Phosphate-modified carbon nanotubes in the oxidative dehydrogenation of isopentanes

Article abstract of DOI:10.1002/cssc.201402457

Ketonic/quinonic C=O groups on the surface of a carbon matrix are capable of abstracting hydrogen in C-H bonds from hydrocarbons and enable them to selectively convert into corresponding unsaturated hydrocarbons; this process is the oxidative dehydrogenation (ODH) reaction. However, a variety of inevitable defects or graphene edges and other oxygen-containing groups on the carbon matrix are detrimental to the selective production of alkenes due to their high activity towards overoxidation. Herein, we show that phosphate can not only impede the total oxidation but also cover the selective C=O groups, hence allowing its use as a modulator to defects and oxygen-containing functional groups on the multiwalled carbon nanotubes, regulating the distribution of active sites and related catalytic targets.

Full text of DOI:10.1002/cssc.201402457

CHEMSUSCHEM
FULL PAPERS
DOI: 10.1002/cssc.201402457
Phosphate-Modified Carbon Nanotubes in the Oxidative
Dehydrogenation of Isopentanes
Rui Huang,[a, b] Hong Yang Liu,[b] Bing Sen Zhang,[b] Xiao Yan Sun,[b] Chang Hai Liang,[a]
Dang Sheng Su,*[b] Bao Ning Zong,[c] and Jun Feng Rong[c]
Ketonic/quinonic C=O groups on the surface of a carbon
matrix are capable of abstracting hydrogen in CÀH bonds from
hydrocarbons and enable them to selectively convert into cor-
responding unsaturated hydrocarbons; this process is the oxi-
dative dehydrogenation (ODH) reaction. However, a variety of
inevitable defects or graphene edges and other oxygen-con-
taining groups on the carbon matrix are detrimental to the se-
lective production of alkenes due to their high activity towards
overoxidation. Herein, we show that phosphate can not only
impede the total oxidation but also cover the selective C=O
groups, hence allowing its use as a modulator to defects and
oxygen-containing functional groups on the multiwalled
carbon nanotubes, regulating the distribution of active sites
and related catalytic targets.
Introduction
Surface modification of catalysts, by means of passivating or
prepoisoning, plays an indispensable role in chemical science
for the synthesis of desirable targets.[1] Heterogeneous metal-
free catalysts, as novel alternatives to the conventional transi-
tion-metal-based catalysts, offer great environmental and eco-
nomical advantages.[2] Among them, surface-modified carbon
materials have been well studied as alternatives to metal oxide
catalysts for oxidative dehydrogenation (ODH) of light (i.e.,
C2–C4) alkanes, such as ethane,[3] propane,[4] n-butane,[5] and
isobutane.[6] However, processes using these catalysts have not
been industrially applied. One (critically important) reason is
deep oxidation of the alkane and of the produced alkene in
consecutive reactions. Higher alkanes (i.e., C4, C5) have an
abundance of secondary and tertiary CÀH bonds (with bond
energies of 401 kJmolÀ1 and 390 kJmolÀ1, respectively) and are
much more reactive towards gaseous oxygen than lower alka-
nes (i.e., C2, C3) (see Supporting Information, Table S1). The
corresponding dehydrogenation products contain the very
active allylic CÀH bond (361 kJmolÀ1), which is easily oxidized
to COx.[7] Numerous studies reveal that the molar ratio of com-
bustion/dehydrogenation (hydrocarbon; HC) products (COx/
HC) in ODH reactions increases with the chain length of reac-
Figure 1. Product distributions during ODH reactions of alkanes (C2–C5)
[7–8]
[9]
[10]
~
&
&
^
over V-Mg-O catalysts. : ethane
;
: propane
;
: butane
, : n-pen-
[12]
tane[11]
;
: isopentane.
*
tants. This tendency is summarized in Figure 1:[7–12] it is more
difficult to suppress the formation of COx during ODH of iso-
pentane than during ODH of lower alkanes (C2–C4). On the
other hand, there is increased demand for C5 olefins and diole-
fins from dehydrogenation of C5 paraffins for use in resins,
elastomers, and fine chemicals.[12] Unfortunately, there are only
few studies on the dehydrogenation of C5 alkanes, and most
of these use metal oxide catalysts: under oxidative conditions,
the selectivity towards the dehydrogenation products is lower
than 65%.[12]
[a] R. Huang, Prof. C. H. Liang
Lab of Advanced Materials & Catalytic Engineering
Dalian University of Technology
No. 2 Linggong Road, Dalian 116024 (PR China)
[b] R. Huang, Dr. H. Y. Liu, Dr. B. S. Zhang, X. Y. Sun, Prof. D. S. Su
Shenyang National Laboratory for Materials Science
Institute of Metal Research, Chinese Academy of Science
No. 72 Wenhua Road, Shenyang 110016 (PR China)
E-mail: dangsheng@fhi-berlin.mpg.de
In carbon technology, phosphorous modification is a well-es-
tablished means of enhancing the oxidation resistance of
carbon materials, especially in case of activated carbon. Phos-
phorus is present in carbon matrices as red phosphorus, in
phosphates, and/or in chemically bonded form, for example in
ÀCÀPÀ bonds or ÀCÀOÀPÀ bonds.[13] Recently, phosphorous
modification was used in carbon catalysis to increase the selec-
tivity of oxidative dehydrogenation, suppressing the deep oxi-
[c] B. N. Zong, J. F. Rong
Sinopec Research Institute of Petroleum Processing
No. 18 Xueyuan Road, Beijing 100083 (PR China)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201402457.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3476

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
dation of reactant/product.[3–6] By studying oxidative dehydro-
genation of isobutane on phosphorous-modified graphitic
mesoporous carbon, Schwartz et al. revealed that phosphorous
groups do not change the nature of the active sites but affect
the availability of the sites, depending on the phosphorous
content loaded onto the carbon catalyst.[6c] The decline of
olefin selectivity at high phosphorous loadings cannot be
clearly assigned to the coverage of selective ketone sites or to
the generation of additional active sites favorable to total oxi-
dation.
Herein, we show for the first time that isopentane can be
dehydrogenated with only carbon as catalyst. By using carbon
nanotubes (CNTs) as catalyst, we systematically studied the
effect of phosphate modification on the oxidative dehydrogen-
ation of isopentane involving the formation of active allylic CÀ
H bonds that can be easily oxidized to CO2. We found that the
surface chemistry and the distributions of active sites of CNTs
are strongly dependent on the loading of phosphate. An addi-
tion of phosphate of up to 1 wt% is capable of inhibiting the
sites for deep oxidation, and thus increasing the selectivity of
isopentane dehydrogenation—the selectivity was doubled.
Phosphorus or phosphorous oxides act as electron-attracting
dopant, inhibiting the generation of electrophilic oxygen spe-
cies and thus suppressing the combustion of hydrocarbons.
However, a higher loading is not beneficial for the reaction.
The nucleophilic ketonic/quinonic groups are blocked by elec-
trophilic phosphate, thus diminishing the number of active
sites for ODH. We reveal that phosphorous oxides not only act
as an inhibitor of non-selective electrophilic oxygen species
but also cover the selective sites (C=O groups) with increasing
loading.
Figure 2. Product selectivity versus isopentane conversion obtained by varia-
tion of reaction temperature, with V-Mg-O as a reference (Supporting Infor-
mation, Figure S1).[12] Reaction conditions: 5P-oCNTs, 200 mg; 3P-oCNTs,
100 mg; 0.5P-oCNTs and 1P-oCNTs, 50 mg; oCNTs, 20 mg; 2.4% isopentane
(i-C5H12), O2/i-C5H12 =2, 16.7 mLminÀ1, 623–673 K. a) Total dehydrogenation
selectivity. b) Carbon oxide selectivity (CO and CO2). c) Mono-olefins selectivi-
&
ty (isoamylene: 2M1B, 2M2B and 3M1B). d) Diolefin selectivity (isoprene).
:
~
~
~
*
*
oCNTs; : 0.5P-oCNTs; : 1P-oCNTs; : 3P-oCNTs; : Mg3V2O8; : Mg2V2O7;
[12]
*
: V-Mg-O.
Results and Discussion
Scheme 1. Schematic illustration of the reaction network of isopentane oxi-
dative dehydrogenation over CNTs.
What level of phosphate loading is favorable for dehydro-
genation?
Pristine CNTs, oxidized CNTs (oCNTs), and phosphate-modified
oCNTs with different loadings (denoted as xP-oCNTs, with x=
0.1–5 wt% P2O5) were employed in the ODH of isopentane. For
comparison with metal oxides, V-Mg-O systems (Mg3V2O8 and
Mg2V2O7) were used. Figure 2 shows that all of the catalysts
are active towards the reaction, but the selectivity towards de-
hydrogenated products (Figure 2a) on xP-oCNTs (x<3 wt%) is
higher than those of oCNTs at the same conversion of isopen-
tane. This demonstrates that phosphate modification of CNTs
is beneficial to inhibit the formation of COx in the ODH of iso-
pentane. As depicted in Figure 2, the selectivity of all catalysts
could not reach 100% when the conversion was extrapolated
to 0, indicating a reaction network with parallel and consecu-
tive hydrocarbon combustions (Scheme 1).[12,14] Figure 2b–d
shows the selectivities towards mono-olefins, diolefins, and
COx versus the isopentane conversion over the different cata-
lysts. When the phosphate loading and the isopentane conver-
sion increases, the main products change from mono-olefins
to diolefins, and then to COx. This indicates the selectivity of
products during ODH is correlated with the phosphate loading.
This correlation did not change with time-on-stream during
a lifetime test of the catalyst (Supporting Information, Fig-
ure S2). Among all of the tested samples, The 1P-oCNTs cata-
lyst exhibits an outstanding performance and stability: the
conversion of isopentane remained almost unchanged (9.5%)
and the alkene selectivity stayed above 62%.
The products distributions over different catalysts, obtained
by sampling at 2.5% isopentane conversion and shown in
Figure 3, reveal that an increase of phosphate loading from 0
to 3 wt% leads to an increase of the selectivity towards mono-
olefins (2M1B, 2M2B, and 3M1B) and diolefins (isoprene),
reaching maximum values of 58.6% and 32.8%, respectively.
However, when the phosphate loading is higher than 3%, the
selectivity of dehydrogenation declines accompanied with an
increase of COx. Ultimately, the isopentane cannot be convert-
ed into any products when the phosphate loading reaches
15 wt%. Interestingly, after sufficient washing with distilled
water by centrifugation, the activity of 15P-oCNTs can match
that of 0.1P-oCNTs. This demonstrates that excessive phos-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3477

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
1 wt% phosphate loading (an increase of activation energy by
31 kJmolÀ1), but a decrease of the activation energy by
19 kJmolÀ1 is observed at higher phosphate loading. Consider-
ing the information in both Figure 3 and 4 leads to the conclu-
sion that the increased selectivity towards mono- and diolefins
at low phosphate loading (0.1–1 wt%) is mainly due to inhibit-
ed C5H10 and C5H8 combustion. At high phosphate loadings (>
1 wt%), the increased formation of diolefins comes at the ex-
pense of mono-olefins, seldom affecting consumption of the
relatively stable paraffin C5H12. Probably, phosphorus-bound
oxygen functional groups or phosphorus oxides activate C5H10,
but this needs further investigation.
Figure 3. Product distributions over CNTs with different phosphate loadings
(wP O =0~15 wt%) in the ODH of isopentane at 2.5% isopentane conver-
sion. Among them, the conversion using the 15P-oCNTs is close to 0. *: After
Structure–performance correlation
2
5
The structural properties, surface properties, and thermal sta-
bility of the catalysts before and after reaction were character-
ized by N2 physisorption, X-ray photoelectron spectroscopy
(XPS), scanning and transmission electron microscopy (SEM
and TEM), Raman spectroscopy, and temperature-programmed
techniques (oxidation: TPO and desorption: TPD). N2 physi-
sorption and surface analysis by XPS (see Table S2, Supporting
Information) show that the addition of phosphorus leads to
a decline of the specific surface area, from 283 m2 gÀ1 on the
oCNTs to 210 m2 gÀ1 on the 5P-oCNTs, accompanied by an in-
crease of the surface oxygen content and P/O atomic ratio
from 4.68% to 5.76% and 0 to 0.20, respectively, while the
pore diameter is nearly constant. This indicates that the added
phosphorus mainly covers the surface of the CNTs, resulting in
rearrangement of the surface oxygen-containing groups (see
Supporting Information, Table S2). It can also be concluded
from SEM and TEM images as well as the Raman results that
the addition of phosphate has an extremely minor impact on
the texture and framework of CNTs, but results in a slightly in-
creased degree of disorder due to carbon debris or phospho-
rus oxides (see the Supporting Information, Figures S3 and S4).
&
&
&
&
&
washing with distilled water. : 3M1B; : 2M2B; : 2M1B; : isoprene; : CO;
*
: CO2.
phate results in full coverage of sites active towards ODH of al-
kanes, but the phosphate covering the dehydrogenation sites
can be easily rinsed out, likely because the nonchemical inter-
actions between the phosphates and dehydrogenation sites
are weak.
Does the nature of the active sites change with phosphate
loading?
To understand the nature of active sites, the steady-state ap-
parent activation energy (Ea) of different C5 hydrocarbons was
obtained by using C5H12, C5H10, and C5H8 as substrates. Figure 4
reveals that the activation energy of C5H8 increases from
121 kJmolÀ1 over oCNTs catalyst to 145 kJmolÀ1 over the 3P-
oCNTs catalyst, as expected considering that the modification
with phosphate suppresses C5H8 combustion. The activation
energy of C5H12 is nearly constant, at about 85Æ5 kJmolÀ1. No-
tably, the activation of C5H10 is suppressed on the oCNTs with
The interactions between phosphates and defect sites
CNTs undergo oxidation (combustion) in the presence of
oxygen at high temperature. The combustion starts at sites on
CNTs that can dissociate the adsorbed oxygen molecules. The
interaction between CNTs and O2 gas was investigated by TG
analysis in air, and the maximum temperatures Tmax of the exo-
thermic differential scanning calorimetry (DSC) curves are sum-
marized in Table S2 (Supporting Information). Phosphate load-
ing increases the Tmax, as expected and well reported in litera-
ture.[15] Phosphorus compounds are known to selectively block
the defect sites,[14] thus improving the carbon materials’ resist-
ance to oxidation.[15] These defect sites convert O2 molecules
to highly reactive electrophilic oxygen species (superoxide,
O2À; peroxide, O22À).[16] They are the starting sites for CNT oxi-
dation or they can attack the electron-rich C=C bonds in the
ODH reaction, leading to the rupture of the carbon skeleton
and subsequent deep oxidation.[5]
Figure 4. Apparent activation energies Ea of different catalysts by variation
The correlation of Tmax with the consumption rate of sub-
strates during ODH reaction in Figure 5 shows that the declin-
of reactants: C5H12, C5H10 (2M2B), and C5H8 were used as a reactant in the
&
&
&
ODH reaction. : oCNTs; : 1P-oCNTs; : 3P-oCNTs.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3478

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Figure 6. Influence of phosphate modification on the surface functional
Figure 5. The consumption rate of reactant (C5H12 and O2) and the maximum
temperature Tmax resistance to oxygen in DSC curves as a function of P2O5
loading, wP O . Reaction conditions: 100 mg, 2.0% isopentane (i-C5H12), O2/i-
groups of CNTs as investigated by ATR-IR. Shadow region: 1000–750 cmÀ1
,
vasC-O-P, vasP-O-P.
2
5
C5H12 =1, 15 mLminÀ1, 673 K. : RC H
;
: RO ; : Tmax.
~
&
*
5
12
2
ing trends of substrate consumption accurately coincides with
the rising tendencies of combustion resistance of phosphate-
modified CNTs: lower loadings of phosphate (0–1 wt%) have
a marked inhibition effect, while the effect is only moderate at
higher loadings (1–5 wt%). This means that active sites on the
surface of CNTs blocked by phosphate have an equivalent
effect on the combustion of CNTs and on the oxidative dehy-
drogenation of isopentane. A loading of up to 1 wt% gives
a pronounced enhancement of the combustion resistance
(Figure 5), while a reduction of COx formation in ODH from
60% COx on oCNTs down to 18% COx on 1P-oCNTs is observed
in Figure 3. Also, an abrupt decline of substrates consumption
is observed in Figure 5. Consistent with the elevated activation
energies of C5H10 and C5H8 in Figure 4, it is deduced that lower
amounts (0–1 wt%) of phosphates preferentially occupy or
cover the defect sites on CNTs where usually the electrophilic
oxygen species were generated.
groups shifts from 1704 cmÀ1 on oCNTs (black line) to
1724 cmÀ1 on phosphate-modified CNTs (colored lines, 1P-, 3P-,
5P-oCNTs).[17] The intensity of the C=O stretching peak, relative
to the intensity of CÀC (1563 cmÀ1) in the aromatic rings of
CNTs,[18] declines with increased loading of phosphate, while
the peak due to CÀOÀP stretching vibrations (1087 cmÀ1, over-
lapping with CÀO stretching vibrations),[19] as well as asymmet-
ric stretching vibrations of CÀOÀP and PÀOÀP (1000–750 cmÀ1
)
increases.[20] This means that the excess phosphate reacts with
the oxygen-containing groups on the CNTs, forming CÀOÀP
bonds.[21] This is most likely due to esterification or dehydration
of ÀCOOH/ÀCÀO bonds on the surface of the CNTs with the À
OH groups of phosphate.[20c] In addition, the linkage of exces-
sive phosphates to the ketonic/quinonic C=O groups cannot
be ruled out, because the dehydrogenation selectivity decreas-
es. Nevertheless, the binding of C=O groups to phosphate
likely involves very weak physical forces, but not chemical
bonding because the activity can be easily recovered after
washing with distilled water, as shown in Figure 3.
In contrast, high phosphate loadings (>1%) have only
a moderate inhibition effect on the consumption of substrates
and the combustion of CNTs (Figure 5, a slower decline of the
red dashed and blue dashed lines, and a relatively gradual rise
of the black lines at wP O >1 wt%). The inset in Figure 5 shows
TPD experiments revealed two significant changes in the
evolution of CO2 desorption (Figure 7a) and CO desorption
(Figure 7b). The first was a gradual shift of the evolution to
higher temperatures when the loading of phosphate increased.
The second change was in the shape of the CO/CO2 evolution
signals: the peak intensity in the lower-temperature regime de-
creased while the evolution in the higher-temperature regime
became stronger. Clearly, phosphate modification retards the
decomposition of the corresponding functional groups. It is
noteworthy that the interaction between phosphate and the
functional groups becomes more and more significant when
the loading of phosphate was more than 1 wt%. In particular,
a sharp peak appears in the CO profiles at about 1180 K at
a loading of more than 3 wt%. The signals due to anhydride
and lactone in CO2 profiles shifted to higher temperatures with
increased phosphate loading. This change of CO/CO2 evolution
at high temperature and at high phosphate loading is assigned
to the presence of highly stable CÀOÀPO3 groups,[15a,22] formed
2
5
that the consumption rate of C5H12 decreased faster than the
consumption rate of oxygen (wP O =1–5 wt%), which indicates
2
5
that the inhibition effect on the consumption of C5H12 is stron-
ger than the consumption of O2. This means that excessive
phosphate (>1 wt%) selectively suppresses the sites active for
dehydrogenation (nucleophilic oxygen species: ketonic/qui-
nonic C=O).
The interaction between phosphates with oxygen-contain-
ing groups
Attenuated total reflection (ATR)–IR spectroscopy was used to
study the interaction between phosphorus species and
oxygen-containing groups. Figure 6 reveals that the C=O
stretching peak of ketones, aldehydes, lactones, or carboxyl
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3479

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Scheme 2. Schematic illustration for the evolution of functional groups on
the surface of CNTs and corresponding reaction routes tuned by phos-
phates.
Figure 7. Deconvolution of TPD profiles for (a) CO2 desorption and (b) CO
desorption.
Conclusions
due to the interaction between phosphorus oxides and
oxygen groups and dissociated at high temperature.[23]
Phosphate-modified carbon nanotubes (CNTs) as novel cata-
lysts for isopentane oxidative dehydrogenation (ODH) shown
a predominant selectivity towards dehydrogenation as com-
pared to the conventional V-Mg-O catalysts. Moreover, the
impact of phosphate on the structure–activity relationship is
investigated in detail. We reveal that different phosphate load-
ings strongly impact on the catalytic properties, either positive-
ly or negatively. Low phosphate loadings (0~0.1 wt%) result in
improved dehydrogenation selectivity by blocking the com-
bustion sites. The catalytic behavior mainly originates from
oxygen-containing groups on the surface of the CNTs. High
loadings (1~5 wt%) lead to a decline of the dehydrogenation
selectivity, due to the partial coverage of selective sites (C=O
groups). Meanwhile, the catalytic behavior is dominated by
a synergistic effect of oxygen groups and phosphorous oxides.
Excessive phosphate loading (15 wt%) results in a full coverage
of active sites from CNTs and no paraffin conversion. Based on
these results, in order to get a desired product we can control
the product distribution of ODH by adjusting the loading of
phosphate on CNT surfaces. Furthermore, our work implies
a universal principle on metal-free nanocarbon catalysis: by
tuning oxygen active sites, we can design a vast variety of cat-
alysts with controllable selectivity involved in the oxidative de-
hydrogenation of light alkanes and alkenes.
The effects caused by phosphate modification were ana-
lyzed in detail by deconvolution of TPD profiles,[24] as shown in
Figure 7. The evolution of CO2 is assigned to strongly acidic
carboxylic groups at ~560 K, weakly acidic carboxylic groups at
~620 K, carboxylic anhydrides at ~750 K, and lactones above
800 K.[25] For CO, the dominant contributions are attributed to
adsorbed CO at ~550 K,[26] carboxylic anhydrides at ~850 K,
phenols at ~1000 K, carbonyls/quinones at ~1150 K, as well as
stable CÀOÀPO3 groups at ~1180 K.[19,20] Although the results
of TPD analysis were not quantified, the decline of carboxyl
acidic groups and carbonyl groups with increased phosphate
loading is apparent. The reduction of carboxyl acidic groups
likely results from neutralization by phosphate ammonia. The
reduction of signals attributable to carbonyls agrees well with
the conclusions derived from IR measurements.
Schematic evolution of different active sites on the surface
of CNTs
Based on the observations mentioned above, a conceivable
mechanism of the role of phosphate-modified CNTs in the
ODH of isopentane is proposed in Scheme 2. With different
phosphate loadings, three distinct stages are observed:
1) At low phosphate loadings (0.1–1 wt%), defect sites
where the highly active oxygen species (superoxide O2À, per-
oxide O22À) could be formed are blocked. This inhibits the
deep oxidation of isopentane to carbon oxides, either directly
or consecutively, and gives a maximum dehydrogenation selec-
tivity (82%). The blocking of such defect sites also renders the
modified CNTs a high oxidation resistance.
Experimental Section
Material synthesis
Pristine CNTs were provided by Tsinghua University (PR China) with
a length in the range of 5–15 mm, average diameter 12Æ4 nm, and
wall thickness 4Æ1 nm. The CNTs were stirred in 5m hydrochloric
acid aqueous solutions (500 mLgÀ1) at room temperature for 12 h
and then were rinsed with distilled water until neutral pH and
dried overnight at 393 K. These samples were denoted as CNTs.
Oxidized CNTs (oCNTs) were obtained by hydrothermal oxidation
with 0.3m nitric acid aqueous solution (60 mLgÀ1) in a 100 mL
Teflon-lined autoclave at 473 K for 2 h. Phosphate-modified sam-
2) At loading of phosphate above 1 wt%, selective C=O sites
are covered: the number of ketone C=O groups is lowered
which results in a decline of the dehydrogenation selectivity.
3) Higher loadings (more than 1 wt%) may lead to the accu-
mulation of phosphorus oxides, POx. The role of the accumu-
lated POx particles in the reaction needs to be explored fur-
ther.
ples were first prepared by impregnation of oCNTs (10 mLgÀ1
)
using aqueous solutions of (NH4)2HPO4 (Sigma Aldrich, >99%) and
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3480

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
were sonicated until all the water had evaporated, and then dried
at 393 K for 24 h. Different samples with P2O5 contents of 0.1, 0.5,
1, 3, and 5 wt% were labeled as 0.1–5 P-oCNT. Mg3V2O7 (pyrovana-
date) and Mg3V2O8 (orthovanadate) samples were synthesized ac-
cording to reference [7]. An appropriate amount of solid MgO was
added into a hot aqueous solution of NH4VO3. The superfluous
water was evaporated and then dried at 353 K. The received solid
was crushed and calcined at 823 K for 6 h. The XRD patterns of
samples shown in Figure S1 (Supporting Information) are consis-
tent with previous studies.[10]
Zailai Xie at the FHI (Berlin) with ATR-IR measurements and Xiaoli
Pan at the IMR (Shenyang) with TEM characterization.
Keywords: active
sites
·
carbon
nanotubes
·
dehydrogenation · hydrocarbons · phosphorous
[1] a) J. R. Rostrup-Nielsen, J. Catal. 1984, 85, 31–43; b) J. F. Shultz, L. J. E.
Hofer, F. S. Karn, R. B. Anderson, J. Phys. Chem. 1962, 66, 501–506; c) C.
Bartholomew, P. Agrawal, J. Katzer, Adv. Catal. 1982, 31, 135–242.
[2] a) V. Schwartz, W. Fu, Y.-T. Tsai, H. M. Meyer, A. J. Rondinone, J. Chen, Z.
Wu, S. H. Overbury, C. Liang, ChemSusChem 2013, 6, 840–846; b) F.
Cavani, J. H. Teles, ChemSusChem 2009, 2, 508–534; c) D. S. Su, J.
Zhang, B. Frank, A. Thomas, X. Wang, J. Paraknowitsch, R. Schlçgl,
ChemSusChem 2010, 3, 169–180; d) X.-K. Kong, C.-L. Chen, Q.-W. Chen,
Chem. Soc. Rev. 2014, 43, 2841–2857; e) D. S. Su, S. Perathoner, G. Centi,
Chem. Rev. 2013, 113, 5782–5816; f) D. Haag, H. Kung, Top. Catal. 2013,
1–12.
ODH reactions of isopentane
On average, 50 mg of catalyst sample (100–400 mm) mixed with
1 g of quartz sand wase placed in a quartz tubular reactor and
kept at 673 K in a 16.7 mLminÀ1 flow with i-C5H12/O2/He=1:2:38
until constant conversion was achieved. The apparent activation
energy was determined by stepwise heating (10 K steps) from 623-
673 K at 25 mLminÀ1 (20 mg oCNTs, 50 mg 1P-oCNTs, 100 mg 3P-
oCNTs, and 100 mg Mg3V2O8).The reactants and products were ana-
lyzed by using an on-line gas chromatography system (Agilent
7890A) with two Hayesep Q columns, a HP-PLOT/Al2O3 column
(50 mꢁ0.32 mmꢁ8 mm), and a molecular sieve column. CO and
CO2 were detected with a thermal conductivity detector and hy-
drocarbons were detected with a flame ionization detector.
[3] B. Frank, M. Morassutto, R. Schomꢂcker, R. Schlçgl, D. S. Su, ChemCatCh-
em 2010, 2, 644–648.
[4] a) Z.-j. Sui, J.-h. Zhou, Y.-c. Dai, W.-k. Yuan, Catal. Today 2005, 106, 90–
94; b) B. Frank, J. Zhang, R. Blume, R. Schlçgl, D. S. Su, Angew. Chem. Int.
Ed. 2009, 48, 6913–6917; Angew. Chem. 2009, 121, 7046–7051.
[5] J. Zhang, X. Liu, R. Blume, A. Zhang, R. Schlçgl, D. S. Su, Science 2008,
322, 73–77.
[6] a) C. Liang, H. Xie, V. Schwartz, J. Howe, S. Dai, S. H. Overbury, J. Am.
Chem. Soc. 2009, 131, 7735–7741; b) J. de Jesffls Díaz Velµsquez, L. M. C.
Suµrez, J. L. Figueiredo, Appl. Catal. A 2006, 311, 51–57; c) V. Schwartz,
H. Xie, H. M. Meyer, S. H. Overbury, C. Liang, Carbon 2011, 49, 659–668;
d) G. K. P. Dathar, Y.-T. Tsai, K. Gierszal, Y. Xu, C. Liang, A. J. Rondinone,
S. H. Overbury, V. Schwartz, ChemSusChem 2014, 7, 483–491.
[7] H. H. Kung, Adv. Catal. 1994, 40, 1–38.
Catalyst characterization
[8] P. M. Michalakos, M. C. Kung, I. Jahan, H. H. Kung, J. Catal. 1993, 140,
202–213.
Fourier-transform infrared spectroscopy was performed using
a Nicolet FTIR infrared microscope with an ATR attachment (ATR-
IR). Specific surface areas (BET) and pore size distributions [Barrett–
Joyner–Halenda (BJH)] of the CNT samples were determined by N2
physisorption at 77 K utilizing a Micromeritics ASAP 3020 BET appa-
ratus. The oxidation stability was carried out by a thermogravimet-
ric analyzer (Netzsch 449 F3). 5–10 mg of sample was heated line-
arly in a corundum crucible from 323 to 1100 K in 50 mLminÀ1 of
10 vol% O2 in Ar. Temperature-programmed desorption (TPD) ex-
periments were performed in He atmosphere. The temperature
program involved heating at 10 KminÀ1 from ambient temperature
to 423 K, that temperature was kept for 30 min, the temperature
was then raised to 1200 K, which was again kept for 30 min, and
ultimately the sample was cooled to room temperature in a helium
atmosphere. The product gases were analyzed by on-line MS
(Pfeiffer-Balzer Omnistar); the masses 28 and 44 were monitored
for the detection of CO and CO2 with a time-resolution of 0.1 s.
Surface analyses by XPS were carried out with an ESCALAB 250 in-
strument. Peak areas were normalized with theoretical cross-sec-
tions to obtain the relative surface compositions. Raman spectros-
copy was conducted with a HORIBA Jobin Yvon spectrometer
using a 633 nm excitation line. Transmission electron microscopy
images were recorded using TEM (FEI Tecnai G2 F20) instruments.
[9] M. A. Chaar, D. Patel, H. H. Kung, J. Catal. 1988, 109, 463–467.
[10] D. Patel, P. J. Andersen, H. H. Kung, J. Catal. 1990, 125, 132–142.
[11] S. A. Korili, B. D. P. Ruiz, Catal. Today 1996, 32, 229–235.
[12] R. X. Valenzuela, J. M. Asperilla, V. C. Corberµn, Ind. Eng. Chem. Res.
2008, 47, 8037–8042.
[13] a) R. Grabowski, Catal. Rev. 2006, 48, 199–268; b) C. TØllez, M. MenØn-
dez, J. Santamaría, J. Catal. 1999, 183, 210–221; c) D. S. H. Sam, V.
Soenen, J. C. Volta, J. Catal. 1990, 123, 417–435.
[14] a) C. Park, R. T. K. Baker, J. Phys. Chem. B 1999, 103, 2453–2459; b) B.
Frank, R. Blume, A. Rinaldi, A. Trunschke, R. Schlçgl, Angew. Chem. Int.
Ed. 2011, 50, 10226–10230; Angew. Chem. 2011, 123, 10408–10413.
[15] a) J. M. Rosas, R. Ruiz-Rosas, J. Rodríguez-Mirasol, T. Cordero, Carbon
2012, 50, 1523–1537; b) Y.-J. Lee, L. R. Radovic, Carbon 2003, 41, 1987–
1997.
[16] a) L. M. Madeira, M. F. Portela, Catal. Rev. 2002, 44, 247–286; b) G. I.
Panov, K. A. Dubkov, E. V. Starokon, Catal. Today 2006, 117, 148–155;
c) F. Atamny, J. Blçcker, A. Dübotzky, H. Kur, O. Timpe, G. Loose, W.
Mahdi, R. Schlçgl, Mol. Phys. 1992, 76, 851–886.
[17] a) D. V. Kosynkin, A. L. Higginbotham, A. Sinitskii, J. R. Lomeda, A.
Dimiev, B. K. Price, J. M. Tour, Nature 2009, 458, 872–877; b) A. M. Puziy,
O. I. Poddubnaya, A. Martínez-Alonso, F. Suµrez-García, J. M. D. Tascón,
Carbon 2002, 40, 1493–1505.
[18] J. Zhang, D. S. Su, R. Blume, R. Schlçgl, R. Wang, X. g. Yang, A. Gajovic,
Angew. Chem. Int. Ed. 2010, 49, 8640–8644; Angew. Chem. 2010, 122,
8822–8826.
[19] a) J. Coates, in Encyclopedia of Analytical Chemistry (Ed.: R. A. Meyers),
Wiley, Chichester, 2000, pp. 10815–10837; b) I. F. Amaral, P. L. Granja,
M. A. Barbosa, J. Biomater. Sci. Polym. Ed. 2005, 16, 1575–1593.
[20] a) S. C. H. Kwok, J. Wang, P. K. Chu, Diamond Relat. Mater. 2005, 14, 78–
85; b) C. Presti, J. G. Alauzun, D. Laurencin, P. H. Mutin, Chem. Mater.
2013, 25, 2051–2055.
[21] a) D. W. McKee, Carbon 1972, 10, 491–497; b) D. W. McKee, C. L. Spiro,
E. J. Lamby, Carbon 1984, 22, 285–290.
[22] a) A. M. Puziy, O. I. Poddubnaya, A. M. Ziatdinov, Appl. Surf. Sci. 2006,
252, 8036–8038; b) A. M. Puziy, O. I. Poddubnaya, R. P. Socha, J. Gurgul,
M. Wisniewski, Carbon 2008, 46, 2113–2123.
Acknowledgements
The authors acknowledge financial support by MOST
(2011CBA00504), the NSFC of China (21133010, 21203215,
51221264, 21261160487),the “Strategic Priority Research Pro-
gram” of the Chinese Academy of Sciences (XDA09030103),
21203214, and Sinopec China. The authors thank Prof. Dr. Robert
Schlçgl for fruitful discussions, and the helpful assistance of Dr.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3481

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
[23] X. Wu, L. R. Radovic, Carbon 2006, 44, 141–151.
[26] J. H. Zhou, Z. J. Sui, J. Zhu, P. Li, D. Chen, Y. C. Dai, W.-K. Yuan, Carbon
2007, 45, 785–796.
[24] a) M. F. R. Pereira, J. J. M. Orfao, J. L. Figueiredo, Appl. Catal. A 1999, 184,
153–160; b) J. L. Figueiredo, M. F. R. Pereira, M. M. A. Freitas, J. J. M.
Orfao, Ind. Eng. Chem. Res. 2007, 46, 4110–4115.
[25] a) M. Almarri, X. Ma, C. Song, Energy Fuels 2009, 23, 3940–3947; b) N. Li,
X. L. Ma, Q. f. Zha, K. S. Kim, Y. S. Chen, C. S. Song, Carbon 2011, 49,
5002–5013.
Received: May 23, 2014
Published online on September 11, 2014
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 3476 – 3482 3482