Active coke: Carbonaceous materials as catalysts for alkane dehydrogenation

Article abstract of DOI:10.1016/j.jcat.2009.11.016

The catalytic dehydrogenation (DH) and oxidative dehydrogenation (ODH) of light alkanes are of significant industrial importance. In this work both carbonaceous material deposited on VO_x/Al₂O₃ catalysts during reaction and unsupported carbon nanofibres (CNFs) are shown to be active for the dehydrogenation of butane in the absence of gas-phase oxygen. Their activity in these reactions is shown to be dependent upon their structure, with different reaction temperatures yielding structurally different coke deposits. Terahertz time-domain spectroscopy (THz-TDS), among other techniques, has been applied to the characterisation of these deposits - the first time this technique has been employed in coke studies. TEM and other techniques show that coke encapsulates the catalyst, preventing access to VO_x sites, without a loss of activity. Studies on CNFs confirm that carbonaceous materials act as catalysts in this reaction. Carbon-based catalysts represent an important new class of potential catalysts for DH and ODH reactions.

Full text of DOI:10.1016/j.jcat.2009.11.016

Journal of Catalysis 269 (2010) 329–339
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Active coke: Carbonaceous materials as catalysts for alkane dehydrogenation
James McGregor a, , Zhenyu Huang , Edward P.J. Parrott , J. Axel Zeitler , K. Lien Nguyen ,
a
a,b
a
a
*
c
d
e,1
e
e
Jeremy M. Rawson , Albert Carley , Thomas W. Hansen , Jean-Philippe Tessonnier , Dang Sheng Su ,
e
e,2
e
e
a
Detre Teschner , Elaine M. Vass , Axel Knop-Gericke , R. Schlögl , Lynn F. Gladden
University of Cambridge, Department of Chemical Engineering and Biotechnology, Cambridge CB2 3RA, UK
University of Cambridge, Cavendish Laboratory, Cambridge CB3 0HE, UK
University of Cambridge, Department of Chemistry, Cambridge CB2 1EW, UK
Cardiff University, School of Chemistry, Cardiff CF10 3AT, UK
Fritz-Haber-Institut der Max-Planck-Gesellschaft, Berlin D-14195, Germany
a
b
c
d
e
a r t i c l e i n f o
a b s t r a c t
Article history:
The catalytic dehydrogenation (DH) and oxidative dehydrogenation (ODH) of light alkanes are of signif-
icant industrial importance. In this work both carbonaceous material deposited on VO /Al catalysts
Received 29 August 2009
Revised 12 November 2009
Accepted 17 November 2009
Available online 6 January 2010
x
2 3
O
during reaction and unsupported carbon nanofibres (CNFs) are shown to be active for the dehydrogena-
tion of butane in the absence of gas-phase oxygen. Their activity in these reactions is shown to be depen-
dent upon their structure, with different reaction temperatures yielding structurally different coke
deposits. Terahertz time-domain spectroscopy (THz-TDS), among other techniques, has been applied to
the characterisation of these deposits – the first time this technique has been employed in coke studies.
Keywords:
Vanadia catalysts
Carbon nanofibres
THz time domain spectroscopy
Coke
Transmission electron microscopy
Butane dehydrogenation
X-ray absorption spectroscopy
x
TEM and other techniques show that coke encapsulates the catalyst, preventing access to VO sites, with-
out a loss of activity. Studies on CNFs confirm that carbonaceous materials act as catalysts in this reaction.
Carbon-based catalysts represent an important new class of potential catalysts for DH and ODH reactions.
Ó 2009 Elsevier Inc. All rights reserved.
1
. Introduction
x 2 3
the activity of coked VO /Al O and carbon nanofibres. To this
end a wide number of advanced techniques are employed includ-
ing the first application of terahertz time-domain spectroscopy
(THz-TDS) in the study of coke.
The role of carbonaceous deposits in DH and ODH reactions has
been the subject of extensive research. In the case of industrial al-
kane dehydrogenation significant coke deposition causes rapid
deactivation of the catalyst. For instance, the CATADIENE process
The catalytic dehydrogenation (DH) of light alkanes in the ab-
sence of oxygen is employed as an industrial process to synthesise
alkenes and alkadienes, important precursor molecules for syn-
thetic rubbers, plastics and a variety of other products. As such, an-
nual consumption (2003) of propane and n-butane feedstocks for
these reactions exceeds 10 Mt [1]. Recent research in this area
has focused on butane DH over alumina-supported vanadia cata-
lysts [2–8]. Additionally, the oxidative dehydrogenation (ODH) of
light alkanes also receives significant academic attention [9–12].
However, the relatively low yields achieved in the latter approach
mean that this process is yet to be implemented commercially
x 2 3
(ABB Lummus) employs a CrO /Al O catalyst operating at 873 K:
this catalyst requires regeneration after ꢀ10 min on stream as a re-
sult of coke build-up causing deactivation [1]. Laboratory studies of
VO
x
/Al
2
O
3
catalysts have drawn correlations between the nature of
units and the structure of the coke formed [5].
the supported VO
x
[
13]. In this work, we show for the first time that carbonaceous
In contrast to many other reaction systems however, in DH reac-
tions polyaromatic coke deposits are not always directly impli-
cated in catalyst deactivation. Recent studies have demonstrated
that such large ensembles play only a minor role in deactivation
during n-butane dehydrogenation. Instead, the observed loss in
activity can be assigned to the presence of tightly bound reaction
intermediates, irreversibly adsorbed onto catalytic active sites un-
der reaction conditions [4]. Furthermore, in the oxidative dehydro-
genation of ethylbenzene, such carbonaceous deposits are not only
not implicated in catalyst deactivation, but instead have been
material, coke, deposited during reaction over VO /Al can
act as an active catalyst in the non-oxidative dehydrogenation of
n-butane. This conclusion is confirmed through a comparison of
x
2 3
O
*
Corresponding author. Fax: +44 (0) 1223 334796.
E-mail address: jm405@cam.ac.uk (J. McGregor).
Present address: Centre for Electron Nanoscopy, Technical University of Denmark,
1
DK-2800 Lyngby, Denmark.
2
Present address: Johnson Matthey Catalysts, P.O. Box 1, Belasis Avenue, Billing-
ham TS23 1LB, UK.
0
021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.11.016

3
30
J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
ꢀ
ꢁ
suggested as being catalytically active in their own right, a
conclusion reinforced by studies on activated carbon catalysts
no: of moles of butane consumed
no: of moles of butane introduced
%
%
Conversion ¼ 100 ꢂ
[
14–21] and nanostructured carbon materials [22–24]. More re-
ð1Þ
ð2Þ
cently carbonaceous materials have also been shown to be active
in the ODH of n-butane [25]. Ketonic C@O functionalities on the
surface of CNTs are believed to be the active site in this system.
Notably, significant deactivation was not observed over a period
of 100 h.
A wide range of experimental techniques have been applied to
the study of coke and carbonaceous materials in the literature.
One reason for this is that no single technique can provide the
wealth of information required to understand the nature of such
complex structures. For this reason, in this work we have applied
a number of complementary techniques to the study of coke
ꢀ
ꢁ
no: of moles of product formed
no: of moles of butane introduced
Yield ¼ 100 ꢂ
ꢀ
ꢁ
no: of moles of product formed
no: of moles of butane consumed
% Selectivity ¼ 100 ꢂ
ð3Þ
2.3. Characterisation techniques
Electron microscopy investigations were carried out in a Philips
CM200 transmission electron microscope operated at 200 kV. The
images were recorded using a Gatan Tridiem GIF. Images were ana-
lysed using the DigitalMicrograph package from Gatan Inc. The
samples were lightly ground using a mortar and pestle and dis-
persed on a carbon-coated copper grid. Elemental mapping was
carried out in the same microscope using an EDAX Genesis EDX
detector. The elemental maps were processed using Image-Pro
Analyzer from Media Cybernetics Inc.
x 2 3
formed over VO /Al O during butane dehydrogenation. One such
method is THz-TDS, and this work represents the first application
of this spectroscopy in the study of coke.
2
. Experimental
2.1. Materials
The principal catalyst employed in the work reported herein is
X-ray diffraction (XRD) measurements were performed in
transmission mode on a STOE STADI P diffractometer equipped
with a primary focusing Ge monochromator (Cu Ka radiation)
1
an alumina-supported vanadia catalyst with a vanadium loading
of 3.5 wt.%, hereafter referred to as VO /Al . This catalyst has
x
2
O
3
previously been described in the literature [2–5,26], and was pre-
pared by incipient wetness impregnation employing aqueous
and position sensitive detector. The samples were mounted in
the form of small amounts of powder sandwiched between two
layers of polyacetate film and fixed with a small amount of X-ray
amorphous grease.
X-ray photoelectron spectra were measured on a Kratos Axis Ul-
tra DLD spectrometer using monochromatised Al Ka radiation and
NH
4 3 2 3
VO (>99%, Aldrich). The alumina support is h-Al O (Johnson
2
ꢁ1
Matthey, UK, BET surface area = 101 m g , pore volume = 0.60
ml g ). Oxalic acid (99% Aldrich) was added to the impregnation
solution to ensure the dissolution of NH VO [NH VO /oxalic
ꢁ1
4
3
4
3
acid = 0.5 (molar ratio)]. After impregnation the catalyst precursor
was mixed thoroughly for 2 h at 350 K to ensure a more homoge-
neous distribution of vanadia on the support. The catalyst was
then dried in air at 393 K overnight and calcined for 6 h at
a pass energy of 40 eV. Data were analysed with CasaXPS and soft-
ware developed in-house at Cardiff University.
1
3
Solid-state C NMR was conducted on the coked catalysts using
1
3
Cross Polarisation Magic Angle Spinning (CP-MAS) at a C operat-
ing frequency of 100.65 MHz on an AV-400 spectrometer (Bruker).
The spinning rate was 14 kHz, and a contact time of 1 ms and a re-
cycle delay of 1 s were used. 50,000 signal scans were averaged for
8
23 K, again in air. Prior to use the catalyst extrudates were
ground and sieved to a particle size of 75–90 m. Additional cata-
l
lytic tests have been conducted over a carbon nanofibre (CNF)
material (PR24-LHT, Pyrograf Products Inc.). Full details about
the synthesis and the structure of these nanofibres can be found
in the review by Tibbetts et al. [27].
A number of model compounds have been studied by THz-TDS
in order to assist in the interpretation of spectra. These are anthra-
cene (99%), phenanthrene (99.5%), 2,3-benzanthracene (98%), 1,2-
benzanthracene (99%), triphenylene (98%), chrysene (98%), coron-
ene (97%) and graphite (all Sigma–Aldrich).
2
each sample. Spectra are referenced to a solid CH adamantane
shift at 38.54 ppm relative to tetramethylsilane (TMS). An expo-
nential line broadening with a factor of 100 Hz was applied to all
data acquired before Fourier transformation to yield the spectral
data. Experimental EPR spectra were measured on a Bruker ER-
200D series EPR spectrometer at room temperature in the region
200–6200 G with a microwave frequency of 9.34 GHz (X-band).
Carbon K-edge X-ray absorption (XAS) experiments were per-
formed in the total electron yield mode in the in situ XPS setup
2.2. Catalytic activity measurements
x 2 3
of FHI at BESSY, Berlin. VO /Al O samples were pressed into a pel-
let and the absorption of the carbonaceous material deposited as a
result of the butane dehydrogenation reaction was recorded under
0.25 mbar He at room temperature. To correct for the absorption of
carbon on the optical elements of the beamline, a reference exper-
iment was carried out with a cleaned (carbon free) Ag foil sample.
For Raman spectroscopy experiments the samples were intro-
duced into an Au coated stainless steel sample holder with a
Catalytic activity data were acquired using a fixed-bed, contin-
uous flow reactor connected to an on-line GC (Agilent 6890 Series,
3
FID, column Agilent HP-5). The catalyst (1.5 cm ) was heated
ꢁ
1
ꢁ1
(
5 K min ) to 973 K in 5% O
2 2
/N (0.5 bar g, 40 ml min ) and held
ꢁ1
at this temperature for 2 h. A flow of He (0.5 bar g, 42 ml min
)
was then established and the temperature adjusted to the desired
reaction temperature (at least 30 min). For reactions over the CNF
the pre-treatment step was conducted in flowing He (0.5 bar g,
2
0.6 mm deep rectangular well covering an area of (12 ꢂ 8) mm .
Raman spectra were measured at room temperature using
ꢁ
1
ꢁ1
4
2 ml min ) in place of O
2
/N
2
to avoid functionalisation or com-
was then
514 nm laser excitation (3 mW) at 5 cm spectral resolution (Kai-
bustion of the nanofibres. In all cases, 3% n-C
introduced (0.5 bar g, 60 ml min ) for a period of 180 min. GC
measurements were taken at regular intervals. Conversion, yield
4
H
10/N
2
ser Optical). Sampling times were typically 50 min. Prior to the
experiments, the Raman spectrometer was calibrated using an Ar
lamp. For background subtraction the spectrum of the bare sample
holder was used. The spectra were fitted with four components (G,
ꢁ
1
4
of C hydrocarbons and selectivity to 1-butene and 1,3-butadiene
ꢁ1
were calculated using Eqs. (1)–(3), respectively. After 180 min
the catalyst was cooled to room temperature in flowing He and
removed for ex situ analysis. Conversion, yield and selectivity are
defined as follows:
D, D3, and D4) [28] without the D2 band at ꢀ1620 cm as this
ꢁ1
band was not distinguishable from the G band at 1600 cm
.
The percentage of carbon, hydrogen and nitrogen present in the
catalysts was determined by microanalysis, performed by the

J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
331
Microanalytical Department, Department of Chemistry, University
of Cambridge.
solid-state NMR and FT-IR spectroscopy [2,3,26,32,33]. These data
demonstrated that the main species present on the catalyst surface
were polymeric vanadates, however these co-existed with smaller
THz-TD spectra were measured on a spectrometer previously
described [29,30]. A Femtosource Ti:sapphire laser (Femtolasers)
pumped by a Millenia Xs CW laser (Spectra Physics) is used to pro-
duce 12 fs laser pulses centred at approximately 800 nm with a
bandwidth >100 nm. This laser beam is then separated by a beam
splitter into a pump beam and a probe beam. The pump beam is
directed towards the surface of a biased GaAs semiconductor emit-
ter, resulting in the generation of pulses of THz radiation of 0.3–
2 5
quantities of both monomeric vanadates and crystalline V O -like
species. Raman spectroscopy allows for the quantitative determi-
nation of the amounts of each of these species present [34]. In this
particular case, the ratio of polymeric vanadates to monomeric
2 5
species to V O is 60:30.2:9.8. That these types of vanadate species
are present on alumina-supported vanadia catalysts is well-estab-
lished. A discussion of the structure of such species and their role
in catalysis is provided by Weckhuysen and Keller [35].
3
.5 THz. The probe beam is used for the detection of THz radiation.
An electro-optic ZnTe crystal is employed as the detector. After
homogeneously mixing the sample powder with polyethylene
powder (Sigma–Aldrich), the mixture was compressed into a flat-
faced pellet. Polyethylene was used as a bulking agent and is trans-
parent to THz radiation in the frequency range of interest. The pre-
pared pellets were scanned one hundred times over a 13 ps time
window with each scan taking approximately 1 min, giving a spec-
tral resolution of 75 GHz. Comparison of the power spectrum of the
sample with that of a reference pellet containing polyethylene
using the Beer–Lambert Law results in a value for the frequency-
dependent absorbance of the sample [31].
x 2 3
3.2. Catalytic activity of VO /Al O
The catalytic performance of VO
723, 823, 873, 898, 923, 948 and 973 K. The yield of C
bons is shown as a function of time-on-stream for a selection of
these temperatures in Fig. 2a, while conversion and selectivity to
1-butene and to 1,3-butadiene are shown in Figs. 2b and c, respec-
/Al
x 2
O
3
has been evaluated at
hydrocar-
4
4
tively. It is apparent that the yield of C products decreases as a
function of time-on-stream at 873 K and below, indicative of the
deactivation of the catalyst. At higher temperatures however, the
catalyst appears to activate with increasing exposure to the reac-
tant and the yield increases. This difference is apparent when com-
paring reaction at 823 and 973 K, with the observed yield being
3
. Results and discussion
4
0% after 180 min at 973 K but only 3% at 823 K. Conversion and
3
.1. Characterisation of VO
x
/Al
2
O
3
selectivity are also greatest at high reaction temperatures. Again
comparing 823 and 973 K, conversions of 13% and 50%, respec-
tively, are achieved after 180 min, while the selectivity to 1-butene
reaches values of 6% and 48%. Butadienes are not observed as a
product in the initial stages of reaction. However, after 180 min,
selectivity towards 1,3-butadiene is comparable to that towards
1-butene with values of 5% and 32% at 823 and 973 K. Note that
TEM and XRD have been applied to the characterisation of the
as-prepared VO /Al catalyst. Additionally, TEM has been em-
x
2 3
O
ployed to study the catalyst after calcination in oxygen prior to
reaction. TEM investigation of the as-prepared sample revealed
no crystalline vanadia particles. However, EDX spectroscopy re-
vealed a vanadium content as expected from the synthesis. These
observations suggest a high dispersion of vanadium in the as-pre-
pared samples. After calcination, VO particles of sizes below 1 nm
x
are observed by TEM, but show no crystallinity. These particles are
uniformly dispersed throughout the support material. To further
4
the low selectivities towards C species at lower temperatures is
due to a greater selectivity to coke deposition at this stage of the
reaction. That butadiene is not produced initially may be indicative
of the formation of more active dehydrogenation sites in situ as the
vanadia species undergo reduction through interaction with n-bu-
tane and coke builds up on the surface. That the increase in buta-
diene production is not concomitant with a decrease in 1-butene
selectivity suggests that the two products are formed from n-bu-
tane at different catalytic sites. Significantly, from between 80
and 110 min on stream at 973 K, selectivity, conversion and yield
all remain constant and do not deactivate with increasing exposure
to n-butane. It is also notable that significant formation of small
hydrocarbons as a result of the cracking of the feed molecule is
x
investigate the dispersion of VO in the sample, elemental mapping
using scanning TEM (STEM) was used. Fig. 1a shows a representa-
tive TEM micrograph and EDX elemental mapping images indicat-
ing the distribution of carbon, oxygen, aluminium and vanadium in
the calcined catalyst particles. Fig. 1b shows the same data for the
catalyst after reaction at 973 K and these data are discussed in Sec-
tion 3.4.2. As expected from the TEM investigations, the mapping
(
Fig. 1a) showed a homogeneous distribution of V on the support
material after calcination. However, the instrument resolution is
insufficient to determine if V is present as a film or small particles.
That no change is apparent in the distribution of vanadium species
after calcination is important, as this process is carried out at
x 2 3
not observed over VO /Al O . This may be due to the removal of
acid sites on the support during the high-temperature calcination
step.
The activation energy for 1-butene formation has been calcu-
lated by means of an Arrhenius relationship, as shown in Fig. 3.
The rate of 1-butene production is calculated (in mmol per gcat)
9
73 K, 10 K above the melting point of bulk V
ported V is however, known to have different physical proper-
ties to bulk V . For instance, TPR experiments show that the
supported material undergoes reduction more easily [26].
In addition to characterising the supported VO species, XRD
2 5
O . Alumina-sup-
2
O
5
2
O
5
in the early stages of reaction, before significant coke forms on
the catalyst surface. From these data an activation energy of
ꢁ1
x
132 ± 16 kJ mol is calculated. This value is similar to values pre-
viously established for alkane dehydrogenation over supported
metal-oxide catalysts. For instance, Airaksinen et al. calculated
has been employed in order to confirm the phase of the alumina
support. Rietveld fits to the acquired data were performed using
the Bruker AXS software Topas, version 3.0. Employing a preferred
orientation model (sixth order spherical harmonics), it was dem-
ꢁ1
activation energies in the range 132–142 kJ mol for isobutene
dehydrogenation over CrO /Al [36].
x
2 3
O
onstrated that the majority of the crystalline fraction is h-Al
however minor quantities of -Al are also present. This is in
agreement with the previous assignment of this material as h-
Al [3–5].
Previous characterisation of this catalyst has been conducted by
2
O
3
,
Additionally, the n-butane dehydrogenation reaction has been
run at 973 K in an empty reactor, i.e. an identical experimental pro-
tocol to that described above was employed but no catalyst was
loaded into the reactor. GC analysis confirms that no conversion
of n-butane occurs, confirming that there is no contribution from
a gas-phase, homogeneous, reaction mechanism to the results de-
scribed herein in the absence of a catalyst.
a
2 3
O
2 3
O
both UV- and vis-Raman spectroscopy, in addition to UV–vis
spectroscopy, temperature-programmed reduction measurements,

3
32
J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
Fig. 1. (a) TEM of VO
particles. (b) TEM of VO
particles.
x
/Al
2
O
3
after calcination at 973 K and EDX elemental mapping images showing distribution of carbon, oxygen, aluminium and vanadium in VO
x
/Al
/Al
2
O
O
3
3
x
/Al O after reaction at 973 K and EDX elemental mapping images showing distribution of carbon, oxygen, aluminium and vanadium in VO
2 3
x
2
3
.3. Characterisation of VO
x
/Al
2
O
3
post-reaction
rather complex, however certain features around g ꢀ 2 are consis-
4+
tent with an axial V ion with evidence of partially resolved
5
1
VO /Al has also been characterised after use for the dehydro-
x
2
O
3
hyperfine coupling to V (100% I = 7/2). Attempted simulation
4
+
genation of n-butane, at temperatures ranging from 723 to 973 K.
TEM and EPR measurements reveal the effect of reaction on the
supported VO
using ‘‘standard parameters” for axial V (g 6 2.0 [37,38]) ex-
plained many of the features in the central region but the peaks
to lower field (an intense peak at ca. 2700 G) and an additional
‘‘half-field” feature around 1450 G are not consistent with the
x
species. Regardless of reaction temperature, TEM
particle
reveals that the catalyst showed no signs of further VO
x
4
+
formation after reaction, while elemental mapping (Fig. 1) does
not show any observable difference in the vanadium distribution
after reaction: the elemental maps of vanadium prior to (Fig. 1a)
and after (Fig. 1b) both show a homogeneous vanadium distribu-
tion across the entire particles.
EPR spectra of catalysts exposed to reaction at 823 and 973 K
reveal that at least some of the vanadium present has undergone
reduction during reaction. Prior to reaction the vanadium is in a
presence of an isolated V ion. Indeed the half-field feature is
consistent with the presence of lower oxidation state V species
3
+
such as V [39]. Vanadia-based catalysts are well known to un-
dergo reduction during dehydrogenation and oxidative dehydro-
genation reactions [5,9,40]. That reduced species are present
after n-butane dehydrogenation over the catalyst employed in
the present work is supported by in situ XPS and NEXAFS studies
which show the average oxidation state of vanadium to be 3.5+
[7,41]. EPR spectroscopy results are therefore in agreement with
these data.
5
+ oxidation state [2,3]. Fig. 4 shows the room temperature X-
band EPR spectra ( = 9.34 GHz) of both samples. The spectra are
m

J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
333
a
b
50
40
30
20
10
0
100
80
60
8
23 K
ꢀ
ꢁ
873 K
973 K
40
20
0
0
50
100
150
0
50
100
150
Time / min
Time / min
c 5
0
0
0
0
0
0
4
3
2
1
0
50
100
Time / min
150
Fig. 2. (a) Yield of gas-phase C4 products; (b) conversion; (c) selectivity to 1-butene (filled symbols) and 1,3-butadiene (open symbols) as a function of time-on-stream for n-
butane dehydrogenation over VO /Al at 823 ( ), 873 ( ) and 973 ( ) K.
x
2 3
O
3.4. Carbon deposits
x
polymeric VO species. Such species are the dominant species on
the surface of the catalyst employed herein [5]. Furthermore, the
mechanism of the formation of such coke has been postulated to
proceed via a polystyrene like intermediate [3]. More recent stud-
ies have suggested that these large polyaromatic networks play
only a minor role in catalyst deactivation during n-butane dehy-
drogenation [4]. It is likely that the precise role that these deposits
play is heavily influenced by reaction conditions. In the present
study the quantity of coke formed is probed by XPS, elemental
microanalysis and C CP-MAS NMR spectroscopy. NMR and TEM
alongside THz-TDS, XAS, Raman and EPR spectroscopies provide
additional insights into the nature of the deposited material. These
techniques show that as the reaction temperature increases more
ordered carbon deposits are formed. Further, we propose that
these structures may have a distinct catalytic function.
The chemical nature of carbonaceous material deposited on
/Al during n-butane dehydrogenation reactions at tempera-
tures of 873 K and below has been the subject of previous investi-
gations. Wu and Stair have demonstrated that two-dimensional
networks of polyaromatic hydrocarbons form in the presence of
VO
x
2 3
O
1
3
0
-
-
-
-
2
4
6
8
3.4.1. Quantification of coke
XPS studies, conducted ex situ, provide information on the
quantity of carbonaceous material present. Fig. 5 shows the varia-
tion in the C(1s)/Al(2p) intensity ratio as a function of reaction
temperature as well as for both as-prepared VO
x 2 3 x
/Al O and VO /
2 3
Al O after calcination at 973 K. These data show a clear and sud-
den increase in coke formation at 873 K, with a continuous in-
crease in carbon coverage in the temperature range 873–973 K.
Information on the quantity of coke deposited is also provided by
microanalysis measurements (Table 1). As can be seen, the quan-
tity of hydrocarbonaceous material increases as a function of reac-
tion temperature, confirming the results of the XPS study. At a
reaction temperature of 723 K only 2.0% of the used catalyst by
mass is carbon and hydrogen; however, at 973 K this rises to
7.4%. In addition to information on the quantity of carbon present
-
10
1
1.1
1.2
000 / T (K)
1.3
1.4
1
Fig. 3. Arrhenius plot of ln k (mmol 1-butene produced per gcat) versus 1000/T for
n-butane DH over VO /Al (s). A linear relationship is observed, the gradient of
which is indicative of the activation energy of the reaction, calculated as
x
2 3
O
ꢁ
1
1
32 ± 10 kJ mol .

3
34
J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
a
b
After reaction at 873 K
200
4600
Magnetic field / G
2600
3100
3600
Magnetic field / G
After reaction at 973 K
g = 2.02
200
4600
Magnetic field / G
2600
3100
3600
Magnetic field / G
Fig. 4. (a) EPR spectrum of VO
x
2
/Al O
3
after reaction at 873 K; (b) EPR spectrum of VO
x 2 3
/Al O after reaction at 973 K. The region 200–4600 G is shown, alongside magnified
images of the region 2600–3600 G from the same spectra.
the conversion of n-butane is significantly higher at 973 K, thus
these selectivities are consistent with the greater quantity of car-
bon retained on the catalyst at higher temperatures.
3
3
.4.2. Nature of coke
.4.2.1. THz time domain spectroscopy. The nature of the coke
deposited during reaction has been probed in detail by THz-TDS
and TEM, while XAS, NMR, EPR and Raman spectroscopies provide
additional insights. Fig. 6a shows the THz-TD spectra of VO /Al O
x 2 3
after calcination at 973 K but prior to reaction, and after reaction at
temperatures from 723 to 973 K. For both the unreacted catalyst,
and those exposed to n-butane at 723, 823 and 873 K, the absorp-
tion of THz radiation is low. For reaction temperatures in excess of
8
73 K however, a dramatic increase in absorption is observed. This
suggests that the nature of the deposited coke changes above this
temperature. A second possibility is that the high absorption arises
from the formation of reduced VO
x
species. For instance, both bulk
Fig. 5. XPS-derived data showing the C(1s)/Al(2p) ratio for (a) fresh VO
x
/Al
2
O
3
2
V O
and V 13, which are formed by the reduction of bulk V O ,
O
3
6
2 5
catalyst (b) calcined at 973 K and after reaction with n-butane at (c) 723 K (d) 823 K
e) 873 K (f) 898 K (g) 923 K (h) 948 K and (i) 973 K.
are conductive at room temperature [42,43]. In order to confirm
that the increase in THz absorption is due to carbonaceous deposits
(
x 2 3
and not reduced vanadate species, THz-TD spectra of VO /Al O
reduced in hydrogen at 973 and 1273 K after calcination were
acquired. Wu et al. have previously demonstrated that reduc-
tion of VO /Al O is achieved at 783–803 K [26]. No detectable
x 2 3
these data also provide insights into the nature of the coke. In par-
ticular the degree of aromaticity is indicated through inspection of
the C/H ratio, also shown in Table 1. A higher C/H ratio suggests
that the coke present is more aromatic in nature. As expected, as
the reaction temperature increases the coke becomes increasingly
aromatic with a C/H mole ratio of 2.6 at 973 K as compared to only
difference was observed between the THz-TD spectra of the
Table 1
0
.4 at 723 K.
Elemental microanalysis data showing wt.% carbon and hydrogen and the C/H mole
These data are further supported by GC analyses which show
x 2 3
ratio after 180 min of reaction over VO /Al O .
that not all of the n-butane consumed is transformed into gas-
phase reaction products detectable by FID. Based on a simple mass
balance, assuming that all of the unaccounted for material is trans-
formed into retained carbon, it is possible to infer selectivity to
coke formation from the data presented in Fig. 2. Perhaps surpris-
ingly, it is at the highest reaction temperature that selectivity to
coke is lowest. After 180 min on stream the calculated selectivity
is 19% at 973 K as opposed to 62% at 823 K. It should be noted that
Catalyst
VO /Al
Reaction temperature (K)
723
wt.% (C + H)
C/H mole ratio
x
2
O
3
2.0
3.1
5.1
5.4
6.6
7.2
7.4
0.4
0.7
1.6
1.5
2.0
1.9
2.6
8
8
23
73
898
9
9
9
23
48
73

J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
335
a 2
.5
b 2.5
V O
2
3
Reaction T
2
2
1.5
1
1.5
1
0.5
0.5
0
0
0
.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
1
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
1
Frequency / THz
Frequency / THz
c 2
.5
2
graphite
1
.5
1
0.5
0
0
.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
1
Frequency / THz
Fig. 6. (a) THz-TD spectra of VO
and 973 K (j). The calcined catalyst prior to reaction is also shown (s). (b) THz-TD spectra of V
benzanthracene (d), 1,2-benzanthracene (ꢀ), triphenylene ( ), chrysene (.), coronene (—) and graphite (}).
x
/Al
2
O
3
employed in the dehydrogenation of n-butane for a time-on-stream of 180 min at 723 (h), 823 (}), 873 (ꢂ), 898 (+), 923 (d), 948 (4)
2 3
O
(s); (c) THz-TD spectra of phenanthrene (h), anthracene (+), 2,3-
calcined and the reduced samples. The THz-TD spectrum of bulk
was also acquired and is presented in Fig. 6b. In contrast to
reduced VO /Al , this material shows strong THz absorption,
teristic of coke deposited at high reaction temperatures. In an at-
2
V O
3
tempt to address this, a number of model coke compounds have
been investigated by THz-TDS. Fig. 6c shows THz-TD spectra of
phenanthrene, anthracene, 2,3-benzanthracene, 1,2-benzanthra-
cene, triphenylene, chrysene and coronene and graphite. All
‘‘molecular” species (i.e. all the model coke compounds bar graph-
ite) show similar, low, THz absorption. Only graphite absorbs THz
radiation strongly, in a similar manner to nanostructured carbons
reported previously [30,44–46]. The low absorption by the molec-
ular aromatic species is similar to the THz spectra observed for
x
2 3
O
due to the presence of a long-range conductive network, absent
in supported nanoparticles. The transition observed above 823 K
in the THz-TD spectra in Fig. 6a can therefore be assigned to a
change in the nature of the carbonaceous deposits. This closely cor-
responds to the temperature at which a sudden increase in the
quantity of deposited coke is observed by XPS (Fig. 5) and is sup-
ported by elemental microanalysis data (Table 1). Two distinct re-
gimes can hence be identified: a high reaction temperature regime
where coke absorbs THz radiation strongly; and a low reaction
temperature regime where only weak absorption occurs. These re-
gimes are clearly distinguished between by THz-TDS
measurements.
VO
x 2 3
/Al O after reaction at temperatures of 873 K or below. This
suggests a structural similarity between coke formed at these con-
ditions and these simple aromatic compounds. In these systems,
the electrons within the aromatic rings can move only within dis-
crete molecules, and the extensive aromatic networks present in
nanostructured carbons are not present. Conversely, above a reac-
THz-TDS has previously been employed in the study of carbona-
ceous materials, in particular CNTs and carbon nanofibres (CNFs)
x 2 3
tion temperature of 873 K the THz-TD spectra of VO /Al O resem-
[
30,44–46]. However, no prior THz-TDS characterisation of coke
ble those of polyaromatic species consisting of an extended two-
dimensional sheet structure, such as graphite. The strong absorp-
tion shown by graphite is predominately due to the presence of
conducting free electrons and phonon mode contributions. The in-
crease in absorption with increasing reaction temperature ob-
deposited during a catalytic reaction has previously been reported.
Studies on nanostructured carbons have clearly demonstrated a di-
rect correlation between the degree of graphitic order of a material
and absorption coefficient in THz-TD spectra. In particular, a higher
absorption coefficient has been shown to be associated with a
higher mobile electron density and a more substantial region of
graphene-like order, with fewer terminations [30,44]. The ability
of THz-TDS to discriminate between such materials arises from
its sensitivity to medium-range interactions such as phonon modes
and other lattice dynamics interactions. It is not however inher-
ently obvious how extensive a graphitic network is required before
carbonaceous material exhibits the high absorption that is charac-
x 2 3
served in the present work for carbon deposits on VO /Al O is
therefore a reflection of the increasing graphitic order of the coke,
and the larger network size of such deposits. While the minimum
size of the graphitic network required for THz radiation to be
strongly absorbed has not been identified, it is apparent that it
must be greater than that present in coronene, the largest of the
model coke compounds studied. Coronene contains seven fused
benzene rings and has a molecular diameter of 0.9 nm.

3
36
J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
3
.4.2.2. Other techniques. Further to analysis by THz-TDS, changes
The increased order of the coke deposited in the high-tempera-
ture regime is confirmed by further spectroscopic analyses. XAS
spectra of reacted catalysts reveal differences between cokes
deposited at different temperatures. Fig. 8 depicts the carbon K-
edge spectrum of the representative samples from the two re-
gimes: 823 and 973 K. The absorption spectrum can be subdivided
in the electronic nature of the coke at different reaction tempera-
tures are also revealed by 13C CP-MAS NMR and EPR spectrosco-
pies, while the increasing order (and therefore increasing C/H
ratio) is confirmed through XPS, Raman spectroscopy, elemental
microanalysis and TEM, vide infra.
Fig. 7 shows 13C CP-MAS NMR spectra obtained from VO
/Al
after reaction at 723, 823, 873 and 973 K. The resonance centred at
130 ppm is associated with carbon nuclei in either aromatic or
into three regions, characterised by specific resonances. The dom-
x
2
O
3
*
inating
p
resonance at 285.5 eV corresponds to the transition of 1s
*
ꢀ
electrons to a
p
antibonding orbital indicating the presence of
polyolefinic environments [47]. If present, carbon in aliphatic envi-
ronments would appear at around 30 ppm, and is observed only
after reaction at 723 and 823 K. Comparing the sample reacted at
unsaturation (p-bonding) in the carbonaceous deposit. Valence
C–H-related transitions occupy the region 287–290 eV. The inten-
sity of these transitions is reduced for the sample reacted at 973 K,
suggesting a lower hydrogen content of the sample, in agreement
7
23 K to that reacted at 823 K it is apparent that the higher reac-
tion temperature results in a 130 ppm peak of greater intensity,
corresponding to a greater quantity of carbonaceous material, in
agreement with XPS and elemental analysis results. The catalysts
reacted at 873 and 973 K do not however produce an observable
NMR spectrum. The loss of NMR signal associated with coked
material is well known. For example, similar results have previ-
ously been observed by Richardson and Haw investigating coke
with the results of microanalysis, Table 1. The third region above
*
the ionization potential is associated with
r
shape resonances.
The small hatched ‘‘peak” in the 973 K sample is typical for gra-
phitic samples, indicating a degree of ordering of the carbonaceous
material. XAS measurements therefore reveal a higher C/H ratio
x 2 3
and greater ordering of the coke deposited on VO /Al O at 973 K
as compared to 823 K.
1
3
deposited on HY zeolites, during reactions of butadiene, by
C
Raman spectroscopy is widely used to investigate carbonaceous
materials with different degrees of (dis)order [5,55–57]. However,
in the present study it does not yield unambiguous results. The
most characteristic part of the Raman spectrum is the 1700–
CP-MAS NMR in which a loss in NMR signal intensity was observed
at high reaction temperatures [48]. This loss in signal correlates
primarily with the presence of organic radicals, and also with a
deficiency of hydrogen nuclei in regions of highly ordered carbon
ꢁ
1
ꢁ1
1000 cm region with the G band at around 1580 cm associated
2
[
49]. Similarly, Meinhold and Bibby observed a loss of NMR signal
from coke derived from the conversion of methanol over H-ZSM-5
50]. This was assigned to the formation of conducting ordered
with the E2g optical mode of graphite or sp carbon materials and
ꢁ1
the D band(s) at ꢀ1350 cm corresponding to disorder-allowed
[
vibration modes. Fig. 9 shows the region of D and G vibrations
coke structures on the catalyst surface detuning the NMR probe.
The NMR probe is similarly detuned in the present study. The
deposited carbon can therefore be considered conducting in nature
and as such highly ordered, in agreement with THz-TDS studies.
Further support for the hypothesis that organic radicals are
present in the deposited carbon is provided by EPR spectroscopy
x 2 3
for representative VO /Al O samples. The spectra were fitted with
four components (G, D1, D3, and D4), which have been described
previously [28]. Note that we have not introduced into the fitting
ꢁ1
procedure the band D2 at around 1620 cm as there is no clear
indication of two peaks around the G band in the recorded spectra.
There are only small differences between the two spectra. The
most obvious is that after reaction at lower temperature the G
and D bands are less separated, the bands are broader and the
(Fig. 4). A sharp resonance at g ꢀ 2.02 is clearly observable in the
sample reacted at 973 K, however only a very small quantity is
present in that reacted at 823 K. This signal most likely arises from
the presence of organic radicals in the coke deposited at high reac-
tion temperature, the g-factor being consistent with previous
observations of such species [21,51]. An EPR resonance has also
been observed at this value for MWCNTs [52,53]. This was assigned
to the presence of holes in the pristine nanotubes. Gas-phase rad-
ical formation is also possible under high-temperature conditions
in the presence of an appropriate catalyst [54].
ꢁ1
D3 band at ꢀ1500 cm is better resolved. There is no significant
difference in the intensity ratio D-to-G and their positions are fairly
constant. Since perfect graphite exhibits only the G band, this ratio
is considered as a good measure of the degree of disorder. How-
ever, Raman active vibrations in the region of the G band have been
observed for benzene, indicating that this band alone cannot prove
7
23 K
5
82
50 °C
3
K
6
00 °C
8
73 K
973 K
700 °C
2
40
200
160
120
80
40
0
ppm
x 2 3
Fig. 8. Carbon K-edge X-ray absorption spectra of VO /Al O employed in the
dehydrogenation of n-butane at reaction temperatures of 823 (a) and 973 K (b) for a
time-on-stream of 180 min. Spectra have been recorded in He at room temperature.
Fig. 7. 13C CP-MAS NMR spectra of VO
n-butane at various reaction temperatures for a time-on-stream of 180 min.
/Al
2
O
employed in the dehydrogenation of
x
3

J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
337
carbon platelets, which are around an order of magnitude greater
in length and more regular in nature than after reaction at 823 K,
can be followed for several tens of nanometres without any signs
of discontinuities at the support particle boundaries. The platelets
even closely follow the valleys where support particles connect
and lower magnification images reveal that support particle
agglomerates are completely encapsulated in carbon, Fig. 10b.
Therefore, after reaction at 973 K it appears that the deposited car-
bonaceous layer completely covers the surface, encapsulating cat-
alyst particles. As a result no accessible vanadia centres are
available to reactant molecules upon the formation of this layer.
These conclusions are based on an analysis of >30 micrographs
per sample, of which those in Fig. 10 are representative examples.
The elemental map of carbon (Fig. 1b) further substantiates the
TEM observation of a uniform carbon layer. Additionally, UV–vis
spectroscopy studies of the catalyst after reaction at 973 K (not
shown) do not present any peaks characteristic of vanadia of units.
In summary, the nature of the carbonaceous material deposited
Fig. 9. Part of the Raman spectra of VO
n-butane at reaction temperatures of 823 (a) and 973 K (b) for a time-on-stream of
80 min. The position of the G and D bands employed in the fitting [28] are
x 2 3
/Al O employed in the dehydrogenation of
x 2 3
on VO /Al O during n-butane dehydrogenation shows a strong
dependence on reaction temperature. At high reaction tempera-
tures highly ordered, conducting networks of polyaromatic hydro-
carbons encapsulate the catalyst particle. At low reaction
temperatures the coke is much less ordered in nature, consisting
of discrete molecular units with substantial areas of the catalyst
surface left exposed and available for reaction.
1
indicated.
the existence of a high degree of order in carbonaceous materials
[
57]. Furthermore, disordered and amorphous materials exhibit
much broader bands than ordered materials and the best parame-
ter (i.e. band position, full width at half maximum or intensity ra-
tio) to reveal the degree of (dis)order is still not fully clear. While
the results of Raman spectroscopy investigations are inconclusive
in isolation, combined with the data acquired through other tech-
niques the somewhat narrower D and G bands of the sample re-
acted at 973 K are clearly consistent with an increase in graphitic
character of the samples.
This increase in the degree of graphitisation is further sup-
ported by TEM data. After reaction, carbonaceous material was ob-
served covering the support particles regardless of reaction
temperature. Samples reacted at 823 K show a carbonaceous sur-
face layer consisting of plate-like carbon following the surface of
x
2 3
3.4.3. Activity of coke deposits on VO /Al O
Spectroscopic and microscopic techniques have identified two
distinct categories of coke deposit over VO /Al O , as described
x
2 3
above. At low reaction temperatures, e.g. 823 K, the coke is poorly
ordered and not present as an extended, graphene-like network. At
high reaction temperatures, e.g. 973 K, much more extensive and
highly ordered coke is deposited. Relating the catalyst activity
shown in Fig. 2, to the nature of the coke, clear trends are apparent.
In the low-temperature regime, where the coke consists of discrete
molecular units and more amorphous carbon, the catalyst deacti-
vates with time-on-stream with both yield and conversion
decreasing. Selectivity towards 1-butene and 1,3-butadiene re-
mains fairly constant throughout. This suggests that the reaction
mechanism is relatively unaffected by coke deposition, and there-
fore the bulk of catalytic activity is at all times provided by sites
present at the start of the reaction, namely VO units on the alu-
mina surface. In the high-temperature regime, where the coke con-
sists of more extended and ordered carbon networks, very different
catalytic behaviour is observed. The calculated yield increases as a
function of time-on-stream indicating activation of the catalyst as
2 3
the Al O support particles (Fig. 10a). The individual platelets are
rarely more than 5 nm long. This surface layer is incomplete, vary-
ing from ꢀ0 to 3 layers of carbon and covers mainly individual sup-
port particles. A few particles consisting of pure carbon were also
observed. For the samples reacted at 973 K (Fig. 10b and c), a sim-
ilar surface covering is observed. However, in these cases the sup-
x
port particles were completely coated by
a carbon layer
continuously covering several support particles. The resultant
Fig. 10. Transmission electron micrographs of VO
x 2 3
/Al O after reaction at 823 K (a) and 973 K (b) and (c). (a) and (c) Show high magnification images revealing the nature of
the deposited carbon while (b) shows that at high-temperature coke encapsulates agglomerates of catalyst particles.

3
38
J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
coke is deposited. Furthermore, after a short induction period, con-
version reaches a constant value. Both these factors demonstrate
that the VO /Al O catalyst does not suffer from deactivation at
x 2 3
previously been shown to exhibit slightly higher selectivity in
the dehydrogenation of alkanes than as-prepared catalysts [5].
However, in that study the possible catalytic action of the coke
was not considered. Increasing the selectivity of zeolite catalysts
through carbon deposition is a strategy employed industrially in
xylene isomerisation [58,59]. Furthermore, the activity of CNTs in
the oxidative dehydrogenation of n-butane has recently been dem-
onstrated [25]. As with the present work little or no deactivation
was observed in the reaction on a carbon-based catalyst over the
timescale investigated. Considering other reactant molecules, the
activity of coke deposits and other carbonaceous materials in the
ODH of ethylbenzene is well established with the activity of these
known since at least the 1970s [60–62]. This area has been re-
viewed by Lisovskii and Aharoni [14] and Cavani and Trifiro [15].
Similarly to the results reported herein, Lisovskii and Aharoni
report that the activity of a metal-oxide catalyst reaches a stea-
dy-state after an induction period in which the entire catalyst sur-
face is covered by carbonaceous material [14]. While the reaction
mechanism may be different between ODH and direct DH reactions
(the active sites for n-butane and ethylbenzene ODH over carbona-
ceous materials are believed to be carboxyl functionalities formed
through oxidation of the carbon surface [14,15,20,22,24,25,61,63])
there are a number of similarities between the nature of the active
carbon. In both cases, highly ordered materials represent excellent
catalysts. Additionally, it has been noted by a number of authors
that the activity of coke deposits in ethylbenzene ODH correlates
with their degree of paramagnetism, with a greater concentration
of unpaired electrons resulting in higher activity [14,21,60]. Such
sites could also provide the active centres in the studies reported
in the present work, since paramagnetic organic radicals were ob-
served by EPR spectroscopy. The unpaired electrons present in
such radicals may be capable of extracting hydrogen atoms from
the n-butane feed, resulting in dehydrogenation. Both the deposi-
tion of new carbonaceous material (with associated paramagnetic
centres) during reaction and the recombination of the extracted H
atoms could then occur in such a way as to replenish the source of
unpaired electrons.
The phenomenon of carbonaceous deposits providing, or
enhancing, catalytic activity is not restricted to DH or ODH reac-
tions. Intriguingly, carbonaceous overlayers formed on mineral
surfaces during Fisher–Tropsch-type reactions have also been pro-
posed as the active phase for catalytic reduction reactions in proto-
stellar nebulae. Organic radicals present in such overlayers have
been suggested as contributing to their catalytic activity [64]. Else-
where, the hydrogenation of carbon–carbon double bonds over
supported metal catalysts has been shown to be strongly depen-
dent upon the nature of any deposited carbonaceous material
[65]. Such reactions have been suggested to proceed, at least in
part, through the transfer of hydrogen from an adsorbed hydrocar-
bonaceous overlayer onto which the reactant is adsorbed [66,67].
Furthermore, carbon deposited during the reaction plays a key role
in the hydrogenation of alkynes over Pd catalysts, forming a cata-
high reaction temperatures, despite a greater quantity of coke
being present as compared to low reaction temperatures. Selectiv-
ity to the dehydrogenation products shows a more marked depen-
dence on time-on-stream than at low temperatures. For example,
selectivity to 1-butene increases from an initial value of 32% to a
steady-state value of 48%. This steady-state value is achieved
around the same time as conversion reaches steady-state at 50%.
A notable decrease in selectivity towards carbon deposition from
6
6% to 20% occurs over the same period. These changes are perhaps
indicative of a change in the reaction mechanism and in the nature
of the catalytically active site.
That a change in the catalytically active site has occurred is con-
firmed by TEM data, which show that the VO
cles are encapsulated in carbon after reaction at 973 K. As such, any
VO sites will be rendered inaccessible to n-butane molecules. De-
spite this, the catalyst remains highly active, even after 180 min on
stream. Indeed the yield of C hydrocarbons reaches its greatest va-
lue (40%) at this time. The initial carbon deposits are most likely
formed through the catalytic action of VO units breaking down
x 2 3
/Al O catalyst parti-
x
4
x
butane molecules. However, after 180 min on stream the carbona-
ceous material has a thickness of multiple layers, despite the inac-
cessible nature of the original catalyst. Further carbon formation
therefore appears to have taken place upon the deposits. These
data therefore suggest the possibility that catalytic activity is pro-
vided by the carbonaceous deposits themselves. TEM data pre-
sented in Section 3.4.2 revealed significant structural similarities
between carbonaceous material deposited at high temperature
x 2 3
over VO /Al O and nanostructured carbons, such as CNFs. Notably,
the deposited coke (Fig. 10) can be characterised as curved ‘‘graph-
ene-like” sheets. CNFs can be considered as graphene sheets
formed into conical sections [27]. As a consequence of the apparent
structural similarity between high-temperature coke deposits and
CNFs, it was decided to study the activity of CNFs towards n-bu-
tane DH. Table 2 compares the selectivity and conversion of VO
Al and the CNFs at a reaction temperature of 973 K. After the
initial induction period, during which coke is deposited on the sur-
face of VO /Al both catalytic systems show very similar behav-
iour. After 180 min on stream VO /Al shows a conversion of
DH products of 40%, a selectivity to 1-butene of
8% and to 1,3-butadiene of 32%. The corresponding values for
x
/
2
O
3
x
2 3
O
x
2 3
O
5
4
0%, a yield of C
4
the CNF are 41%, 32%, 46% and 34%, respectively. This similarity
in catalytic behaviour between coked VO /Al and the CNF sup-
x
2 3
O
ports the hypothesis that carbonaceous deposits on the surface of
the supported metal-oxide catalyst are indeed active for n-butane
dehydrogenation. Neither the CNFs nor the coked VO /Al O
x 2 3
exhibit deactivation as a function of increasing exposure to
n-butane. It is also worth noting that very slight increases in the
1
-butene selectivity are observed towards the end of the reaction
period at low temperatures over VO /Al . These changes may also
x
2 3
O
be due to the build-up of small amounts of active coke deposits.
lytically active Pd-C
x
surface phase [68,69]. Carbonaceous deposits
While their role in the direct dehydrogenation of n-butane has
not previously received attention, coked VO /Al O catalysts have
x 2 3
and related materials therefore show activity in a wide number of
reactions. The above-mentioned data clearly show that the dehy-
drogenation of n-butane can be added to the list of reactions in
which carbonaceous materials can provide high activity and
selectivity.
Table 2
Yield of C
obtained at 180 min during the direct dehydrogenation of n-butane over CNFs and
VO /Al at 973 K.
4
DH products, conversion and selectivity to 1-butene and to 1,3-butadiene
x
2 3
O
4
. Conclusions
Catalyst
Reaction
temperature (%)
K)
Yield Conversion Selectivity to Selectivity
(%)
1-butene (%) to 1,
3-butadiene (%)
It has been clearly demonstrated that highly ordered coke
(
deposits are active for the dehydrogenation of n-butane. This con-
clusion is confirmed by complementary studies on unfunctiona-
lised CNFs which at 973 K demonstrate almost identical activity
VO
x
/Al
2
O
3
973
40
32
50
41
48
46
32
34
PR24-LHT 973

J. McGregor et al. / Journal of Catalysis 269 (2010) 329–339
339
to coked VO
x
2
/Al O
3
. Organic radicals associated with the carbona-
[23] N. Keller, N.I. Maksimova, V.V. Roddatis, M. Schur, G. Mestl, Y.V. Butenko, V.L.
Kuznetsov, R. Schlögl, Angewandte Chemie: International Edition 41 (2002)
ceous material may play a role in the reaction mechanism. As this
reaction proceeds successfully in the absence of gas-phase oxygen
this activity is wholly attributable to the action of the carbona-
ceous material. Deactivation of the reaction is not observed in
the timescale investigated, in stark contrast to behaviour at lower
temperatures where rapid deactivation occurs. It is notable that
commercially, alkane dehydrogenation also suffers from rapid
deactivation at these lower reaction temperatures.
1885.
[
24] G. Mestl, N.I. Maksimova, N. Keller, V.V. Roddatis, R. Schlögl, Angewandte
Chemie: International Edition 40 (2001) 2066.
[
[
25] J. Zhang, X. Liu, R. Blume, A. Zhang, R. Schlögl, D.S. Su, Science 322 (2008) 73.
26] Z.L. Wu, H.-S. Kim, P.C. Stair, S. Rugmini, S.D. Jackson, Journal of Physical
Chemistry B 109 (2005) 2793.
[27] G.G. Tibbetts, M.L. Lake, K.L. Strong, B.P. Rice, Composites Science and
Technology 67 (2007) 1709.
[
28] A. Sadezky, H. Muckenhuber, H. Grothe, R. Niessner, U. Poschl, Carbon 43
2005) 1731.
(
Additionally, THz spectroscopy is shown to be a valuable new
resource in the catalyst characterisation toolkit. THz-TDS has the
capability to distinguish between polyaromatic carbons on the ba-
sis of their long-range order and conductivity. Such discrimination
is difficult to achieve by other methods, and as such THz-TDS will
likely play an increasingly important role in future studies.
[29] P.C. Upadhya, K.L. Nguyen, Y.C. Shen, J. Obradovic, K. Fukushige, R. Griffiths, L.F.
Gladden, A.G. Davies, E.H. Linfield, Spectroscopy Letters 39 (2006) 215.
[
30] E.P.J. Parrott, J.A. Zeitler, J. McGregor, S.-P. Oei, H.E. Unalan, S.-C. Tan, W.I.
Milne, J.-P. Tessonnier, R. Schlögl, L.F. Gladden, The Journal of Physical
Chemistry C 113 (2009) 10554.
[
[
[
[
[
31] J.A. Zeitler, P.F. Taday, D.A. Newnham, M. Pepper, K.C. Gordon, T. Rades, Journal
of Pharmacy and Pharmacology 59 (2007) 209.
32] Z.L. Wu, H.-S. Kim, P.C. Stair, in: S.D. Jackson, J.S.J. Hargreaves (Eds.), Metal
Oxide Catalysis, Wiley-VCH, Weinheim, 2008.
33] J. McGregor, Solid-state NMR of oxidation catalysts, in: S.D. Jackson, J.S.J.
Hargreaves (Eds.), Metal Oxide Catalysis, Wiley-VCH, Weinheim, 2008.
34] Z.L. Wu, P.C. Stair, S. Rugmini, S.D. Jackson, Journal of Physical Chemistry C 111
Acknowledgments
(
2007) 16460.
Support for this work was provided through the ATHENA pro-
ject (EPSRC grant GR/R47523/01) funded by the Engineering and
Physical Sciences Research Council and Johnson Matthey plc, and
through Research Councils UK Basic Technology Programme
35] B.M. Weckhuysen, D.E. Keller, Catalysis Today 78 (2003) 25.
[36] S.M.K. Airaksinen, M.E. Harlin, A.O.I. Krause, Industrial and Engineering
Chemistry Research 41 (2002) 5619.
[
37] M. Baltes, P. van der Voort, B.M. Weckhuysen, R.R. Rao, G. Catana, R.A.
Schoonheydt, E.F. Vansant, Physical Chemistry Chemical Physics 2 (2000)
2673.
(
Grants GR/R87086/02 and EP/E048811/1). Frank Girgsdies (FHI)
[
[
38] R. Gallay, J.J. Vanderklink, J. Moser, Physical Review B 34 (1986) 3060.
39] J.M. Rojo, J.L. Mesa, R. Calvo, L. Lezama, R. Olazcuaga, T. Rojo, Journal of
Materials Chemistry 8 (1998) 1423.
is acknowledged for XRD measurements of the catalysts, Genka
Tzolova-Mueller (FHI) for UV–vis spectroscopy, while Christian
Hess is thanked for assistance in conducting the Raman spectros-
copy experiments. The authors would also like to thank Professor
Peter Stair, Dr. Zili Wu (Northwestern University, USA), Professor
S. David Jackson and Dr. Sreekala Rugmini (University of Glasgow,
UK) for helpful discussions. The BESSY staff are gratefully acknowl-
edged for their continuing support of our measurements.
[40] F. Cavani, F. Trifiro, Catalysis Today 24 (1995) 307.
[
[
41] E.M. Vass, D. Teschner, A. Knop-Gericke, R. Schlögl, unpublished work.
42] H. Bosch, B.J. Kip, J.G. Vanommen, P.J. Gellings, Journal of the Chemical Society
Faraday Transactions 80 (1984) 2479.
[43] D.B. McWhan, J.P. Remeika, Physical Review B 2 (1970) 3734.
[
44] E.P.J. Parrott, J.A. Zeitler, J. McGregor, S.-P. Oei, W.I. Milne, J. Tessonnier, D.S. Su,
R. Schlögl, L.F. Gladden, Advanced Materials 21 (2009) 3953.
[
[
45] N. Hisaaki, M. Nobutsugu, S. Ryo, Applied Physics Letters 91 (2007) 011108.
46] K. Chul, M. In Hee, O. Seung Jae, L. Seong Chu, A. Kay Hyeok, L. Young Hee, S.
Joo-Hiuk, Physical Review B (Condensed Matter and Materials Physics) 75
(2007) 085410.
References
[
47] B. Paweewan, P.J. Barrie, L.F. Gladden, Applied Catalysis A: General 185 (1999)
2
59.
[
[
[
1] R.E. Kirk, D.F. Othmer (Eds.), Kirk-Othmer Encyclopedia of Chemical
Technology, John Wiley and Sons Inc., Hoboken, 2004.
2] J. McGregor, Z. Huang, G. Shiko, L.F. Gladden, R.S. Stein, M.J. Duer, Z.L. Wu, P.C.
Stair, S. Rugmini, S.D. Jackson, Catalysis Today 142 (2009) 143.
[
[
48] B.R. Richardson, J.F. Haw, Analytical Chemistry 61 (1989) 1821.
49] J.L. Bonardet, M.C. Barrage, J. Fraissard, Journal of Molecular Catalysis A:
Chemical 96 (1995) 123.
50] R.H. Meinhold, D.M. Bibby, Zeolites 10 (1990) 121.
51] C.J. Rhodes, C.S. Hinds, Molecular Engineering 4 (1994) 119.
52] J. Tarabek, L. Kavan, M. Kalbac, P. Rapta, M. Zukalova, L. Dunsch, Carbon 44
[
[
[
3] S.D. Jackson, S. Rugmini, P.C. Stair, Z.L. Wu, Chemical Engineering Journal 120
(
2006) 127.
[
[
[
4] S.D. Jackson, S. Rugmini, Journal of Catalysis 251 (2007) 59.
5] Z.L. Wu, P.C. Stair, Journal of Catalysis 237 (2006) 220.
6] Z. Huang, J. McGregor, P. Steiner, L.F. Gladden, S. Rugmini, S.D. Jackson,
Magnetic Resonance Imaging 25 (2007) 562.
7] E.M. Vass, M. Havecker, S. Zafeiratos, D. Teschner, A. Knop-Gericke, R. Schlögl,
Journal of Physics: Condensed Matter 20 (2008) 184016.
8] S.D. Jackson, P.C. Stair, L.F. Gladden, J. McGregor, in: S.D. Jackson, J.S.J.
Hargreaves (Eds.), Metal Oxide Catalysis, Wiley-VCH, Weinheim, 2008.
9] A. Christodoulakis, M. Machli, A.A. Lemonidou, S. Boghosian, Journal of
Catalysis 222 (2004) 293.
(
2006) 2147.
53] V. Likodimos, S. Glenis, N. Guskos, C.L. Lin, Physical Review B 68 (2003)
45417.
[
0
[
[
[
[
54] F. Cavani, F. Trifirò, Catalysis Today 51 (1999) 561.
[
[
[
55] Y.T. Chua, P.C. Stair, Journal of Catalysis 213 (2003) 39.
56] S. Kuba, H. Knozinger, Journal of Raman Spectroscopy 33 (2002) 325.
57] J. Schwan, S. Ulrich, V. Batori, H. Ehrhardt, S.R.P. Silva, Journal of Applied
Physics 80 (1996) 440.
58] F. Bauer, W.-H. Chen, Q. Zhao, A. Freyer, S.-B. Liu, Microporous and Mesoporous
Materials 47 (2001) 67.
59] J.S. Beck, R.A. Crane, M.F. Mathias, J.A. Kowalski, D.N. Lissy, D.L. Stern, WO
Patent 99 52842, 1999 to Mobil Oil Corporation.
60] R. Fiedorow, W. Przystajko, M. Sopa, I.G. Dalla Lana, Journal of Catalysis 68
[
[
[
[
[
[
[
[
10] O.R. Evans, A.T. Bell, T.D. Tilley, Journal of Catalysis 226 (2004) 292.
11] M.A. Bañares, Catalysis Today 51 (1999) 319.
12] T. Blasco, J.M.L. Nieto, Applied Catalysis A: General 157 (1997) 117.
13] F. Cavani, N. Ballarini, A. Cericola, Catalysis Today 127 (2007) 113.
14] A.E. Lisovskii, C. Aharoni, Catalysis Reviews: Science and Engineering 36
(
1981) 33.
[
[
[
[
61] A. Schraut, G. Emig, H.G. Sockel, Applied Catalysis 29 (1987) 311.
62] T.G. Alkhazov, A.E. Lisovskii, Kinetika i Kataliz 17 (1976) 434.
63] G. Emig, H. Hofmann, Journal of Catalysis 84 (1983) 15.
64] J.A. Nuth, N.M. Johnson, S. Manning, Astrophysical Journal Letters 673 (2008)
L225.
65] J. McGregor, L.F. Gladden, Applied Catalysis A: General 345 (2008) 51.
66] S.J. Thomson, G. Webb, Journal of the Chemical Society: Chemical
Communications (1976) 526.
(
1994) 25.
[
[
[
15] F. Cavani, F. Trifiro, Applied Catalysis A: General 133 (1995) 219.
16] A. Guerrero-Ruiz, I. Rodriguez-Ramos, Carbon 32 (1994) 23.
17] M.F.R. Pereira, J.J.M. Orfao, J.L. Figueiredo, Applied Catalysis A: General 184
1999) 153.
18] M.F.R. Pereira, J.J.M. Orfao, J.L. Figueiredo, Applied Catalysis A: General 196
2000) 43.
19] M.F.R. Pereira, J.J.M. Orfao, J.L. Figueiredo, Applied Catalysis A: General 218
2001) 307.
20] J.A. Macia-Agullo, D. Cazorla-Amoros, A. Linares-Solano, U. Wild, D.S. Su, R.
Schlögl, Catalysis Today 102–103 (2005) 248.
21] L.E. Cadus, L.A. Arrua, O.F. Gorriz, J.B. Rivarola, Industrial and Engineering
Chemistry Research 27 (1988) 2241.
22] D.S. Su, N. Maksimova, J.J. Delgado, N. Keller, G. Mestl, M.J. Ledoux, R. Schlögl,
Catalysis Today 102–103 (2005) 110.
(
[
[
[
[
[
[
[
(
[
[
67] S.D. Jackson, C.A. Hamilton, G.J. Kelly, D. de Bruin, Reaction Kinetics and
Catalysis Letters 73 (2001) 77.
(
68] D. Teschner, E.M. Vass, M. Havecker, A. Zafeiratos, P. Schnorch, H. Sauer, A.
Knop-Gericke, R. Schlögl, M. Chamam, A. Wootsch, A.S. Canning, J.G. Gamman,
S.D. Jackson, J. McGregor, L.F. Gladden, Journal of Catalysis 242 (2006) 26.
69] D. Teschner, J. Borsodi, A. Wootsch, Z. Revay, M. Havecker, A. Knop-Gericke,
S.D. Jackson, R. Schlögl, Science 320 (2008) 86.
[