Synthesis of palladium nanoparticles over graphite oxide and carbon nanotubes by reduction in ethylene glycol and their catalytic performance on the chemoselective hydrogenation of para-chloronitrobenzene

Article abstract of DOI:10.1016/j.apcata.2015.11.048

Pd nanoparticles have been synthesized over carbon nanotubes (CNT) and graphite oxide (GO) by reduction with ethylene glycol and by conventional impregnation method. The catalysts were tested on the chemoselective hydrogenation of p-chloronitrobenzene and the effect of the synthesis method and surface chemistry on their catalytic performance was evaluated. The catalysts were characterized by N₂ adsorption/desorption isotherms at 77 K, TEM, powder X-ray diffraction, thermogravimetry, infrared and X-ray photoelectron spectroscopy and ICP-OES. It was observed that the synthesis of Pd nanoparticles employing ethylene glycol resulted in metallic palladium particles of smaller size compared to those prepared by the impregnation method and similar for both supports. The presence of oxygen groups on the support surface favored the activity and diminished the selectivity. It seems that ethylene glycol reacted with the surface groups of GO, this favoring the selectivity. The activity was higher over the CNT-based catalysts and both catalysts prepared by reduction in ethylene glycol were quite stable upon recycling.

Full text of DOI:10.1016/j.apcata.2015.11.048

Accepted Manuscript
Title: Synthesis of palladium nanoparticles over graphite
oxide and carbon nanotubes by reduction in ethylene glycol
and their catalytic performance on the chemoselective
hydrogenation of para-chloronitrobenzene.
Author: A.B. Dongil L. Pastor-P e´ rez J.L.G. Fierro N.
Escalona A. Sep u´ lveda-Escribano
PII:
S0926-860X(15)30313-6
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2015.11.048
APCATA 15711
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
6-10-2015
25-11-2015
27-11-2015
Please cite this article as: A.B.Dongil, L.Pastor-P e´ rez, J.L.G.Fierro, N.Escalona,
A.Sep u´ lveda-Escribano, Synthesis of palladium nanoparticles over graphite oxide and
carbon nanotubes by reduction in ethylene glycol and their catalytic performance on
the chemoselective hydrogenation of para-chloronitrobenzene., Applied Catalysis A,
General http://dx.doi.org/10.1016/j.apcata.2015.11.048
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Synthesis of palladium nanoparticles over graphite oxide and carbon nanotubes
by reduction in ethylene glycol and their catalytic performance on the
chemoselective hydrogenation of para-chloronitrobenzene.
*a
b
c
d,e
b
A.B. Dongil , L. Pastor-Pérez , J.L.G. Fierro , N. Escalona , A. Sepúlveda-Escribano
a
Universidad de Concepción, departamento de físicoquímica, laboratorio de catálisis
por metales, Edmundo Larenas 129, Concepción, Chile.
b
Laboratorio de Materiales Avanzados, Departamento de Química Inorgánica-Instituto
Universitario de Materiales de Alicante, Universidad de Alicante,
Apartado 99, E-03080 Alicante, Spain
c
Instituto de Catálisis y Petroleoquímica CSIC, Grupo de Energía y Química
Sostenible, C/Marie Curie2, Cantoblanco, Madrid, Spain
d
Departamento de Ingeniera Química y Bioprocesos, Pontificia Universidad Católica
de Chile, Avda. Vicuña Mackenna 4860, Macul, Santiago, Chile.
e
Facultad de Ciencias Químicas, Pontificia Universidad Católica de Chile.
(*) corresponding author: adongil@udec.cl

Graphical abstract
GO
Pd(acac) Ethylene glycol
Pd/GOEG
Pd/CNTEG
Cl
2
CNT
373 K, 2 h
Cl
2
0 bar H , 298 K
1
20
00
2
1
Pd/GOEG
NO
2
NH
2
Pd/CNTEG
80
60
40
20
0
Cycle 1
Cycle 2
Cycle 3
Highlights
The synthesis of palladium nanoparticles by reduction of the organometallic precursor
Pd (acac) with ethylene glycol and their deposition over carbon nanotubes (CNT) and




2
graphite oxide (GO) was studied.
Well-dispersed metallic palladium nanoparticles of similar average size were obtained
over Pd/CNTEG and Pd/GOEG, this suggesting that the morphology and surface
chemistry do not affect the size of the nanoparticles obtained by this methodology
The catalytic performance of Pd/CNTEG and Pd/GOEG was studied on the para-
chloronitrobenzene chemoselective hydrogenation and compared with analogous
catalysts prepared by traditional impregnation.
The activity of Pd/CNTEG and Pd/GOEG was higher than that obtained with the catalysts
prepared by impregnation and both catalysts could be reused on successive cycles.

ABSTRACT
Pd nanoparticles have been synthetized over carbon nanotubes (CNT) and graphite oxide
(GO) by reduction with ethylene glycol and by conventional impregnation method. The
catalysts were tested on the chemoselective hydrogenation of p-chloronitrobenzene and
the effect of the synthesis method and surface chemistry on their catalytic performance
was evaluated. The catalysts were characterized by N adsorption/desorption isotherms
2
at 77 K, TEM, powder X-ray diffraction, thermogravimetry, infrared and X-ray
photoelectron spectroscopy and ICP-OES. It was observed that the synthesis of Pd
nanoparticles employing ethylene glycol resulted in metallic palladium particles of
smaller size compared to those prepared by the impregnation method and similar for both
supports. The presence of oxygen groups on the support surface favored the activity and
diminished the selectivity. It seems that ethylene glycol reacted with the surface groups
of GO, this favoring the selectivity. The activity was higher over the CNT-based catalysts
and both catalysts prepared by reduction in ethylene glycol were quite stable upon
recycling.
Keywords: Carbon nanotubes; graphite oxide; palladium; ethylene glycol; para-
chloronitrobenzene; hydrogenation.
1
. INTRODUCTION
The interest in heterogeneous catalysts being palladium the active phase has increased
recently because of its lower price and better resistance to deactivation compared to
platinum, among other reasons [1-2]. However, the conventional methods to obtain
supported nanoparticles for catalytic applications do not offer good metal dispersion, and
alternative methods as the colloidal route have been developed. It consists on the dispersion

of the metal precursor into an appropriate solution to which the reducing agent and the
support are added under specific conditions. A surfactant or a polymer is generally necessary
to avoid aggregation of the nanoparticles [3]. However, in order to optimize the use of the
active phase these stabilizers must be removed, which may cause agglomeration of the
nanoparticles and, therefore it would be interesting to develop methodologies that avoid the
use of these capping agents. With that aim, the use of a reducing agent and a metal precursor
that can be employed under mild conditions to avoid the agglomeration of the nanoparticles
is desirable. Organometallic precursors allow for colloidal synthesis at lower temperatures
using alcohols as reducing agents, although aggregation cannot always be avoided [4].
Among these methods, the polyol process is based on the decomposition of the polyol
(commonly ethylene glycol) by treatment under a certain temperature in a solution
containing the metal precursor salt [5-7] to which the support can be also added [8]. In this
way, Pt nanoparticles supported on carbon nanotubes (CNT) have been synthesized [9-10].
It has been suggested that the polyol acts as solvent and reducing agent [11], though an
additional reducing agent is frequently needed [12-13].
The selective catalytic hydrogenation of aromatic haloamines has been extensively
studied to displace the environmentally unfriendly traditional synthetic routes. Although the
activation of the N=O bond is relatively easy, the main problem associated with this reaction
is the accumulation of intermediates and, depending on the metal, the cleavage of the
halogen. The accumulation of hydroxylamine may lead to rapid exothermic decomposition
and/or formation of condensation products. The most studied metals, i.e. Ni, Pt and Pd, offer
good activities for this hydrogenation, but to achieve a high yield of the desired product it
might be necessary to work at high temperatures, to add promoters or to employ bimetallic
catalysts [14-15]. Palladium is in general less selective than platinum for this reaction due to
parallel dehalogenation, and complete inhibition of the C-Cl scission is difficult [16-17]. In

this sense it has been reported that the use of a hydrophobic support as activated carbon may
inhibit the formation of aniline [17] and an alternative support as CNT can offer further
improvement of the catalytic performance [18]. Moreover, recent literature points to
graphene oxide as an ideal candidate to act as support for the synthesis Pd nanoparticles [19-
2
1]. This has been achieved by chemical [22], electrochemical [23] or microwave-assisted
methods [24]. Some of these methods may also reduce the parent graphene oxide in such a
way that the aromatic structure is recovered.
Therefore, we have explored the synthesis of palladium nanoparticles over graphite
oxide by the polyol method and compared it with the synthesis over carbon nanotubes. The
catalytic performance in the hydrogenation of p-chloronitrobenzene (p-ClNB) was tested
and compared with analogous catalysts prepared by conventional impregnation over GO,
CNT and previously oxidized CNT.
2
2
. EXPERIMENTAL
.1 Catalysts preparation
Graphite oxide (GO) was synthesized from graphite powder (99.999% stated purity, 200
mesh, Alfar Aesar) following a modification of the Brodie´s method [25]. Graphite was
added to a reaction flask containing fuming HNO (20 mL/g of support) which was
3
previously cooled to 273 K in an iced bath. Then, potassium chlorate (8 g/g of support) was
slowly added. The reaction was left to proceed for 24 h under stirring, and the final solid was
filtered, extensively washed with deionized water until neutral pH, dried under vacuum at
3
23 K overnight and kept in a desiccator. High purity commercial multi-wall carbon
nanotubes (CNT), supplied by Nanocyl (Nanocyl 3100, > 95% purity), were also employed.
A portion of the CNTs was oxidized with concentrated HNO for 24 h at 403 K (CNTox).
The catalysts were prepared employing the necessary amount of the precursor, palladium
3

acetylacetonate Pd(acac) (Sigma Aldrich, 99%), to obtain a 1 wt.% Pd over GO and CNT.
2
To prepare Pd/GOEG and Pd/CNTEG the support and the Pd precursor were dispersed in
ethylene glycol and heated at 373 K for 2 h, filtered off, washed extensively with ethanol
and dried under vacuum at 323 K for 16 h. Catalysts Pd/CNT, Pd/CNTox and Pd/G were
prepared by incipient impregnation in toluene, followed by drying at 373 K for 24 h and
heating under Ar (50 mL/min) at 673 K for 2 h, with a heating ramp of 5 ºC/min. The samples
prepared by impregnation were reduced under H
2
(50 mL/min) at 673 K for 2 h prior to the
reaction.
2
.2 Characterization
The specific surface areas (SBET) of the catalysts were determined from the N
2
adsorption isotherms at 77 K, which were obtained with a Micromeritics ASAP 2010
equipment. Prior to the measurements, the samples were degassed at 473 K for 2 h.
TEM images were obtained with a JEOL electron microscope (model JEM-2010)
working at 200 kV. It was equipped with an INCA Energy TEM 100 analytical system and
a SIS MegaView II camera. Samples for analysis were suspended in ethanol and placed on
copper grids with a holey-carbon film support. The average particle size was obtained from
the measurement of 200 nanoparticles.
X-ray powder diffraction patterns were recorded on a Rigaku diffractometer
equipped with a nickel-filtered CuK1 radiation (= 1.5418˚A), using a 2º/min scanning
rate.
The actual Pd loading on the different catalysts was determined by ICP-OES in a Perkin–
Elmer device (Optimal 3000). To this end, the metal was extracted from the catalysts by
digestion in HNO
3
/H
2
2
O (4:1) for 30 min, in a microwave oven at 473 K.

Thermogravimetric analyses were performed with a Mettler Toledo Thermogravimetric
TGA/SDTA 851. The samples were treated under flowing He for 2 h and then heated at a 10
K/min rate up to 1123 K.
X-ray photoelectron spectra (XPS) of the catalysts were obtained on a
MicrotechMultilb 3000 spectrometer, equipped with a hemispherical electron analyzer and
MgKα (hν = 1253.6 eV) photon source. Prior to the analysis, the catalysts were in
situ reduced under the same conditions employed before the catalytic tests. The surface O/C
and Pd/C atomic ratios were estimated from the integrated intensities of Pd 3d5/2, C 1s, and
O 1s lines after background subtraction and correction by the atomic sensitivity factors. The
spectra were fitted to a combination of Gaussian–Lorentzian lines of variable proportions.
Infrared spectra were collected on a NicoletNexus FT-IR in the middle range
−
1
(
4000 – 400 cm ), and recorded by a DTGS detector from 128 scans and with a resolution
-1
of 4 cm . GO and its derivatives were mixed with pre-dried potassium bromide to a final
concentration of approximately 1% (w/w).
2
.3 Hydrogenation tests
Catalytic tests were performed in a stainless steel Parr-type batch reactor,
equipped with a glass sleeve and a magnetic stirrer set at 1000 rpm. The hydrogenations
were performed with 50 mg of catalyst, 320 mg of p-chloronitrobenzene and 50 mL of
ethanol were stirred at 20 bar and 298 K. For the recycling studies the amounts were
recalculated for each cycle. The concentrations of reactants and products were monitored by
gas chromatography mass spectrometry using a GC–MS instrument (Shimadzu GCMS-
QP5050) with chiral -Dex 225, a 30-m column (Supelco), and He as the carrier gas.
Additional experiments were performed to verify whether the reaction was heterogeneously

catalyzed: the reaction was stopped at 50% conversion and the liquid filtrate was submitted
to the reaction conditions.
3
3
. Results
.1 Characterization
The estimated BET surface areas are reported in Table 1. It can be seen that the
surface area of the supports slightly decreased after the incorporation of Pd. On the other
2
-1
hand, the BET surface area of Pd/G was much higher, 670 m ·g as a consequence of the
expansion of the GO layers during the thermal exfoliation treatment applied in the synthesis
of this catalyst. The adsorption isotherms of samples GO and Pd/GOEG could not be
measured because the pretreatment changed the structure of the support, as it was verified
by XRD analysis performed after the adsorption experiment.
Fig. 1 shows representative TEM images of the catalysts. It can be observed that
Pd/GOEG, Fig. 1a, presented transparent areas corresponding to exfoliated graphene sheets,
along with zones where the sheets seemed to be thicker. However, sample Pd/G, Fig. 1b,
was less homogeneous and, although it also displayed some exfoliated areas, a high
proportion of the images showed dark regions that resembled amorphous carbon. From these
images it can be concluded that palladium nanoparticles densely covered the surface in Pd/G
and Pd/GOEG catalysts Pd nanoparticles on Pd/GOEG were mostly smaller than 2 nm, and
they could not be properly measured, although some particles of around 8-10 nm could also
be observed. On the other hand, sample Pd/G displayed a heterogeneous particle size
distribution with an estimated average size of 4.1 nm. For the CNT-based catalysts different
nanoparticle sizes distribution was also found, and the average particle size, reported in
Table 1, followed the trend Pd/CNTEG < Pd/CNTox < Pd/CNT. On catalyst Pd/CNT, Fig.
1
d, the nanoparticles mainly formed large agglomerates, the estimated average particle size
being 9.4 nm. The surface of catalysts Pd/CNTEG and Pd/CNTox was more homogeneously

covered with Pd nanoparticles, the mean metal particle size for Pd/CNTox was 2.7 nm. The
particle sizes obtained for Pd/GOEG and Pd/CNTEG were the smallest among the prepared
catalysts, in the range of previously reported values in literature for Pd nanoparticles (1.6
wt.%) supported on graphene and graphene oxide [14], and smaller than the values reported
for non-supported palladium nanoparticles stabilized by an organic moiety [26].
The XRD patterns of GO and GO-based catalysts are shown in Fig. 2. The position
and wideness of the diffraction peaks provide an estimation of the crystallinity an exfoliation
of these laminar materials. For the sample GO, the intense and sharp peak of the (001)
reflection centered at 2θ = 15.4° corresponding to an interlayer distance of 0.57 nm, and the
disappearance of the peak at 26º due to the reflection of the (002) of graphite confirm a high
yield in the GO synthesis [27]. The synthesis of palladium nanoparticles over GO afforded
new materials with different morphology, as their XRD profile showed. The sample
Pd/GOEG displayed a new diffraction peak at 2 = 11.2º and a wider peak centered at 2
22°. The diffraction peak at lower angle indicates that the material was partially exfoliated,
the new interlayer distance being 0.79 nm. The exfoliated layers could be filled with
palladium nanoparticles or residues of ethylene glycol. The broader peak could be an
indication of the presence of graphene layers with different interlayer distance, which could
be turbostratic graphite as some TEM images suggested [28-29]. On the other hand, for
sample Pd/G the characteristic peak of GO disappeared and it displayed the broad peak at
2
 22°, which is ascribed to turbostratic graphite. The XRD profiles of the GO-based
catalysts corroborate the TEM observations, which suggested that Pd/GOEG was more
exfoliated than Pd/G. The XRD of CNT-based catalyst (not shown) displayed the
characteristic peak of graphite at 2 = 26º, wider than the one observed for graphite as a
result of the rolling structure. Finally, none of the diffractograms showed the characteristics

peaks of palladium or its oxide, probably because of the low percentage of metal on the
samples and/or to the small particle size.
The thermogravimetric analyses (Fig.3) were performed to study the thermal
stability of the samples and the presence of organic residues. The thermogram of GO showed
a main weight loss with maximum rate at 534 K, which corresponds to the release of H
CO and CO during the reductive exfoliation [26]. The weight loss of this stage was 25%.
At higher temperatures, the weight loss followed a slower rate and a total mass loss of about
6% was reached. The weight loss and the temperature at which it was produced were similar
2
O,
2
4
to others reported for GO prepared by this method [26][30]. The thermogravimetric profile
of Pd/GOEG was shifted to lower degradation temperature, 489 K, and the weight loss at this
stage was similar to that of GO. The presence of ethylene glycol molecules inside the layers
may be responsible of the lower thermal stability, as previously reported for other solvents
[
31], and it would be in agreement with the XRD results that showed an increase in the
interlayer distance. Furthermore, modifications on the type of solvent-GO interaction during
the thermal analysis could be responsible for the degradation of the solvent molecules at
higher temperature (>523 K), and it would explain the increase of the weight loss at this
stage compared to GO. The thermogram of Pd/G showed much higher thermal stability with
barely any mass loss until 655 K, which was followed by a smoother decay up to 29 % of
total weight loss. Nonetheless, the high total weight loss observed for Pd/G compared to
Pd/CNT suggested that there is a large proportion of carbon on defects on this sample, as
observed by TEM and XRD. The thermal analyses of the CNTs and their corresponding
catalysts are shown in Fig. 3b. It can be observed that the Pd/CNT catalyst showed barely
any mass loss along the whole range of temperature. On the contrary, catalysts Pd/CNTox
and Pd/CNTEG displayed a total mass loss of 10 and 12 % respectively, which can be ascribed

to the decomposition of the oxygen groups that are present after the reduction treatment and
the degradation of adsorbed ethylene glycol, respectively.
Infrared spectra of GO and the GO-based catalysts are shown in the supplementary
information, Fig. S1. GO displayed the characteristic bands of C-OH groups at 3627, 3490,
-1
-1
1
620 and 1383 cm [28][32-33], C=O bond of carbonyl and/or carboxyl at 1713 cm and
-1
the band at 1063 cm due to deformation of the C-O bond of epoxy functionalities. The
infrared spectra of the sample Pd/GOEG showed a decrease of the bands at 3627, 3490, 1713
-1
-1
and 1620 cm , while new bands appeared at 2930 and 2853 cm assigned to asymmetric
group and to the vibration of C sp -C sp2
2
and symmetric stretching vibration of CH
2
-1
2
2
respectively, and a small band at 1562 cm assigned to the vibration of C sp -Csp bonds
characteristic of the skeletal vibrations of graphitic domains. These changes point to the
contribution of ethylene glycol molecules on the Pd/GOEG catalyst by reaction with the
hydroxyl groups on the parent GO, in agreement with the XRD profiles and the thermal
analysis, and to the partial recovery of the aromaticity. On the other hand, sample Pd/G
-1
2
2
displayed the band at 1578 cm ascribed to the vibration of C sp -C sp bonds, which
confirms the reduction of this material. In addition, qualitative changes could be observed
for sample Pd/G in the low frequency range, which also suggest modifications on the surface
chemistry.
Analyses of the XPS spectra of the catalysts were performed. The detailed values
for the C 1s and O 1s regions and their proportions, which are included in the supporting
information, are in quite good agreement with literature [27-28] [34]. The C 1s and O1s
region of GO and the GO-based catalysts are shown in Fig. 4a and 4c. As expected, the
position of the maximum of the C 1s spectra for GO appears at higher B.E, 286.0 eV, than
2
the corresponding to graphite which is ascribed to sp carbon atoms, i.e. 284.6 eV, and the
*
characteristic π → π peak of polyaromatic structures is absent. This confirms that the main

3
contribution is associated to sp carbon atoms. The C 1s spectra of Pd/GOEG slightly changed
2
compared to that for GO, but the proportion of the contribution ascribed to sp carbon is
higher than for GO, in agreement with the IR spectra. On the other hand, for catalyst Pd/G
3
the contribution of sp carbon has clearly decreased, this indicating that the support in Pd/G
recovered the aromaticity. As far as the O 1s region is concerned, the maximum for GO was
at 532.0 eV and it displayed the contribution of three components assigned to oxygen in
carbonyl groups (531.2 eV), oxygen atoms in C-O bonds like hydroxyl and ether groups
(532.6 eV) and oxygen atoms in acidic carboxyl groups (533.5 eV). The maximum of the O
1
s peaks was shifted to higher binding energies for the Pd/G and Pd/GOEG catalysts
compared to GO, 532.7-533.6 eV, this clearly indicating a change in the chemical state of
the oxygen species. For the catalyst Pd/GOEG, bearing in mind the thermal analysis and the
XRD profile that suggested the presence of intercalated solvent, it seems plausible that the
hydroxyl groups of the support had reacted with the hydroxyl groups of the solvent
molecules via hydrogen bond or condensation, as observed when other molecules were
intercalated within the layers, and this would explain the more oxidized state of oxygen on
these samples [27] [35]. The C 1s region of CNT displayed the maximum at 284.6 eV
2
ascribed to sp carbon, and the typical contributions of carbon materials. No significant
changes were observed for the C 1s region of the CNT-based catalysts. The O 1s region of
CNT and CNTox displayed the contributions ascribed to oxygen in carbonyl groups (531.6-
5
31.7eV), oxygen atoms in C-O bonds (533.1-533.2 eV) and oxygen atoms in acidic
carboxyl groups (534.5-534.7 eV). It was observed that the contribtion ascribed to carboxylic
acid increased upon the oxidation treatment, and then it decreases for the Pd/CNTox catalyst.
For catalysts Pd/CNTEG and Pd/CNT no significant differences were observed in the relative
contributions of each component. The relative amounts of C, O and Pd were calculated from
the corresponding peak areas divided by the sensitivity factors, and the most relevant data

are shown in Table 1. It was observed that the O/C ratio for sample Pd/GOEG (Table 1)
increased compared to GO, which confirms the presence of remaining solvent molecules in
this material, as other characterization results showed. Opposite to this, the O/C ratio
decreased for Pd/G compared to GO, in agreement with the infrared spectrum and the
thermal analysis. The O/C ratio for Pd/CNT was lower compared to Pd/CNTox and
Pd/CNTEG, which confirms that these latter catalysts presented oxygen groups and ethylene
glycol, respectively, on the surface as the thermal analyses suggested. Moreover, all the
catalysts displayed a new peak around 335 eV corresponding to the Pd 3d5/2 level
(Supporting Information, Fig. S2), and the BEs values are shown in Table 1. It is important
to remark that the deconvolution of this region displayed only one contribution ascribed to
0
Pd for all the catalysts except Pd/G, despite the mild conditions employed for the synthesis
of the nanoparticles on Pd/GOEG and Pd/CNTEG. For the Pd/G catalyst two components have
0
+2
to be considered at 335.4 and 337.7 eV, which are ascribed to Pd and Pd respectively [36-
3
7]. The contribution of each component is also included in brackets in Table 1, and it
indicates that palladium was reduced to a larger extent. The Pd/C ratio was also estimated
and the values are shown in Table 1. Most of the samples displayed a Pd/C ratio in the range
0
.001-0.002 except for Pd/CNTox, with a ratio of 0.007. Considering the superficial
character of the XPS technique and the ICP analyses in Table 1, the higher surface Pd/C
ratio for this sample compared to Pd/CNT would be in line with the smaller palladium
particle size obtained in Pd/CNTox, as observed by TEM. On the other hand, the similar
Pd/C ratio for catalyst Pd/CNTEG compared to Pd/CNT, despite the higher dispersion of Pd
nanoparticles on Pd/CNTEG, can be explained by its lower content on Pd, and similar
conclusions can be made for the Pd/C ratio of catalysts Pd/G and Pd/GOEG. However, as the
TEM images suggested, it is plausible to consider that a proportion of the palladium oxide
is encapsulated in the carbon structure for the Pd/G catalyst, and this could also explain the

similar Pd/C ratio compared to Pd/CNT and the lower reduction degree of this sample, as
less particles would be accessible to the hydrogen reduction flow.
3
.2 Reaction
The catalysts were tested in the liquid phase hydrogenation of p-
chloronitrobenzene. The evolution of the conversion vs time is shown in Fig. 5a and the most
relevant results are shown in Table 2. All the catalysts were active, and the activity calculated
as mol of p-ClNB converted per mol of Pd (obtained from ICP-OES) and reaction time at
5
0% conversion in the first cycle followed the trend Pd/CNTEG > Pd/CNTox > Pd/GOEG >
Pd/CNT >> Pd/G. The activity achieved with Pd/GOEG and Pd/CNT was in the range of the
values reported for palladium supported over activated carbon, while those obtained with
Pd/CNTox and Pd/CNTEG were nearly one order of magnitude higher [17]. As far as the
selectivity is concerned, p-ClAN was the main product of the reaction for all the catalysts
and, under the experimental conditions employed, no intermediates were observed. The
selectivity to p-ClAN was 100% with all the catalysts except for Pd/CNTox, which afforded
a selectivity of 78% when 100% conversion was reached, the secondary product being
exclusively aniline (AN). In addition, the stability of Pd/CNTEG, and Pd/GOEG was studied
by re-using the catalysts in three cycles, and the evolution of the conversion with time on
stream is shown in Fig. 5b and 5c, respectively. The activity decreased in the second cycle,
this decrease being more pronounced for Pd/GOEG (70 %) than for Pd/CNTEG (44%).
However, the activity was stable in the third cycle for both catalysts, and the selectivity to
p-ClAN remained 100% upon recycling within the conversion range. Additional
experiments, described in the Experimental Section, were performed with catalysts
Pd/CNTEG and Pd/GOEG to assess that the reaction was heterogeneously catalysed. The

absent of reaction with the liquid filtrate confirmed that the reaction was heterogeneously
catalysed despite the possible leaching of palladium nanoparticles.
The hydrogenation paths of p-ClNB are summarized on Scheme 1. The catalytic reduction
of p-ClNB may proceed through the hydrogenation of the nitro group to give the desired
product p-ClAN. Then, p-ClAN may suffer dehalogenation to produce AN. The
dehalogenation may also take place on the p-ClNB molecule producing nitrobenzene (NB)
which could then be hydrogenated to AN [15]. It has been suggested that the dehalogenation
reaction occurs via electrophilic attack of the activated hydrogen on the absorbed aromatic
halides [38].
The obtained results reflect the influence of the synthesis method and surface chemistry of
the support on the activity and selectivity in this reaction. On one hand, the activity achieved
with Pd/CNTox was higher than that obtained with Pd/CNT, as well as the calculated
intrinsic activity considering the palladium dispersion estimated by TEM. It has been
proposed in the literature that the hydroxyl groups of the support interact with the nitro
groups of the reactant, this activating the N=O bond [39]. Alternatively, it has been proposed
that the electron rich acid groups of the surface may be the adsorption sites for the aromatic
ring, this increasing the activity [40].
Concerning the selectivity it is interesting to remark the high value obtained with Pd/CNT,
as Pd has been generally considered as a non-selective metal for p-ClNB hydrogenation due
to the formation of AN. For that reason, it has been employed in combination with other
elements, either by forming bimetallic catalyst (Pd/Au and Pd/Ni supported on alumina [41][
1
5]) or by the use of additives in the reaction medium [42]. Apparently, the addition of a
second metal activates the N=O bond and hinders the adsorption of the p-ClAN avoiding its
subsequence dehalogenation. However, in the case of carbon nanotubes, the  cloud of the
support may repel the Cl atom, this favoring an increase of selectivity [43], similarly to what

has been observed in the hydrogenation of cinnamaldehyde [44]. Moreover, the oxygen
groups on the catalyst surface, which are present on Pd/CNTox even after the reduction
treatment, may favor the adsorption of the formed p-ClAN on the support and this would
enable its consecutive dehalogenation [39]. Therefore, the presence of oxygen groups on the
surface seems to favor the activity while diminishes the selectivity.
It has been observed that Pd nanoparticles in Pd/CNTEG and Pd/GOEG displayed similar
particle size, this suggesting that the morphology and surface chemistry do not affect the size
of the nanoparticles obtained by this methodology. Nonetheless the activity was higher when
the catalyst Pd/CNTEG was employed, which can tentatively be related to its higher surface
area. Unfortunately, the specific surface area of Pd/GOEG could not be properly measured,
as explained in Section 3.1. These structural changes upon heating can be explained
considering that the exfoliated structure is preserved due to the intercalated ethylene glycol,
which seem to be decomposed already at the pretreatment temperature of the adsorption
isotherm analysis. Furthermore, both catalysts displayed high selectivity to p-ClAN. The
high selectivity obtained with Pd/GOEG, along with the characterization results, confirms the
proposed chemical interaction between the surface OH and the ethylene glycol molecules.
This would avoid the interaction between the surface oxygen groups which, according to the
characterization results, are present on the Pd/GOEG catalyst, and the reactant molecules and,
thus, hinder the formation of aniline, the undesired product. In addition, the Pd/CNTEG and
Pd/GOEG catalysts proved to have good recyclability after the first cycle. The loss of activity
observed between the first and the second cycle might be due to the leaching of adsorbed Pd
nanoparticles as the Pd/C ratio of the fresh and the used catalysts obtained by XPS showed.
The conversion profiles in the second and third cycle seem to display an induction period
for both Pd/CNTEG and Pd/GOEG, which might be due to the progressive reduction of the
palladium oxide layer under the reaction conditions [19]. The similar O/C ratio in the used

Pd/GOEG catalyst and in the fresh one seems to confirm the above mentioned chemical
interaction between ethylene glycol and the oxygen groups on the surface of GO which
avoids the desorption of the ethylene glycol, unlike to what is observed for the used
Pd/CNTEG catalyst, for which the O/C ratio is lower than in the fresh catalyst.
Finally, the activity achieved with the Pd/G catalyst was much lower than that obtained with
Pd/CNT, despite the higher surface area on the former. For this sample the characterization
studies suggested that Pd was not fully reduced, and that a given amount of the Pd
nanoparticles may be encapsulated inside the carbon structure and, therefore, they might not
be accessible to reactants. In the literature, the synthesis of metal nanoparticles supported on
graphene oxide by impregnation followed by heating under air or nitrogen flow has been
reported for Pd, Pt, Ru, Au or Ag among others metals. Those catalysts were active, however
no comparison with other systems was made and the metal loading was higher than that
employed in the present work, i.e. 10-25 wt.% [45-46]. Alternatively, the impregnation could
have been performed before the reduction of GO but, in that case, the inert surface may also
lead to low dispersion.
4
. Conclusions
We have synthetized Pd nanoparticles over CNT and GO employing ethylene glycol
as reducing agent, and compared their catalytic performance in the chemoselective
hydrogenation of p-ClNB with analogous catalysts prepared by traditional impregnation and
over previously oxidized CNT. The characterization studies showed that the polyol method
was effective for obtaining metal palladium nanoparticles over CNT and GO, and the
catalysts Pd/CNTEG and Pd/GOEG displayed similar Pd particle size which was smaller than
those obtained by the traditional impregnation method. The characterization performed

seems to indicate that under the synthesis conditions employed, ethylene glycol reacted with
the oxygen groups of the parent GO surface and, thus, Pd/GOEG holds intercalated solvent.
Also, according to the characterization studies, a given amount of the Pd precursor and/or
the Pd oxide was encapsulated inside the structure of the reduced graphene oxide in the
catalyst prepared by impregnation over GO, Pd/G. This may hinder the accessibility of H
2
making more difficult the reduction of palladium oxide.
The catalysts were tested in the chemoselective hydrogenation of p-ClNB. It was
observed that the activity was higher over Pd/CNTox catalyst than over Pd/CNT, what can
be ascribed to the presence of oxygen groups on the Pd/CNTox surface. However the
selectivity to the desired halo-aniline, p-ClAN, was lower over Pd/CNTox, and this has been
explained also by the presence of oxygen groups on the surface. On the other hand, it was
found that Pd/CNTEG was more active than Pd/GOEG, which can be due the different surface
area. The passivation of the oxygen groups on the Pd/GOEG surface by molecules of ethylene
glycol seem to favor the selectivity of this catalyst. The traditional impregnation of the
metallic precursor over GO followed by hydrogen reduction under heating hinders the
reduction of palladium oxide and promotes the encapsulation of palladium oxide and/or
palladium precursor, this leading to a less active catalyst. Finally the stability of both
Pd/CNTEG and Pd/GOEG was proved after an initial loss of activity due to the partial leaching
of Pd nanoparticles.
In summary, the oxygen groups on the surface of the support were found to be
detrimental for the selectivity. However, as observed by TEM, the dispersion of the metal
nanoparticles was improved. In order to increase the dispersion of palladium without
performing the oxidation of the surface the synthesis of nanoparticles can be performed by
reduction with ethylene glycol, this avoiding the previous oxidation step.

Acknowledgements
Financial support for the present study was received from CONICYT Chile, project 3130483
(Postdoctoral), Generalitat Valenciana, Spain (PROMETEOII/2014/004) is also gratefully
acknowledged.
References
[
1] S.I. El-Hout, S.M. El-Sheikh, H.M.A. Hassan, F.A. Harraz, I.A. Ibrahim, E.A. El-
Sharkawy, Appl.Catal. A: General 503 (2015) 176–185.
[
[
[
[
[
[
2] N. Marín-Astorga, G. Alvez-Manoli, P. Reyes, J. Mol. Cat. A: Chem. 226 (2005) 81–88.
3] M.L. Toebes, J.A. van Dillen, K. P. de Jong, J. Mol. Cat. A: Chem. 173 (2001) 75-98.
4] J. Athilakshmi, S. Ramanathan, D.K. Chand, Tetrahed. Lett. 49 (2008) 5286–5288.
5] L.K. Kurihara, G.M. Chow, P.E. Schoen, NanoStruct. Mater. 5 (1995) 607-613.
6] F.A. Harraz S.E. El-Hout, H.M. Killa, I.A. Ibrahim, J. Catal. 286 (2012) 184–192.
7] I.S. Armadi, Z.L.Wang, T.C. Green, A. Henglein, M.A.E. Sayed, Science 272 (1996)
1
924-1925.
[
[
[
8] A. Miyazaki, I. Balint, K.I. Aika, Y. Nakano, J. Catal. 203 (2001) 364.
9] V. Lordi, N. Yao, J. Wei, Chem. Mater. 13 (2001) 733-737.
10] W.Z. Li, C.H. Liang, J.S. Qiu, W.J. Zhou, H.M. Han, Z.B. Wei, G.Q. Sun, Q. Xin,
Carbon 40 (2002) 791-794.
[
[
[
[
11] L.K. Kurihara, G.M. Chow, P.E. Schoen, NanoStruct. Mater. 5 (1995) 607-613.
12] C.C Yeh, D.H. Chen, Appl. Catal. B: Environ. 150– 151 (2014) 298– 304.
13] H.T. Zhu, C.Y.Zhang, Y.S. Yin, J. Cryst. Growth 270 (2004) 722–728.
14] H.U. Blaser, A. Schnyder, H. Steiner, F. Rössler, P. Baumeister in Handbook of
Heterogeneous Catalysis (Wiley, 2008), vol. 8 pp. 3284-3307
15] F. Cárdenas-Lizana, S. Gómez-Quero, C. Amorim, M.A. Keane, Appl. Catal. A: General
73 (2014) 41–50.
16] H. Liu, M.Liang, C. Xiao, N. Zheng, X. Feng, Y. Liu, J. Xie, Y.Wang J. Mol. Catal. A:
General 308 (2009) 79-86.
17] V. Krathy, M. Kralik, M. Mecarova, M. Stolcova, L. Zalibera, M. Hronec, Appl. Catal.
A, 235 (2002) 225-231-
18] A.B. Dongil, C. Rivera-Cárcamo, L. Pastor-Pérez, A. Sepúlveda-Escribano, P.Reyes,
Catal. Today, 249 (2015) 72-78.
19] K.H.Lee, S.W.Han, K.Y. Kwon, J. B. Park, J. Coll. Interf. Sci 403 (2013) 127–133.
[
4
[
[
[
[

[
[
20] R. Awasthi, R.N. Singh, Carbon 51 (2013) 282 –289.
21] X. Chen, G.Wu, J.Chen, X. Chen, Z. Xie, X. Wang, J. Am. Chem. Soc. 133 (2011) 3693–
3
695
22] Á. Mastalir, Z. Király, Á. Patzkó, I. Dékány, P. L’Argentiere, Carbon 46 (2008) 1631-
637.
23] Y. Wang, Y. Zhao, J.Yin, M. Liu, Q. Dong, Y. Su, Int. J. Hydrogen Energy 39 (2014)
325-1335.
24] A.R. Siamaki, Abd El Rahman S. Khder, V.Abdelsayed, M. Samy El-Shall, B. Frank
Gupton J. Catal. 279 (2011) 1–11.
25] A.B.Dongil, B.Bachiller-Baeza, A. Guerrero-Ruiz, I.Rodríguez-Ramos, Catal. Comm.
6 (2012) 149-154.
26] S.Jansat, M. Gómez , K. Philippot , G. Muller , E.Guiu ,C. Claver, S.Castillón,
B.Chaudret, J. Am. Chem. Soc., 126 (2004) 1592–1593.
[
1
[
1
[
[
2
[
[
(
[
[
(
[
27] A.B. Dongil, B. Bachiller-Baeza, A. Guerrero-Ruiz, I. Rodríguez-Ramos, J. Catal. 282
2011) 299-309.
28] Z.Q. Li, C.J. Lu, Z.P. Xia, Y. Zhou, Z. Luo, Carbon 45 (2007) 1686-1695.
29] A.B. Bourlinos, D. Gournis, D. Petridis, T. Szabo, A. Szeri, I. Dekany, Langmuir, 19
2003) 6050-6055.
30] F. Barroso-Bujans, S. Cerveny, R. Verdejo, J.J. del Val, J.M. Alberdi, A. Alegría, J.
Colmenero, Carbon 48 (2010) 1079–1087.
[
(
[
[
31] F. Barroso-Bujans, J.L. G. Fierro, A. Alegría, J.Colmenero, Thermochim. Acta 526
2011) 65– 71.
32] M. Mermoux, Y. Chabre, A. Rousseau, Carbon 29 (1991) 469-474.
33] T. Szabó, O. Berkesi, P. Forgó, K. Josepovits, Y. Sanakis, D. Petridis, and I. Dékény,
Chem. Mater. 18 (2006) 2740.
34] N.I.Kovtyukhova, P.J. Olliver, B.R. Martin, T.E. Mallouk, S.A. Chizhik, E.V. Buzaneva,
A.D. Gorchinskiy, Chem. Mater. 11 (1998) 771-778
[
[
[
[
(
[
35] Y.Matsuo, Y.Sakai, T. Fukutsuka, Y.Sugie, Carbon 47 (2009) 804–811.
36] J.C Hierso, C.Satto, R. Feurer, P. Kalck, Chem Mater 8 (1996) 2481-2485.
37] W.Sun, X.Lu, Y. Tong, Z. Zhang, J.Lei, G.Nie, C.Wang, Int. J. Hydrogen Energy 39
2014) 9080–9086.
38] F. Notheisz, M. Bartok, in Fine Chemicals Through Heterogeneous Catalysis (Wiley
2
001) New York, 415-426.

[
39] A. Nieto-Marquez, S. Gil, A. Romero, J. L.Valverde, S. Gómez-Quero, M.A. Keane,
Appl. Catal. A: General 363 (2009) 188–198.
[
[
40] M. V.Rahaman, M. A. Vannice, J. Catal. 127 (1991) 267-275.
41] F. Cardenas-Lizana, S. Gomez-Quero, A. Hugon, L. Delannoy, C. Louis, M. A. Keane J.
Catal. 262 (2009) 235–243.
42] Q.Xu, X.M. Liu, J.R. Chen, R.X. Li, X.J. Li, J. Mol. Catal. A: Chemical 260 (2006)
99–305.
43] N. Mahata, A.F. Cunha, J.J.M. Órfão, J.L. Figueiredo, Catal. Comm 10 (2009) 1203-
206.
44] J.P. Tessonnier, L. Pesant, G. Ehret, M. J. Ledoux, C. P.Huu, Appl. Catal. A: General
88 (2005) 203–210.
45] K.Gotoh, K. Kawabata, E. Fujii, K. Morishige, T. Kinumoto, Y. Miyazaki, H. Ishida,
Carbon 47 (2009) 2112-2142.
46] K. Gotoh, T.Kinumoto, E. Fujii, A. Yamamoto, H. Hashimoto, T. Ohkubo, A. Itadani, Y.
Kuroda, Hiroyuki Ishida, Carbon 49 (2011) 1118-1125.
[
2
[
1
[
2
[
[

Figure 1. TEM micrographs of a) Pd/GOEG ; b) Pd/G; c) Pd/CNTEG; d) Pd/CNT;
e) Pd/CNTox.
a)
b)
c)
d)
f)
e)

Figure 2. XRD patterns of the GO-based catalysts.
Graphite
GO
Pd/GOEG
Pd/G
0
10
20
30
40
50
60
2
(degree)

Figure 3. Thermal analyses of the a) GO ( ), Pd/GOEG ( ), Pd/G ( ) and b) Pd/CNT
(
), Pd/CNTox ( ) and Pd/CNTEG
(
).
100
100
b)
a)
90
80
70
60
50
40
90
80
70
60
50
40
373 473 573 673 773 873 973 1073
373 473 573 673 773 873 973 1073
Temperature (K)
Temperature (K)

Figure 4. XPS spectra of the C 1s region a) GO and GO-based catalysts; b) CNT,
CNTox and CNT-based catalysts and the O 1s region c) GO and GO-based catalysts, d)
CNT, CNTox and CNT-based catalysts.
a)
b)
GO
Pd/GOEG
CNT
CNTox
Pd/G
Pd/CNTEG
Pd/CNT
Pd/CNTox
2
94 292 290 288 286 284 282 280
294 292 290 288 286 284 282 280
B.E. (eV)
B.E. (eV)
d)
c)
CNT
GO
CNTox
Pd/CNTEG
Pd/CNT
Pd/GOEG
Pd/G
Pd/CNTox
540
538
536
534
532
530
528
540
538
536
534
532
530
528
B.E. (eV)
B.E. (eV)

Figure 5. Conversion of p-cloronitrobenzene vs time a) First cycle; b) Pd/CNTEG and c)
Pd/GOEG on successive cycles.
100
a)
80
60
40
20
0
Pd/CNTox
Pd/CNTEG
Pd/GOEG
Pd/CNT
Pd/G
0
50
100
150
200
250
300
Time (min)
100
b)
80
60
40
20
0
Cycle 1
Cycle 2
Cycle 3
0
20
40
60
80
100
120
140
Time (min)

100
c)
80
60
40
20
0
Cycle 1
Cycle 2
Cycle 3
0
50
100
150
200
250
300
Time (min)

Scheme 1. Reaction paths for the para-chloronitrobenzene hydrogenation.
Cl
Cl
NH2
+
N
NH2
-
O
O
+
N
-
O
O

Table 1. Physico-chemical properties of supports and catalysts.
S
BET
-1
dp TEM
(nm)
Pd 3d5/2
Sample
O/C
Pd/C Pd (%) ICP
2
(
m g )
eV
GO
-
-
-
-
0.355
0.011
0.088
-
-
-
-
-
-
CNT
300
319
-
-
-
CNTox
-
Pd/GOEG
Pd/GOEG- used
Pd/G
< 2
-
335.1
0
.474
0.002
0.62
-
-
3
35.2
0.455
0.084
0.001
0.002
3
3
35.4 (73)
670
288
-
4.1
< 2
-
0.90
0.51
-
37.1 (27)
335.2
Pd/CNTEG
Pd/CNTEG- used
Pd/CNT
Pd/CNTox
0.096
0.029
0.022
0.071
0.002
0.001
0.001
0.007
335.2
335.9
335.7
285
289
9.4
2.7
1.02
0.91