Identifying active sites of CoNC/CNT from pyrolysis of molecularly defined complexes for oxidative esterification and hydrogenation reactions

Article abstract of DOI:10.1039/c5cy01349f

As an emerging catalyst based on earth-abundant metals, Co₃O₄ supported on carbon materials, derived from the pyrolysis of metal salts and nitrogen-containing ligands, shows excellent performance in hydrogenation, hydrogenated coupling, oxidative esterification and oxidation of amines. Herein, we unravel the real active sites of this catalyst through a combination of XRD, XPS, HRTEM, EELS and poisoning experiments with sulfur-containing compounds. The oxidative esterification of benzyl alcohol, hydrogenation of nitrobenzene and hydrogenated coupling of nitrobenzene with benzaldehyde were used as probe reactions. By comparing the catalytic performance before and after HCl washing, it was demonstrated that particulate Co₃O₄ has a marginal effect on the catalysis, and the highly dispersed CoN_x on carbon nanotubes created during the pyrolysis is responsible for the activity. It was proposed that cobalt chelate complexes bonded to 2 to 3 nitrogens in the graphene lattice, probably like the pyridinic vacancy, might be responsible for the activity.

Full text of DOI:10.1039/c5cy01349f

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Identifying active sites of CoNC/CNT from
pyrolysis of molecularly defined complexes for
oxidative esterification and hydrogenation
reactions†
Cite this: DOI: 10.1039/c5cy01349f
a
a
a
a
Tianyuan Cheng, Hao Yu, Feng Peng, Hongjuan Wang, Bingsen Zhang b and
*
*
Dangsheng Sub
As an emerging catalyst based on earth-abundant metals, Co3O4 supported on carbon materials, derived
from the pyrolysis of metal salts and nitrogen-containing ligands, shows excellent performance in hydroge-
nation, hydrogenated coupling, oxidative esterification and oxidation of amines. Herein, we unravel the real
active sites of this catalyst through a combination of XRD, XPS, HRTEM, EELS and poisoning experiments
with sulfur-containing compounds. The oxidative esterification of benzyl alcohol, hydrogenation of nitro-
benzene and hydrogenated coupling of nitrobenzene with benzaldehyde were used as probe reactions. By
comparing the catalytic performance before and after HCl washing, it was demonstrated that particulate
Co3O4 has a marginal effect on the catalysis, and the highly dispersed CoNx on carbon nanotubes created
during the pyrolysis is responsible for the activity. It was proposed that cobalt chelate complexes bonded
to 2 to 3 nitrogens in the graphene lattice, probably like the pyridinic vacancy, might be responsible for the
activity.
Received 18th August 2015,
Accepted 28th October 2015
DOI: 10.1039/c5cy01349f
www.rsc.org/catalysis
chemically functionalized and/or decorated with metallic com-
plexes and enzymes to improve the catalytic activity or to act as
1. Introduction
Catalysts are essential for chemical transformations in the
chemical and pharmaceutical industries. Numerous molecu-
lar catalysts with earth-abundant metals as the core and suit-
able ligands have been developed and applied in diverse
chemical synthesis, such as oxidative esterification,1,2 hydro-
genation reactions,3,4 epoxidation of olefins,5–7 Diels–Alder
reaction8,9 and so on. However, homogeneous catalysts suffer
from difficulties in recycling and separation from products.
To overcome this weakness, it is highly desirable to immobi-
lize molecular catalysts on solid supports to form their
heterogeneous counterparts without losing activity and
specificity.
biosensors.16–18 Recently, a thermolysis approach has been pro-
posed by Beller M. and co-workers for synthesizing iron/cobalt
oxide catalysts on carbon black.19–26 Through the pyrolysis of
metal salts and nitrogen-containing ligands, represented by
1,10-phenanthroline, blended with carbon black, very active
catalysts have been produced for the heterogeneous hydrogena-
tion of nitroarenes,20,21 oxidation of amines to nitriles,24 oxida-
tive esterification of alcohols,19 and hydrogenated coupling of
carbonyl compounds.23,26 Heterogeneous catalysts are inexpen-
sive, easily reusable and environmentally benign.
Despite these achievements, it currently remains unclear
what exactly are the active sites of those emerging catalysts.
In one of the early publications in this field, Westerhaus F. A.
et al.21 defined the structure of a hydrogenation catalyst,
formed by pyrolysis of a cobalt-phenanthroline complex
adsorbed on activated carbon, as Co3O4 particles with a very
wide size distribution, on the surface of which three nitrogen
atoms are statistically bound to one cobalt atom. A similar
catalyst, Co3O4–N supported on Vulcan XC72R carbon black,
was proved to be active in oxidative esterification of alco-
hols.19 In a recent work by Deng J. et al., the formation of
Co0 during pyrolysis was also emphasized.27 In the case of
iron oxide catalysts produced by a similar method, their
excellent hydrogenation activity has been attributed to the
So far, various homogeneous catalysts have been immo-
bilized on resins, zeolites, metal oxides, etc. by ion
exchange,10,11 encapsulation,12,13 grafting14,15 and so on. Car-
bon materials are regarded as excellent supports, which can be
a School of Chemistry and Chemical Engineering, South China University of
Technology, Guangzhou 510640, PR China. E-mail: yuhao@scut.edu.cn;
Fax: +86 20 87114916; Tel: +86 20 87114916
b Catalysis and Materials Division, Shenyang National Laboratory for Materials
Science, Institute of Metal Research, Chinese Academy of Sciences, Shenyang
110016, PR China. E-mail: bszhang@imr.ac.cn; Tel: +86 24 83970027
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c5cy01349f
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
formation of active Fe2O3 particles surrounded by 3–5 layers
of nitrogen-doped graphene, on which FeNx structure can be
detected.20 Recently, Beller and co-workers defined these
structures as core–shell-structured metal oxide@N-doped
graphene supported on carbon materials.23,26 Nevertheless,
some fundamental issues on the nature of active sites are not
yet fully understood. It is still unclear what the role of metal
oxide is – whether and how the oxide core participates in the
catalysis, as it is encapsulated in a carbon shell. Secondly,
previous works show that active catalysts usually have a wide
size distribution, from sub-nanometer structures to large par-
ticles over 100 nm. Are there any different catalytic behaviors
on them? In addition, disclosure of other possible types of
active sites deserves attention, because many compounds
may be produced through the pyrolysis of a system con-
taining a transition metal, nitrogen and carbon, such as
oxides, nitrides, oxynitrides and carbonitrides, analogized to
the complicated cases of transition metal catalysts for oxygen
reduction reaction.28
In this work, an effort was made to identify the active sites
responsible for the excellent activity of CoNC/carbon. Carbon
nanotubes (CNTs) were used as a support to immobilize the
Co complex by pyrolysis, because the well-defined structure
and mesoporous feature of CNTs are beneficial for the unam-
biguous identification of cobalt species. The major concern
was to evaluate the role of cobalt oxide, which can be selec-
tively removed by an acid, allowing for a comparative study
of the catalysts with or without cobalt oxide. The catalysts
were subjected to the oxidative esterification of benzyl alco-
hol by molecular oxygen, the selective reduction of nitroben-
zene by hydrogen and the hydrogenated coupling of nitroben-
zene and benzaldehyde as probe reactions. Across the
spectrum of reactions, new insights into the active sites of
CoNC/CNT were revealed.
For comparison, CNTs were replaced with carbon black
(XC-72R) to synthesize the CoNC catalysts using the same
procedures. The resulting catalysts are denoted as CoNC/CB,
CoNC/CB-HCl and CoNC/CB-HNO3, respectively.
2.2 Catalyst characterizations
Brunauer–Emmett–Teller (BET) specific surface areas (SBET
)
were measured by N2 adsorption measurement at 77 K in an
ASAP 2010 analyzer. X-ray diffraction (XRD) patterns were
recorded on a Bruker D8 ADVANCE diffractometer equipped
with a rotating anode using Cu Kα radiation (40 kV, 40 mA).
X-ray photoelectron spectroscopy (XPS) was performed in a
Kratos Axis ultra (DLD) spectrometer equipped with an AlKα
X-ray source. The binding energies were referenced to the
C1s peak at 284.8 eV. High-resolution transmission electron
microscopy (HRTEM) images and electron energy loss
spectroscopy (EELS) spectra were recorded by using a FEI
Tecnai G2 F20 microscope. Raman spectra were recorded on
a LabRAM Aramis micro Raman spectrometer with a 633 nm
wavelength laser and a 2 μm spot size. Elemental analysis
was conducted using a Vario EL III (Elementar, Germany).
The contents of element Co were measured using atomic
absorption spectroscopy (Hitachi Z-5000). H2S temperature-
programmed desorption (H2S-TPD) was carried out in a
TP5080 adsorption instrument equipped with a thermal con-
ductivity detector (TCD). Before H2S-TPD experiments, ∼0.05 g
of sample was pretreated in N2 (30 mL min−1) at 300 °C for
1 h, then exposed to 1024 ppm H2S/N2 atmosphere at 200 °C
for 2 h. After purging with He at 50 °C for additional 1 h,
TPD profiles were recorded from room temperature to 600 °C
at a ramping rate of 10 °C min−1 under flowing He.
2.3 Catalytic test
The oxidative esterification of benzyl alcohol (BA) was carried
out in a 60 ml stainless steel autoclave lined with Teflon to
minimize the effect of metallic wall. 0.4 ml of BA, 32 ml of
methanol, 1 ml of anisole as internal standard, 0.1 g of
K2CO3 and 200 mg of catalyst were added into the reactor
and heated to 60 °C with stirring at 1000 rpm. Then O2
(2 MPa) was fed into the reactor (defining t = 0) to initiate
the reaction. After a certain reaction time, liquid products
mixed with catalysts were transferred from the reactor to stop
the reaction. The products were analyzed by gas chromatogra-
phy (GC) with a HP-5 ms capillary column (30 m, DF = 0.25
mm, 0.25 mm i.d.).
For the hydrogenation of nitrobenzene (NB), 0.4 ml of NB,
32 ml of ethyl alcohol, 1 ml of octane as internal standard
and 200 mg of catalyst were added into the reactor and
heated to 110 °C with stirring at 1000 rpm. Then the hydroge-
nation reaction was carried out at 2 MPa H2. For the hydroge-
nated coupling of NB and benzaldehyde (BDH), 0.4 ml of
NB, 1 ml of BDH, 30 ml of ethyl alcohol, 0.5 ml of octane as
internal standard and 400 mg of catalyst were used. The reac-
tion was carried out at 110 °C and 2 MPa H2 with stirring at
1000 rpm.
2. Experimental
2.1 Catalyst preparation
The CoNC/CNT catalyst was prepared as described in the lit-
erature.19 Typically, 125 mg of CoIJOAc)2·4H2O and 198 mg of
1,10-phenanthroline (Co : phenanthroline = 1 : 2 (mol)) were
dissolved in 30 mL of ethanol and stirred for 30 min at room
temperature. Then, 650 mg of CNTs (provided by Cnano Co.
Ltd., as received, >99.9% purity, synthesized using
a
supported Fe catalyst) was added to form a suspension under
agitation for 5 hours at 60 °C. After being cooled to room
temperature, the suspension was dried at 70 °C for 12 h. The
obtained solids were ground to fine powder, and then
annealed at 800 °C for 2 h under flowing Ar at 100 Ncm3
min−1.
The CoNC/CNT catalyst was thoroughly washed with 2 M
hydrochloric acid at room temperature for 5 h. The resulting
sample is denoted as CoNC/CNT-HCl. The CoNC/CNT catalyst
was also washed with 9 M nitric acid at 120 °C, and then
annealed at 800 °C for 2 h. The resulting sample is denoted
as CoNC/CNT-HNO3.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
The poisoning effects of KSCN and H2S on the CoNC cata-
lysts in the oxidative esterification of BA were investigated. In
the case of KSCN as a poison, 1 g of KSCN was added into
the reactants to conduct the reaction. In the case of H2S, the
catalyst was pretreated by flowing 1024 ppm H2S/N2 at 100
Ncm3 min−1 at 200 °C for 2 h in a horizontal tubular furnace.
3. Results and discussion
Scheme 1
The catalytic performance of CoNC/CNT and CoNC/CB was
evaluated in the oxidative esterification of BA. As shown in
Table 1, all cobalt containing catalysts display considerable
activity. However, the control experiments without any cata-
lyst (entry 1), with CNTs (entry 8), N-doped CNTs (entry 9)
and phenanthroline modified CNTs (entry 10) are completely
inactive, indicating that cobalt species are essential for the
catalysis, and the nitrogen functionalities on CNTs cannot
catalyze the reaction alone. Compared to XC-72R carbon
black, CNTs displayed comparative performance as a support
for CoNC (entries 2 and 5). This result validates the scenario
using CNTs as a substitute for CB to facilitate the identifica-
tion of active sites. It should be mentioned that the CoNC/
CNT catalyst is highly active, even being comparable to
homogeneous noble metal catalysts.30 Considering the lack
of optimized reaction conditions at present, we believe that
there are still many possibilities to enhance the performance
of CoNC/CNT. We also compared the performance of CoNC/
CNT and FeNC/CNT. The cobalt catalyst displayed a higher
activity than the iron-based one, showing the promise of
CoNC catalysts (see Table S1†).
It is surprising that HCl washing has no notable effect on
the conversion of BA and MB selectivity over CoNC/CNT
(entries 2 and 3), suggesting that the real active sites for the
oxidative esterification are too stable to be removed by HCl.
Compared with CoNC/CNT, the content of Co decreased from
1.81 w% to 0.42 w% after HCl washing, indicating that about
80% of cobalt are washable and redundant for the reaction.
In the case of the CB-supported catalyst, although the initial
Co content was higher, a similar fraction (~80%) of Co was
washable. When a harsher washing with HNO3 was con-
ducted, the content of Co further decreased, because oxida-
tive HNO3 may destroy the surfaces of carbon and modify the
surfaces of CNTs,31 evidenced by the increase of oxygen con-
tent from 4.5% for CoNC/CNT-HCl to 5.0% for CoNC/CNT-
HNO3 measured by XPS. Significant loss of activity was
observed after HNO3 washing (entries 4 and 7), indicating
that HNO3 could remove active cobalt species.
Kinetics measurements were carried out to compare the
catalytic performances of CoNC/CNT, CoNC/CNT-HCl and
CoNC/CNT-HNO3. As shown in Table 2, CoNC/CNT and
CoNC/CNT-HCl show very close reaction rate constants,
confirming that they qualitatively and quantitatively have the
same active sites for the reaction. A drastic drop of reaction
rate constant and an increase of activation energy by about
20 kJ mol−1 were observed for CoNC/CNT-HNO3, indicating
the destruction of active sites under concentrated HNO3.
Table 2 Pseudo-second-order reaction rate constants and apparent
activation energies of CoNC/CNT, CoNC/CNT-HCl and CoNC/CNT-
HNO3 for the oxidative esterification of BA with methanola
Catalyst
kb (L mol−1 h−1
)
Ea (kJ mol−1
)
CoNC/CNT
CoNC/CNT-HCl
CoNC/CNT-HNO3
9.2
8.7
0.49
101.5
98.6
122.0
a
b
See Fig. S1 and S2 for detailed experimental data. Reaction rate
constant at 60 °C.
Table 1 Catalytic performances of CoNC/CNT and CoNC/CB catalysts in the oxidative esterification of BA (Scheme 1)a
Composition
Co/Fe/N (w%)
SBET
(m2 g−1
Conv.
(%)
Selectivity (%)
BDH
Entry
Catalysts
)
MB
1
2
3
4
5
6
7
8
9
Blankb
—
—
0
0
0
CoNC/CNT
CoNC/CNT-HCl
CoNC/CNT-HNO3
CoNC/CB
1.81/—/1.42
0.42/—/1.71
0.14/—/0.31
3.01/—/n.d.
0.74/—/n.d.
0.05/—/n.d.
—
155.9
144.8
210.0
123.1
109.8
210.3
204.1
60.4
88.1
88.4
42.1
91.4
81.8
47.4
0
8.4
7.4
14.5
3.0
5.7
24.2
0
91.6
92.6
85.5
97.0
94.3
75.8
0
CoNC/CB-HCl
CoNC/CB-HNO3
CNTs
NCNTsc
3.44d
—
0
0
0
0
0
0
10
N/CNTe
189.0
a
b
c
Reaction conditions: 0.4 ml of BA, 32 ml of CH3OH, 0.1 g of K2CO3, 0.2 g of catalyst, 60 °C, 2 MPa O2, 12 h. Without a catalyst. Prepared
by a CVD method using xylene as the carbon source in NH3 atmosphere at 800 °C. See our previous work for details.29 d Atomic concentration
of nitrogen measured by XPS. e Prepared using the same procedure but without adding CoIJOAc)2·4H2O.
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
More reactions were conducted to compare the activities
pristine CNTs, indicating that they were derived from the car-
bonization of 1,10-phenanthroline. Although no Co particles
can be found in this area, EDS analysis showed considerable
content of Co. The distribution of C, N, O and Co was investi-
gated by energy-filtered transmission electron microscopy
(EFTEM). As shown in Fig. 2b, N, O and Co elements are
homogeneously distributed on the CNTs, indicating that the
active cobalt is highly dispersed in the form of sub-
nanometer clusters or single atom. A similar result can be
obtained by STEM an EDS mapping (see Fig. S5†). It is obvi-
ous that the nitrogen comes from the graphene derived from
1,10-phenanthroline revealed by HRTEM. The strong spatial
relationship between N and Co indicates that the active
cobalt might be trapped on the N-doped graphene wrapped
around CNTs, not on Co3O4.
EELS spectroscopy was applied to reveal the valence state
of N and Co. As depicted in Fig. 2c, two sharp π* peaks at
about 394 and 401 eV in the N-K edge can be observed, which
are due to the sp2 bonding of N with carbon. The splitting
feature of the π* state is evidence of the co-existence of two
types of nitrogen, namely graphitic and pyridinic nitro-
gen.32,33 The broad peak centered around 410 eV corresponds
to the σ* contribution. The shoulder at 408 eV is a signature
of pentagonal defect, i.e. pyrrolic nitrogen.34 Fig. 2d shows
the EELS spectrum of the Co-L2,3 edge of CoNC/CNT-HCl.
The L3/L2 intensity ratio and distance between L3 and L2 lines
can be used to estimate the valence state of Co.35,36 The L3/L2
ratio is determined to be 3.19, and the closeness between
them is 15.22 eV (see the ESI† for details). According to the
correlations established by Müller P. et al.,36 the valence state
of Co in CoNC/CNT-HCl can be estimated to be between +2
and +2.66.
of CoNC/CNT and CoNC/CNT-HCl. Two typical hydrogenation
reactions, i.e. the hydrogenation and hydrogenated coupling
of nitrobenzene, were used to test the catalytic performances.
Being similar to oxidative esterification, hydrogenation reac-
tions display very close conversion and selectivity (see
Schemes 2 and 3), indicating that the active sites that sur-
vived from HCl washing have diverse catalytic functions.
Obviously, it is helpful for unravelling the active sites of these
catalysts to characterize the Co species before and after acid
washing.
TEM was used to distinguish the washable inactive cobalt
and the stable active one. As shown in Fig. 1a, the as-
synthesized CoNC/CNT consists of quite large particles. How-
ever, this particulate matter completely disappeared after HCl
or HNO3 washing (Fig. 1b and c), suggesting that the large
particles are not responsible for the catalytic activity. XRD
was used to identify the composition of the large particles. As
shown in Fig. 1d, the peaks of Co3O4 can be detected in
CoNC/CNT. However, these peaks disappeared in CoNC/CNT-
HCl and CoNC/CNT-HNO3, suggesting that the washable
cobalt is mainly particulate Co3O4.
Westerhaus F. A. et al.21 have observed similar large Co3O4
particles on carbon-black-supported CoNC catalysts. They
concluded that the Co3O4 particles with N-doped graphene
shells are responsible for the working active sites.23,24 A more
detailed HRTEM measurement was carried out to identify the
core–shell structure. Unfortunately, this structure was so
occasionally observed (see Fig. S3†) that it can only be a triv-
ial contributor to activity. It was noticed that a large amount
of graphene-like debris can be found on the outer surfaces of
CoNC/CNT-HCl (Fig. 2a and S5a†), which were absent in
Fig. 3 shows the N1s and Co2p3/2 regions of the XPS spec-
tra of CoNC/CNT, CoNC/CNT-HCl and CoNC/CNT-HNO3.
Four distinct N-functionalities can be observed in the N1s
spectra of CoNC/CNT and CoNC/CNT-HCl, while no N1s sig-
nal can be detected on CoNC/CNT-HNO3 within the analytical
range of the XPS technique, indicating that the HNO3 treat-
ment destroyed the N-doped graphene. These N species can
be attributed to NP (pyridinic N, 398.3 eV), CoNx (cobalt-coor-
dinated N, 399.2 eV),37 NQ (quaternary N, 400.9 eV) and Nox
(N-oxides, 402.1 eV). It is interesting that CoNC/CNT-HCl has
a higher N content of 3.06% than CoNC/CNT (2.53%). It may
be caused by the re-exposure of N-functionalities to CNTs
after removing Co3O4 particles. For both CoNC/CNT and
CoNC/CNT-HCl, the nitrogen bonded to cobalt dominates
among the four N species, indicating that the real active sites
are CoNx. The Co 2p peaks of CoNC/CNT and CoNC/CNT-HCl
show the presence of Co2+ (780.6 eV) and Co3+ (779.6 eV) on
catalyst surfaces, accompanied by a shake-up satellite of Co2+
(786.3 eV).38 The amount of Co decreased from 1.04 a% to
0.64 a% after HCl treatment. It was noticed that the ratio of
decrements of Co2+ and Co3+ was approximately 1 : 2,
suggesting a stoichiometry of Co3O4 for the washable cobalt
species. Moreover, although a mixed valence of Co was
observed, a lower oxidation state of CoNC/CNT-HCl can be
Scheme 2
Scheme 3
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
Fig. 1 TEM images of (a) CoNC/CNT, (b) CoNC/CNT-HCl, and (c) CoNC/CNT-HNO3, and (d) their XRD patterns.
determined compared with CoNC/CNT, because of the higher
fraction of Co2+, which is consistent with the EELS results.
Poisoning experiments were carried out to further prove
the involvement of the highly dispersed Co in the catalysis.
CN−, SCN− and H2S have been demonstrated to be capable of
bonding with transition metal-centered active sites and
deactivating them.39–41 In this work, we examined the active
cobalt sites by poisoning them with KSCN and H2S. As shown
in Fig. 4a, the presence of KSCN in the reactants and pre-
adsorbed H2S significantly deactivates the CNT-supported
catalysts, suggesting the interaction between the sulfur-
containing poisons and active sites containing cobalt. More-
over, HRTEM measurement and EFTEM analysis of the cata-
lyst poisoned by KSCN revealed a homogeneous profile of sul-
fur over CNTs (see Fig. S6†). The high spatial relevance with
Co is indicative of the adsorption of SCN− on Co sites.
H2S-TPD was carried out to investigate the interaction
between the active sites and H2S as a poison. As shown in
Fig. 4b, two distinct desorption peaks, centered at 320 °C and
405 °C, can be observed for CoNC/CNT. After washing with
HCl, the two peaks remain, but shift toward higher tempera-
tures by about 20 °C, indicating stronger adsorption. A simi-
lar H2S-TPD behavior has been reported by Singh D. et al. for
FeNC catalysts for ORR reaction.40 The peak at 320 °C
originates from H2S adsorbed on N-functionalities of
N-CNTs. It can be supported by the TPD profile of N/CNTs,
which shows a similar desorption peak. However, no peaks
can be detected for pristine CNTs (see Fig. S7†). This result
also suggests that the peak at higher temperature is due to
the stronger adsorption of H2S on Co sites, which may inter-
fere with the coordination between the transition metal and
pyridinic nitrogen and deactivate the catalyst, being similar
to the deactivation behavior of FeNC for ORR.40 It is notice-
able that the intensity of the high-temperature peak remains
the same after HCl washing, indicating the high stability of
the active sites as revealed by the catalytic results.
The above results clearly indicate that, although there are
two types of Co species on CNTs, i.e. the large Co3O4 particles
and sub-nanometer CoNC, the active sites are composed of
cobalt bonded to nitrogen of N-doped graphene derived from
1,10-phenanthroline. G. Zhu et al.42 also raised the similar
active sites which are CoNx (maybe Co–N4) supported on
porous materials synthesized by the carbonization of a Co-
MOF. A schematic mechanism was proposed to elucidate the
formation of the active sites. As shown in Fig. 5, 1,10-
phenanthroline is carbonized to form N-doped graphene
sheets wrapped on CNTs at 800 °C. Because of the strong
chelation between Co2+ and 1,10-phenanthroline, cobalt can
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Fig. 2 (a) HRTEM and (b) the corresponding EFTEM images of CoNC/CNT-HCl. A similar result with a lower magnitude is shown in Fig. S4.† The
white arrows indicate the defective graphene on CNTs. (c) and (d) show the EELS spectra of nitrogen and cobalt.
Fig. 3 (a) N 1s and (b) Co 2p3/2 XPS spectra of CoNC/CNT, CoNC/CNT-HCl and CoNC/CNT-HNO3.
be stabilized by the nitrogen sites, probably pyridinic nitro-
gen, during the pyrolysis, which results in the generation of
homogeneously distributed Co sites on the atomic/sub-
nanometer scale. These sites are characterized by high dis-
persion, bonding to N atoms and thus excellent catalytic
activity analogous to well-defined molecular catalysts. On the
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
Fig. 4 (a) Effect of KSCN and H2S as poisons on BA conversion in its oxidative esterification with methanol over CNT-supported catalysts. Reaction
conditions: 0.4 ml of benzyl alcohol, 32 ml of CH3OH, 0.1 g of K2CO3, 60 °C, 2 MPa O2, 12 h. (b) H2S-TPD profiles of CoNC/CNT, CoNC/CNT-HCl
and N/CNT.
Fig. 5 Schematic diagram for the formation of CoNC active sites for BA oxidative esterification, NB hydrogenation and hydrogenated coupling
with BDH.
other hand, the decomposition of excess CoIJOAc)2 leads to
the formation of particulate Co3O4, around which the
graphene sheets derived from 1,10-phenanthroline can also
be formed. It cannot be excluded that CoNx embedded in
graphene shells encapsulating Co3O4 particles might contrib-
ute to the activity to some extent; however, its effect should
be marginal because of the relatively low surface area of
Co3O4 particles. The Co3O4 particles and graphene on their
surfaces cannot survive from HCl washing, after which the
catalysts with homogeneous Co active sites can be obtained.
This catalyst is active in diverse organic synthetic reactions.
Washing with HNO3 is detrimental to the activity, because
HNO3 can oxidatively remove the graphene containing active
Co. This can be supported by the reduction of Co content
and the decreased defectiveness of carbon revealed by Raman
spectroscopy (see Fig. S7†).
It is highly desired to understand the molecular structure
of active sites. A deduction was made through the quantita-
tive XPS results. Assuming that all Co atoms on CoNC/CNT-
HCl are active, the N/Co ratio of the active sites can be esti-
mated to be 2.2 (Table 3), where only the nitrogen bonded to
Co is considered. Although the precise molecular structure of
the active sites is unclear yet, this value implies that a cobalt
chelate complex with 2 to 3 nitrogens in the graphene lattice,
probably like the pyridinic vacancy, might be responsible for
the activity. By combining the quantitative XPS data with the
specific surface area, the intrinsic catalytic activity for the oxi-
dative esterification of BA can be estimated. High TOF values
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 3 Quantitative XPS analysis for CoNC/CNT and CoNC/CNT-HCl
Coa (%)
Co2+a (%)
Co3+a (%)
Na (%)
NP (%)
CoNx (%)
NQa (%)
Nox (%)
TOFb (h−1
)
a
a
a
Catalyst
CoNC/CNT
CoNC/CNT-HCl
1.04
0.64
0.51
0.39
0.53
0.25
2.53
3.06
0.21
0.46
1.06
1.41
0.69
1.02
0.57
0.17
3315
5207
a
b
Atomic concentration determined by quantitative XPS analysis. The turnover frequency for the oxidative esterification of BA is defined as:
. The BA conversion was controlled at lower than 10%.
up to 3300 and 5200 h−1 were measured for CoNC/CNT and
CoNC/CNT-HCl, respectively, as shown in Table 3. More
experiments are needed to resolve the structure in the future.
9 J. L. Carsten Bolm, J. L. Paih and L. Zani, Chem. Rev.,
2004, 104, 6217–6254.
10 A. Vaccari, Catal. Today, 1998, 41, 53–71.
11 A. Vaccari, Appl. Clay Sci., 1999, 14, 161–198.
12 F. Bedioui, Coord. Chem. Rev., 1995, 144, 39–68.
13 R. F. Parton, I. F. J. Vankelecom, D. Tas, K. B. M. Janssen,
P.-P. Knops-Gerrits and P. A. Jacobs, J. Mol. Catal. A: Chem.,
1996, 113, 283–292.
14 R. Augustine, S. Tanielyan, S. Anderson and H. Yang, Chem.
Commun., 1999, 1257–1258.
15 C. Freire, C. Pereira and S. Rebelo, Catalysis, 2012, 24,
116–203.
16 A. Morozan, P. Jegou, B. Jousselme and S. Palacin, Phys.
Chem. Chem. Phys., 2011, 13, 21600–21607.
17 S. Deng, P. Yuan, X. Ji, D. Shan and X. Zhang, ACS Appl.
Mater. Interfaces, 2015, 7, 543–552.
18 W. Feng and P. Ji, Biotechnol. Adv., 2011, 29, 889–895.
19 R. V. Jagadeesh, H. Junge, M. M. Pohl, J. Radnik, A.
Bruckner and M. Beller, J. Am. Chem. Soc., 2013, 135,
10776–10782.
4. Conclusions
In conclusion, a simple acid-washing method was used to
distinguish the heterogeneity of cobalt species of the CoNC/
CNT catalyst. It was found that Co3O4 particles that can be
removed by HCl washing have no effect on the activity. The
real active cobalt species are highly dispersed on CNTs on
the atomic or sub-nanometer scale, which are stable under
HCl conditions. The structure of the active sites is proposed
as a cobalt chelate complex embedded in pyridine-like vacan-
cies of N-doped graphene created by the thermolysis of cobalt
salts and 1,10-phenanthroline. The new insight into the active
sites of carbon supported CoNx catalysts may lead to a ratio-
nal design of high-performance catalysts in the future.
Acknowledgements
20 R. V. Jagadeesh, A. E. Surkus, H. Junge, M. M. Pohl, J.
Radnik, J. Rabeah, H. Huan, V. Schunemann, A. Bruckner
and M. Beller, Science, 2013, 342, 1073–1076.
21 F. A. Westerhaus, R. V. Jagadeesh, G. Wienhöfer, M.-M. Pohl, J.
Radnik, A.-E. Surkus, J. Rabeah, K. Junge, H. Junge, M. Nielsen,
A. Brückner and M. Beller, Nat. Chem., 2013, 5, 537–543.
22 D. Banerjee, R. V. Jagadeesh, K. Junge, M. M. Pohl, J.
Radnik, A. Bruckner and M. Beller, Angew. Chem., Int. Ed.,
2014, 53, 4359–4363.
23 T. Stemmler, F. A. Westerhaus, A.-E. Surkus, M.-M. Pohl, K.
Junge and M. Beller, Green Chem., 2014, 16, 4535–4540.
24 R. V. Jagadeesh, H. Junge and M. Beller, ChemSusChem,
2015, 8, 92–96.
25 R. V. Jagadeesh, T. Stemmler, A.-E. Surkus, H. Junge, K.
Junge and M. Beller, Nat. Protoc., 2015, 10, 548–557.
26 S. Pisiewicz, T. Stemmler, A.-E. Surkus, K. Junge and M.
Beller, ChemCatChem, 2015, 7, 62–64.
27 J. Deng, H.-J. Song, M.-S. Cui, Y.-P. Du and Y. Fu,
ChemSusChem, 2014, 7, 3334–3340.
28 Y. Wu, Q. Shi, Y. Li, Z. Lai, H. Yu, H. Wang and F. Peng,
J. Mater. Chem. A, 2015, 3, 1142–1151.
29 Y. Cao, H. Yu, J. Tan, F. Peng, H. Wang, J. Li, W. Zheng and
N.-B. Wong, Carbon, 2013, 57, 433–442.
30 C. Liu, J. Wang, L. Meng, Y. Deng, Y. Li and A. Lei, Angew.
Chem., Int. Ed., 2011, 50, 5144–5148.
This work was supported by the National Natural Science Foun-
dation of China (No. 21133010, 21273079), the Guangdong Pro-
vincial Natural Science Foundation (No. S20120011275,
2014A030312007), the Program for New Century Excellent
Talents in University (NCET-12-0190) and the Fundamental
Research Funds for the Central Universities of China (No.
2014ZG0005, 2015PT012).
References
1 Y. Zhu and Y. Wei, RSC Adv., 2013, 3, 13668–13670.
2 X.-F. Wu and C. Darcel, Eur. J. Org. Chem., 2009, 1144–1147.
3 Y. Li, S. Yu, X. Wu, J. Xiao, W. Shen, Z. Dong and J. Gao,
J. Am. Chem. Soc., 2014, 136, 4031–4039.
4 K. Junge, K. Schroder and M. Beller, Chem. Commun.,
2011, 47, 4849–4859.
5 F. G. Gelalcha, B. Bitterlich, G. Anilkumar, M. K. Tse and M.
Beller, Angew. Chem., Int. Ed., 2007, 46, 7293–7296.
6 G. Anilkumar, B. Bitterlich, F. G. Gelalcha, M. K. Tse and M.
Beller, Chem. Commun., 2007, 289–291.
7 K. Schröder, S. Enthaler, B. Join, K. Junge and M. Beller,
Adv. Synth. Catal., 2010, 352, 1771–1778.
8 S. Kanemasa, Y. Oderaotoshi, S. Sakaguchi, H. Yamamoto, J.
Tanaka, E. Wada and D. P. Curran, J. Am. Chem. Soc.,
1998, 120, 3074–3088.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
31 L. Jie, A. G. Rinzler, D. Hongjie, J. H. Hafner, R. K. Bradley,
P. J. Boul, A. Lu, T. Iverson, K. Shelimov, C. B. Huffman, F.
Rodriguez-Macias, S. Young-Seok, T. R. Lee, D. T. Colbert
and R. E. Smalley, Science, 1998, 280, 1253–1256.
32 M. Terrones, P. M. Ajayan, F. Banhart, X. Blase, D. L. Carroll,
J. C. Charlier, R. Czerw, B. Foley, N. Grobert, R.
Kamalakaran, P. Kohler-Redlich, M. Rühle, T. Seeger and H.
Terrones, Appl. Phys. A: Mater. Sci. Process., 2002, 74,
355–361.
36 P. Müller, M. Meffert, H. Stormer and D. Gerthsen, Microsc.
Microanal., 2013, 19, 1595–1605.
37 J. M. Ziegelbauer, T. S. Olson, S. Pylypenko, F. Alamgir, C.
Jaye, P. Atanassov and S. Mukerjee, J. Phys. Chem. C,
2008, 112, 8839–8849.
38 T. J. Chuang, C. R. Brundle and D. W. Rice, Surf. Sci.,
1976, 59, 413–429.
39 M. S. Thorum, J. M. Hankett and A. A. Gewirth, J. Phys.
Chem. Lett., 2011, 2, 295–298.
33 H. Lin, R. Arenal, S. Enouz-Vedrenne, O. Stephan and A.
Loiseau, J. Phys. Chem. C, 2009, 113, 9509–9511.
40 D. Singh, K. Mamtani, C. R. Bruening, J. T. Miller and U. S.
Ozkan, ACS Catal., 2014, 4, 3454–3462.
34 R. Arenal, K. March, C. P. Ewels, X. Rocquefelte, M. Kociak,
A. Loiseau and O. Stéphan, Nano Lett., 2014, 14, 5509–5516.
35 Z. L. Wang, J. S. Yin and Y. D. Jiang, Micron, 2000, 31,
571–580.
41 S. Gupta, C. Fierro and E. Yeager, J. Electroanal. Chem.
Interfacial Electrochem., 1991, 306, 239–250.
42 G. Yu, J. Sun, F. Muhammad, P. Wang and G. Zhu, RSC Adv.,
2014, 4, 38804–38811.
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.