Influence of Carbon Supports on Palladium Nanoparticle Activity toward Hydrodeoxygenation and Aerobic Oxidation in Biomass Transformations

Article abstract of DOI:10.1002/ejic.201900190

Three palladium catalysts at similar loadings supported on few-layers graphene (FLG), carbon nanotubes (CNT) and carbon nanofibers (CNF) have been prepared by wet impregnation of palladium nitrate with the purpose of determine the influence of the support on Pd catalytic activity. The supports and catalysts have been characterized by chemical analysis, Raman spectroscopy, XRD, electron microscopy and XPS. The average Pd particle size depends on the carbon support, ranging from 1.6 nm for CNF to 2.6 nm for FLG. The catalytic activity of these catalysts was evaluated for two different reactions of interest for biomass transformations, namely hydrodeoxygenation of vanillin to 2-methoxy-4-methyl-phenol (creosol) that requires a bifunctional catalyst with hydrogenating and Lewis acid sites, and aerobic oxidation of 5-hydroxymethylfurfural to 2,5-furandicarboxylic acid. Both compounds have application either as food flavouring additive and polyester co-monomer. For the two reactions the activity order of the fresh catalyst was Pd/FLG > Pd/CNF > Pd/CNT, indicating that FLG contributes favorably to the activity in spite of the larger Pd size of the nanoparticles on this support, a fact that has been attributed to the interaction with the prismatic planes on where Pd nanoparticles are located.

Full text of DOI:10.1002/ejic.201900190

DOI: 10.1002/ejic.201900190
Full Paper
Supported Catalysts
Influence of Carbon Supports on Palladium Nanoparticle
Activity toward Hydrodeoxygenation and Aerobic Oxidation in
Biomass Transformations
Juan C. Espinosa,[a][‡] Rubén Castro Contreras,[b][‡] Sergio Navalón,[a] Camila Rivera-Cárcamo,[b]
Mercedes Álvaro,[a] Bruno F. Machado,[c] and Philippe Serp,*[b]Hermenegildo Garcia*
Abstract: Three palladium catalysts at similar loadings sup- oxygenation of vanillin to 2-methoxy-4-methyl-phenol (creosol)
ported on few-layers graphene (FLG), carbon nanotubes (CNT) that requires a bifunctional catalyst with hydrogenating and Le-
and carbon nanofibers (CNF) have been prepared by wet im- wis acid sites, and aerobic oxidation of 5-hydroxymethylfurfural
pregnation of palladium nitrate with the purpose of determine to 2,5-furandicarboxylic acid. Both compounds have application
the influence of the support on Pd catalytic activity. The sup- either as food flavouring additive and polyester co-monomer.
ports and catalysts have been characterized by chemical analy- For the two reactions the activity order of the fresh catalyst
sis, Raman spectroscopy, XRD, electron microscopy and XPS. was Pd/FLG > Pd/CNF > Pd/CNT, indicating that FLG contributes
The average Pd particle size depends on the carbon support, favorably to the activity in spite of the larger Pd size of the
ranging from 1.6 nm for CNF to 2.6 nm for FLG. The catalytic nanoparticles on this support, a fact that has been attributed
activity of these catalysts was evaluated for two different reac- to the interaction with the prismatic planes on where Pd nano-
tions of interest for biomass transformations, namely hydrode- particles are located.
Introduction
cles (NPs).[5] Since different allotropic forms of carbon have be-
come commercially available, and considering the influence
that the support plays on the catalytic activity and stability of
supported metal nanoparticles (MNPs), there have been an in-
terest in comparing the performance of these MNPs supported
on different carbon nanoforms.[6] In this regard it has been
found that carbon nanotubes (CNTs) and few-layers graphenes
(FLG) can be used as supports of MNPs due to the interaction
between the extended π orbital of the CNT and FLG and the
partially empty d orbitals of transition metals.[7] In addition,
carbon nanofibers (CNFs) consisting in 1D thick carbon materi-
als with long aspect ratio can be advantageous also from the
point of view of their availability.[6]
For the sake of sustainability, there is much current interest in
developing processes to obtain bulk chemicals from bio-
mass.[1,2] Most of the biomass transformation reactions require
catalysts to achieve high conversion and selectivity towards the
target product. Since there are numerous reactions in biomass
transformations consisting in hydrogenations and aerobic oxid-
ations, many of these processes can be promoted by supported
palladium catalysts.[3,4] Among the preferred supports, carbon
derived materials have been widely used due to their availabil-
ity, large surface area, tunability of surface properties and the
ability of these carbon supports to stabilize small Pd nanoparti-
In the present study a series of three Pd catalysts supported
on CNT, FLG or CNF has been prepared with the purpose of
evaluating the influence of the nature of the carbon support
on the palladium catalyst for two different reaction types of
interest in biomass transformation. Although there are some
reports on the activity of other bimetallic catalysts for these
reactions,[4] since palladium is among the most efficient cata-
lysts, the focus of the present study is the influence of the na-
ture of the carbon supports on the catalytic activity of Pd nano-
particles for two important reactions in the transformation of
biomass. Specifically, the reactions under study are shown in
Scheme 1. As it can be seen there, one of them is the hydrode-
oxygenation of vanillin (V), an abundant starting material,[8] to
2-methoxy-4-methylphenol (MMP, creosol) that is an important
product in the food flavoring industry. This reaction requires a
bifunctional catalyst having hydrogenating activity as well as
some acidity.[9] In the literature there are precedents reporting
[a] Departamento de Química and Instituto de Tecnología Química CSIC-UPV,
Universitat Politècnica de València, Consejo Superior de Investigaciones
Científicas,
Av. De los Naranjos s/n, 46022 Valencia, Spain
E-mail: hgarcia@qim.upv.es
http://itq.upv-csic.es/empleado/garcia-gomez-hermenegildo
[b] LCC-CNRS 8241, Université de Toulouse, CNRS, INPT.
Toulouse, France
E-mail: Philippe.saerp@ensiacet.fr
https://www.lcc-toulouse.fr/auteur222.html?lang=fr
[c] Laboratory of Separation and Reaction Engineering - Laboratory of
Catalysis and Materials (LSRE-LCM), Chemical Engineering Department,
Faculty of Engineering, University of Porto,
Rua Dr. Roberto Frias s/n, 4200-465 Porto, Portugal
[d] Center of Excellence for Advanced Materials Research, King Abdulaziz
University,
Al Ehtifalat St, Jeddah 21589 , Saudi Arabia
[‡] Both are first authors.
Supporting information and ORCID(s) from the author(s) for this article are
available on the WWW under https://doi.org/10.1002/ejic.201900190.
Eur. J. Inorg. Chem. 0000, 0–0
1
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 1. Two selected biomass transformations that can be catalysed by Pd NPs supported on carbon materials.
the catalytic hydrodeoxygenation of vanillin, a reaction requir- as support for palladium deposition. CNT, CNF and FLG sup-
ing bifunctional acid/base and hydrogenating sites, using sup- ports were previously oxidized with nitric acid to increase the
ported noble metal nanoparticles such as Pt, Au and Pd. Of concentration of surface carboxylic groups, which should act as
them, Pd was among the most active catalyst.[10] The reported anchoring sites for MNPs.[16] The elemental analysis, XPS and
studies deal with the catalytic activity of Pd nanoparticles on XRD patterns, and nitrogen adsorption data of these three sup-
metal oxides and large area active carbons. However, the possi- ports are listed in Table 1. TEM micrographs of the supports are
ble role of the nature of the carbon support has not been yet provided in Figure S1.
addressed. This is important in order to understand better the
role of the carbon support on the catalytic function. The second
reaction consists in the aerobic oxidation of 5-hydroxymethyl-
furfural (HMF) to furan dicarboxylic acid (FDCA) that can be a
biomass-derived monomer for the preparation of sustainable
polyesters.[11,12] HMF is the main platform chemical obtained
from cellulose that accounts for 40 % of all the biomass feed-
stock.[1,13] Pd containing catalysts have also been used to pro-
mote the aerobic oxidation of HMF.[14] In this reaction, the main
issue was to achieve high selectivity towards the wanted di-
carboxylic acid. Also for this reaction, comparison with other
noble metals such as Pt or Au shows the higher activity of
Pd.[11] Although active carbon has also been used as support
for Pd nanoparticles in HMF oxidation, a detailed study on the
influence of different type of carbons has still not been made,
this being the main objective of the present study.[15] The data
obtained have allowed to rationalize the activity of Pd nanopar-
ticles and their stability as depending on the basal/prismatic
plane ratio of the carbon support.
These materials show different arrangements of the graph-
ene layers. Among the filamentous carbon materials, we can
distinguish between CNTs, for which the graphene layers are
parallel to the filament axis, and CNFs, for which an angle exists
between the filament axis and the graphene layers. The average
angle with respect to the filament long axis is ≈ 26° for CNF.
For this latter support, disordered amorphous carbon can be
occasionally observed at its surface, resulting in a mixture of
edges and disordered carbon (Figure S2). The mean external
diameter of CNFs is 50 nm, and their mean inner diameter is
6 nm. The CNTs are multi-walled, with an average external
diameter of 15 nm, and an internal diameter of 7 nm. They are
composed of 5–10 graphene layers. The FLG consists in gra-
phitic flakes of turbostratic carbon of around 1 μm2, the num-
ber of graphene layers ranges between 5 and 50. The elemental
analysis shows a high level of purity for these materials (90–
97 %). The treatment with nitric acid, for a given type of mate-
rial, results in an increase of the amount of surface oxygen, as
revealed by XPS. The percentage of oxygen determined by XPS
for the purified supports was: 0.8 for CNTs, 0.9 for FLG and 1.5
for CNFs. The textural properties of the supports have been
evaluated from nitrogen adsorption/desorption isotherms.
These carbon materials present pore volumes of 0.2 to
1.2 cm3 g–1 and are predominantly mesoporous. For the fila-
Results and Discussion
Catalyst Support and Catalyst Characterization
Three kinds of carbon materials presenting different arrange- mentous carbons, the pores arise from interstices between en-
ment of the graphene layers, CNTs, CNFs, and FLG were used tangled fibers or tubes. The specific surface area of the supports
Table 1. Chemical, textural, and crystallite parameters of the carbon supports used in the present study.
Supports
Elemental analysis
XPS analysis
Textural properties
Crystallite parameters
C
[%]
C
[%]
O
[%]
BET surface
Pore volume
d002
[nm]
LC(002)
[nm]
La(110)
[nm]
area [m2 g–1
]
[cm3 g–1
]
CNTCOOH
FLGCOOH
CNFCOOH
90.4
97.1
95.7
94.3
96.2
93.4
5.7
3.8
6.6
220
51
71
1.2
0.2
0.3
0.3407
0.3393
0.3434
4.48
10.34
4.64
7.95
16.05
10.95
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
2
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ranged between 50 and 220 m2 g–1. Type IV isotherms with edges and disordered carbon, as already discussed (Figure S2).
type H2 hysteresis loops were obtained (Figure S3), suggesting From this characterization, we can anticipated that the CNF sup-
that the supports are mainly mesoporous and contain, as ex- port with high proportion of edges and defects should not fa-
pected, a small number of micropores. Since the ratio between vor metal sintering due to a limited metal diffusion on such a
the basal plane and “non-basal plane” surface areas should be surface.
an important factor in determining the surface reactivity of
these materials, DFT calculations, derived from nitrogen adsorp-
tion data, were used for the determination of surface hetero-
geneities in these supports by adsorptive potential distribu-
tion.[17] Thus, the absolute and relative proportion of basal
plane and “non-basal plane” surface areas can be determined
from the nitrogen adsorption measurement, followed by an
evaluation of the adsorption data using DFT.[18] The “non-basal
plane surfaces” can be further subdivided into prismatic sur-
faces and “defect surfaces” (the latter contain surface groups
and other surface defects). Figure 1 depicts the percentages of
prismatic: basal: defect surfaces of the three supports. The high-
est proportion of basal surface area is observed for CNTs, fol-
lowed by FLG. The fact that the produced FLG flakes present a
significant number of layers (5–50) and defect density (see Ra-
man spectra) could explain this result. As expected, CNF
presents a significant amount of defects. This should be related
to the surface of the CNF being composed of a mixture of
The d002 graphene interlayer distances, the average crystal-
lite size considering the particles as parallelepiped in the basal
plane (parallel to the hexagonal layers, La) and parallelepiped
height (perpendicular to the hexagonal layers, Lc). These param-
eters were determined from the position of the corresponding
peaks in XRD.[19] The corresponding diffractograms are given in
Figure S4. The highest intensity for the graphite (002) peak
(2θ = 26°) and therefore d002 values close to the value for graph-
ite (0.335 nm), was observed for the FLG support. This support
shows also the largest crystallite size.
The Raman and detailed XPS characterizations of the carbon
supports are reported in Table 2. The best fit for the main C1s
spectrum (Figure S5) was obtained by deconvoluting the exper-
imental high resolution XPS peak into seven line shapes: i) peak
i, at around 283.7 eV, is attributed to the presence of point
defects (C vacancies, pentagon and heptagon rings, Stone-
Wales defects); ii) peak ii, at 284.6 eV, represents sp2 C–C bonds;
iii) peak iii, at 285.1–285.2 eV, is assigned to sp3 carbon species;
iv–vi) the peaks at 286.0–286.2 (iv), 287.3–287.5 (v), and 288.4–
289.2 eV (vi), correspond to carbon atoms attached to different
oxygen-containing groups; and vii) the π–π∗ transition loss
peak is detected at 290.6–291.2 eV.[20] For all the investigated
supports, significant amounts of oxygenated species were
detected (peaks iv, v and vi in XPS C1s peak), resulting from
HNO3 oxidation. The amount of sp3 carbon follows the order
CNF > CNT > FLG. The amount of sp2 carbon follows the oppo-
site order. FWHM of the C 1s peak has been used to assess the
graphitic nature of the supports, with a larger FWHM corre-
sponding to a more disordered structure.[21]
Raman spectroscopy was also used to evaluate the degree
of structural order of the supports. It provides analogous infor-
mation to XRD, although it has the advantage of surface speci-
ficity, thus allowing the study of very heterogeneous materials.
The most important parameter calculated Raman spectroscopy
is the ratio between the integrated intensities of the D band
(ID) at ≈ 1380 cm–1, attributed to the defects of the graphitic
structure, and G band (IG) at ≈ 1580 cm–1, which is ascribed to
a graphitic (ordered) structure, both bands belonging to the
first-order Raman spectrum for carbon materials. Another pa-
rameter, measurable by Raman spectroscopy, which is relevant
to catalyst preparation, is the LD; LD is a typical inter-defect
distance, with the defect being considered a point-like (zero-
dimensional) structure.[22] A quantification of the long-range or-
Figure 1. Percentage of prismatic (black): basal (light grey): defect surfaces
(dark grey) surface area of the supports. a) CNT; b) FLG, and c) CNF.
Table 2. Raman analyses [intensity ratio of the Raman G and D bands (ID/IC) and inter-defect distance (LD) and microcrystallite size (La)] and XPS C1s peak
components of the carbon supports employed in the present study.
Supports
Raman analysis
XPS analysis, C1s peak
ID/IG
LD
[nm]
La
[nm]
FWHM
[eV]
i
ii
iii
iv
v
vi
vii
CNTCOOH
FLGCOOH
CNFCOOH
1.01
0.64
1.24
11.9
15.0
10.7
19
30
15
0.72
0.69
0.76
1.3
0.9
4.6
56.6
62.1
52.1
9.8
7.0
10.5
10.0
9.6
10.4
7.0
6.0
7.2
5.6
5.1
6.3
9.6
9.3
9.0
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
3
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
der in the support can also be made from the graphitic in-plane (XRD, Raman, and XPS analyses). Similar dispersion were ob-
microcrystallite size, La.[23] The Raman analyses confirm the XPS tained on CNTs and FLG. This results might appear surprising,
results in the sense that both analyses point to a higher order considering the fact that CNTs have a larger surface area
in FLG: low ID/IG, large La and LD, and narrow FWHM. The most (220 m2 g–1) compared to FLG (50 m2 g–1), and larger amount
defective support is CNF, and CNT constitutes an intermediate of surface oxygen groups (Table 1). We propose that this result
situation.
From all these analyses, some conclusions can be drawn re-
arises from the surface on which these MNPs are deposited:
mainly prismatic surface for FLG, and basal surface for CNTs.
Figure 2b clearly shows that Pd NPs are preferentially deposited
on the edges and prismatic planes of the FLG support.
garding palladium catalyst preparation and use. Firstly, if we
consider the prismatic: basal: defect surfaces of the supports,
the palladium–carbon support interface/interaction should be
different for these catalysts. It is expected to have a higher pro-
portion of Pd-prismatic surface interaction for FLG, and a higher
proportion of Pd-defect surface interaction for CNF. CNTs consti-
tuting an intermediate situation. Secondly, considering the sur-
face area of the supports, and the long range order according
to Raman, XRD and XPS analyses, it can be anticipated that
palladium surface migration, and thus possible sintering should
be favoured on FLG. On CNFs, the presence of disordered
carbon on their surface should limit this unwanted agglomera-
tion. Finally, considering the high proportion of oxygenated sur-
face species, the palladium dispersion should be relatively high
for the three supports.
The 2 w/w% Pd catalysts were prepared by wet impregnation
using hydrated Pd nitrate. This precursor was chosen because
its reduction yields impurity-free Pd particles. Table 3 shows the
characterization data for the Pd catalysts. The specific surface
area of the catalysts did not change significantly compared to
the bare supports. XRD analyses show a very small and broad
Pd(111) peak at 2θ = 40.1° characteristic of metallic palladium
(Figure S6), pointing to a small Pd particle size. The palladium
particle size was measured from TEM micrographs, and the
metal dispersion was calculated from an equation for metal dis-
persion from TEM, assuming a spherical particle geometry, as
described elsewhere.[24]
Figure 2. Representative micrographs and particle size distribution for: a) Pd/
CNT–HRTEM (scale bar = 50 nm); b) Pd/FLG–STEM-HAADF (scale bar = 20 nm);
and c) Pd/CNF–HRTEM (scale bar = 10 nm).
Table 3. Metal loading, surface area, Pd mean size and dispersion for the
investigated catalysts.
Pd
[%]
BET surface
area
Particle
size
Pd
The catalysts have also been analyzed by XPS (Figure S7 and
Table 4). The XPS Pd 3d peak for all catalysts was wide, indicat-
ing either multiple oxidation states of the Pd species, or several
different interactions of the oxidized carbon support. Peaks at
335.5 and 340.5 eV, were assigned to Pd0 3d5/2 and 3d3/2, re-
spectively, while the 3d5/2 and 3d3/2 peaks for Pd2+ are at
dispersion
[%]
[m2 g–1
]
[nm][a]
Pd/CNTCOOH
Pd/FLGCOOH
Pd/CNFCOOH
2.04
1.87
1.70
232
50
82
2.2 1.1
2.6 1.6
1.6 0.9
49
43
67
Representative micrographs, and particle size distributions 337.8 eV and 343.4 eV. The XP spectra of palladium catalyst
are shown in Figure 2. It is worth mentioning that for CNTs that showed peaks for Pd 3d5/2 and Pd 3d3/2 with Pd0 and Pd2+
present an inner cavity, some particles should be located in this combined lines, but Pd2+ line as dominant. The presence of
cavity. The higher palladium dispersion was obtained on CNFs, higher Pd oxidation states are indicated by the Pd 3d5/2 peak
presumably because of a higher concentration in surface oxy- at a binding energy of 339 eV, which can be assigned to PdO2
genated species (Table 1), and a higher concentration of defects or Pd(OH)4. From the deconvolution pf the XPS Pd 3d peak it
Table 4. XPS Pd 3d peak components and the ratios between reduced and oxidized Pd atoms for each catalyst.
Catalyst
XPS analysis, Pd 3d peak
i
ii
iii
iv
v
vi
Pd0/Pd2+
Pd0/Pdox
Pd/CNTCOOH
Pd/FLGCOOH
Pd/CNFCOOH
12.30
12.30
18.01
8.18
8.18
11.97
39.92
39.44
38.37
26.54
26.22
25.50
7.85
8.33
3.70
5.22
5.53
2.46
0.31
0.31
0.47
0.26
0.26
0.43
[a] From TEM.
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
4
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
appears that, while Pd/CNT and Pd/FLG have a similar distribu- higher in Pd/FLG compared to the other two catalysts (Pd dis-
tion of Pd0 vs. oxidized Pd atoms, Pd/CNF has a larger popula- persion on FLG is lower than for the other two carbon sup-
tion of reduced Pd0.
ports). Thus, other properties of FLG as support and, particu-
larly, the presence of prismatic planes on which the Pd NPs are
interacting with the support should be playing a role determin-
ing Pd catalytic activity. Thus, considering that Pd NPs are the
active sites and carbon supports play a major role interacting
with those Pd NPs, it is proposed that the main factor responsi-
ble for the higher activity of Pd/FLG compared to the other two
catalysts would be the specific location of the Pd NPs on the
prismatic planes present on FLG.
Stability of carbon-supported Pd catalysts was studied by
performing a series of catalytic tests. The results are presented
in Figure 4 (Pd/CNF) and S8 (Pd/FLG)–S9 (Pd/CNT) in the sup-
porting information. As it can be seen in these Figures one
of the most surprising results was that, although the temporal
profiles of three consecutive runs for vanillin disappearance us-
ing Pd/FLG as catalyst were coincident, indicating that the cata-
lytic activity for hydrogenation of V into VA is maintained, the
rate of the second step of hydrodeoxygenation decreases con-
siderably upon reuse, particularly from the fresh to the first
used catalyst. Although, as it can be seen in Table 5, a slight
growth of the average Pd NP size was observed for the three
supported Pd, the size increase was about 1 nm. Thus, it seems
that this decay in activity cannot be attributed to the minor
changes in the Pd NP size distribution. Therefore, other causes
have to be invoked to explain catalytic activity decay.
Catalytic Reactions
The three supported Pd catalysts were used for the two proc-
esses of relevance in biomass transformation indicated in
Scheme 1. As commented in the introduction, one of the reac-
tions under study was the hydrodeoxygenation of vanillin to 2-
methoxy-4-methylphenol. This transformation encompasses
two consecutive steps, the first one being hydrogenation of
conjugated aldehyde group to a benzylic alcohol requiring
hydrogenation sites followed by hydrodeoxygenation of the
hydroxyl group by different sites. Both steps, hydrogenation of
carbonyl groups and removal of oxygen are very common in
biomass transformation by bifunctional catalysts and, therefore,
the results achieved for the transformation of vanillin to MMP
with the different Pd catalysts should also be readily expanded
to many other hydrogenations and hydrodeoxygenations. The
reaction was carried out under moderate hydrogen pressure
(5 bar) in water as solvent at the boiling temperature in the
presence of the three Pd-catalysts under study.
The only product observed in the process as reaction inter-
mediate is vanillyl alcohol (VA). Figure 3 shows the temporal
evolution of vanillin conversion as well as the yields of VA inter-
mediate and the wanted MMP. As it can be seen in this Figure 3,
even though the palladium content is similar for the three ma-
terials, the solids exhibit contrasting catalytic activity. The order
of activity based on conversion of V being Pd-FLG > Pd-CNF >
Pd-CNTs. Besides conversion of V, subsequent disappearance of
vanillyl alcohol and MMP also follows the same order as the
disappearance of vanillin. While the relative activity of Pd/CNF
vs. Pd/CNT can be explained based on the smaller particle size
of Pd NPs in Pd/CNF compared with Pd/CNT, the high activity
of Pd/FLG cannot be predicted based on Pd particle size that is
Figure 4. Vanillin conversion (a), vanillyl alcohol evolution (b) and 2-methoxy-
4-methylphenol (c) formation using Pd/CNF as catalyst during four consecu-
tive uses. Legend: First use (■), second use (B), third use (▲) and fourth use
(P). Reaction conditions: Catalyst (10 mg), vanillin (20 g L–1; 0.13
M), 2 mL of
H2O, 5 bar H2, 100 °C.
Since it seems that the second hydrodeoxygenation step is
the one limiting the overall conversion of V into MMP, it could
be that Pd/FLG becomes deactivated due to the strong adsorp-
tion of the aromatic intermediate VA or MMP. To support this
possibility, the TPD profile of the used Pd/FLG catalyst after
exhaustive washings was performed and compared to that of
the fresh Pd/FLG sample. The results are presented in Figure
S10 confirming that the spent Pd/FLG catalyst desorbs in the
range of temperature from 250 to 700 °C a considerable
Figure 3. Vanillin conversion (a), vanillyl alcohol evolution (b) and 2-methoxy-
4-methylphenol (c) formation using Pd/CNT (■), Pd/CNF (B) and Pd/FLG (▲)
as catalysts. Reaction conditions: Catalyst (10 mg), vanillin (20 g L–1; 0.13
2 mL of H2O, 5 bar H2, 100 °C.
M),
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
5
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 5. Characterization data of four-times used palladium catalysts for vanillin hydrodeoxygenation or 5-hydroxymethylfurfural oxidation, including particle
size, metal leaching and estimated TON and TOF for these reactions.
Entry
Catalyst
Reaction
Particle size
[nm]
Pd Leaching
[%]
TON[a]
TOF[b]
[h–1
]
1
2
3
4
5
6
Pd/CNT
Pd/CNT
Pd/FLG
Pd/FLG
Pd/CNF
Pd/CNF
Hydrodeoxygenation of V
Oxidation of HMF
Hydrodeoxygenation of V
Oxidation of HMF
Hydrodeoxygenation of V
Oxidation of HMF
3.4 3.4
3.9 3.6
2.3 0.8
15.9 11.3
2.8 1.1
6.7 11.3
0.1
3.0
2.3
6.6
0.5
2.1
140
80
149
85
164
94
21
813
78
829
72
901
[a] TON: mol of converted substrate (V or HMF) divided mol of supported palladium after one use. [b] TOF: mol of converted V or HMF divided mol of
supported palladium at 1 and 0.083 h, respectively.
amount of adsorbed reaction intermediates that are absent in bases are not needed. A control experiment in the present case
the fresh catalysts. However, control experiments, performing using Pd/FLG in the absence of base for the oxidation of HMF
the reaction under the general conditions but adding on pur- was unsuccessful, meaning that bases are needed under the
pose at initial reaction time 30 % of VA or MMP, did not show present reaction conditions. After a series of preliminary trials,
any sign of catalyst deactivation either in the temporal profile K2CO3 was found a convenient base for the process in our case.
of vanillin conversion or in the evolution of reaction intermedi- The role of the base is to promote the formation of alcoholate
ates. These poisoning experiments rule out VA or MMP as cata- that should bind to the surface of Pd NPs stronger than the
lyst poison. It is proposed that other by-products could be re- corresponding alcohol. The reaction is carried out in water as
sponsible for this deactivation. This proposal is somehow sup- solvent at 160 °C. During the course of the reaction hydroxy-
ported by comparison of the composition by elemental analysis methylfuroic acid (HMFCA) was observed as the major reaction
of the three palladium catalysts, fresh or after the reaction (see intermediate, accompanied by lesser proportions of the corre-
Table S1 in the supporting information). In the cases of Pd/ sponding aldehydes DFF and FFCA (Scheme 1b). A summary of
CNT and Pd/CNF the differences in the carbon and hydrogen the results is presented in Figure 5, where the relative activity
between the fresh and used materials were not relevant. In con- order can also be inferred.
trast, for Pd/FLG a decrease in the carbon content of about 5 %
upon catalyst use was measured, suggesting that this material
has adsorbed reaction intermediates, products or substrates,
causing its deactivation. Therefore, it seems that deactivation in
the case of Pd/FLG could occur by strong adsorption of by-
products.
In contrast to Pd/FLG, for Pd/CNT and, particularly, Pd/CNF
(compare Figure 4 with S8 and S9), although the fresh materials
are initially less active, both, Pd/CNT (Figure S9) and Pd/CNF
(Figure 4), maintain a significant degree of catalytic activity in
Figure 5. 5-hydroxymethylfurfural (HMF) oxidation (a) and 2,5-furandicarbox-
the two steps, indicating their higher stability under the reac-
tion conditions. In this regard, Pd/CNF combines both a reason-
able catalytic activity and a notably high reusability. Chemical
analysis of the liquid phase after removal of the solid Pd/CNF
catalyst showed that the Pd content in the liquid was only
about 0.1 wt.-% of the initial supported palladium content, indi-
cating the heterogeneous nature of the catalytic process.
Higher palladium leaching was observed when using the Pd/
CNT (0.3 wt.-%) sample and, particularly, when using Pd/FLG
(2.3 wt.-%). Heterogeneity of hydrodeoxygenation of vanillin to
MMP was also supported by performing a hot filtration test in
which the reaction was carried out in the presence of Pd/CNF
solid under the general reaction conditions and, then, filtering
the catalyst after depressurization of the reaction, while the
mixture was still hot (≈ 80 °C) at 30 % conversion. It was ob-
served that the reaction completely stops after filtration of cata-
lyst (Figures S11a–c).
ylic acid (FDCA) formation (b) using Pd/FLG (■), Pd/CNT (B) and Pd/CNF (▲)
as catalyst. Reaction conditions: Catalyst (10 mg), HMF (9.5 g L–1; 75 m
),
K2CO3 (21 g L–1; 152 m
), 2 mL of H2O, 5 bar O2, 160 °C. Note that plots a)
and b) display different reaction times due to the occurrence of two consecu-
tive steps.
M
M
From the temporal evolution of HMF conversion and FDCA
formation it can be deduced that Pd/FLG is again more active
than Pd/CNF, Pd/CNT being the less active material in the series.
Heterogeneity of the process in the case of Pd/FLG was con-
firmed by performing a hot filtration test at 30 % conversion
and by analysis of the Pd in liquid phase at final reaction time.
The results presented in supplementary information (Figure
S11d and e) shows that, although formation of FDCA does not
progress upon removal of the solid catalyst, oxidation of HMF
does not stop completely. This further progress of HMF oxid-
ation could be attributed to the small percentage of Pd leached
to the liquid phase (about 2 % of the total initial Pd content in
The second reaction under study was the aerobic oxidation the Pd/FLG sample). Thus, both measurements hot filtration test
of HMF to 2,5- FDCA using molecular oxygen under pressure. and Pd analysis suggest the occurrence of a minor contribution
This oxidation generally requires the presence of a base in order to the catalysis of a homogeneous component due to the pres-
to take place, although there are scattered examples for which ence of leached Pd.
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
6
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Stability of the catalyst was also studied by performing a lyst for both processes, but it undergoes deactivation in signifi-
series of consecutive reuses under the same reaction conditions cant extent upon reuse. Among the causes of deactivation, poi-
(Figures S12–S15). As in the case of the hydrodeoxygenation of soning in the case of hydrodeoxygenation and Pd particle size
vanillin (V), Pd/FLG undergoes a significant decay in the cata- growth in the case of the aerobic oxidation appear the most
lytic activity in the reuse, Pd/CNF being again the most stable probable ones. In this regard, CNFs as support presents a com-
material. Pd/CNT exhibits an intermediate performance be- promise between activity of the fresh material and stability
tween those of Pd/FLG and Pd/CNF. Reusability tests for upon reuse. This role of CNFs has been interpreted as derived
Pd/CNF are presented in Figure 6, while supporting information from the stabilization of particle size and the lack of poisoning
contains information for the other two catalysts (Figures S12– by adsorption as indicated by elemental analysis. Characteriza-
S15).
tion of the supported Pd catalyst has shown that for similar
particle size and loading, it is not only the Pd dispersion the
only parameter controlling the catalysis, but the nature and
proportion of the basal, non-basal and defects planes of the
support can have also a decisive importance in determining the
activity and stability of the Pd NPs. In this regard, based on the
characterization of CNF by electron microscopy, it is proposed
that the superior catalytic activity of Pd NPs on CNF derives
from a good balance with amorphous regions on the surface
and basal and non-basal planes to stabilize Pd NPs.
Figure 6. 5-hydroxymethylfurfural (HMF) conversion (a) and 2,5-furandicarb-
oxylic acid (FDCA) formation (b) upon three consecutive uses of Pd/CNF as
catalyst. Legend: First use (■), second use (B), third use (▲) and forth use
Experimental Section
(P). Reaction conditions: Catalyst (10 mg), HMF (9.5 g L–1; 75 m
M), K2CO3
Materials: Vanillin, vanillyl alcohol, 2-methoxy-4-methyl-phenol, 5-
(hydroxymethyl)furfural, 5-hydroxymethyl-2-furancarboxylic acid, 5-
formyl-2-furoic acid, 2,5-furandicarboxaldehyde, 2,5-furandicarbox-
ylic acid and palladium(II) nitrate dihydrate were supplied by Sigma-
Aldrich. Other reagents or solvents employed in the present work
were analytical or HPLC grade.
(21 g L–1; 152 m
M), 2 mL of H2O, 5 bar O2, 160 °C.
In order to understand the origin of deactivation, TEM im-
ages of fresh and used supported Pd catalyst were compared.
The particle size distributions for Pd NPs on the three carbon
supports after use in 5-hydroxymethylfurfural oxidation are
shown in Figures S16–S18, while Table 5 shows the distributions
for all supported Pd catalysts. It was observed that in the case
of Pd/FLG a remarkable increase in the Pd NP particle size from
2.6 1.6 to 15.9 11.3 occurs after the 5-hydroxymethylfurfural
oxidation reaction. In the case of Pd/CNT, some scattered big
Pd NPs are also observed in the used sample . Considering that
carboxylic acids are formed in the oxidation process and that
by Coulombic interactions these compounds can increase the
size of metal NPs, the data obtained by TEM images for Pd/FLG
and Pd/CNT indicating a certain growth of Pd NPs are not with-
out precedent. In the case of Pd/CNF the increase in Pd particle
size is also clear, although it seems that the average is still suffi-
ciently low (6.7 nm) to exhibit a good catalytic performance.
Not surprisingly, considering the mobilizing effect of carboxylic
acids for metal ions, palladium leaching from the solid catalyst
to the solution was significantly higher for the oxidation of HMF
than for the hydrodeoxygenation of vanillin (see Table 5), reach-
ing in the worst case a Pd leaching of 6.6 % of the initial Pd
content of the Pd/FLG.
Material Preparation: CNTCOOH synthesis. An AlCoFeO4 catalyst was
reduced in a fluidized bed reactor under a nitrogen (225 mL min–1
)
and hydrogen (150 mL min–1) flow at 650 °C. After the reduction
step, the ethylene flow was adjusted to 225 mL min–1
for 30 min to produce CNT. The CNT were recovered and then puri-
fied using an aqueous solution (50 vol.-% H2SO4) under reflux at
140 °C for 3 h. The acidic solution was then filtered and the solid
washed with distilled water. The resulting solid was dried in an oven
at 80 °C overnight. The purified CNT was then functionalized with
HNO3 under reflux at 140 °C for 3 h. The acidic solution was filtered
and washed with distilled water. The resulting solid was dried in an
oven at 80 °C overnight to produce functionalized CNT (CNTCOOH).
FLGCOOH synthesis. A CoFe2O4 catalyst was reduced in a fluidized
bed reactor under a nitrogen (225 mL min–1) and hydrogen
(150 mL min–1) flow at 650 °C. After reduction step, the ethylene
flow was adjusted to 225 mL min–1 for 30 min to produce FLG. The
FLG was recovered and purified at room temperature overnight us-
ing HCl. The acidic solution was then filtered and washed. The re-
sulting solid was dried in an oven at 80 °C overnight. The purified
FLG was then functionalized with HNO3 under reflux at 140 °C for
3 h. The acidic solution was filtered and washed. The resulting solid
was dried in an oven at 80 °C overnight to produce functionalized
FLG (FLGCOOH).
Conclusions
CNFCOOH synthesis: A CoFe2O4 catalyst was reduced in a fluidized
bed reactor under a nitrogen (225 mL min–1) and hydrogen
(150 mL min–1) flow at 450 °C. After the reduction step, the ethylene
flow was adjusted to 225 mL min–1 for 30 min to produce CNF.
The CNF were recovered and then purified at room temperature
overnight using HCl. The acidic solution was then filtered and
washed. The resulting solid was dried in an oven at 80 °C overnight.
The purified CNF was functionalized using HNO3 under reflux at
The present results have shown that although regardless the
support Pd NPs exhibit a general activity for catalyzing hydro-
deoxygenations and aerobic oxidations of interest in biomass
transformation, the nature of the carbon support plays a signifi-
cant role determining the activity of the fresh samples and the
stability of Pd NPs under the reaction conditions. It has been
found that Pd supported on FLG is initially the most active cata- 80 °C for 3 h. The acidic solution was then filtered and washed. The
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
7
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
resulting solid was dried in an oven at 80 °C overnight to produce
functionalized CNF (CNFCOOH).
then, pressurized at 5 bar. The system, whilst stirring, was heated
at 100 °C employing a preheated oil bath. This event was consid-
ered the 0 reaction times. At the required times, aliquots were taken
from the reactor (100 μL), diluted with H2O (100 mL) and filtered
(0.2 μm Nylon filter) before analysis by HPLC.
Pd/carbon nanomaterials. A wet impregnation method was used
to prepare the palladium catalysts supported on carbon nanomate-
rials. The desired amount of Pd(NO3)2·2H2O was added to an acet-
one solution (100 mL) containing 1 g of carbon nanomaterial, so as
to introduce 2 w/w% of metal phase. The solution was then soni-
cated at room temperature for 1 h and magnetically stirred over-
night. The solution was then filtered and washed with acetone that
ensures that only strongly bound Pd species remains adsorbed on
the support. The main factors responsible for Pd loading are specific
surface area and surface acidity of the carbon materials.[25] The re-
sulting solid was dried in an oven at 120 °C overnight. Finally, the
catalyst was reduced in a horizontal tube oven under a nitrogen
and hydrogen flow (20 vol.-% H2) at 300 °C for 2 h.
The oxidation of HMF was carried out in a pressurized glass reactor
with head stirrer. HMF (19 mg) was dissolved in H2O (2 mL) contain-
ing K2CO3 as base (42 mg) and, then, the required amount of cata-
lyst was added (i.e. 10 mg). The mixture was sonicated for 30 min
with a power of 450 W. This relatively low sonication power ensures
good dispersion of the catalyst in the reaction mixture without pro-
ducing structural damage. The reactor was purged with oxygen five
times and pressurized with oxygen (5 bar) before heating the sys-
tem at 160 °C, whilst stirring, in a preheated oil batch. The initial
reaction rate was considered when the reactor was introduced into
the bath.
Materials Characterization: N2 sorption: The surface area was
measured using the Brunauer, Emmet, and Teller (BET) method. The
pore radius distribution was determined by the Barrett, Joyner, and
Halenda (BJH) method. TEM micrographs were taken with a JEOL
JEM 1011 microscope operating at 100 kV. Palladium particle size
distribution was calculated using the ImageJ software measuring at
least 250 particles. XRD patterns were recorded at room tempera-
ture with a Panalytical X'PERT PRO diffractometer, employing a
Cu-Kα radiation (λ = 1.54 Å) and a parabolic MPD-mirror for Cu
radiation. The diffractograms were acquired in 2θ range of 5–90°,
using a continuous scan mode with an acquisition step size of
0.0170° and a counting time of 299.72 s. Raman spectra were meas-
ured in a micro Raman Spectrometer HR 800 Jobin Yvon Horiba
using a laser of 532 nm wavelength as excitation source. Nitrogen
adsorption measurements were performed at –196 °C using a 3Flex
Surface Characterization Analyzer (Micromeritics GmbH). The sur-
face area was measured using Brunauer, Emmet and Teller (BET)
method. The pore radius distribution was determined by the Bar-
rett, Joyner and Halenda (BJH) method. Before analysis, the samples
were degassed at 150 °C using a Micromeritics VacPrep 061 Sample
Degas System until a static pressure of less than 0.01 Torr
(0.0133 mbar) was reached. The elemental results were obtained
using a CHN Perkin-Elmer elemental analyzer. Palladium loadings
were determined using an iCAP 6300 ICP Spectrometer, by diges-
tion of the samples in hot aqua regia. The adsorptive potential dis-
tributions were calculated from the adsorption isotherms using the
standard software implemented in the apparatus (3Flex Version
4.04). The calculation of the DFT, basal plane and “non-basal plane”
surfaces as well as of the respective ratios has been performed as
described previously.[18] The surface chemical composition of the
carbon materials was analyzed by XPS (Kratos AXIS Ultra HSA spec-
trometer), using a Mg-Kα radiation (1486.7 eV). The existing ele-
ments, and their corresponding concentrations, were determined
by recording a survey spectra up to a binding energy (BE) of
1300 eV. High-resolution XPS spectra of the C1s, and Pd3d core
levels and full-width-at-half-maximum (FWHM) values were used to
assess the chemical state of these elements at the catalyst surface,
according to the NIST database. Inductively coupled plasma-atomic
emission spectroscopy (ICP-AES) was employed to determine the
amount of possible Pd leaching from the catalyst to the solution
upon reaction.
Reusability experiments for each reaction were carried out under
the corresponding conditions indicated above. At the end of the
reaction the catalyst was recovered by filtration employing a nylon
membrane (0.2 μm). The used catalyst was thoroughly washed with
water and ethanol and, then, dried in an oven at 100 °C. The result-
ing solid was employed in a subsequent catalytic cycle. This proce-
dure was repeated three times.
The heterogeneity of the reaction was tested by performing the
hot filtration test. This test consists in removing the catalyst from
the reaction mixture after depressurization, filtering the suspension
without cooling (≈ 80 °C) at a certain conversion. The hot filtrate
was allowed to react further, in the absence of solid, under the
reaction conditions.
Analysis of the Reaction: The reagents and products of both reac-
tions under study (Scheme 1) were analyzed by HPLC using a pho-
todiode array as detector. Previously filtered reaction aliquots were
injected in the HPLC system equipped with a reverse-phase chro-
matography column [Kromasil-CN for vanillin (V) reaction or Krom-
asil-C18 for HMF reaction] as stationary phase. Mobile phase work-
ing under isocratic conditions for compounds V and HMF were
H2O:CH3CN (40:60 vol.-%) or H2O, respectively.
Acknowledgments
Financial support by the Spanish Ministry of Economy and
Competitiveness (Severo Ochoa GTQ2015-65163-C02-R1 and
CTQ2014-53292-R) is gratefully acknowledged. Generalidad
Valenciana is also thanked for funding (Prometeo 2017/063).
S. N. thanks financial support by the Fundación Ramón Areces
(XVIII Concurso Nacional para la Adjudicación de Ayudas a la
Investigación en Ciencias de la Vida y de la Materia, 2016).
C. R. C. thanks CONICYT for the financial support (Becas de
doctorado en el extranjero “Becas Chile” - n° 72170200). The
authors thank Dr. Tobias Placke (Universität Münster, Germany)
for nitrogen adsorption measurements and adsorptive potential
distributions calculations.
Keywords: Heterogeneous catalysis · Palladium
nanoparticles · Carbon support · Biomass transformation
Catalytic Tests: Hydrodeoxygenation of vanillin was carried out in
a pressurized glass reactor with head stirrer. Briefly, vanillin (40 mg)
dissolved in H2O (2 mL) was placed together with the required
amount of catalyst (i.e. 10 mg). The mixture was sonicated in an
ultrasound bath (450 W) for 30 min to ensure a good dispersion of
the catalyst. Then, the reactor was purged five times with H2 and,
[1] A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411–2502.
[2] G. W. Huber, S. Iborra, A. Corma, Chem. Rev. 2006, 106, 4044–4098; R. M.
Navarro, M. A. Peña, J. L. G. Fierro, Chem. Rev. 2007, 107, 3952–3991;
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
8
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, B. M. Weckhuysen, Chem.
Rev. 2010, 110, 3552–3599.
J. N. Chheda, G. W. Huber, J. A. Dumesic, Angew. Chem. Int. Ed. 2007, 46,
7164–7183; Angew. Chem. 2007, 119, 7298.
C.-H. Zhou, X. Xia, C.-X. Lin, D.-S. Tong, J. Beltramini, Chem. Soc. Rev. 2011,
40, 5588–5617.
S. Crossley, J. Faria, M. Shen, D. E. Resasco, Science 2010, 327, 68–72; X.
Xu, Y. Li, Y. Gong, P. Zhang, H. Li, Y. Wang, J. Am. Chem. Soc. 2012, 134,
16987–16990.
[16] B. F. Machado, M. Oubenali, M. R. Axet, T. T. NGuyen, M. Tunckol, M.
Girleanu, O. Ersen, I. C. Gerber, P. Serp, J. Catal. 2014, 309, 185–198.
[17] T. Placke, V. Siozios, R. Schmitz, S. F. Lux, P. Bieker, C. Colle, H.-W. Meyer,
S. Passerini, M. Winter, J. Power Sources 2012, 200, 83–91.
[18] Placke, T., Siozios, V., Rothermel, S., et al. (2015). Assessment of Surface
Heterogeneity: a Route to Correlate and Quantify the 1st Cycle Irreversi-
ble Capacity Caused by SEI Formation to the Various Surfaces of Graphite
Anodes for Lithium Ion Cells. Themed Issue: Lithium Ions in Solids – Be-
tween Basics and Better Batteries / Paul Heitjans. Zeitschrift für Physika-
lische Chemie, 229(9), pp. 1451–1469. Retrieved 22 Mar. 2019, from
doi:10.1515/zpch-2015-0584.
[19] T. V. Reshetenko, L. B. Avdeeva, Z. R. Ismagilov, V. V. Pushkarev, S. V.
Cherepanov, A. L. Chuvilin, V. A. Likholobov, Carbon 2003, 41, 1605–
1615.
[20] H. Estrade-Szwarckopf, Carbon 2004, 42, 1713–1721; K. Ganesan, S.
Ghosh, N. G. Krishna, S. Ilango, M. Kamruddin, A. K. Tyagi, Phys. Chem.
Chem. Phys. 2016, 18, 22160–22167; T. I. T. Okpalugo, P. Papakonstan-
tinou, H. Murphy, J. McLaughlin, N. M. D. Brown, Carbon 2005, 43, 153–
161.
[3]
[4]
[5]
[6]
[7]
E. Lam, J. H. Luong, ACS Catal. 2014, 4, 3393–3410.
S. Navalon, A. Dhakshinamoorthy, M. Alvaro, H. Garcia, Coord. Chem. Rev.
2016, 312, 99–148; G. G. Wildgoose, C. E. Banks, R. G. Compton, Small
2006, 2, 182–193; J. Zhu, A. Holmen, D. Chen, ChemCatChem 2013, 5,
378–401.
F. G. Calvo-Flores, J. A. Dobado, ChemSusChem 2010, 3, 1227–1235; B. M.
Upton, A. M. Kasko, Chem. Rev. 2016, 116, 2275–2306.
[8]
[9] F. Zhang, Y. Jin, Y. Fu, Y. Zhong, W. Zhu, A. A. Ibrahim, M. S. El-Shall, J.
Mater. Chem. A 2015, 3, 17008–17015.
[10] J. L. Santos, M. Alda-Onggar, V. Fedorov, M. Peurla, K. Eranen, P. Maki-
Arvela, M. A. Centeno, D. Y. Murzin, Appl. Catal. A 2018, 561, 137–149; P.
Hao, D. K. Schwartz, J. W. Medlin, Appl. Catal. A 2018, 561, 1–6; F. Zhang,
S. Zheng, Q. Xiao, Y. Zhong, W. Zhu, A. Lin, M. Samy El-Shall, Green Chem.
2016, 18, 2900–2908.
[21] J.-O. Müller, D. S. Su, U. Wild, R. Schlögl, Phys. Chem. Chem. Phys. 2007,
9, 4018–4025.
[22] L. G. Cançado, A. Jorio, E. H. Martins Ferreira, F. Stavale, C. A. Achete, R. B.
Capaz, M. V. O. Moutinho, A. Lombardo, T. S. Kulmala, A. C. Ferrari, Nano
Lett. 2011, 11, 3190–3196.
[23] L. G. Cançado, K. Takai, T. Enoki, Appl. Phys. Lett. 2006, 88, 163106.
[24] S. Domínguez-Domínguez, A. Berenguer-Murcia, B. K. Pradhan, A. Lin-
ares-Solano, D. Cazorla-Amorós, J. Phys. Chem. C 2008, 112, 3827–3834.
[25] M. L. Toebes, J. A. van Dillen, K. P. de Jong, J. Mol. Catal. A 2001, 173,
75–98; D. J. Suh, T.-J. Park, S.-K. Ihm, Carbon 1993, 31, 427–435; M. Kang,
M. W. Song, K. L. Kim, React. Kinet. Catal. Lett. 2002, 76, 207–212.
[11] S. E. Davis, M. S. Ide, R. J. Davis, Green Chem. 2013, 15, 17–45.
[12] S. Dutta, S. De, B. Saha, ChemPlusChem 2012, 77, 259–272; Z. Zhang,
K. Deng, ACS Catal. 2015, 5, 6529–6544.
[13] S. P. Teong, G. Yi, Y. Zhang, Green Chem. 2014, 16, 2015–2026.
[14] B. Siyo, M. Schneider, J. Radnik, M.-M. Pohl, P. Langer, N. Steinfeldt, Appl.
Catal. A 2014, 478, 107–116.
[15] P. V. Rathod, V. H. Jadhav, ACS Sustainable Chem. Eng. 2018, 6, 5766–
5771.
Received: February 18, 2019
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
9
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Supported Catalysts
Carbon nanofibers appears to be the
best choice among few-layers graph-
ene and carbon nanotubes as support
of Pd nanoparticles for the two differ-
ent reactions shown in the scheme
J. C. Espinosa, R. Castro Contreras,
S. Navalón, C. Rivera-Cárcamo,
M. Álvaro, B. F. Machado, P. Serp,*
H. Garcia* .......................................... 1–10
Influence of Carbon Supports on
Palladium Nanoparticle Activity to-
ward Hydrodeoxygenation and Aer-
obic Oxidation in Biomass Transfor-
mations
DOI: 10.1002/ejic.201900190
Eur. J. Inorg. Chem. 0000, 0–0
www.eurjic.org
10
© 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim