Non-noble metal catalysts for hydrogenation: A facile method for preparing Co nanoparticles covered with thin layered carbon

Article abstract of DOI:10.1016/j.jcat.2016.04.006

Metallic cobalt nanoparticles with surface CoO_x patches covered by thin layered carbon (named Co@C) have been directly synthesized by thermal decomposition of Co-EDTA complex. Raman spectra and HRTEM images suggest that discontinuities can be found in the disordered layered carbon. XPS shows that the CoO_x patches in the Co@C nanoparticles can reduced to metallic Co by H₂ under reaction conditions (7 bar at 120 °C), and H₂-D₂ exchange experiments show that the reduced metallic Co nanoparticles covered by carbon layers can dissociate H₂. The Co@C nanoparticles show excellent activity and selectivity during chemoselective hydrogenation of nitroarenes for a wide scope of substrates under mild reaction conditions. Based on the results from DRIFTS adsorption experiments, we propose that metallic Co in the Co@C nanoparticles is the active phase. The role of the carbon layers is to protect the Co from overoxidation by air, leading to the chemoselective hydrogenation of nitroarenes.

Full text of DOI:10.1016/j.jcat.2016.04.006

Journal of Catalysis 340 (2016) 1–9
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Non-noble metal catalysts for hydrogenation: A facile method
for preparing Co nanoparticles covered with thin layered carbon
⇑
Lichen Liu, Patricia Concepción, Avelino Corma
Instituto de Tecnología Química, Universidad Politécnica de Valencia-Consejo Superior de Investigaciones Científicas (UPV-CSIC), Av. de los Naranjos s/n, 46022 Valencia, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Metallic cobalt nanoparticles with surface CoOx patches covered by thin layered carbon (named Co@C)
have been directly synthesized by thermal decomposition of Co-EDTA complex. Raman spectra and
HRTEM images suggest that discontinuities can be found in the disordered layered carbon. XPS shows
that the CoOx patches in the Co@C nanoparticles can reduced to metallic Co by H2 under reaction condi-
tions (7 bar at 120 °C), and H2–D2 exchange experiments show that the reduced metallic Co nanoparticles
covered by carbon layers can dissociate H2. The Co@C nanoparticles show excellent activity and selectiv-
ity during chemoselective hydrogenation of nitroarenes for a wide scope of substrates under mild reac-
tion conditions. Based on the results from DRIFTS adsorption experiments, we propose that metallic Co in
the Co@C nanoparticles is the active phase. The role of the carbon layers is to protect the Co from overox-
idation by air, leading to the chemoselective hydrogenation of nitroarenes.
Received 17 December 2015
Revised 29 March 2016
Accepted 9 April 2016
Keywords:
Co nanoparticles
Layered carbon
Chemoselective hydrogenation
Nitroarenes
Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction
transformation of CoOx to metallic Co during the selective hydro-
genation of nitroarenes under 30 bar of H2, although the catalyst
The chemoselective hydrogenation of nitroarenes is an impor-
tant hydrogenation reaction for production of fine and bulk chem-
icals, with wide industrial and pharmaceutical applications [1].
Noble metals such as Au, Pt, Ru, Pd, and Ir have been proved to
be active components for hydrogenation of nitroarenes [2,3]. By
controlling the metal–support interaction and particle size, the
chemoselective hydrogenation of nitroarenes can be carried out
with high chemoselectivity [4,5]. However, considering the high
price and limited availability of noble metals, it is of interests to
develop non-noble metal catalysts for chemoselective hydrogena-
tion reactions.
In the last years, non-noble metal catalysts have attracted much
attention due to their comparable properties with those of noble
metal catalysts in photocatalysis, in electrocatalysis, and in homo-
geneous and heterogeneous catalysis [6–9]. Recently, the applica-
tion of CoOx@N-doped carbon and Fe2O3@N-doped carbon
materials for chemoselective hydrogenation of nitroarenes has
been reported [10,11]. In this work, the catalysts work under
high-pressure conditions (50 bar of H2), and the authors claim that
metal oxide nanoparticles covered by carbon layers were the active
species for the chemoselective hydrogenation and hydrogenation-
transfer reaction. In a recent paper, Wang et al. observed the in situ
was a mixture of CoOx and metallic Co [12]. As substitutes for noble
metal catalysts, the non-noble metal catalysts should work
under conditions similar to those for Pt (3–6 bar of H2) and Au
(9–15 bar of H2) catalysts. However, the reported catalysts are
working under much higher H2 pressure than noble metal
catalysts.
From the above reports, it is still not clear whether metallic Co
or CoOx is the active phase for the selective hydrogenation reaction.
Moreover, if one considers that in situ dynamic transformation of
metal species can occur under reaction conditions, the real active
species in the hydrogenation reaction may not be any of the start-
ing materials. Furthermore, the role of the layered carbon materials
in the catalytic mechanism during this catalytic hydrogenation has
not been clarified. In some works, the carbon layers can protect
metal NPs. For example, Ding et al. prepared supported Ni NPs
embedded in carbon nitride layers for hydrogenation of nitroben-
zene under strong acid conditions [13]. The carbon layers may also
block the access of substrates to the metal NPs. In some previous
works [10–12], it was proposed that the hydrogenation reactions
occur on the surface of carbon layers. However, considering the
impermeability of graphene, H2 cannot diffuse directly through
perfect graphene layers, making it even more difficult to unreveal
the role of carbon layers in hydrogenation reactions [14,15].
In the first part of this work, by following the thermal
decomposition of a Co-EDTA complex, we prepare monodispersed
⇑
Corresponding author.
E-mail address: acorma@itq.upv.es (A. Corma).
http://dx.doi.org/10.1016/j.jcat.2016.04.006
0021-9517/Ó 2016 Elsevier Inc. All rights reserved.

2
L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
Co nanoparticles with a thin carbon shell (Co@C). Compared with
conventional method for monodispersed Co nanoparticles, organic
ligands are not required here to stabilize the Co nanoparticles. The
resultant material can catalyze the chemoselective hydrogenation
of nitroarenes under mild reaction conditions similar to Au cata-
lysts (7–10 bars of H2) with high activity and selectivity (>93%).
In the second part of this work, with the help of in situ
spectroscopic characterizations, we will show that the selective
adsorption of reactants occurs on the surface of cobalt
nanoparticles. Metallic Co, instead of CoOx, is the active phase for
the hydrogenation reaction, and the role of the carbon layers is
to protect the metallic Co NPs from overoxidation.
annular dark-field detector (HAADF), which allows Z-contrast
imaging.
Field-emission scanning electron microscopy (FESEM) measure-
ment was performed with a ZEISS Ultra 55 FESEM. The solid pow-
der sample was adsorbed on conductive carbon tape.
X-ray photoelectron spectra of the catalysts were recorded with
a SPECS spectrometer equipped with a Phoibos 150MCD-9 multi-
channel analyzer using nonmonochromatic MgKa (1253.6 eV) irra-
diation. Spectra were recorded using an analyzer pass energy of
30 eV and an X-ray power of 100 W and under an operating pres-
sure of 10ꢀ9 mbar. The fresh Co@C NPs sample was reduced by H2
(7 bar) at 120 °C for 60 min in a high-pressure catalytic cell con-
nected, under ultrahigh vacuum, to the XPS analysis chamber. Peak
intensities were calculated after nonlinear Shirley-type back-
ground subtraction and corrected by the transmission function of
the spectrometer. During data processing of the XPS spectra, bind-
ing energy (BE) values were referenced to the C1s peak (284.5 eV).
CasaXPS software was used for spectra treatment [16].
2. Experiments
2.1. Preparation of Co@C NPs
The Co@C NPs were prepared through the reduction of Co-EDTA
complex by H2. The Co-EDTA complex was prepared through a
hydrothermal process. First, 6.98 g Co(NO3)2, 4.47 g Na2EDTA,
and 0.96 g NaOH were dissolved in 20 mL H2O. Then, 10 mL metha-
nol was added to the mixed aqueous solution temperature under
stirring at room temperature. After the formation of a homoge-
neous solution, 23 mL of the purple solution was transferred into
a 35 mL stainless steel autoclave, followed by static hydrothermal
processing at 200 °C for 24 h. After cooling to room temperature,
the generated precipitates were filtered and washed with deion-
ized water and acetone several times, followed by drying at
100 °C in air for 16 h. The obtained complex was denoted as Co-
EDTA. Then Co@C NPs were prepared by reduction of Co-EDTA in
H2 (50 mL/min) at 450 °C for 2 h with a ramp rate of 10 °C/min
from room temperature to 450 °C. After the H2 reduction process
at 450 °C, the sample was cooled down to room temperature in
H2 atmosphere. Then the black solid product was stored in a glass
vial in ambient environment. The ICP analysis shows that the
amount of cobalt in the Co@C NPs is P95 wt.%. The Co@C–250Air
and Co@C–450Air were prepared through the calcination of a
Co@C sample in air (50 mL/min) at 250 and 450 °C for 2 h with a
ramp rate of 5 °C/min.
Raman spectra were recorded at ambient temperature with a
785 nm HPNIR excitation laser on a Renishaw Raman spectrometer
(‘‘Reflex”) equipped with an Olympus microscope and a CCD detec-
tor. The laser power on the sample was 15 mW and a total of 20
acquisitions were taken for each spectrum.
Hydrogen/deuterium (H/D) exchange experiments were carried
out in a flow reactor at 25 and 80 °C. The feed gas consisted of
4 mL/min H2, 4 mL/min D2, and 18 mL/min argon, and the total
weight of catalyst was 180 mg. Reaction products (H2, HD, and
D2) were analyzed with a mass spectrometer (Omnistar, Balzers).
The Co@C sample was reduced in situ at 450 °C for 2 h with a ramp
rate of 10 °C/min from room temperature to 450 °C. Then the tem-
perature was decreased to 25 °C and, once stabilized, the H2 feed
was changed to the reactant gas composition. The temperature
was increased to 80 °C and maintained for 1 h.
DRIFT spectra were recorded at room temperature with a Nexus
8700 FTIR spectrometer using a DTGS detector at 4 cmꢀ1 resolu-
tion. Prior to the adsorption experiments, the ex situ reduced sam-
ples were reduced in situ at 120 °C under H2 for 2 h. After
activation, the sample was evacuated at 10ꢀ2 mbar and nitroben-
zene and/or styrene was adsorbed until sample saturation, fol-
lowed by evacuation (10ꢀ2 mbar) in order to remove physisorbed
species. A commercial DRIFT cell (SPECAC) was used. Spectra were
acquired in Kubelka–Munk units.
2.2. Catalytic studies
Powder X-ray diffraction (XRD) was performed in a HTPhilips
X’Pert MPD diffractometer equipped with a PW3050 goniometer
using CuKa radiation and a multisampling handler.
The chemoselective hydrogenation of nitroarenes was per-
formed in batch reactors. The reactant, internal standard (dode-
cane), solvent (toluene or THF), and powder catalyst, as well as a
magnetic bar, were added into the batch reactor. After the reactor
was sealed, air was purged by flushing two times with 10 bar of
hydrogen. Then the autoclave was pressurized with H2 to the cor-
responding pressure. The stirring speed was kept at 800 rpm and
the size of the catalyst powder was below 0.02 mm to avoid either
external or internal diffusion limitation. Finally, the batch reactor
was heated to the target temperature. For the kinetic studies,
3. Results and discussions
3.1. Catalyst preparation and characterization
The Co@C NPs were prepared by thermal decomposition of Co-
EDTA complex under H2 at 450 °C. The morphological characteriza-
tion of Co-EDTA complex can be found in the Supporting Informa-
tion (see Fig. S1). The morphology of Co@C NPs was characterized
by FESEM and TEM. As shown in Fig. S2 and Fig. 1a, the Co@C mate-
rial is formed by monodisperse metal NPs ranging from ca. 20 to ca.
150 nm. Those Co NPs are covered by carbon layers with thickness
ranging from ca. 1 to ca. 10 nm (Fig. 1b). From the HRTEM images
(Fig. 1c and d), the lattice fringe of metallic Co can be seen [17]. A
further look to the surface structures of the Co NPs makes it possi-
ble to observe the crystal lattice fringes of Co3O4, as shown in
Fig. S3 [18]. Therefore, one can assume, as a first approximation,
that the Co NP can be formed by a core–shell structure with metal
Co at the core and CoOx as the shell. In order to know the chemical
composition of the Co@C sample, STEM-HADDF elemental
50
lL of the mixture was taken out for GC analysis at different
reaction times. For the scope studies, 100
lL of the mixture was
taken out for GC analysis. The products were also analyzed by
GC–MS.
2.3. Characterization techniques
Samples for electron microscopy studies were prepared by
dropping the suspension of Co@C NPs using CH2Cl2 as the solvent
directly onto holey-carbon-coated nickel grids. All the measure-
ments were performed in a JEOL 2100F microscope operating at
200 kV both in transmission (TEM) and scanning-transmission
modes (STEM). STEM images were obtained using a high-angle

L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
3
Fig. 1. Morphological characterizations of Co@C NPs. (a and b) Low-magnification TEM image of Co@C. The size of Co NPs ranges from ca. 20 to ca. 150 nm. Monodisperse Co
NPs are separated by carbon layers, as shown in (b). (c and d) HRTEM images of Co NPs covered with carbon layers. (e and f) Profile of the carbon layers in selected areas in (c)
and (d), respectively. The distance between two carbon layers is about 0.36–0.37 nm. (g) STEM-HADDF image of Co@C NPs. (h–j) Elemental mapping of Co, C, and O. (k)
Schematic illustration of the structure of Co@C NPs. Blue balls stand for metallic Co and yellow balls stand for CoOx. The surface of Co NP is mainly made up of CoOx, while the
core is metallic Co. The green circle surrounding the particle is the carbon layers with cracks. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
mapping was performed and the results are displayed in Fig. 1g–j.
They confirm that the Co NPs are covered by carbon layers. Mean-
while, oxygen can also be found in the sample, which may come
from surface CoOx and/or from the oxygenated groups in carbon
layers. Contrast profiles of the carbon layers over Co NPs were also
obtained, as presented in Fig. 1e and f. It can be seen there that the
distance between the carbon layers is 0.34–0.37 nm, which corre-
sponds to the distance between graphene layers. The thickness of
most carbon layers is between 1 and 5 nm, with several layers of
graphene [19]. A schematic illustration of the structure of Co@C
NP is shown in Fig. 1k, indicating that the particle contains metallic
Co as the core, some CoOx patches on the surface, and carbon layers
surrounding.
From the HRTEM images (Figs. S4–S6 in the Supporting Infor-
mation), it can also be seen that the thin carbon layers over Co
NPs are not closed, and some cracks can be found, implying that
reactant molecules can have access to the Co NPs through the car-
bon layers. This can explain why the Co@C NPs could be partially
oxidized by air under ambient conditions after the preparation
procedure. In a recent work, Bao and his co-workers prepared a
sample with CoNi alloy NPs totally encapsulated by several layers
of graphene [20]. Those CoNi NPs were not soluble in a strong acid
environment because of the protection of the carbon layers. How-
ever, in our case, Co NPs covered with cracked carbon layers can be
almost totally dissolved in aqueous H2SO4 (photographs of the dis-
solving process can be seen in Fig. S7 in the Supporting Informa-
tion), suggesting that Co@C NPs are not totally covered by carbon
layers.
In general, due to the instability of metallic Co NPs, their prepa-
ration for heterogeneous catalytic application still remains a chal-
lenge. So far, metallic Co NPs have usually been prepared through a
wet-chemistry method [21–23]. In these methods, Co NPs are
capped with organic ligands to protect them from agglomeration
and oxidation by air, but, at the same time, these long-chain
organic ligands decrease the activity of Co NPs. Furthermore, Co
NPs prepared by wet-chemistry methods are usually not stable
under ambient conditions because of the oxidation of Co by air.
In contrast, with the protection of carbon layers, the Co NPs pre-
pared by our method can be stored under ambient conditions for
over one month without obvious morphological changes. Only
the oxidation of Co NPs at the surface occurs, and CoOx patches
are formed that can easily be reduced by H2 to metallic Co. Com-
pared with previous reported pyrolysis methods (calcination at
800 °C), well-defined Co@C NPs with carbon layers can be directly
synthesized at relatively lower temperatures. These air-stable
Co@C NPs with CoOx patches can be prepared on a large scale (sev-
eral grams) in a facile way.
The bulk and surface properties of the Co@C NPs were also
investigated. The XRD pattern of Co@C NPs is shown in Fig. 2a.
Only the X-ray diffraction patterns of metallic Co can be observed
with no X-ray diffraction peaks corresponding to other species,
although the formation of CoOx patches could be observed by
HRTEM. The fact that CoOx is not detected by XRD should be due
to the small number of CoOx patches in the Co@C sample. In the
case of metallic Co, besides the cubic Co phase (PDF code 96-
900-8467), a small amount of hexagonal Co phase (PDF code 96-
900-8493) can also be observed in the XRD pattern. For metallic
Co, the hexagonal close-packed (hcp) phase is more stable at lower
temperature than the face-centered cubic (fcc) phase [24]. Since
the Co-EDTA complex was decomposed and reduced at 450 °C, it
is not surprising to see that both phases can exist [25].
The surface structures of Co@C NPs, as well as the carbon layers,
were also studied by Raman spectroscopy. As indicated in Fig. 2b,
vibration modes of Co3O4 (F2g, Eg, and A1g) can be observed [26].
Besides, the typical Raman signals of layered carbon, bands at
1318 cmꢀ1 and 1595 cmꢀ1, can be observed, which corresponding
to the D and G bands, respectively [27]. No peak related to the 2D
band is detected between 2500 and 2800 cmꢀ1. The intensity ratio
of the G band (IG) to the D band (ID) is ca. 0.8, suggesting that there
is a large percentage of disorder in the structured carbon in Co@C
NPs [28]. Combining these results and those from HRTEM images,
it can be speculated that the degree of graphitization is relatively
low in the carbon layers around Co NPs. Therefore, based on the
structural characterizations, we can propose that Co@C NPs have
core–shell structures with metallic Co as the core, CoOx on the sur-
face of Co crystallites, and thin layered carbon as the shell.
X-ray photoelectron spectroscopy (XPS) was employed to study
the chemical state of Co in the as-prepared Co@C NPs. The Co@C
NPs were reduced by H2 at 7 bar and 120 °C (that condition will
also be used in the hydrogenation reaction) within a prereactor
integrated with the apparatus, and the sample is maintained iso-
lated from any external contact during the entire process. The

4
L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
Fig. 2. Structural characterizations of Co@C NPs. (a) XRD pattern of fresh Co@C NPs, (b) Raman spectrum of Co@C NPs, (c) XPS spectra of Co2p region, and (d) Co L3VV Auger
spectra of the fresh Co@C NPs and the sample after ex situ reduction by 7 bar of H2 at 120 °C.
XPS spectra of the Co2p region and the Auger spectra of Co L3VV
are shown in Fig. 2c and d, respectively. In the as-prepared Co@C
sample, only CoOx can be observed, which can be ascribed to the
CoOx patches on the surfaces of Co NPs [29]. After H2 reduction
treatment (7 bar of H2 at 120 °C), the CoOx is totally transformed
into metallic Co, as confirmed by both Co2p XPS and Co L3VV Auger
spectra. In the fresh as-prepared catalyst, N is presented in the
sample, as shown by elemental mapping (see Fig. S8 in the Sup-
porting Information). However, after the ex situ reduction treat-
ment, no N can be found in the sample by XPS, indicating that
there are practically no N species in the working catalyst.
With the characterization results presented up to now, we can
say that during preparation of Co@C NPs, Co NPs of 20–150 nm
were generated, which have a metallic Co core with CoOx on the
surface, surrounded by disordered carbon layers. Moreover, under
reduction conditions, which are the same that are used for per-
forming the catalytic test, the CoOx patches can be totally reduced
to metallic Co.
conditions than those reported for the hydrogenation of
3-nitrostyrene with Co- and Fe-based catalysts [10,11]. The reac-
tion conditions and catalytic results are given in Fig. 3. Ii can be
seen there that fresh Co@C NPs are active and selective catalyst
for the hydrogenation of 3-nitrostyrene. When the conversion of
3-nitrostyrene is 95%, the selectivity to 3-aminostyrene is 93%,
which is comparable to the values obtained with noble metal
catalysts [4,5]. The initial TOF calculated based on surface Co atoms
is ca. 8.2 hꢀ1 according to the particle size distribution of Co@C
NPs. Notably, a reaction induction period of about 20–30 min can
be observed in the kinetic curve (Fig. S9 in the Supporting Informa-
tion) starting with the fresh catalyst that contains a core of metallic
cobalt and a shell of CoOx. In accord with XPS results, this induction
period could be caused by the time required for the reduction of
surface CoOx to metallic Co under the reaction conditions. As pre-
sented in Fig. S10 in the Supporting Information, the morphology
and particle size distribution after reduction under reaction condi-
tions are similar to those for the fresh Co@C sample, suggesting
that no structural changes or agglomeration have occurred. HRTEM
images of the used sample also confirm that metallic Co NPs are
still surrounded by carbon layers. Therefore, the above morpholog-
ical characterization indicates that the nanoscale structures of the
Co@C NPs sample are stable under reaction conditions.
3.2. Catalytic results
At this point, the chemoselective hydrogenation of 3-
nitrostyrene was chosen as a model reaction to study the catalytic
properties of Co@C NPs and to see if the non-noble metal cobalt
nanoparticles can behave as a chemoselective catalyst. To test this,
the same reaction conditions that showed chemoselective hydro-
genation when noble metal catalysts were used, i.e., 120 °C and
7 bar of H2, were selected [4,5]. Notice that these are much milder
In a recent work, Zhang et al. have demonstrated that atomi-
cally dispersed Co species in the carbon matrix can serve as the
active sites for oxidation reactions [30]. Here, in order to exclude
the role of atomically dispersed Co species in the hydrogenation
reaction, we have also measured the residual solid after the acid

L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
5
Fig. 3. (a) Reaction scheme of the chemoselective hydrogenation of 3-nitrostyrene. Reaction conditions: 0.5 mmol 3-nitrostyrene, 30 mg Co@C NPs as catalyst, 2 mL toluene
as solvent, 30 L dodecane as internal standard. (b) Catalytic performance of Co@C, Co@C–250Air, and Co@C–450Air in chemoselective hydrogenation of 3-nitrostyrene to 3-
aminostyrene. The black and red legends correspond to conversion and selectivity, respectively. (c) Catalytic performance of Co@C–H2, Co@C–250Air–H2, and Co@C–450Air–
2 in chemoselective hydrogenation of 3-nitrostyrene to 3-aminostyrene. The black and red legends correspond to conversion and selectivity, respectively. (For interpretation
l
H
of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4. H2–D2 exchange and DRIFTS adsorption experiments. (a) Enhancement of the ion current in the mass signal of HD during the H2–D2 exchange experiment over Co@C–
H2, Co@C–250Air, and Co@C–250Air–H2 NPs at 80 °C. (b) DRIFTS of adsorbed nitrobenzene, (c) DRIFTS of adsorbed styrene, and (d) DRIFTS of co-adsorbed nitrobenzene and
styrene on Co@C NPs and Co@C–250Air sample.
leaching treatment (as shown in Fig. S7 in the Supporting Informa-
tion). No activity can be observed in the hydrogenation of 3-
nitrostyrene, suggesting that the metallic Co NPs should be the
active sites for hydrogenation reactions.
The recyclability of Co@C NPs in the hydrogenation of 3-
nitrostyrene was also tested. Since metallic Co NPs are paramag-
netic, the separation of the solid catalyst from the liquid was very
facile with the help of a magnetic bar [31,32] (see Fig. S11 in the

6
L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
Supporting Information). After being washed several times with
toluene, the catalyst can be reused without further treatment,
and the Co@C catalyst shows good activity and selectivity for five
recycles (see Table S1 in the Supporting Information).
3.3. Role of the carbon layers in the catalyst
At this point, we know that the reduced Co@C NPs are active
and chemoselective for reduction of substituted nitroarenes into
corresponding anilines under mild reaction conditions. To investi-
gate if the carbon layers play an intrinsic role in the chemoselective
hydrogenation reaction, calcination of the Co@C NPs at 250 and
450 °C in air was performed to remove the carbon layers on the
nanoparticles. The samples after calcination in air are denoted as
Co@C–250Air and Co@C–450Air, respectively. From the Raman
spectra (Fig. S12 in the Supporting Information), we find that in
Co@C–250Air, bands at 1318 and 1595 cmꢀ1 are still visible, indi-
cating that part of the carbon layers is still preserved. In Co@C–
450Air, the above Raman bands are already very weak, suggesting
that almost all the carbon layers have already been removed. As
shown in Fig. 3b, it is possible to see that activity and selectivity
drop when Co@C–250Air is used, while Co@C–450Air shows no
activity in the hydrogenation reaction. The product distribution
in the hydrogenation of 3-nitrostyrene with a Co@C–250Air sam-
ple as the catalyst is shown in Fig. S13. To explain these changes,
TEM is used to clarify the structural transformation occurring with
Co@C NPs after calcination in air. As shown in Fig. S14, metallic Co
NPs crack into smaller CoOx NPs after calcination in air at 250 °C,
which has also been observed in other works [33]. The crystal lat-
tice fringes of CoOx NPs can be observed in the HRTEM images,
though there are still some particles with metallic Co phase in
the Co@C–250Air sample. Some carbon layers can still be found
on Co-CoOx and CoOx NPs in Co@C–250Air, as confirmed by Raman
spectra. It should be noted that part of the carbon layers were
burned during the calcination in air, resulting in the formation of
naked Co and CoOx NPs.
In the Co@C–450Air sample (shown in Fig. S15 in the Support-
ing Information), Co NPs are totally converted into Co3O4 NPs. The
phase composition of the samples after calcination in air can also
be measured by selective area electron diffraction (SAED). As can
be seen in Fig. S16, metallic Co NPs transform into a mixture of
Co and Co3O4 when calcined at 250 °C, and to pure Co3O4 when cal-
cined at 450 °C. Considering the significant decrease of activity
observed when the Co@C sample was calcined at 250 °C and even
more when the temperature was 450 °C, it can be deduced that
CoOx or Co3O4 cannot be the catalytically active phase.
Co@C, Co@C–250Air, and Co@C–450Air NPs were reduced in the
batch reactor under higher temperature and H2 pressure (200 °C,
10 bar H2) before the hydrogenation reaction to reduce the CoOx
species to metallic Co. After the reduction pretreatment, 3-
nitrostyrene was added and the hydrogenation reaction was per-
formed at 7 bar of H2 and 120 °C. As presented in Fig. 3c, Co@C–
H2 gives a kinetic curve similar to that for pristine Co@C, except
for the disappearance of the induction period due to the reduction
pretreatment. However, the situation changes a lot in the case of
Co@C–250Air and Co@C–450Air. The Co@C–250Air–H2 sample
shows higher activity than pristine Co@C NPs and very good selec-
tivity. This would indicate that the metallic Co NPs formed after
reduction of Co@C–250Air should be responsible for the high activ-
ity and selectivity in hydrogenation reaction observed with
Co@C–250Air–H2. What is more, the reduction pretreatment also
has a significant effect on Co@C–450Air. After H2 reduction, the
Co@C–450Air–H2 sample shows moderate activity and high selec-
tivity, which further indicates that the chemoselective hydrogena-
tion of 3-nitrostyrene is related to metallic Co and the role of
Fig. 5. Proposed reaction pathways of hydrogenation of 3-nitrostyrene on Co@C
NPs with coverage of carbon layers (a) and defective coverage of carbon layers (b).
In schematic illustration (a), the chemoselective hydrogenation of ANO2 groups can
be divided into the following steps: the diffusion of H2 from the gas phase up to the
surface of metallic Co NPs (1), activation of H2 (2), adsorption of the nitro aromatic
(3), and H-transfer from Co NPs to ANO2 groups (4) for selective hydrogenation of
ANO2 groups. In the scheme presented in (b), a sample with Co3O4 patches after
removal of part of the carbon layers, hydrogenation of ANO2 and @C groups can be
divided into following steps: diffusion of H2 to the surface of metallic Co NPs (1),
activation of H2 (2), hydrogen transfer (3), and hydrogenation of both ANO2 and
C@C groups (4). In schematic illustration (c), for the in situ reduced sample after
removal of part of carbon layers, the chemoselective hydrogenation of ANO2 groups
can be divided into several steps, including the diffusion of H2 to the surface of
metallic Co NPs (1), activation of H2 (2), H-transfer from Co NPs to ANO2 groups (3),
and selective hydrogenation of ANO2 groups (4).
carbon layers in Co@C NPs is mainly to protect Co NPs from
overoxidation by air.
3.4. In situ adsorption spectroscopy
In order to explain the activity and high selectivity of Co@C NPs,
H2–D2 exchange and the adsorption of substrates on Co@C NPs
followed by diffuse reflectance infrared Fourier transform

L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
7
spectroscopy (DRIFTS) were carried out. The H2–D2 exchange rates
on Co NPs and Co@C–250Air (with and without H2 prereduction)
are shown in Fig. 4a. H2–D2 exchange can be observed on Co@C
NPs. Interestingly, as displayed in Fig. S17 in the Supporting Infor-
mation, a large amount of H2 was released from Co NPs when the
atmosphere was changed from H2 to Ar after the in situ reduction
process. The H2 released should come from the H2 adsorbed by Co
NPs [34,35] It appears then that H2 diffuses through the potential
cracks present in carbon layers and is activated on the surface of
Co NPs. In the case of Co@C–250Air, no H2–D2 exchange can be
observed when the sample is not prereduced. After H2 pretreat-
ment, the H2–D2 exchange rate is higher on Co@C–250Air–H2 than
on Co@C, suggesting that more Co sites are exposed after the
removal of carbon layers and the reduction of CoOx. The H2–D2
exchange experiments further confirm that metallic Co are the
active sites for H2 activation and carbon layers are not playing a
direct role in the catalytic process.
We have found in previous work that the way that the reactant
is adsorbed on metal catalysts can determine the chemoselectivity
during hydrogenation of substituted nitroaromatics [36]. Thus, in
order to explain different catalytic behavior of Co@C NPs and
Co@C–250Air samples for chemoselective hydrogenation of 3-
nitrostyrene, we have performed in situ IR adsorption experiments
by DRIFTS, with nitrobenzene and styrene as probe molecules. For
comparison, the IR spectra of nitrobenzene and styrene on KBr
substrate have also been measured (see Fig. S18 in the Supporting
Information). The adsorption spectra of the probe molecules on
Co@C and Co@C–250Air are shown in Fig. S19. The IR adsorption
spectra of nitrobenzene on Co@C and Co@C–250Air samples show
wide IR bands between 1600 and 1250 cmꢀ1 in both samples (see
Fig. 4b). The vibration bands of ANO2 groups can be observed at ca.
1525 and 1360 cmꢀ1 [37,38]. The intensity of the IR bands is sim-
ilar on both samples, suggesting that the adsorption spectra of
Co@C and Co@C–250Air for nitrobenzene are similar. On the other
hand, the adsorption spectra of styrene on Co@C and Co@C–250Air
(Fig. 4c) are significantly different. Indeed, as can be seen in Fig. 4c,
the IR band corresponding to the aromatic ring can be observed
between 1580 and 1450 cmꢀ1, with the vibration mode of the
C@C bond at ca. 1625 cmꢀ1 [39]. Co@C–250Air shows much stron-
ger adsorption of styrene than the Co@C sample, indicating that the
adsorption of styrene onto CoOx NPs will be enhanced when metal-
lic Co NPs are oxidized. We have also studied the co-adsorption of
nitrobenzene and styrene on Co@C and Co@C–250Air by DRIFTS. As
can be seen in Fig. 4d, only the adsorption of nitrobenzene can be
observed on Co@C, while strong adsorption of styrene occurs on
Co@C–250Air. In the case of Co@C–450Air, only the adsorption of
styrene can be observed after the evacuation (see Fig. S20 in the
Supporting Information). Therefore, based on the IR adsorption
results, we can conclude that the chemical states of Co have a
significant influence on the adsorption properties of substrate
Table 1
Scope of the hydrogenation of nitroarenes catalyzed by Co@C NPs.
Entry
1
Substrate
Product
Time (h)
6
H2 pressure (bar)
7
Temp. (°C)
Con. (%)
93
Sel. (%)
95
120
2y
12
10
100
95
95
3
4
5
10
10
140
140
99
95
97
99
10
5
6
8
10
10
140
140
99
99
97
95
12
7
8
10
10
8
8
120
120
99
91
99
93
9
15
10
140
98
95
10
15
10
10
10
140
140
99
97
95
99
11à
12
10
10
140
97
99
Notes: Reaction conditions: Co@C NPs is 30 mg. 0.5 mmol nitroarenes, 30
lL dodecane as the internal standard, 2 mL toluene as the solvent.
y
100 mg Co@C NPs as the catalyst.
à
2 mL THF as the solvent.

8
L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
molecules in Co@C NPs. For metallic Co NPs, only ANO2 groups will
be preferentially adsorbed. In contrast, when the Co NPs are par-
tially oxidized, the C@C instead of ANO2 groups will be preferen-
tially adsorbed. The selective adsorption observed in above
experiments can explain the differences for the chemoselective
hydrogenation of 3-nitrostyrene on metallic and partially oxidized
cobalt nanoparticles.
small cracks. H2 can diffuse through the cracks in the thin carbon
layers and be activated at room temperature. The existence of car-
bon layers can protect the Co NPs from overoxidation by air. This
type of material shows good activity and selectivity for hydrogena-
tion of nitroarenes, with wide scope and good tolerance. It can also
give high yields of heteroaromatic amines. The adsorption proper-
ties of substrate molecules will be affected by the chemical state of
Co. Combining with in situ IR adsorption experiments and results
from electron microscopy, a reaction mechanism is proposed in
which the important role of metallic cobalt and the negative role
of CoOx in the selective hydrogenation reaction is shown. Nondi-
rect involvement of the carbon in the reaction is also proposed.
The role of carbon layers is as protection of the Co NPs from getting
deeply oxidized. This work provides new insight into design of
non-noble metal catalysts for heterogeneous catalytic applications.
3.5. Catalytic mechanism
It appears then that unless Co nanoparticles are reduced to
metallic nanoparticles, their activity and selectivity will be rela-
tively low. In the case where partial (surface) oxidation occurs,
not only is the hydrogenation activity low, but also it will favor
adsorption of styrene versus nitrobenzene, which can further favor
the loss of chemoselectivity during the hydrogenation of 3-
nitrostryrene.
Acknowledgments
Based on the above catalytic results and spectroscopic charac-
terizations, we propose a reaction mechanism of chemoselective
hydrogenation of 3-nitrostyrene on Co@C NPs covered with disor-
dered carbon layers. Three cases with different oxidation states of
Co and coverages of carbon layers are considered. As shown in
Fig. 5, the first step in the hydrogenation reaction catalyzed by
both Co@C and Co@C–250Air is the diffusion of H2 to the surface
of Co NPs under the cover of carbon layers. Then H2 will be acti-
vated by Co NPs and form active H species. The third step will be
different for Co@C and Co@C–250Air due to their different adsorp-
tion properties to the substrate molecules. In the case of Co@C
(Fig. 5a), 3-nitrostyrene will be adsorbed onto metallic Co prefer-
entially through the ANO2 groups, which will be reduced by the
active H species formed on Co NPs. The adsorption of C@C bond
on the Co@C sample is relatively weak. However, in the case of
Co@C–250Air (Fig. 5b), both C@C and ANO2 groups can have direct
access to the surface of Co NPs. As a consequence, both ANO2 and
C@C groups can be hydrogenated by H2, resulting in lower selectiv-
ity. When Co@C–250Air is reduced by H2 before the catalytic test
(Fig. 5c), more metallic Co sites will selectively absorb ANO2
groups instead of C@C bonds, leading to high activity and
selectivity.
L.L. thanks ITQ for a contract. The European Union is also
acknowledged by ERC-AdG-2014-671093-SynCatMatch. The
authors also thank the Microscopy Service of UVP for kind help
with TEM and STEM measurements.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2016.04.006.
References
[1] H.-U. Blaser, U. Siegrist, H. Steiner (Eds.), Aromatic Nitro Compounds: Fine
Chemicals Through Heterogeneous Catalysis, Wiley-VCH, Weinheim,
Germany, 2001.
[2] A. Corma, P. Serna, Chemoselective hydrogenation of nitro compounds with
supported gold catalysts, Science 313 (2006) 332–334.
[3] H.-U. Blaser, H. Steiner, M. Studer, Selective catalytic hydrogenation of
functionalized nitroarenes: an update, ChemCatChem 1 (2009) 210–221.
[4] A. Corma, P. Serna, P. Concepcion, J.J. Calvino, Transforming nonselective into
chemoselective metal catalysts for the hydrogenation of substituted
nitroaromatics, J. Am. Chem. Soc. 130 (2008) 8748–8753.
[5] P. Serna, M. Boronat, A. Corma, Tuning the behavior of Au and Pt catalysts for
the chemoselective hydrogenation of nitroaromatic compounds, Top. Catal. 54
(2011) 439–446.
3.6. Scope of the catalyst
[6] J. Ran, J. Zhang, J. Yu, M. Jaroniec, S.Z. Qiao, Earth-abundant cocatalysts for
semiconductor-based photocatalytic water splitting, Chem. Soc. Rev. 43 (2014)
7787–7812.
[7] H.R. Byon, J. Suntivich, Y. Shao-Horn, Graphene-based non-noble-metal
catalysts for oxygen reduction reaction in acid, Chem. Mater. 23 (2011)
3421–3428.
[8] R.V. Jagadeesh, H. Junge, M. Beller, Green synthesis of nitriles using non-noble
metal oxides-based nanocatalysts, Nat. Commun. 5 (2014) 4123.
[9] B. Su, Z.C. Cao, Z.J. Shi, Exploration of earth-abundant transition metals (Fe Co,
and Ni) as catalysts in unreactive chemical bond activations, Acc. Chem. Res.
48 (2015) 886–896.
[10] F.A. Westerhaus, R.V. Jagadeesh, G. Wienhofer, M.M. Pohl, J. Radnik, A.E.
Surkus, J. Rabeah, K. Junge, H. Junge, M. Nielsen, A. Bruckner, M. Beller,
Heterogenized cobalt oxide catalysts for nitroarene reduction by pyrolysis of
molecularly defined complexes, Nat. Chem. 5 (2013) 537–543.
[11] R.V. Jagadeesh, A.E. Surkus, H. Junge, M.M. Pohl, J. Radnik, J. Rabeah, H. Huan, V.
Schunemann, A. Bruckner, M. Beller, Nanoscale Fe2O3-based catalysts for
selective hydrogenation of nitroarenes to anilines, Science 342 (2013) 1073–
1076.
The scope of Co@C NPs in hydrogenation of substituted
nitroarenes was investigated by reacting substituted nitrobenzene
with halogens, amides, and ester groups. In all cases, activity and
chemoselectivity were high under mild reaction conditions
(Table 1). Selective hydrogenation of ANO2 groups can be per-
formed in the presence of groups such as olefins, alkynes, and car-
bonyls under much lower H2 pressure (7–10 bar) than in previous
reports (30–50 bar) [10–12]. As can be seen in Table 1, all the sub-
strates can be selectively transformed with over 93% selectivity.
Most of the entries show over 90% yield for both electron-
deficient and electron-rich substituents. Co@C NPs are also active
and selective for nitroarenes, with easily reducible moieties includ-
ing alkynes, ketones, and halogens. In previous work, the yields of
heteroaromatic amines were not very high (from 53% to 75%). With
Co@C NPs, we can get much higher yields of heteroaromatic ami-
nes (Entries 9 and 10 in Table 1). Thus, Co@C NPs are superior cat-
alysts for chemoselective hydrogenation of nitroarenes under mild
conditions.
[12] Z. Wei, J. Wang, S. Mao, D. Su, H. Jin, Y. Wang, F. Xu, H. Li, Y. Wang, In situ-
generated Co0-Co3O4/N-doped carbon nanotubes hybrids as efficient and
chemoselective catalysts for hydrogenation of nitroarenes, ACS Catal. 5 (2015)
4783–4789.
[13] T. Fu, M. Wang, W. Cai, Y. Cui, F. Gao, L. Peng, W. Chen, W. Ding, Acid-resistant
catalysis without use of noble metals: carbon nitride with underlying nickel,
ACS Catal. 4 (2014) 2536–2543.
[14] V. Berry, Impermeability of graphene and its applications, Carbon 62 (2013) 1–
10.
[15] J.S. Bunch, S.S. Verbridge, J.S. Alden, A.M. van der Zande, J.M. Parpia, H.G.
Craighead, P.L. McEuen, Impermeable atomic membranes from graphene
sheets, Nano Lett. 8 (2008) 2458–2462.
4. Conclusions
In this work, we present a strategy for the synthesis of Co@C
NPs. Metallic Co NPs are covered with thin carbon layers with
[16] N. Fairley, CasaXPS Version 2.3.14, Casa Software Ltd., 2008.

L. Liu et al. / Journal of Catalysis 340 (2016) 1–9
9
[17] W. Zhong, H. Liu, C. Bai, S. Liao, Y. Li, Base-free oxidation of alcohols to esters at
room temperature and atmospheric conditions using nanoscale co-based
catalysts, ACS Catal. 5 (2015) 1850–1856.
[18] M.R. Ward, E.D. Boyes, P.L. Gai, In situ aberration-corrected environmental
TEM: reduction of model Co3O4 in H2 at the atomic level, ChemCatChem 5
(2013) 2655–2661.
catalysts for CO2 hydrogenation: synthesis, characterization, and catalytic
reactions, Nano Lett. 12 (2012) 3091–3096.
[30] L. Zhang, A. Wang, W. Wang, Y. Huang, X. Liu, S. Miao, J. Liu, T. Zhang, Co–N–C
catalyst for C–C coupling reactions: on the catalytic performance and active
sites, ACS Catal. 5 (2015) 6563–6572.
[31] D. Wang, D. Astruc, Fast-growing field of magnetically recyclable
nanocatalysts, Chem. Rev. 114 (2014) 6949–6985.
[32] M.B. Gawande, P.S. Branco, R.S. Varma, Nano-magnetite (Fe3O4) as a support
for recyclable catalysts in the development of sustainable methodologies,
Chem. Soc. Rev. 42 (2013) 3371–3393.
[33] H. Wang, C. Chen, Y. Zhang, L. Peng, S. Ma, T. Yang, H. Guo, Z. Zhang, D.S. Su, J.
Zhang, In situ oxidation of carbon-covered cobalt nanocapsules creates highly
active cobalt oxide catalysts for hydrocarbon combustion, Nat. Commun. 6
(2015) 7181.
[34] C.H. Bartholomew, R.C. Reuel, Cobalt–support interactions: their effects on
adsorption and carbon monoxide hydrogenation activity and selectivity
properties, Ind. Eng. Chem. Prod. Res. Dev. 24 (1985) 56–61.
[19] L. Qiu, J.Z. Liu, S.L. Chang, Y. Wu, D. Li, Biomimetic superelastic graphene-based
cellular monoliths, Nat. Commun. 3 (2012) 1241.
[20] J. Deng, P. Ren, D. Deng, X. Bao, Enhanced electron penetration through an
ultrathin graphene layer for highly efficient catalysis of the hydrogen
evolution reaction, Angew. Chem. Int. Ed. 54 (2015) 2100–2104.
[21] S. Sun, C.B. Murray, Synthesis of monodisperse cobalt nanocrystals and their
assembly into magnetic superlattices, J. Appl. Phys. 85 (1999) 4325.
[22] S. Guo, S. Zhang, L. Wu, S. Sun, Co/CoO nanoparticles assembled on graphene
for electrochemical reduction of oxygen, Angew. Chem. Int. Ed. 51 (2012)
11770–11773.
[23] Y. Lu, X. Lu, B.T. Mayers, T. Herricks, Y. Xia, Synthesis and characterization of
magnetic Co nanoparticles: a comparison study of three different capping
surfactants, J. Solid State Chem. 181 (2008) 1530–1538.
[35] I.C. Yates, C.N. Satterfield, Intrinsic kinetics of the Fischer–Tropsch synthesis
on a cobalt catalyst, Energy Fuels 5 (1991) 168–173.
[24] X.Q. Zhao, S. Veintemillas-Verdaguer, O. Bomati-Miguel, M.P. Morales, H.B. Xu,
Thermal history dependence of the crystal structure of Co fine particles, Phys.
Rev. B 71 (2005).
[36] M. Boronat, P. Concepcion, A. Corma, S. Gonzalez, F. Illas, P. Serna, A molecular
mechanismfor the chemoselective hydrogenation of substituted nitroaromatics
with nanoparticles of gold on TiO2 catalysts: a cooperative effect between gold
and the support, J. Am. Chem. Soc. 129 (2007) 16230–16237.
[37] I. Ahmad, T.J. Dines, C.H. Rochester, J.A. Anderson, IR study of nitrotoluene
adsorption on oxide surfaces, Faraday Trans. 92 (1996) 3225.
[38] C.A. Koutstaal, P.A.J.M. Angevaare, E.J. Grootendorst, V. Ponec, On the
adsorption and surface reactions of nitro- and nitrosobenzene on oxides: an
IR study, J. Catal. 141 (1993) 82–93.
[25] V.C.A. de la Peña O0Shea, P.R.R. de la Piscina, N. Homs, G. Aromí, J.L.G. Fierro,
Development of hexagonal closed-packed cobalt nanoparticles stable at high
temperature, Chem. Mater. 21 (2009) 5637–5643.
[26] V.G. Hadjiev, M.N. Iliev, I.V. Vergilov, The Raman spectra of Co3O4, J. Phys. C
Solid State Phys. 21 (1988) L199–L201.
[27] A.C. Ferrari, J. Robertson, Raman spectroscopy of amorphous, nanostructured,
diamond-like carbon, and nanodiamond, Phil. Trans. R. Soc. A 362 (2004)
2477–2512.
[28] Y. Wang, D.C. Alsmeyer, R.L. McCreery, Raman spectroscopy of carbon materials:
structural basis of observed spectra, Chem. Mater. 2 (1990) 557–563.
[29] V. Iablokov, S.K. Beaumont, S. Alayoglu, V.V. Pushkarev, C. Specht, J. Gao, A.P.
Alivisatos, N. Kruse, G.A. Somorjai, Size-controlled model Co nanoparticle
[39] B. Bachiller-Baeza, J.A. Anderson, FTIR and reaction studies of styrene and
toluene over silica–zirconia-supported heteropoly acid catalysts, J. Catal. 212
(2002) 231–239.