrovanadate and the asymmetric stretching of
the VO4 group in the orthovanadate, respectively. In the high
vanadium loading, 50-VMgO, both pyro- and ortho-vanadate
became even more manifested as shown by Fig. 1(d). The band at
663 cmÀ1 is assignable to the asymmetric vibration of the VOV
group [1,8], and that at 825 cmÀ1 is attributable to the asymmetric
vibration of the VO4 group of the orthovanadate [24]. The slight
shift of these two bands towards lower wave numbers implies
more interaction between these phases at such high vanadium
concentration. It can be inferred from the above that both
pyrovanadate and orthovanadate are present in all the catalysts.
Moreover, these phases are more manifested with increasing the
vanadium concentration. No band attributable to V2O5 was
observed. In fact it has been reported that the interaction between
vanadium and magnesium inhibits the formation of V2O5 [26]. The
3.2. Characterization
As Table 1 reveals, there was no continuous trend between the
vanadium concentration and the resultant surface area, but it can
be noted that the incorporation of vanadium onto the magnesium
oxide causes an increase in the surface area to a certain point (15%)
and a noticeable decrease takes place thereafter.
3.2.1. Infrared spectroscopy
As Fig. 1 displays, the infrared spectrum of the 5-VMgO showed
two weak bands at around 816 and 877 cmÀ1. According to Sam et
al. [8], the band around 877 cmÀ1 is indicative of the orthovana-
date (asymmetric stretching of the VO4 group). The band at
816 cmÀ1 is attributable to the pyrovanadate phase [1,24]. This
indicates that both phases are present in the 5-VMgO. However,
Fig. 1. IR spectra of (a) 5-VMgO, (b) 15-VMgO, (c) 25-VMgO, and (d) 50-VMgO.

E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
125
incorporation of vanadium onto MgO caused an increase in the
surface area (Table 1). When reaching a higher vanadium
concentration (15-VMgO), the micrographs reflect a rough surface
with relatively small platelets and the vanadium seems to be
uniformly distributed over the catalyst surface. At higher
vanadium concentrations, i.e. 50-VMgO and to some extent 25-
VMgO, vanadium dispersion on the surface seems to be negatively
affected as white particles appear, implying that vanadium starts
to agglomerate on the catalyst surface. Generally, these SEM
images indicate that 15-VMgO possesses the most uniform
vanadium distribution on the catalyst surface. As Fig. 4 displays,
the TEM micrographs of the MgO shows stacked big rounded
particles. This feature is partly shown in the micrograph of the 5-
VMgO indicating that the surface is only sparingly covered with
vanadium. The micrographs of the 15- and 25-VMgO showed a
uniform distribution of the particles, while that of 50-VMgO
exhibits relatively large particles. As mentioned in Section 3.2.2,
there was a good comparison between the crystallite sizes
calculated by the Scherrer equation and the particle sizes
estimated from the TEM micrographs as shown in Table 1.
Fig. 2. XRD diffractograms; (a) MgO, (b) 5-VMgO, (c) 15-VMgO, (d) 25-VMgO, and
3.2.4. Temperature-programmed reduction
(e) 50-VMgO.
As appears from the TPR profiles in Fig. 5, the maximum
temperature for reduction increases parallel to the vanadium
concentrations, indicating that total reduction becomes more
difficult as the vanadium loading rises. With the exception of 5-
VMgO, however, the onset temperatures are within a comparable
range (Table 2). According to the assignments reported by Chang et
al. [30], our XRD investigations on the reduced catalysts (Fig. 6 and
Table 2) suggest, withthe exceptionof the 5-VMgO, the formation of
thespinelmagnesiumvanadate(Mg2VO4),indicatingthatvanadium
waspredominantlyreducedfromV5+ toV4+. Theexceptionrelatedto
5-VMgO was also reflected in its peculiar profile which shows two
peaks (and a shoulder) compared to one main peak for the rest. Each
of the 15- and the 25-VMgO TPR profiles show one main peak.
However, the peaks’ shapes, along with the hydrogen consumption
and the corresponding oxidation states of the reduced vanadium
(Table 2), indicatethatvanadium was predominantlyreduced toV4+
and that minor further reduction to V3+ was also took place. In the
lowest (5-VMgO) and the highest (50-VMgO) vanadium concentra-
tions, however, the situation was a little different, as these two
catalysts showed two peaks with a shoulder and a peak with a
shoulder, respectively. For the 5-VMgO, the first peak at 590 8C and
theshoulderat670 8Careprobablyduetothereductionofvanadium
in slightly different environments and, as indicated by the hydrogen
consumption, they are likely of the same type of reduction (V5+ to
V4+). The second peak, however, is presumably for a further
reduction step, as inferred from the hydrogen consumption and
from the absence of lines attributable to the formation of V4+ species
in the XRD of the reduced material (Table 2). The absence of the V4+
species in the reduced sample of this catalyst may be attributed to
the further reduction of V4+ and to low vanadium concentration in
this catalyst. In the extreme vanadium loading (50-VMgO), the main
peak at 755 8C and the shoulder at around 825 8C indicate that
vanadium is present in two different coordination environments.
This is in agreement with the remark that high vanadium
concentrations lead to the formation of octahedrally coordinated
species inaddition tothe tetrahedrallycoordinated ones[31]. Infact,
at such high vanadium concentration, coupled with low surface area
(41 m2/g), the presence of surface and bulk vanadium species is
conceivable; the shoulder may be attributed to the reduction of the
bulk vanadium species. The TPR of MgO did not show any peaks,
indicating that all the observed peaks are due to the reduction of
vanadium.
exposed V55O bond in this oxide is thought to be responsible for
oxygenates formation and deep oxidation that leads to the
formation of COx [13].
3.2.2. Powder X-ray diffraction
Fig. 2 displays the diffraction patterns of the four catalysts, as
well as magnesium oxide. The diffraction patterns of the catalysts
5-, 15-, and 25-VMgO show two lines characteristic of magnesium
´
˚
orthovanadate [1,27] at 2u equal 348 (d spacing of 3.04 A) and at
´
˚
428 (d spacing of 2.5 A). In all three catalysts no band characteristic
of pyrovanadate was recorded; IR, however, showed its existence.
This implies that pyrovanadate in these three catalysts is either of
low concentrations or poorly crystalline and highly dispersed. In
the extreme vanadium loading (50-VMgO), however, a crystalline
phase of pyrovanadate was detected, along with the orthovana-
date, by the lines at 2
respectively) [27]. The diffraction patterns of all the catalysts show
the existence of magnesium oxide with its characteristic line at 2
equal 508 (d spacing of 2.1 A). Again, and as shown by IR, no line
attributable to V2O5 or magnesium metavanadate was recorded.
The average crystallite sizes (Table 1) show irregular trends,
though they fall within narrow limits of 6.4–8.2 nm, with the
vanadium concentrations; an observation similar to this was
reported by Vidal-Michel and Hohn [28]. However, a drawback of
the line broadening technique is that it gives only the crystallite
´
˚
u of 33 and 368 (d spacings of 3.12 and 2.91A,
u
´
˚
size in the direction perpendicular to the plane of the chosen 2u,
besides the difficulty in estimating the broadening due to the
crystal deformation [29]; it is useful, therefore, to compare the
sizes obtained by this technique with sizes obtained by other
techniques such as TEM. The fair agreement between these average
sizes and particle sizes estimated by the TEM (Table 1) implies
that, in addition to the particles’ uniformity, there was no
significant contribution from the crystal deformation to the line
broadening used for the crystallite size calculation.
3.2.3. Electron microscopy
As portrayed by the SEM micrographs (Fig. 3), in the pure MgO
the surface displays large stacked plates and with the low
vanadium concentration (5-VMgO) these plates become a bit
smaller, probably by undergoing a fragmentation due to the
incorporation of vanadium. In support of this is that MgO showed
the highest crystallite and particle sizes and also the initial
Considering the charge balance of the reduction V5+ to V4+, and
assuming that the same type of reduction is likely to take place

126
E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
Fig. 3. SEM images (15,000 times magnification) of (a) MgO, (b) 5-VMgO, (c) 15-VMgO, (d) 25-VMgO, and (e) 50-VMgO.
during the catalyst testing, the VMgO system is likely to furnish the
less nucleophilic OÀ specie which is thought to enhance ODH
products over oxygenates’ formation [12]. In support of this is that
during the testing of the catalysts reported here only traces of
oxygenates were observed.
24 h. To investigate the catalysts’ stability, the temperature was
dropped down to 350 8C and ramped again to 550 8C and results
very similar to the original ones were obtained. The GHSV was kept
at 6000 hÀ1 throughout this study which corresponds to the flow
rates of 9, 54 and 37 ml/min for gaseous n-octane, air and nitrogen,
respectively.
3.2.5. Temperature-programmed desorption
Table 3 shows that magnesium oxide, though known for its
Brønsted basicity, possesses some sort of surface acidity, which is
probably of a Lewis nature. Indeed, ithas been reported that the MgO
surface appears to show some acidity and that ammonia can be
adsorbed ontothe MgO surfaceby theinteractionof itslonepair with
Mg2+ cations [32–34]. Each of the VMgO catalysts exhibits two types
of acidic sites; one around 450 8C (nominally denoted as weak acidic
sites), and the other at around 650 8C (strong acidic sites). As Table 3
shows, increasing vanadium loading causes the total acidity to
increase, as is the case for 5- and 15-VMgO, but with more vanadium
loadings (25- and 50-VMgO) the acidity started to drop. This
decreaseintheaciditymaybeattributedtothephasiccompositionof
the catalyst (namely the coexistence of pyro and orthovanadate)
and/or the decrease of the support (MgO) contribution to the acidity
as a result of the support coverage by the added vanadium.
3.3.1. Catalytic activity
Fig. 7a shows the variation of conversion with temperature for
different VMgO catalysts. For all the catalysts, the conversion
increases as the temperature rises. At low temperatures (350 and
400 8C), 5-VMgO shows highest catalytic activity. This activity at
low temperatures (especially at 350 8C) can be attributed to the
reducibility of this catalyst as reflected by its TPR profile (Fig. 5a)
which exhibits an early reduction peak at around 590 8C with an
onset temperature around 340 8C. However, as the temperature
increases, i.e. 450 8C and above, the catalyst 15-VMgO dominates
the other catalysts as it shows a noticeably higher conversion.
Unlike the 5-VMgO catalyst, the other three catalysts showed very
low conversion at 350 8C (around 2–3%). This is parallel to their
comparable onset temperatures as shown by the TPR results
(Fig. 5 and Table 2). An assessment of the oxygen conversion,
Fig. 7b, indicates that the system was operating under aerobic
conditions, which supports the assumption that this reaction is
predominantly oxidative, as it is believed that the non-oxidative
dehydrogenation over VMgO is likely to start when oxygen is
totally consumed [35].
3.3. Catalytic testing
The catalytic testing was carried out over the temperature
range of 350–550 8C with each temperature being held for around

E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
127
Fig. 4. TEM micrographs (500,000 times magnification) of (a) MgO, (b) 5-VMgO, (c) 15-VMgO, (d) 25-VMgO, and (e) 50-VMgO.
3.3.2. Product selectivity on VMgO catalysts
As Fig. 8a shows, octenes selectivity over the four catalysts is
generally high when compared to the aromatics selectivity
(Fig. 8b). This might be attributable to the basic nature of the
MgO. Apparently paradoxically, at low temperatures (350–450 8C)
and with the only exception of 50-VMgO at 400 8C, the octenes
selectivity increases as the vanadium loading increases (i.e. as MgO
concentration decreases). Alkenes selectivity in general is attri-
butable to a short residence time on the catalysts’ surface as the
quick desorption minimizes the chances for further undesirable
reactions, viz., cracking or secondary combustion to COx [1,3,11].
This short residence time on the catalyst surface at the high
vanadium loading (50% and to some extent 25% vanadium loading)
could either due to the acid–base character and/or to the phasic
composition of the catalyst. As Table 3 reveals, the total acidity
increases when moving from 5 to 15% V2O5 nominal concentration,
and as the V2O5 concentration reaches 25 and 50%, the acidity
starts to drop. This low acidity at high vanadium loadings is likely
to enhance the desorption of the formed octenes, which eventually
leads to a high selectivity to these compounds on the 25- and 50-
VMgO catalysts. In addition to the above, the IR and XRD
investigations in this study suggest that the magnesium pyrova-
nadate phase is manifested, along with the presence of orthova-
nadate, as the vanadium concentration increases. The possible
synergistic coexistence between these two phases could also
enhance the fast desorption of the formed octenes from the
catalyst surface and thereby increase their selectivity. At high
temperatures, 500 and 550 8C, this trend is somewhat disrupted, as
Fig. 5. TPR profiles of (a) 5-VMgO, (b) 15-VMgO, (c) 25-VMgO, and (d) 50-VMgO.

128
E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
Table 2
Summary of the peaks and the resultant phase in the TPR experiments.
´
a
˚
Catalyst
Onset temperature
Peak 1
Peak 2
Peak 3
H2 consumed
mol/g)
Oxidation state
of reduced V
XRD lines for Mg2VO4 in A
(
m
5-VMgO
340
475
490
510
590
638
685
755
670
805
–
332b and 431c
850
4.0 and 3.7
–
´
˚
15-VMgO
25-VMgO
50-VMgO
–
3.9
3.6
3.5
4.91 A
–
–
1980
4.86, 2.98,1.62
4.88, 2.98, 1.62
825 (shoulder)
–
4003
a
d-Spacing of the lines assigned to Mg2VO4 [30].
b
c
Hydrogen consumed for the first two peaks.
Total hydrogen consumed at the end of the TPR experiment.
the 15-VMgO showed the lowest octenes selectivity; however, 50-
VMgO still maintains the highest octenes selectivity. This
discontinuity in the octenes selectivity trend with vanadium
concentrations is likely due to the aromatics formation at these
high temperatures. In support of this is that 15-VMgO showed the
highest aromatics selectivity and correspondingly the lowest
octenes selectivity. The opposite was true for the 50-VMgO, as it
exhibited the highest octenes selectivity and the lowest selectivity
to aromatics. This suggests that octenes are precursors to the
aromatics, especially at high temperatures where the provision of
the energy required for this transformation is likely to be met.
Interestingly, only 5-VMgO forms aromatics at 3508 (Fig. 8b) as the
other three catalysts form aromatics from 400 8C and above. This is
probably because of the high reducibility (low reduction tem-
perature) of the 5-VMgO as inferred from its TPR profile (Fig. 5a).
As shown in Fig. 9a, selectivity of COx shows a slight decrease as
the temperature rises. This is possibly due to the formation of other
products; especially the aromatics. The stable aromatic nucleus in
styrene, ethylbenzene and xylene represents a driving force for the
formation of these compounds and thereby reduces the production
of carbon oxides, especially the contribution from secondary
combustion. Consistent with this is that there was a reverse
relation between aromatics and COx selectivity as shown in Figs. 8b
and 9a; 15-VMgO shows the highest aromatics selectivity increase
coupled with the most noticeable decrease in COx selectivity. The
same opposing trend could also be seen for COx and aromatics
selectivity over 50-VMgO. It may be concluded from this that
cyclization, which eventually leads to the aromatics formation,
provides an energetically favourable pathway for the reaction,
which in turn decreases the formation of carbon oxides. The
cracking products (C1 to C7 alkanes and alkenes) exhibit low
selectivity, around 2–3%, up to 500 8C (Fig. 9b), a steep increase,
however, takes place above this temperature.
3.3.3. Product selectivity at isoconversion
At isoconversion at 450 8C, as Fig. 10 reveals, octenes selectivity
increases, though slightly, as the vanadium concentrations
increase with the 50-VMgO showing the highest selectivity and
5-VMgO the lowest. This is presumably because of quick
desorption of the octenes from the catalyst surface as the
vanadium concentration rises. On the other hand, C8 aromatics
generally show an opposite trend to that exhibited by the octenes,
Fig. 6. XRD diffractograms of the reduced catalysts after TPR experiments, (a) the
sample holder, (b) 5-VMgO, (c) 15-VMgO, (d) 25-VMgO, and (e) 50-VMgO.
Table 3
Acidity of the different VMgO catalysts.
Catalyst
Weak acidic
sites (
Strong acidic
sites (
Total acidity
a
)
b
)
(mmol NH3/g)
(mmol NH3/g)
(mmol NH3/g)
MgO
0.713
1.054
1.127
0.911
1.121
0.702
0.778
0.730
0.880
0.367
1.415
1.832
1.857
1.790
1.488
5-VMgO
15-VMgO
25-VMgO
50-VMgO
Fig. 7. Variation of conversion with temperature at GHSV of 6000 hÀ1 for (a) n-
octane (b) oxygen.

E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
129
Fig. 10. Products selectivity on different catalysts at 450 8C and isoconversion
(ꢀ19%).
which again suggests that octenes are aromatics precursors, and
therefore fast desorption from the catalyst surface (high octenes
selectivity), as is the case at high vanadium concentration (50-
VMgO) will result in low aromatics selectivity. In contrast, at low
vanadium loadings, 5- and 15-VMgO, a relatively slow desorption
of the (basic) octenes from the catalyst surface lowers their
selectivity but at the same time causes the aromatics selectivity to
increase by providing more chance for the formed octenes to be
oxidized further to give the corresponding aromatics, and as
mentioned in Section 3.3.2. The fast (or slow) desorption of the
octenes from the catalyst surface could be attributed to the acid–
base character of the catalyst and/or its phasic composition.
Interestingly, the selectivity to octenes and C8 aromatics at this
isoconversion (19%) correlates well with both the acid–base
character of the catalysts (Table 3) and their phasic composition.
An interesting feature shown in Fig. 10 is that 15-VMgO showed
the highest selectivity for C8 aromatics and the lowest selectivity
for carbon oxides which, as mentioned in Section 3.3.2, indicates
that the aromatics formation provides an energetically favourable
pathway for the reaction, which in turn lowers the COx formation.
Cracking products selectivity for all the four catalysts was minimal.
At these comparative conversions (around 19%), the selectivity
to the octene isomers was found to be influenced by vanadium
loadings (Table 4). 15-VMgO gives the highest selectivity to 1-
octene. This selectivity, however, drops as the vanadium loadings
increase to 25 and 50%. 2-Octene (combined cis and trans isomers)
was shown to be the dominant isomer over all four catalysts with
its selectivity steadily increasing with increasing vanadium
loadings. As Table 4 shows, the selectivity to trans-2-octene is
always slightly higher than that to the cis-2-octene, this may be
attributed to the relative thermodynamic stability of trans isomers
in general compared to the corresponding cis isomers [36]. 3-
Octenes, on the other hand, exhibit a noticeably higher selectivity
on 50-VMgO if compared to the other three catalysts. Interestingly,
4-octenes were found to be the least produced isomers over all the
catalysts, indicating that n-octane activation at C4 was unfavour-
able despite the fact that the C–H bond at this carbon is equivalent
in energy to the other methylenic C–H bonds at C2 and C3 [37]. The
rate-determining step in alkane activation is the homolytic rupture
Fig. 8. Selectivity at 6000 hÀ1 for (a) octenes (b) C8 aromatics (styrene,
ethylbenzene and o-xylene).
Table 4
Selectivity of the octenes at 450 8C and isoconversion (ꢀ19%).
1-Octene trans-2-octene cis-2-octene 3-Octenes 4-Octenes
5-VMgO
10.8
7.3
7.4
8.2
9.2
6.3
7.2
7.4
8.1
10.5
10.2
10.1
13.7
3.5
3.5
3.9
4.7
15-VMgO 11.8
25-VMgO 11.3
50-VMgO 10.0
Fig. 9. Selectivity at 6000 hÀ1 for (a) COx and (b) cracking products (C1 to C7).

130
E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
of the C–H bond to form the surface alkyl species, which is then
followed by a fast elimination of a second hydrogen from a
neighbouring carbon to form the olefenic bond [37]. However, the
low selectivity to 4-octenes suggests that, beside the thermo-
dynamic element, factors like steric effects and the number of
active centers also influence the selectivity. In the activation of n-
octane, the steric effects are likely to play a vital role as the
accessibility to the active sites by such a large molecule is likely to
become a limiting factor and thereby have a significant impact on
the catalytic behaviour. The steric effects, besides the thermo-
dynamic factor, can also account for the observation that 3- and 4-
octenes, unlike 1-octene, are largely produced by the 50-VMgO
catalyst, as the presumably high VO4 density in this catalyst is
likely to enhance the activation at sterically hindered positions like
those of C4. Considering the alkane activation mechanism and by
inspecting the relative abundance of octenes isomers (Table 4), it is
plausible to assume that n-octane activation has occurred
predominantly at C2 and C3 and to a far lesser extent at C4.
Moreover, this activation is better explained by considering both
the thermodynamic factors and the steric effects.
Concerning the relative abundance of the C8 aromatics
(Fig. 11), styrene was constantly produced at higher selectivity,
followed by ethylbenzene and then o-xylene. The latter shows a
noticeably low selectivity over all four catalysts. A remarkable
feature is that 15-VMgO shows superior styrene selectivity over
the other three catalysts.
3.3.4. 1-Octene and styrene production
Fig. 12. The yield at 6000 hÀ1 (a) 1-octene and (b) styrene.
1-Octene and styrene are highly demanded products, and
therefore a special focus has been given to their production. As
Fig. 12a reveals, 15-VMgO is superior to the other three catalysts in
1-octene production, except at 350 8C where 5-VMgO shows the
highest yield, plausibly because of its high reducibility, as indicated
by its TPR profile, which allows the reaction to start at a relatively
low temperature. The yield of 1-octene over the 15-VMgO shows a
steady increase until 500 8C (2.6%) and a slight drop thereafter.
Generally, all the catalysts showed a sharp increase in the 1-octene
yield as the temperature rises from 350 to 450 8C and after that no
significant change was observed. As for the case for 1-octene, 15-
VMgO shows (Fig. 12b) the highest yield to styrene, showing a very
sharp increase from 400 to 500 8C (1.0–3.6%). A noticeable feature
shown in Fig. 12b is that 5-VMgO reaches a maximum styrene yield
at 450 while 25-VMgO shows a maximum yield at 500 8C. 50-VMgO,
on the other hand, shows an irregular behaviour in styrene
production with temperature. The irregularity in the catalytic
behaviour was a common feature for the catalyst 50-VMgO, as it
appears in the octenes and C8 aromatics selectivity (Fig. 8), as well
forCOx formation(Fig. 9a). Thisirregularityisprobablyrelatedtothe
extremely high vanadium concentration and, as shown by the XRD,
that a crystalline phase of Mg2V2O7 was formed only at this high
concentration, alongwithMg3V2O8. Itwasreportedthatmagnesium
pyrovanadate showed some structural instability and that during
typical experimental conditions some structural changes took place
[38]. In a conceivably similar behaviour, under the dynamic
conditions during the catalytic testing of the 50-VMgO, some
structural changes might be induced in the crystalline phase of
Mg2V2O7, which in turn lead to the observed irregularity in the
catalytic behaviour. In support of this is that the XRD trace of the
used 50-VMgO, unlike those of theother three catalysts, was marked
by the disappearance of two lines characteristic of magnesium
´
˚
pyrovanadate, at 2
u
of 338 (d spacing of 3.12 A) and 368 (d spacing of
´
˚
2.91 A), and moreover, a new line attributable to the pyrovanadate
´
˚
u equal 228 (d spacing of 4.68 A). This is in
was observed at 2
agreement with the observation reported in [38] that magnesium
pyrovanadate undergoes structural changes during the catalyst
testing. These structural changes are plausibly the cause of the
irregularity observed during the catalytic testing of the 50-VMgO.
3.3.5. Modes of cyclization
In the case of n-octane, the cyclization to form a six-member
carbon ring, and eventually the corresponding aromatics, could
either take place via C1 to C6 or C2 to C7 bonding. The former is
believed to lead to the formation of ethylcyclohexane, ethylbenzene
or styrene, while the latter will result in o-xylene formation [39,40].
Fig. 13 shows the combined percentage of the 1–6 cyclization mode
productsout of the selectivityof the totalcyclization products. In the
ODH of n-octane (Fig. 13), the cyclization has taken place
predominantly through the C1 to C6 mode; at low temperatures
(350–450 8C) more than 90% of the cyclization follows this mode of
cyclization. The tendency towards the C2 to C7 cyclization mode,
however, generally increases as the temperature rises. Noticeably,
the 50-VMgO catalyst showed the highest tendency towards C2 to
C7 cyclization, while15-VMgOexhibited the lowesttendencyto this
mode, especially between the temperatures 450 and 550 8C. It could
Fig. 11. Selectivity to C8 aromatics at 450 8C and isoconversion.

E.A. Elkhalifa, H.B. Friedrich / Applied Catalysis A: General 373 (2010) 122–131
131
Acknowledgements
We would like to thank the NRF and THRIP for support. Thanks
are also extended to the Electron Microscopy Unit at the University
of Kwazulu-Natal (Westville).
References
[1] M.A. Chaar, D. Patel, M.C. Kung, H.H. Kung, J. Catal. 105 (1987) 483–498.
[2] H.H. Kung, M.C. Kung, Appl. Catal. A 157 (1997) 105–116.
[3] T. Blasco, J.M. Lopez Nieto, Appl. Catal. A 157 (1997) 117–142.
[4] C. Pak, A.T. Bell, T.D. Tilley, J. Catal. 206 (2002) 49–59.
[5] W. Oganowski, W. Mista, Bull. Pol. Acad. Sci., Chem. 32 (1984) 181–193.
[6] M.A. Chaar, D. Patel, H.H. Kung, J. Catal. 109 (1988) 463–467.
[7] W. Oganowski, J. Hanuza, L. Kepinski, W. Mista, M. Maczka, A. Wyrostek, Z.
Bukowski, J. Mol. Catal. A 136 (1998) 91–104.
[8] D. Siew Hew Sam, V. Soenen, J.C. Volta, J. Catal. 123 (1990) 417–435.
[9] G.V. Isaguliants, I.P. Belomestnykh, Catal. Today 100 (2005) 441–445.
[10] A. Pantazidis, A. Burrows, C.J. Kiely, C. Mirodatos, J. Catal. 177 (1998) 325–334.
[11] T. Blasco, J.M. Lopez Nieto, A. Dejoz, M.I. Vazquez, J. Catal. 157 (1995) 271–282.
[12] A. Corma, J.M. Lopez Nieto, N. Paredes, J. Catal. 144 (1993) 425–438.
[13] L. Balderas-Tapia, I. Hernandez-Perez, P. Schacht, I.R. Cordova, G.G. Aguilar-Rios,
Catal. Today 107–108 (2005) 371–376.
Fig. 13. The percentage of the 1–6 cyclization mode out of the total cyclization.
[14] J. Fung, I. Wang, J. Catal. 130 (1991) 577–587.
be concluded from the above that C1 to C6 was the favourable
cyclization mode and the chance for the apparently unfavourable
mode (C2 to C7) increases as the temperature rises, presumably
because raising the temperature will provide more energy to
overcome any activation barrier.
[15] A. de Lucas, P. Sanchez, F. Dorado, M.J. Ramos, J.L. Valverde, Appl. Catal. A 294
(2005) 215–225.
[16] A. Szechenyi, F. Solymosi, Appl. Catal. A 306 (2006) 149–158.
[17] R. Subramanian, G.J. Panuccio, J.J. Krummenacher, I.C. Lee, L.D. Schmidt, Chem.
Eng. Sci. 59 (2004) 5501–5507.
[18] K.A. Williams, L.D. Schmidt, Appl. Catal. A 299 (2006) 30–45.
[19] H.B. Friedrich, A.S. Mohamed, Appl. Catal. A 347 (2008) 11–22.
[20] M.M. Bhasin, J.H. McCain, B.V. Vora, T. Imai, P.R. Pujado, Appl. Catal. A 221 (2001)
397–419.
4. Conclusions
[21] L. Albaric, N. Hovnanian, A. Julbe, G. Volle, Polyhedron 20 (2001) 2261–2268.
[22] I.V. Mishakov, A.A. Vedyagin, A.F. Bedilo, V.I. Zaikovskii, Catal. Today 144 (2009)
278–284.
[23] A. Corma, J.M. Lopez Nieto, N. Paredes, Appl. Catal. A 104 (1993) 161–174.
[24] J. Hanuza, B. Jerowska-Trzebiatowska, W. Oganowski, J. Mol. Catal. 29 (1985)
109–143.
[25] K. Nakamoto, Infrared Raman Spectra of Inorganic and Coordination Compounds,
fourth ed., John Wiley and Sons, United States of America, 1986.
[26] L. Balderas-Tapia, J.A. Wang, I. Hernandez-Perez, G.G. Aguilar-Rios, P. Schacht,
Mater. Lett. 58 (2004) 3034–3039.
[27] R. Burch, E.M. Crabb, Appl. Catal. A 100 (1993) 111–130.
[28] R. Vidal-Michel, K.L. Hohn, J. Catal. 221 (2004) 127–136.
[29] J.R. Anderson, M. Boudart, Catalysis Science and Technology, 5, Springer-Verlag,
Berlin, Heidelberg, 1984, pp. 221–273.
Both textural and chemical properties of the VMgO catalysts
were found to be affected by vanadium loadings. The crystalline
Mg3V2O8 phase was formed at all vanadium loadings, while
crystalline Mg2V2O7 was detected only at high vanadium
concentration (50-VMgO); however, the formation of its amor-
phous form could be inferred from the IR. TPR showed that Tmax of
reduction increases as the vanadium loading increases. However,
the onset temperatures of reduction were comparable, except for
5-VMgO, and they explained better the catalytic activity.
The partial oxidation of n-octane over VMgO catalysts produced
mainly (in addition to carbon oxides) octenes, styrene, ethylben-
zene and to a lesser extent xylene and some cracking products. 2-
octenes were found to be the dominant octane isomers while 4-
octenes were the least formed ones, indicating that n-octane
activation was favourable at C2 and C3, and unfavourable at C4.
The relative abundance of different octene isomers was found to be
affected by vanadium concentration, e.g. 1-octene was the
dominant isomer over 15-VMgO, while over 50-VMgO 3-octene
was the dominant isomer with 1-octene being the least formed
one. 50-VMgO showed the highest selectivity to the octenes and
correspondingly the lowest to the aromatics. Moreover, cyclization
occurred predominantly via C1 to C6 which led to the predomi-
nance of styrene and ethylbenzene over xylene.
[30] W.S. Chang, Y.Z. Chen, B.L. Yang, Appl. Catal. 124 (1995) 221–243.
[31] O.R. Evans, A.T. Bell, T.D. Tilley, J. Catal. 226 (2004) 292–300.
[32] S. Coluccia, E. Garrone, E. Borello, J. Chem. Soc., Faraday Trans. 1 (79) (1983) 607–
613.
[33] S. Coluccia, S. Lavagnino, L. Marchese, J. Chem. Soc., Faraday Trans. 1 (83) (1987)
477–486.
[34] M. Calatayud, A. Markovits, M. Menetrey, B. Mguig, C. Minot, Catal. Today 85
(2003) 125–143.
[35] N. Ballarini, A. Battisti, F. Cavani, A. Cericola, C. Cortelli, M. Ferrari, F. Trifiro, P.
Arpentinier, Appl. Catal. A 307 (2006) 148–155.
[36] R.T. Morrison, R.N. Boyd, Organic Chemistry, Sixth ed., Prentice-Hall, New Jersey,
1992, pp. 317–366.
[37] B.K. Hodnett, Heterogeneous Catalysis Oxidation: Fundamental and Technologi-
cal Aspects of the Selective and Total Oxidation of Organic Compounds, Wiley,
2000, pp. 65–101.
[38] S. Sugiyama, Y. Hirata, K. Nakagawa, K.-I. Sotowa, K. Maehara, Y. Himeno, W.
Ninomiya, J. Catal. 260 (2008) 157–163.
[39] P. Meriaudeau, A. Thangaraj, C. Naccache, S. Narayanan, J. Catal. 146 (1994) 579–
582.
15-VMgO was found to superior to the other catalysts in activity
as well as in the production of 1-octene and styrene.
[40] B. Shi, B.H. Davis, J. Catal. 157 (1995) 626–630.