Support-induced oxidation state of catalytic Ru nanoparticles on carbon nanofibers that were doped with heteroatoms (O, N) for the decomposition of NH3

Article abstract of DOI:10.1002/cctc.201300455

Ru nanoparticles were supported on monoliths that were coated with variously functionalized carbon nanofibers (CNFs), that is, un-doped CNFs, CNFs that had been post-treated with H₂O₂, and CNFs that had been doped with nitrogen during their growth. The Ru uptake (by equilibrium adsorption) onto N-doped CNFs was larger compared to that on their un-doped and O-doped counterparts. The functionalization of the CNF support did not play a significant role in determining the size of the deposited Ru nanoparticles, but it substantially impacted on the sintering under the reaction conditions and on the electron density of the reduced metal. Among the catalysts on the different CNF supports, Ru on N-CNF exhibited the highest H₂ productivity from ammonia decomposition, which pointed to electronic effects that were induced by functionalization of the support. What a state of affairs: High NH ₃-decomposition conversions were obtained at low temperatures by using a supported Ru catalyst on a honeycomb monolith that was coated with N-doped carbon nanofibers (N-CNF). The higher negative charge on the Ru center accounted for its enhanced performance compared to its un-doped CNF-supported counterpart. Copyright

Full text of DOI:10.1002/cctc.201300455

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300455
Support-Induced Oxidation State of Catalytic Ru
Nanoparticles on Carbon Nanofibers that were Doped with
Heteroatoms (O, N) for the Decomposition of NH3
Yanila Marco, Laura Roldꢀn, Sabino Armenise, and Enrique Garcꢁa-Bordejꢂ*[a]
Ru nanoparticles were supported on monoliths that were
coated with variously functionalized carbon nanofibers (CNFs),
that is, un-doped CNFs, CNFs that had been post-treated with
H2O2, and CNFs that had been doped with nitrogen during
their growth. The Ru uptake (by equilibrium adsorption) onto
N-doped CNFs was larger compared to that on their un-doped
and O-doped counterparts. The functionalization of the CNF
support did not play a significant role in determining the size
of the deposited Ru nanoparticles, but it substantially impact-
ed on the sintering under the reaction conditions and on the
electron density of the reduced metal. Among the catalysts on
the different CNF supports, Ru on N-CNF exhibited the highest
H2 productivity from ammonia decomposition, which pointed
to electronic effects that were induced by functionalization of
the support.
Introduction
Carbon nanofibers (CNFs) or carbon nanotubes (CNTs) that are
doped with nitrogen have shown superior performance to
their un-doped counterparts as catalyst supports for different
reactions. For instance, they have been used to support Ru cat-
alysts for ammonia decomposition,[1,2] to support Pt electroca-
talysts[3] or Fe catalysts[4] for the oxygen-reduction reaction, as
Pd supports for the direct synthesis of H2O2 from H2 and O2,[5]
as Pd supports for the hydrogenation of cynamaldehyde,[6] as
supports of Pd and Au for the selective aerobic oxidation of al-
cohols,[7] and as supports of Au for the preferential oxidation
of CO.[8] There are several benefits to N-doped carbon sup-
ports, such as improved activity, owing to better dispersion of
the catalyst,[1,7] increased selectivity,[6,9] owing to specific inter-
actions between the doped support and the metal, and en-
hanced stability, because the nitrogen groups contribute to
keeping the metal particles well-dispersed[8,10] and in a reduced
state.[11,12] This enhanced performance could be attributed to
either geometric effects, because the doping can govern the
size and structure of the metal nanoparticles, or to electronic
effects on the metal nanoparticles, owing to the high electron-
ic conductivity of the graphitic carbon atoms.
whereas H2 requires very low temperatures (20 K) and expen-
sive tanks to adopt its liquid form. Alternative ammonia sour-
ces, both in the liquid and solid states, have also been pro-
posed to generate H2, such as urea[14] or metal ammines.[15]
Thus, the safety issues with regards to its storage and handling
are well-established.[15,16] More importantly, the product stream
(hydrogen/nitrogen) is CO-free and the amount of unconvert-
ed ammonia can be decreased to less than 200 ppb by using
a suitable adsorbent. Therefore, the H2 stream can be fed di-
rectly into a fuel cell without further purification steps, which
are needed for H2 that is generated from hydrocarbon reform-
ing. The decomposition of NH3 is a key step in the implemen-
tation of this technology. Therefore, it is necessary to develop
robust catalysts that are active at low temperatures.
To shed some light on the effect of the N doping of a CNF
support on the metal catalyst and its catalytic performance in
ammonia decomposition, herein, we have used un-doped
CNFs and CNFs that were doped with two heteroatoms (O, N)
as a Ru support. Doping with O and N atoms was either per-
formed by postfunctionalization with H2O2 or by in situ doping
during CNF growth, respectively. The catalysts were character-
ized by STEM, XPS, and TPR. CNF functionalization did not sig-
nificantly affect the Ru-particle size. On the contrary, the oxida-
tion state of the impregnated fresh catalyst and the electronic
charge on the reduced Ru catalyst markedly depended on the
doping of the CNF support. The characterization results can ex-
plain the higher H2 productivity with the N-CNF-supported cat-
alyst.
NH3 has been put forward as a very promising H2-storage
media for small-scale fuel cells in some niche applications.[13]
Ammonia has high gravimetric (17.7 wt.%) and volumetric H2
capacities, the latter of which is about 45% larger than that of
liquid hydrogen. It poses no danger of explosion and it can be
easily stored as a liquid at room temperature and at 8.6 bar,
[a] Y. Marco, Dr. L. Roldꢀn, Dr. S. Armenise, Dr. E. Garcꢁa-Bordejꢂ
Instituto de Carboquimica (ICB-CSIC)
Miguel Luesma Castꢀn 4, 50018 Zaragoza (Spain)
Fax.: (+34)976733318
E-mail: jegarcia@icb.csic.es
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300455.
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3829 – 3834 3829

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
piece of the monolith (0.5 cm) was cut off and characterized. This
sample was denoted by the name of the support followed by
“fresh”. The metal uptake was calculated from the analysis of the
initial and final solutions by inductive coupled plasma-optical spec-
trometry (ICP-OES).
Experimental Section
Growth of un-doped CNFs and N-doped CNFs on cordierite
monoliths
The CNFs and N-CNFs were grown on cordierite monoliths accord-
ing to literature procedures.[17,18] Cordierite monoliths (diameter:
1 cm, length: 5 cm, 400 cpsi; Corning) were wash-coated with alu-
mina by using a dip-coating method that was similar to the sol-gel
coating method described by Nijhuis et al.[19] In this method, a sol
was prepared from pseudoboehmite (AlOOH, Pural; Sasol), urea,
and a 0.3m aqueous solution of nitric acid with a weight ratio of
2:1:5. After stirring for 30 min, the dried monolith was dipped in
the sol. The liquid inside the monolith channels was removed by
thoroughly flushing with pressurized air and then drying at RT for
24 h, with continuous rotation around its axis. Finally, the monolith
was calcined in air at a rate of 1 KminÀ1 up to 873 K to obtain the
g-alumina wash-coating.
Nickel was deposited by adsorption from a pH-neutral solution ac-
cording to a literature procedure.[20] Ni(NO3)2·6H2O (29 g; Sigma–Al-
drich), NH4NO3 (80 g; Sigma–Aldrich), and ammonia solution (25%,
4 mL) were mixed in a 1 liter bottle. The monolith samples were
kept overnight in this solution under a continuous flow of the
liquid through their channels. Then, the monoliths were rinsed
thoroughly with deionized water, followed by drying first at RT
overnight and then at 373 for 1 h. Subsequently, the monoliths
were calcined in a flow of nitrogen gas (1 KminÀ1) up to 873 K, fol-
lowed by a dwell time of 2 h. The Ni content in the monoliths was
measured by inductive coupled plasma-optical emission spectros-
copy (ICP-OES).
Subsequently, the monolith was rinsed with copious amounts of
water, dried at 380 K, calcined under a N2 atmosphere, and re-
duced in a flow of H2 at 473 K by using a heating rate of 1KminÀ1
.
The sample after this calcination and reduction treatment was de-
noted by the name of the support followed by “reduced”.
Characterization
Ex situ XPS spectra were recorded on an ESCAPlus Omnicrom
system that was equipped with an AlKa radiation source to excite
the sample. Calibration of the instrument was performed with the
Ag 3d5/2 line at 368.27 eV. All of the measurements were performed
under ultrahigh vacuum (UHV, <10À10 Torr). Internal referencing of
the spectrometer energies was performed by using the dominant
C 1s peak of the support at 284.6 eV and the Al 2p peak at 74.3 eV.
The program that was used to perform the curve-fitting of the
spectra was CasaXPS after performing a baseline Shirley method.
Because the binding energy of Ru 3d overlapped with the C 1s
region (284.6), we used the Ru 3p3/2 region (458–468 eV) to deter-
mine the atomic content of ruthenium species in the catalyst. The
analysis of the Ru XPS peak enabled the determination of Ru8,
Ru4+, and Ru hydrate at 461.2–461.6, 463.4–463.8, and 465.7–
466.1 eV, respectively.[21,22]
In situ XPS was also performed. To this end, prior to recording the
XPS spectra, the passivated catalyst was reduced in situ with a flow
of H2 at 473 K. Therefore, the reduced sample was not exposed to
air before the XPS measurements, thereby avoiding re-oxidation.
The size of the Ru nanoparticles on the carbon nanofibers was
studied by using scanning transmission electron microscopy
For the growth of the CNFs and N-CNFs, the monoliths were fitted
in a quartz reactor by wrapping in a quartz band. The reduction of
the calcined catalyst was performed under a hydrogen atmosphere
at 823 K for 120 min (5 KminÀ1). Then, the monolith was heated
(5 KminÀ1) to 873 K. Once this temperature had been reached,
(STEM) on
a FEI TECNAI F30 electron microscope that was
a
gaseous mixture of C2H6/H2 (50:50) or C2H6/NH3 (50:50,
100 mLminÀ1, standard temperature and pressure) was fed into
the reactor for the growth of samples, which were denoted as
CNFs and N-CNF, respectively. The growth of the CNFs was allowed
to proceed for 2 h, following up by cooling under an inert atmos-
phere. This reaction time was long enough for the catalyst to be
deactivated, as corroborated by gas chromatography analysis of
the flue gas.
equipped with a Gatan Energy Filter and a cold-field emission gun
(FEG) operating at 300 kV with a lattice resolution of 1.5 ꢄ. TEM
specimens were prepared by ultrasonically dispersing the powder
that was retrieved from the monoliths in EtOH. A drop of the sus-
pension was applied to a holey carbon support grid. The particle-
size distribution was calculated by statistical analysis of 300 parti-
cles on the CNF from about 20 images.
Temperature-programmed reduction (TPR) profiles were measured
on a Micromeritics PulseChemisorb 2700. After placing the sample
in the holder and flowing a mixture of 10% H2 in Ar, the tempera-
The CNF-monolith was also subjected to oxidation treatment with
H2O2 at RT for 20 h. During this treatment, H2O2 was continuously
flowed through the channels of the monolith. This sample was de-
noted as O-CNF.
ture was increased from 313 to 573 K at a rate of 5 KminÀ1
.
Catalytic testing
Adsorption of the Ru precursor onto the CNF-based
supports
To compare the intrinsic activity of the catalysts for NH3 decompo-
sition, the monoliths were crushed into powders with particle
size<100 mm. The powdered catalyst (200 mg) was diluted with
SiC and placed between quartz wool in the middle of a reactor
tube (6 mm i.d.). Subsequently, the reactor was placed in the con-
stant-temperature zone of a furnace that was equipped with a tem-
perature controller (Eurotherm). The flow rate of the feed gas (5%
anhydrous ammonia in Ar) that was used for the catalyst testing
was fixed at 33 mLminÀ1 by using Bronkhorst mass-flow control-
lers. The reactant dilution minimized the increase in volume at
high conversion. The absence of diffusional limitations was as-
sessed by applying the Weisz–Prater criterion.[23] The absence of
external diffusion limitations was corroborated because increasing
the flow rate did not lead to a further increase in activity.
For the adsorption of the metal-nanoparticle precursor onto the
CNF-, O-CNF-, and N-CNF-coated monoliths, an equilibrium-adsorp-
tion method was used. First, an aqueous solution of Ru nytrosil ni-
trate ([Ru(NO)(NO3)], 40 mL) was prepared. The amount of metal
that was added into the solution was calculated to be 5 wt.% with
respect to the weight of the CNFs, which led to a ruthenium con-
centration of 500–700 ppm in the solution. The CNF-, N-CNF-, and
O-CNF-coated monoliths were introduced into a tube with the
metal solution and the vial was continuously rotated overnight
perpendicular to the axial direction. This procedure guaranteed the
continuous flow of the liquid through the channels of the mono-
lith and the homogenization of the solution. After this step, a small
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3829 – 3834 3830

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
The passivated catalysts were in-situ reduced at 473 K and subse-
quently tested in the decomposition of NH3 at temperatures be-
tween 573–773 K. First, the conversion was measured at the high-
est temperature until it stabilized. Then, the temperature was de-
creased in 50 K steps and the conversion measured until it stabi-
lized. Subsequently, the conversion was also measured by stepwise
increasing the temperature to check the repeatability of the experi-
ment. The outlet gases were analyzed on an Agilent Micro GC
3000A; H2 and N2 were analyzed by using a molsieve column and
NH3 was analyzed by using a Plot-Q column.
Results and Discussion
The size of the Ru particles on the supports was characterized
by using STEM. On all of the supports, the particle-size distri-
bution (Figure 1) was very narrow and all of the Ru particles
were very small, with average sizes of (1.42Æ0.022), (0.95Æ
0.017), (1.28Æ0.018) nm for Ru on CNF, O-CNF, and N-CNF, re-
spectively.
Table 1 shows the metal uptake of Ru on the CNFs with dif-
ferent surface chemistry. The metal uptake was 25% larger on
N-CNFs than on un-doped CNFs and O-CNF. In a previous
report,[24] we observed that the amount of pyridinic groups on
N-CNF decreased after metal deposition, thus suggesting that
these groups could play a role in the adsorption of the metal
precursor. This result was in agreement with other reports,[1]
which showed that the pyridinic groups interacted strongly
with Ru3+. They found that the average size of the Ru particles
decreased with increasing content of pyridinic nitrogen atoms.
In contrast, we did not observe any significant effect of nitro-
gen groups on the Ru-particle size.
To study the reduction state of the metal at the different
preparation stages, the catalysts were characterized by ex situ
XPS (Figure 2) and TPR (Figure 3) upon impregnation of the
metal precursor (fresh) and after reduction and exposure to air
(reduced). Quantitative XPS analysis of the reduced catalysts
revealed that the O/C ratios (at.%) were 16:1, 24:1, and 7.2:1
for the CNF-, O-CNF-, and N-CNF-supported catalysts, respec-
tively. After the reaction, the O/C ratios decreased slightly to
13:1, 17:1, and 6.5:1, respectively. The XPS N/C ratio (at.%) of
the reduced N-CNF catalyst was 5.2:1. The N/C ratio remained
unchanged (within experimental error) after reduction and
after the reaction (see the Supporting Information). The pro-
portion of each type of nitrogen species changed after the re-
Figure 1. Representative STEM images and particle-size distributions of re-
duced Ru catalysts that were supported on a) un-doped CNF, b) O-CNF, and
c) N-doped CNF. Scale bars=20 nm.
action, with the peak for the pyridine oxide species disappear-
ing, whereas the amount of pyridinic and pyrrolic nitrogen
species increased.
The quantitative results of Ru XPS peak deconvolution are
also listed in Table 1. Upon impregnation of Ru onto the sup-
ports, the percentage of metallic catalyst, that is, the percent-
age of Ru in its zero oxidation state (Figure 2 and Table 1), was
Table 1. Quantification of the XPS spectra and TPR profiles of fresh and reduced Ru catalysts on both CNF/monolith and N-CNF/monolith supports.
Sample
Ru uptake
per CNF
[wt.%]
XPS peak position
[eV] (abundance [%])
RuO2
TPR
H2 consumed
[cm3 gÀ1Ru]
Peak maximum
[K]
Abundance of
oxidized Ru[a] [%]
Ru0
RuO2·H2O
Ru/CNF fresh
3.9
3.7
5.1
3.9
3.7
5.1
461.0 (0)
461.8 (9)
462.6 (18)
461.8 (43)
461.8 (19)
461.8 (54)
463.8 (46)
464.3 (57)
464.0 (64)
463.4 (38)
463.4 (53)
463.4 (34)
466.0 (54)
466.0 (34)
465.8 (18)
465.7 (19)
465.4 (28)
465.4 (12)
416.8
405.0
407.6
353
348
338
2323
1260
2333
1323
782
100
100
100
56.9
62.0
9.7
Ru/O-CNF fresh
Ru/N-CNF fresh
Ru/CNF reduced
Ru/O-CNF reduced
Ru/N-CNF reduced
226
[a] Percentage of Ru that was oxidized upon exposure to air, which was calculated as H2 consumption in reduced catalyst with respect to H2 consumption
in the fresh catalyst.
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3829 – 3834 3831

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
the reduced catalysts were subsequently exposed to air, part
of metal was re-oxidized, as evidenced by a small low-tempera-
ture reduction peak in the TPR spectrum of the reduced cata-
lyst between 338–353 K. Both ex situ XPS and TPR techniques
revealed that the extent of re-oxidation was lower if supported
on N-doped CNFs compared to its un-doped and oxidized
counterparts. This result was corroborated by integration of
the TPR peaks of the reduced catalyst (Table 1), which showed
that 90% Ru was in its metallic state for Ru on N-doped CNF,
whereas only 43% and 38% of Ru was metallic if supported
on un-doped and oxidized CNFs, respectively. From the XPS
deconvolution, 54%, 43%, and 19% of Ru was metallic Ru on
N-doped, un-doped, and oxidized CNFs, respectively. The dif-
ferences between these percentages of the reduced metal as
measured by XPS were smaller than those measured by TPR.
The degree of re-oxidation as measured by XPS was generally
larger than that measured by TPR, thus indicating that the re-
oxidation mainly occurred at the catalyst surface. Both tech-
niques confirmed that the degree of re-oxidation of the two
metals upon exposure to air followed the order: N-CNF !
CNF<O-CNF. Therefore, nitrogen groups contributed to keep-
ing the metal in its reduced state on exposure to an O2 atmos-
phere, which could be attributed to the electron-donating
properties of nitrogen on highly conducting graphitic struc-
tures.[25] Clearly, under the reaction conditions (reducing at-
mosphere, high temperature), the metal must be in its reduced
state. The extent of re-oxidation at room temperature is con-
sidered herein as a descriptor of electron donation from sup-
ports with different types of functionalization. Operando spec-
troscopic techniques are planned to confirm the electron don-
ation from the support under these reaction conditions.
Figure 2. XPS Ru 3p3/2 spectra of the metal catalysts that were adsorbed
onto un-doped CNF, N-doped CNF, and O-CNF after two (ex situ) preparation
stages, that is, just after impregnating the precursor (fresh) and after reduc-
tion at 473 K and exposure to air at room temperature (reduced).
Besides ex situ XPS characterization, in situ XPS characteriza-
tion of Ru/CNF and Ru/N-CNF was also performed (Figure 4).
During this characterization process, the catalyst was reduced
in the XPS chamber at 473 K and XPS spectra were recorded
without exposing the sample to the atmosphere. Only one
peak, which corresponded to reduced Ru, was observed for
both catalysts. The center of the peak was shifted by 0.3 eV for
Ru supported on CNF compared to that on N-CNF, which indi-
Figure 3. TPR profile of the Ru catalysts that were adsorbed onto un-doped
CNF, N-doped CNF, and O-CNF after two preparation stages, that is, just
after impregnating the precursor (fresh) and after reduction at 473 K and ex-
posure to air at room temperature (reduced).
about 18% on N-doped CNFs, whereas it was significantly
lower or zero on the other functionalized CNFs. This result sug-
gests that nitrogen groups cooperate to reduce the Ru2+ pre-
cursor upon impregnation and, hence, fixing it onto the sup-
port. This process can account for the higher metal loadings
on N-CNF and guarantees that the catalyst precursor is uni-
formly distributed along the N-CNF monolith, as observed by
TEM at different monolith locations.
The impregnated catalyst that was prepared as described
above was reduced at 473 K in a flow of H2. TPR measurements
of the fresh catalysts (Figure 3) demonstrated that this temper-
ature was high enough to reduce the Ru precursor. However, if
Figure 4. XPS Ru 3p3/2 spectra of Ru/N-CNF and Ru/CNF that were reduced
in situ at 473 K.
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3829 – 3834 3832

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
cated that the electronic density on the ruthenium was slightly
larger if supported on N-CNF for the fully reduced catalyst.
These Ru catalysts were tested in the decomposition of NH3
for in situ H2 generation. A plot of conversion versus tempera-
ture is shown in Figure 5. Ru supported on N-CNF (Ru/N-CNF)
showed significantly higher conversion than Ru on the other
perimental error). Statistical analysis of the STEM images of the
catalysts after the reaction (see the Supporting Information) re-
vealed an average Ru size of (1.35Æ0.020), (1.47Æ0.027), and
(1.19Æ0.023) nm for Ru on CNF, O-CNF, and N-CNF,
respectively. Compared with the Ru mean size of the reduced
catalysts before the reaction (Figure 1), the Ru particle sizes
did not increase after the reaction, except for the catalyst that
was supported on O-CNF, which underwent a 50% increase in
particle size. The initial decrease in activity could be explained
by initial restructuring of the metal catalyst or support on ex-
posure to the reaction conditions.
To gain insight into the interactions between the metal par-
ticles on the N-doped carbon support, a survey of the litera-
ture regarding the interactions of metal nanoparticles with un-
doped carbon materials was performed. The interactions of
metal particles with un-doped carbon materials has been well-
described elsewhere.[27] For un-doped activated carbon materi-
als, unsaturated surface carbon atoms on the edges or defects
of graphite planes have been found to form very strong bonds
with metal atoms.[28] At very low loadings, that is, small nano-
particles, molecular-orbital analysis indicated that metal atoms
on activated carbon surface were positively charged, owing to
the extremely effective delocalized donor–acceptor interac-
tions with the p system of the support. Increasing the loading,
that is, employing larger nanoparticles, caused a decrease in
the corresponding net charge. The positive charge on Pd
agreed with the highest Pd 3d binding energies for the Pd
cluster on graphite with a mean diameter of less than 1 nm rel-
ative to the bulk metal.[29] Larger Pd clusters (>5 nm) showed
binding energies that were close to the value of the bulk
metal (335.1 eV). The shift in the binding energy relative to the
bulk metal was as large as +1.2 eV in clusters of diameter
<1 nm. According to the literature reports mentioned above,
smaller metal particles on un-doped carbon materials gave rise
to higher binding energies, owing to electron transfer from
the metal to the support. Comparing the binding energies of
the metals on the different CNF-based supports that were pre-
pared herein, the smallest positive charge on the metal atoms
on N-CNFs could not be explained by metal-particle-size ef-
fects because STEM analysis showed that the metal particles
were very similar and there were no particles larger than
2.5 nm for all of the catalysts. Thus, the difference in the oxida-
tion state of Ru on the different CNF supports should be attrib-
uted to the effect of electronic donation of the support rather
than to particle-size effects.
Figure 5. Plot of NH3 conversion as a function of temperature in the steady
state (after an initial stabilization period) for the different catalysts: Ru/CNF
(~), Ru/O-CNF (&), and Ru/N-CNF (*); dashed line denotes the N-CNF sup-
port. Inset shows the NH3 conversions over 72 h at 723 K.
supports at the same temperature. On approaching complete
conversion, the differences in conversion between the catalysts
became less apparent, because the reaction rate has been re-
ported to be proportional to the concentration of NH3 and in-
hibited by the H2 product.[26] At NH3 conversions close to
100%, the influence of the concentrations of the reactants and
products prevailed over the intrinsic catalyst activity.
Because the Ru loading in Ru/N-CNF was larger than in the
other samples, the specific H2 productivity (per mg of Ru) at
723 K was calculated to enable a comparison between all of
the catalysts (Table 2). Normalized to the amount of Ru cata-
lyst, the Ru/N-CNF catalyst still exhibited higher H2 productivi-
ty, which could be attributed to the higher intrinsic activity of
the catalyst that was supported on N-CNF compared to those
supported on CNF with other types of functionalization.
The inset in Figure 5 shows the NH3 conversion at 723 K
over 72 h for the different catalysts. The conversion decreased
at the start of experiment and then remained stable (within ex-
Recently, Chen et al. also reported that Pd metal was in
a more-reduced state on N-CNT than on O-CNT for both the
fresh and reduced catalysts.[30] In addition, they observed
smaller Pd particle size for the
N-CNT-supported catalyst at sim-
Table 2. Catalytic activity for H2 production from the catalytic decomposition of NH3.
ilar metal loadings. Herein, Ru
showed a larger metal loading
on N-CNF than on CNF and
O-CNF, which may also affect the
metal-particle size. Despite its
higher metal loading, after the
reaction, Ru on N-CNF had suf-
Catalyst
Ru loading per
weight of CNF
[wt.%]
Ru loading per
weight of monolith
[wt.%]
Conversion
at 723 K
[%]
Specific H2 productivity
at 723 K
[mmolmgRuÀ1 hÀ1
]
Ru/CNF reduced
Ru/O-CNF reduced
Ru/N-CNF reduced
3.9
3.7
5.1
0.267
0.253
0.309
72.1
69.3
86.4
8.95
9.08
9.27
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3829 – 3834 3833

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[3] C. H. Hsu, H. M. Wu, P. L. Kuo, Chem. Commun. 2010, 46, 7628.
fered less sintering than Ru on O-CNF. Thus, although the Ru-
particle size was smaller on O-CNF for the reduced catalyst
before the reaction, after the reaction, the smallest metal-parti-
cle size corresponded to the catalyst on N-CNF.
[4] G. C.-K. Liu, J. R. Dahn, Appl. Catal. A 2008, 347, 43.
[5] S. Abate, R. Arrigo, M. E. Schuster, S. Perathoner, G. Centi, A. Villa, D. Su,
R. Schlçgl, Catal. Today 2010, 157, 280.
[6] K. Chizari, I. Janowska, M. Houllꢂ, I. Florea, O. Ersen, T. Romero, P. Bern-
hardt, M. J. Ledoux, C. Pham-Huu, Appl. Catal. A 2010, 380, 72.
[7] A. Villa, D. Wang, P. Spontoni, R. Arrigo, D. Su, L. Prati, Catal. Today
2010, 157, 89.
Conclusions
[8]
Monoliths that were coated with variously functionalized
carbon nanofibers, that is, un-doped CNFs, CNFs that had been
post-treated with H2O2, and CNFs that were doped with nitro-
gen during the growth process, have been used as supports
for Ru catalysts. After impregnating a ruthenium-nitrate precur-
sor onto the different supports by equilibrium adsorption, the
fresh and reduced catalyst were characterized by STEM, TPR,
and ex situ and in situ XPS. XPS characterization of the fresh
catalyst showed that the ruthenium-nitrate precursors had
become partially reduced (ca.18%) upon impregnation onto
N-doped CNF, whereas the extent of metal reduction upon im-
pregnation on un-doped or O-doped CNFs was significantly
lower. This support-mediated metal reduction could account
for the larger metal uptake on N-doped CNFs compared to the
other supports. Likewise, N-doped CNFs adsorbed 25% more
Ru metal than un-doped and O-doped CNFs. Moreover, XPS
characterization demonstrated that electron donation from the
support to the metal was larger for N-doped CNFs compared
to the un-doped and O-doped CNFs. Very small particle sizes
(mean size: 1 nm) and narrow particle-size distribution were
found for all of the supports, regardless of the type of func-
tionalization. After the reaction, Ru on O-CNF suffered more
sintering than Ru on CNF or N-CNF. Ru that was supported on
N-CNF induced significantly higher conversion in the decom-
position of NH3 than that on the other supports. The catalysts
were stable for (at least) 72 h reaction time at 723 K. Therefore,
Ru/N-CNF is an excellent candidate catalyst for H2 generation
from NH3 to feed fuel cells. Furthermore, its monolithic shape
endows it with robustness and portability.
E. Castillejos, R. Chico, R. Bacsa, S. Coco, P. Espinet, M. Perez-Cadenas, A.
Guerrero-Ruiz, I. Rodriguez-Ramos, P. Serp, Eur. J. Inorg. Chem. 2010,
5096.
[9] Y. Motoyama, Y. Lee, K. Tsuji, S. H. Yoon, I. Mochida, H. Nagashima,
ChemCatChem 2011, 3, 1578.
[10] H. J. Schulte, B. Graf, W. Xia, M. Muhler, ChemCatChem 2012, 4, 350.
[11] S. Armenise, M. Nebra, E. Garcꢁa-Bordejꢂ, A. Monzꢆn, Stud. Surf. Sci.
Catal. 2010, 175, 483.
[12] V. G. Ramu, A. Bordoloi, T. C. Nagaiah, W. Schuhmann, M. Muhler, C.
Cabrele, Appl. Catal. A 2012, 431–432, 88.
[13] F. Schꢇth, R. Palkovits, R. Schlçgl, D. S. Su, Energy Environ. Sci. 2012, 5,
6278.
[14] A. N. Rollinson, J. Jones, V. Dupont, M. V. Twigg, Energy Environ. Sci.
2011, 4, 1216.
[15] A. Klerke, C. H. Christensen, J. K. Norskov, T. Vegge, J. Mater. Chem.
2008, 18, 2304.
[16] C. H. Christensen, R. Z. Sorensen, T. Johannessen, U. J. Quaade, K. Hon-
kala, T. D. Elmoe, R. Kohler, J. K. Norskov, J. Mater. Chem. 2005, 15, 4106.
[17] E. Garcꢁa-Bordejꢂ, I. Kvande, D. Chen, M. Ronning, Adv. Mater. 2006, 18,
1589.
[18] L. Roldꢀn, S. Armenise, Y. Marco, E. Garcia-Bordeje, Phys. Chem. Chem.
Phys. 2012, 14, 3568.
[19] T. A. Nijhuis, A. E. W. Beers, T. Vergunst, I. Hoek, F. Kapteijn, J. A. Moulijn,
Catal. Rev. Sci. Eng. 2001, 43, 345.
[20] N. Jarrah, J. van Ommen, L. Lefferts, Catal. Today 2003, 79–80, 29.
[21] R. Chetty, W. Xia, S. Kundu, M. Bron, T. Reinecke, W. Schuhmann, M.
Muhler, Langmuir 2009, 25, 3853.
[22] R. K. Raman, A. K. Shukla, A. Gayen, M. S. Hegde, K. R. Priolkar, P. R.
Sarode, S. Emura, J. Power Sources 2006, 157, 45.
[23] For an interpretation of measurements in experimental catalysis, see:
P. B. Weisz, C. D. Prater in Advances in Catalysis, Vol. 6 (Ed.: W. G. Frank-
enburg), Academic Press, 1954, p. 143.
[24] S. Armenise, L. Roldꢀn, Y. Marco, A. Monzꢆn, E. Garcꢁa-Bordejꢂ, J. Phys.
Chem. C 2012, 116, 26385.
[25] R. Czerw, M. Terrones, J. C. Charlier, X. Blase, B. Foley, R. Kamalakaran, N.
Grobert, H. Terrones, D. Tekleab, P. M. Ajayan, W. Blau, M. Rꢇhle, D. L.
Carroll, Nano Lett. 2001, 1, 457.
[26] S. Armenise, E. Garcia-Bordeje, J. L. Valverde, E. Romeo, A. Monzon,
Phys. Chem. Chem. Phys. 2013, 15, 1463–9076.
[27] J. Zhu, A. Holmen, D. Chen, ChemCatChem 2013, 5, 378.
[28] I. Efremenko, M. Sheintuch, J. Catal. 2003, 214, 53.
[29] H. N. Aiyer, V. Vijayakrishnan, G. N. Subbanna, C. N. R. Rao, Surf. Sci.
1994, 313, 392.
Acknowledgements
Financial support from the European Commission is acknowl-
edged (FP7, Grant agreement no. 280658).
[30] P. Chen, L. M. Chew, A. Kostka, M. Muhler, W. Xia, Catal. Sci. Technol.
2013, 3, 1964.
Keywords: doping · heterogeneous catalysis · nanostructures ·
ruthenium · supported catalysts
[1] J. Chen, Z. H. Zhu, S. Wang, Q. Ma, V. Rudolph, G. Q. Lu, Chem. Eng. J.
2010, 156, 404.
Received: June 12, 2013
Revised: July 23, 2013
[2] F. R. Garcꢁa-Garcꢁa, J. ꢅlvarez-Rodrꢁguez, I. Rodrꢁguez-Ramos, A. Guer-
rero-Ruiz, Carbon 2010, 48, 267.
Published online on September 23, 2013
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3829 – 3834 3834