Nitrogen-doped carbon materials derived from acetonitrile and Mg-Co-Al layered double hydroxides as electrocatalysts for oxygen reduction reaction

Article abstract of DOI:10.1016/j.electacta.2016.06.150

N-doped carbon materials of various morphologies (carbon platelets, carbon nanotubes) were prepared by simultaneous carbonization of CH₃CN and decomposition of Mg-Co-Al layered double hydroxides (LDHs) at 600, 700 and 800?°C followed by treatme

Full text of DOI:10.1016/j.electacta.2016.06.150

Electrochimica Acta 212 (2016) 47–58
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Nitrogen-doped carbon materials derived from acetonitrile and Mg-Co
-Al layered double hydroxides as electrocatalysts for oxygen reduction
reaction
Aleksandra Pacułaa,b,c,*, Kohei Uosaki
a,b,d
c
c
, Robert P. Socha , El z_ bieta Biela nꢀ ska ,
e
c
Piotr Pietrzyk , Małgorzata Zimowska
a
International Center for Materials Nanoarchitectonics (WPI-MANA), National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044,
Japan
b
c
Division of Chemistry, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan
Jerzy Haber Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, Niezapominajek 8, 30-239 Kraków, Poland
Global Research Center for Environment and Energy based on Nanomaterials Science (GREEN), National Institute for Materials Science (NIMS), 1-1 Namiki,
d
Tsukuba, Ibaraki 305-0044, Japan
e
Jagiellonian University, Faculty of Chemistry, Ingardena 3, 30-060 Kraków, Poland
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 28 December 2015
Received in revised form 27 June 2016
Accepted 28 June 2016
N-doped carbon materials of various morphologies (carbon platelets, carbon nanotubes) were prepared
by simultaneous carbonization of CH CN and decomposition of Mg-Co-Al layered double hydroxides
3
ꢀ
(LDHs) at 600, 700 and 800 C followed by treatment with HCl.
Available online 29 June 2016
Several techniques including powder X-ray diffraction, thermogravimetric analysis, elemental analysis,
Raman spectroscopy, X-ray photoelectron spectroscopy, Fourier transform infrared spectroscopy,
scanning electron microscopy and nitrogen sorption were applied to describe their physicochemical
features. The electrocatalytic performance of N-doped carbon materials for oxygen reduction reaction
Keywords:
carbon nanotube
layered double hydroxide
nitrogen-doped carbon
catalytic chemical vapour deposition
oxygen reduction reaction
(ORR) was evaluated using rotating disk electrode (RDE) and cyclic voltammetry (CV) measurements in
alkaline solution.
ꢀ
ꢀ
The samples prepared at 600 and 700 C appear to be more active for ORR than that prepared at 800 C.
Higher activity of C-600 and C-700 is associated with higher contribution of carbon platelets providing
higher specific surface area and offering higher concentration of the active sites. The contribution of
carbon tubular nanostructures in C-600, C-700 and C-800 increases with increasing synthesis
temperature. The presence of carbon nanotubes seems to be not advantageous for ORR performance.
The N-doped carbon materials are able to catalyze the oxygen reduction reaction by 2 and 4 electrons.
The order of selectivity for 4-electron reduction is the following: C-800 < C-700 < C-600. Among studied
samples, C-600 has the highest concentration of pyridinic nitrogen being responsible for 4-electron
reduction and the lowest concentration of O-containing groups, in particular carbonyl and carboxyl,
being responsible for 2-electron reduction.
ã 2016 Elsevier Ltd. All rights reserved.
1. Introduction
preparation of the electrocatalysts for ORR to replace Pt-based
catalysts [1,2,6,7,8].
Nitrogen (N) and cobalt (Co) or iron (Fe) bonded together on the
surface of the carbon materials are claimed to be active for oxygen
reduction reaction (ORR) [1–5]. Therefore, doping carbon materials
with N, Co or Fe has been intensively investigated for the
3
Acetonitrile (CH CN) is a convenient carbon (C) and nitrogen
(N) source for the synthesis of N-doped carbon materials by
chemical vapour deposition method.
3
The application of CH CN as C/N precursor for the preparation
of the electrocatalysts for ORR was started by Dodelet and his co-
workers almost 2 decades ago [9,10]. Their first attempt involved
3
carbonization of CH CN with the addition of other organic
ꢀ
compounds at 1000 C for 2 h in the presence of 2 % Fe/C [9].
They obtained N-doped carbon nanotubes (CNTs) (N content
*
Corresponding author at: Niezapominajek 8, 30-239 Kraków, Poland.
E-mail address: ncpacula@cyfronet.pl (A. Pacuła).
http://dx.doi.org/10.1016/j.electacta.2016.06.150
013-4686/ã 2016 Elsevier Ltd. All rights reserved.
0

4
8
A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
determined by XPS was 3.6–5.5 %). The obtained electrocatalysts
without purification) appeared to be active and stable in ORR in
acidic environment. The catalytic activity of the studied materials
was not related to metallic iron being present in form of particles
2
Other supports such as silica (SiO ) and magnesia (MgO) were
(
also explored for the preparation of the electrocatalysts by Ozkan’s
ꢀ
group [14,15]. They carried out CVD experiments at 900 C for 2 h
with the use of SiO
comparison, they carried out the pyrolysis of CH
2
and MgO impregnated with 2 % Fe. For
CN with the use of
encased in the carbonaceous materials. Fe-N
detected in the samples.
x
species was not
3
pure supports and supports impregnated with Ni or Co. The
ꢀ
Then, they prepared the catalysts by carbonization of CH
3
CN at
impregnated supports were additionally heated at 550 C under
ꢀ
various temperatures (400, 600, 800 and 1000 C) for 2 h with the
hydrogen flow. After carbonization the as-prepared materials were
purified and then, they were examined for ORR in acidic medium.
ꢀ
use of 10 % Fe/C, which was additionally heated at 600 C under
hydrogen flow [10]. The obtained materials were additionally
washed with acid. Depending on the pyrolysis temperature, the
samples contained Fe particles surrounded either by carbonaceous
The samples derived from SiO
areas than those of the samples derived from Al
morphology of carbon particles synthesized with the application of
SiO and MgO differed from that generated in the presence of
Al . SiO and MgO did not lead to the formation of carbon
nanofibres like in the case of Al . However, the presence of Fe
and Co on both supports (SiO and MgO) led to the formation of
carbon nanofibres like in the case of 2 % Fe/Al . SiO appeared to
2
and MgO had lower specific surface
2 3
O
. Moreover, the
ꢀ
material (600 C) or encased in tubular structures (800 and
2
ꢀ
ꢀ
1
000 C). The samples prepared at 600, 800 and 1000 C were
2
O
3
2
active for ORR in acidic medium. Only the electrocatalysts
2 3
O
ꢀ
prepared at 800 and 1000 C were stable. Fe-containing particles
2
did not possess the catalytic activity.
2
O
3
2
Since 2006, Ozkan and her co-workers have been done a lot of
be better support than MgO for the preparation of the electro-
work to develop the electrocatalysts derived from CH
3
CN [11–17].
catalysts in terms of their activities. The catalyst prepared by the
ꢀ
At the beginning, they obtained a series of the electrocatalysts
pyrolysis of CH
3
CN at 900 C for 2 h in the presence of 2 % Fe/SiO
2
ꢀ
at 600, 700, 800, 900 C in the presence of 2 % Fe/C [11]. They also
appeared to be the most active among studied catalytic materials.
Based on the studies carried out in Ozkan’s group over 6 years
(2003–2009), the authors claimed that iron was not a center or
even a component of ORR active sites [16,17]. There was no
evidence for the formation of nitrogen-coordinated Fe centers. Iron
might play a role in the formation of the active sites through the
carbon nanostructure growth mechanism. The role of iron was to
catalyze the formation of carbon nanostructures with increased
number of N-functionalities on the surface.
modified the duration of carbonization, i.e. 20 min., 2 h and 12 h. In
addition, they synthesized an analogue set of the catalysts in the
presence of 2 % Ni/C. They observed either carbon tubular
structures in Fe-derived electrocatalysts or carbon spherical
particles in Ni-derived electrocatalysts. The as-prepared electro-
catalysts (without purification) were examined in ORR in acidic
medium. The electrocatalysts derived from 2 % Fe/C appeared to be
more active than those derived from 2 % Ni/C. The sample prepared
ꢀ
by the pyrolysis of CH
3
CN at 900 C for 2 h in the presence of 2 %
Ozkan and her co-workers made also an effort to synthesize the
ꢀ
Fe/C was the most active among studied catalytic materials. Its
electrocatalysts by carbonization of CH
3
CN at 800 C for 2 h over
2
ꢁ1
specific surface area was 127 m g , whereas N content deter-
10 % Co/C and 10 % Co/MgO [18]. The as-prepared catalysts were
washed with acid and then, they were examined for ORR in acidic
medium. The catalysts derived from 10 % Co/MgO appeared to be
more active than those derived from 10 % Co/C.
mined by XPS was 3.5 %.
Then, they synthesized the electrocatalysts by carbonization of
ꢀ
CH
% Ni/Al
with acid and then, they were examined for ORR in acidic medium.
The electrocatalysts derived from 2 % Fe/Al were more active
than those derived from 2 % Ni/Al . The electrocatalyst derived
from Al was also active, which led to the conclusion that the
3
CN at 900 C for 2 h in the presence of Al
2
O
3
, 2 % Fe/Al
2
O
3
and
2
2
O
3
[12]. The as-prepared electrocatalysts were washed
3
Muhler and his co-workers used CH CN and Co-based catalyst
to synthesize N-CNTs and evaluate their activity for ORR [19]. They
ꢀ
2
O
3
carried out the carbonization for 16 h either at 550 or at 750 C. The
2
O
3
as-prepared samples were washed with acid. The N-CNTs prepared
ꢀ
2
O
3
at 550 C had a higher degree of structural defects (I
D
/I
G
ratio of
presence of metal was not responsible for the electrocatalytic
activity. Number of electrons involved in ORR was 3.7, 3.9 and 3.6
1.25) and higher amount of pyridinic groups, whereas the N-CNTs
ꢀ
prepared at 750 C had a lower degree of structural defects (I
D
/I
G
for the electrocatalysts derived from Al
Al , respectively. The carbonaceous deposit in the catalysts
prepared over different supports had various morphologies. The
samples derived from 2 % Fe/Al and 2 % Ni/Al nearly
exclusively contained the carbon fibers. However, the structure of
the carbon fibers differed significantly depending on whether Fe or
Ni was used. The carbon fibers formed in the presence of Fe had
mostly a stacked cup structure, which meant that tube walls were
not parallel to the central carbon fiber axis. Alumina resulted in the
formation of the carbon fibers similar to those observed in the
2
O
3
, 2 % Fe/Al
2
O
3
and 2 % Ni/
ratio of 1.17) and lower amount of pyridinic groups. Number of
electrons involved in ORR carried out in acidic medium was 3.92
and 2.48 for the sample prepared at 550 and 750 C, respectively. It
was revealed that cobalt nanoparticles being covered by carbona-
ceous layers were not accessible by the electrolyte and they were
not a part of the active sites. The contribution of Co-containing
active sites such as CoN C or CoN C was also excluded. It was
2 x 4 y
proved that the pyrolysis temperature was the most important
parameter determining the amount of active sites.
2 3
O
ꢀ
2
O
3
2
O
3
3
Tammaveski and his co-workers used CH CN to grow N-CNTs
samples derived from 2 % Fe/Al
2
O
3
. The carbon fibers grown in the
over GC impregnated with Co/Al salts [20]. Despite they carried out
ꢀ
presence of 2 % Ni/Al had tube walls parallel to the central fiber
2
O
3
the carbonization at various temperatures of 700, 800 and 900 C
ꢀ
axis. It was revealed that increase in exposure of graphite edge
planes and increase in content of pyridinic nitrogen resulted in
enhanced catalytic activity.
Ozkan and her co-workers also checked the influence of Fe
loading on the electrocatalytic performance for ORR in acidic
only the sample synthesized at 800 C could be examined as
electrocatalysts for ORR in acidic and alkaline media.
According to the above-mentioned literature, MgO and Al
can be useful for the preparation of N-doped carbon nano-
structures by carbonization of CH CN. We previously applied Mg-
Al LDHs, as precursor of MgO and Al
2 3
O
3
medium. They prepared the electrocatalyst by carbonization of
2
O
3
, to synthesize N-doped
ꢀ
ꢀ
CH
3
CN at 900 C for 2 h with the use of 10 % Fe/Al
2
O
3
[13].
carbon materials via CVD at 600, 700 and 800 C [21]. SEM images
revealed that the carbon nanostructures consisted of carbon plate-
like particles exposing edge planes.
Increasing Fe content from 2 to 10 % did not lead to enhanced
electrocatalytic performance, moreover the activity of the sample
prepared with 10 % Fe/Al
Al
2
O
3
was lower than that prepared with
LDHs are convenient and inexpensive metal/metal oxide
precursors. Decomposition of LDHs at high temperatures leads
2 3
O .

A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
49
to a mixture of metal/metal oxides showing large surface area, high
component dispersion and stability against sintering.
as-prepared samples are able to catalyze the reduction of oxygen
only by 2 electrons.
Then, we also applied Mg-Co-Al LDHs to prepare N-doped
carbon nanostructures by carbonization of CH
3
CN [22]. Application
2. Experimental
of LDHs containing Co2 ions surrounded by Mg and Al ions
successfully results in obtaining nano-scale Co species. Co nano-
particles participate in the formation of carbon nanotubes and
influence the incorporation of N-containing functionalities (being
responsible for activity of N-doped carbons in ORR) into the carbon
framework as previously described in the literature [2,16,23]. Oh
and Kim [7] claim that cobalt favors the formation of quaternary
+
2+
3+
2.1. Materials
Magnesium nitrate hexahydrate (Mg(NO
3
)
2
ꢂ
6H
6H
9H
sodium hydroxide (NaOH, JIS special grade), potassium hydroxide
(KOH, JIS special grade), acetone ((CH CO, 99.0 %), hydrochloric
acid (HCl, 35.0–37.0 % JIS special grade), sulphuric acid (H SO
2
O, JIS special
O, JIS special
O, 99.9 %),
grade), cobalt (II) nitrate hexahydrate (Co(NO
3
)
2
ꢂ
2
grade), aluminium nitrate nonahydrate (Al(NO
3
)
3
ꢂ
2
(
denoted as N-4) and pyridinic (denoted as N-6) and suppresses
3 2
)
the formation of pyrrolic (denoted as N-5), whereas an opposite
opinion prevails in other works, i.e. that cobalt favors the formation
of N-5 species [24,25].
2
4
,
96.0-98.0 %, super special grade) were purchased from Wako Pure
Chemical Industries Ltd. All chemicals were used as received.
Our previous publication describes the electrocatalysts for ORR
derived from CH CN and Mg-Co-Al LDHs [22]. The synthesis of the
3
Alumina powder for polishing (particle size, 0.3
m
m) was
purchased from Buehler. Nafion was purchased from Aldrich.
Nitrogen (N , 99.99 %) was purchased from Air Water, whereas
other gases, i.e. argon (Ar, 99.999 %) and oxygen (O , 99.95 %) were
1
electrocatalysts was carried out at various temperatures, i.e. 600,
2
ꢀ
7
00 and 800 C, to check the influence of the temperature on the
2
formation of Co nanoparticles, the morphology and porosity of
carbon deposit and the contribution of N-containing functionali-
ties and therefore the influence on electrocatalytic activity.
Contrary to the previous works, we used in CVD as-prepared
LDHs (without calcination and reduction), which were synthesized
under mild conditions (at room temperature, in air, with short
ageing). The electrocatalysts were obtained by simultaneous
decomposition of Mg-Co-Al LDHs and acetonitrile at high temper-
atures. This method resulted in obtaining the composites
containing carbonaceous component intermixed with non-carbo-
naceous component. During high-temperature synthesis cobalt
oxides derived from Mg-Co-Al LDHs should be reduced to metallic
cobalt, which may react with N-containing groups in carbon
deposit. Due to the presence of a lone pair of electrons, nitrogen has
a tendency to form complexes with transition metals (Co). So, the
creation of Co-N-C species on the surface of the carbon materials
may influence their electrocatalytic properties for ORR.
purchased from Linde Gas. Water was purified using a Milli-Q
purification system (Millipore).
2.2. Preparation of the N-doped carbon materials
The sample of Mg-Co-Al layered double hydroxides (LDHs) used
as cobalt nanoparticle precursor was synthesized via method
outlined in our previous publication [22]. The preparation of the N-
doped carbon materials involved simultaneous decomposition of
ꢀ
Mg-Co-Al LDHs and carbonization of CH
3
CN at 600, 700 and 800 C.
LDHs (1 g) in an alumina boat was placed in a horizontal tube
ꢁ
1
furnace and heated under a flow of nitrogen (30 mL min ) to the
ꢀ
required temperature (600, 700 and 800 C). Then the nitrogen
3
stream was saturated with CH CN vapor and the sample
maintained at the target temperature for 3 h. After cooling under
nitrogen flow, the resulting samples were treated with hydro-
2
chloric acid (HCl:H O 2:1 v/v) for 48 h and ultrasonically agitated
In the present work, we report on the electrocatalysts derived
from CH CN and Mg-Co-Al LDHs, prepared according to the
3
procedure previously described [22], but additionally treated with
acid. So, the fabricated materials are called acid-treated samples in
contrast to as-prepared ones. We removed inorganic part (MgO, Co,
for 2 h to remove inorganic moiety. Finally, the resulting samples
were separated, washed thoroughly with water and dried in a dry
box at 50 C. The weight loss after leaching process was 68, 59,
ꢀ
34 wt. % for the acid-treated samples synthesized at 600, 700 and
ꢀ
800 C. The obtained samples are further denoted as C-600, C-700,
CoO, Al
2
O
3
and MgAl
2
O
4
) from the as-prepared samples by
and C-800, respectively. Thermogravimetric (TG) analysis (Fig. S1
(a) in Supplementary materials) of the acid-treated samples
indicates a residual mass, i.e. 12, 14 and 15 wt. % at 1000 C for C-
600, C-700 and C-800, respectively. Thus, TG analysis reveals that
the N-doped carbon materials still contain some non-carbona-
ceous component derived from Mg-Co-Al LDHs.
dissolution in the acid and examined the performance of
carbonaceous component for ORR separately. The electroreduction
of oxygen was chosen as a test reaction probing the access of
oxygen to ORR active sites. The electrocatalytic activity of the acid-
treated samples for ORR was evaluated in alkaline solution and the
results were compared with those of the as-prepared samples
described previously [22]. According to our best knowledge no
report on the application of acetonitrile and Mg-Co-Al LDHs for the
preparation of the electrocatalysts may be found in the literature.
Here, we demonstrate that the acid-treated samples are more
active than the as-prepared ones. The increase in activity could be a
consequence of elimination of inorganic compounds from the as-
prepared materials which block the access to the active species
occupied on carbonaceous component. The as-prepared materials
are a mixture of two components – carbonaceous (14–54 %
according to EA) and non-carbonaceous (46–86 % according to EA).
In contrast, the acid-treated materials consist mainly of one,
carbonaceous constituent (81–86 % according to EA).
ꢀ
2.3. Characterization of the physicochemical properties
Thermogravimetric (TG) analysis was performed from ambient
ꢀ
temperature to 1000 C under static air condition with a heating
ꢀ
ꢁ1
rate of 5 C min , using a Thermo plus TG8120 (Rigaku) analyzer.
Powder X-ray diffraction (XRD) analysis was performed using a
MultiFlex Rigaku powder diffractometer with Cu K
a radiation
(40 kV, 20 mA). The XRD patterns were recorded in the range of 5–
ꢀ
ꢀ
ꢁ1
ꢀ
139 at scan rate of 4.8 min with a step of 0.05 .
Elemental analysis was carried out using a CHNS analyzer
(MICRO CORDER JM10). Raman spectroscopic measurements were
performed by means of a Renishaw InVia spectrometer equipped
with Ar ion laser (514 nm, power 50 mW), a Leica DMLM confocal
microscope, and a CCD detector. The Raman spectra were recorded
In addition, treating the as-prepared materials with acid
eliminates exposed cobalt, therefore the acid-treated materials
become less selective for 2-electron reduction than the corre-
sponding as-prepared ones. The acid-treated samples are able to
catalyze the reduction of oxygen by 2 and 4 electrons, whereas the
ꢁ
1
using 5 % of the power, in the spectral range of 200–3500 cm
ꢁ1
with a resolution of 3 cm . Nine scans were accumulated to
ensure a sufficient signal-to-noise ratio.

5
0
A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
X-ray photoelectron spectroscopic (XPS) measurements were
performed using a XPS-7000 spectrometer (Rigaku) equipped with
Mg K X-ray source.
(CVs) were recorded at potential sweep rates of 5, 10, 20, 50,
ꢁ1
100 mV s in 0.1 M KOH solution. Linear polarization measure-
ꢁ
1
a
ments were conducted at a sweep rate of 5 mV s using rotating
disk electrode (RDE) with rotation rate from 400 to 1600 rpm in
alkaline medium. The electrolytes were aqueous solutions bubbled
with either oxygen or argon. The bubbling time with oxygen or
argon was roughly 30 minutes. CVs were also recorded in 0.1 M
Fourier transform infrared spectroscopic (FT-IR) measurements
were performed by means of Nicolet 380 FT-IR spectrometer. The
amount of the sample was 10 times lower than in the standard
procedure of measurement (1 mg) in order to ensure transparency
of the disks formed by pressing the mixture of the sample (0.1 mg)
and KBr (200 mg). The FT-IR spectra were recorded in the spectral
2 2
KOH containing hydrogen peroxide (1.2 mM H O , equal to the
solubility of oxygen) bubbled with argon before the measure-
ꢁ1
range of 400–4000 cm . The FT-IR spectra were obtained by co-
ments. All the electrochemical experiments were carried out at
ꢁ1
ꢀ
addition of 64 scans recorded with a resolution of 4 cm . Spectra
25 ꢃ1 C. All potentials are referred to Ag/AgCl/KClsat
.
TM
processing was performed using Thermo Scientific OMNIC
software package.
3. Results and discussion
Nitrogen adsorption/desorption analysis was carried out at ꢁ
ꢀ
1
96 C using an Autosorb-6B, Quantachrome surface area and pore
3.1. Physicochemical properties of the N-doped carbon materials
size analyzer. Prior to volumetric sorption measurement, the
ꢀ
samples were preheated and degassed under vacuum at 100 C for
4
The SEM images shown in Fig.1 (a,b,c,d,e,f) illustrate that C-600,
C-700 and C-800 contain carbon tubular nanostructures with
hollow channels, which were previously identified to be carbon
nanotubes (CNTs) [22]. Carbon nanotubes are formed on cobalt
nanoparticles (derived from Mg-Co-Al LDHs) acting as the catalysts
for CNT growth. Besides CNTs, the samples also contain large
aggregates of oriented platelets. SEM images reveal that the
fraction of CNTs increases with increasing synthesis temperature
for C-600, C-700 and C-800, thereby C-600 has less CNTs than C-
700 and C-800. In the case of C-600, carbon plate-like particles
dominate over other observed carbon nanostructures.
h.
Scanning electron microscopy (SEM) images were recorded
using a JEOL JSM–7500 F field emission apparatus (with an
operating voltage of 15 keV). Energy dispersive spectroscopy
(
EDS) analysis of SEM micrographs was performed.
2.4. Characterization of the electrocatalytic activity/selectivity
Electrochemical measurements were performed by means of
potentiostat/galvanostat (AUTOLAB PGSTAT204) using an electro-
chemical cell in conventional three-electrode configuration. A Pt
wire and an Ag/AgCl/KClsat electrode were used as a counter and
reference electrodes, respectively. Glassy carbon (GC) disk (5 mm
in diameter) was used as a substrate for electrochemical measure-
ments. A GC substrate was polished in aqueous alumina suspen-
sion on a polishing cloth, and then cleaned by ultrasonic treatment
in Milli-Q water and acetone. The electrode materials (2.5 mg)
The XRD patterns of C-600, C-700 and C-800 (Fig. S2 in
Supplementary materials) display two reflections at 25.9 and
ꢀ
43.3 , which are due to the (002) and (101) diffraction planes from
graphite-like structure [23–25]. These reflections are broad, which
implies that the carbon materials are more turbostratic than
crystalline (graphitic) as we previously observed for the as-
prepared samples [26]. The disorder of graphite-like structure may
be caused by the incorporation of nitrogen atoms into the carbon
framework.
were ultrasonically mixed with 1 wt. % Nafion suspension (250
for 20 min. Then the resulting dispersion (5 L) was spread on a GC
disk and dried in air at room temperature. The loading of the
mL)
m
Inspection of the data (N percentage and C/N molar ratio) in
Table 1 shows that the extent of N-doping depends on
ꢁ2
studied materials on GC was 0.25 mg cm . Cyclic voltammograms
Fig. 1. The SEM images of the N-doped carbon materials denoted as C-600 (a, b), C-700 (c, d) and C-800 (e, f).

A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
51
Table 1
Bulk (calculated from elemental analysis) CNH composition of the samples
prepared at various temperatures.
Sample
Elemental analysis (wt. %)
C
N
H
C/Na
C/Ha
C + N + H
C-600
C-700
C-800
78.4
81.5
78.1
6.4
3.1
2.0
0.7
0.8
0.7
14.3
30.7
45.6
9.4
8.6
9.4
85.5
85.4
80.8
a
molar ratio.
carbonization temperature. Comparison of the data (N percentage
and/or C/N molar ratio) from Table 1 (elemental analysis) and
Table 2 (XPS) implies that nitrogen is almost uniformly distributed
throughout the N-doped carbon framework. Note also that the
percentages of carbon determined by elemental analysis (Table 1)
and XPS (Table 2) are very similar to each other which precludes
the possibility that doping atoms (nitrogen and oxygen) are
concentrated at the surface of the carbonaceous component.
Partial graphitization of the carbonaceous component in C-600,
C-700 and C-800 can be deduced from the elemental analysis data
Fig. 2. The Raman spectra of the N-doped carbon materials denoted as C-600, C-
7
00 and C-800.
The ratio of ID1/(I + I + ID2) reaches the values of 0.68, 0.63 and
G
D1
0.56 for C-600, C-700 and C-800, respectively, and indicates that a
gradual increase in synthesis temperature diminishes the extent of
(Table 1), in particular from the C/H molar ratio. A fully crystallized
defects. In addition, the ID2/I ratio decreases from 0.7 via 0.4 down
G
graphitic carbon should have a high C/H molar ratio, whereas in the
case of a single hydrogen-terminated graphene ribbon (one ring
wide and infinite in length) would be 2.0. The studied samples
show intermediate values of C/H molar ratio comparing to the
above mentioned extremes, ranging from 9.4 via 8.6 to 9.4 for C-
to 0.3 for C-600, C-700 and C-800, respectively, which implies that
the graphitic domains subsequently become thicker as carboniza-
tion temperature increases [29].
The analysis of the Raman spectra reveals that besides ordered/
disordered graphitic carbon, some amorphous carbon is also
present in all the samples. The contribution of amorphous carbon
in C-600, C-700 and C-800 decreases with increasing synthesis
temperature as indicated by the ID3/IG ratio, i.e. 1.3, 0.9 and 0.5,
respectively.
600, C-700 and C-800, respectively.
The XPS results confirm that carbon dominates on the surface of
the samples as its concentration reaches 78.8, 82.7 and 79.4 wt. %
Table 2) for C-600, C-700 and C-800, respectively. The C 1 s spectra
(
of C-600, C-700 and C-800 are displayed in Fig. S3 (a) in
Supplementary materials. All C 1 s spectra may be deconvoluted
The Raman spectra of C-600, C-700 and C-800 in the region of
-1
2500–3200 cm show two very weak peaks centered at 2710 and
2930 cm- (Fig. 2). The former can be attributed to the first
overtone of the D1 band whereas the latter can be assigned to a
combination of the G and D modes characteristic for disturbed
graphitic structures [29]. According to the literature [30], N-doping
1
into six peak components. The main peak component at 284–
2
2
85 eV can be assigned to the sp –hybridized graphitic carbon
(
7
C
¼
C) [27,28]. The contribution of C
¼
C component for C-600, C-
00 and C-800 attaining 47.9, 51.7 and 53.7 %, respectively, which
-1
reveals that the contribution of graphitic lattice on the surface of
the samples gradually increases with synthesis temperature.
Raman spectroscopy proves that the degree of graphitization of
the carbonaceous component in C-600, C-700 and C-800 increases
with carbonization temperature. The Raman spectra of C-600, C-
may suppress the intensity of the band at ꢄ 2700 cm . Therefore
this band is less seen for C-600 than for C-700 and C-800.
Higher degree of graphitization of the carbon materials should
manifest in higher combustion temperature [20]. Thermal analysis
indicates that the onset temperature for the combustion of the
carbonaceous component in C-600, C-700 and C-800 gradually
increases as carbonization temperature becomes higher (Fig. S1 (a)
in Supplementary materials).
-
1
7
00 and C-800 (Fig. 2) in the region of 1000–1800 cm exhibit two
-
1
peaks centered at about 1360 and 1590 cm . These signals can be
considered as a combination of four Lorentzian-shaped bands (G,
-1
D1, D2, and D4) at about 1580, 1350, 1620 and 1200 cm ,
DTG (differential thermogravimetric) profiles (Fig. S1 (b) in
Supplementary materials) illustrate that all the samples has
multiple oxidation peaks. It is not clear whether these peaks
originated from various types of carbonaceous species (amorphous
carbon, disordered graphitic carbon, carbon plate-like/tubular
structures) or are caused by metal particles (Co) in contact with
carbon. The temperature of the most pronounced mass loss
respectively, with a Gaussian-shaped band (D3) at 1500 cm-
1
[29]. These bands can be attributed to an ideal graphitic lattice (G),
graphene layer edges (D1), graphene layer surface (D2), other types
of disordered graphitic lattice (D4), and amorphous carbon (D3).
The decomposition of the Raman spectra of C-600, C-700 and C-
800 are shown in Fig. S4 (a,b,c) in Supplementary materials.
ꢀ
maximum shifts towards higher values from 382 via 433 to 527 C
ꢀ
for the samples prepared at 600, 700 and 800 C, respectively.
Washing the as-prepared samples with acid insufficiently
eliminates non-carbonaceous species, namely Mg-, Co- and Al-
containing compounds derived from Mg-Co-Al LDHs [22]. Matter
et al. reported that the washed electrocatalysts still contained
Table 2
Surface (derived from XPS analysis) chemical composition of the samples prepared
at various temperatures.
Sample
XPS analysis (wt. %)
2
metal particles (Fe, Co, Ni), metal oxides (SiO and MgO) wrapped
C
N
O
Mg
Al
Co
Cl
C/Na
with carbonaceous deposit [14,15].
C-600
C-700
C-800
78.8
82.7
79.4
5.8
2.5
2.4
7.0
6.7
7.6
2.5
7.7
8.3
1.0
0.1
2.0
2.08
0.02
0.03
2.82
0.28
0.27
15.8
38.6
38.6
The XRD patterns (Fig. S2 in Supplementary materials) reveal
the presence of metallic cobalt (Co) in C-600, C-700 and C-800 (the
ꢀ
reflections at 2
(Co
aluminium hydroxide (Al(OH)
u
of 44.3, 51.9, 76.0 and 93.0 ) [22,31], cobalt carbide
a
ꢀ
molar ratio.
x
C) in C-600, C-700 and C-800 (the reflection at 2
u
of 43 ) [32],
3
) in C-700 and C-800 (the shoulder

5
2
A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
ꢀ
ꢀ
at 2
C-800 (the shoulders at 2
In addition, the Raman spectra (Fig. 2) reveal the presence of
in C-600, C-700 and C-800 (the bands at about 478, 520, 617
u
of 18.3 ) [33] and magnesium aluminium oxide (MgAl
2
ꢀ
O
4
) in
N-CNTs prepared at 550 and 750 C with the use of CH
based catalyst [19].
3
CN and Co-
u
of 31.6, 37.3, 44.4, 60.0 and 65.4 ) [22].
It is clear that the trend in the performance of C-600, C-700 and
C-800 for ORR depends on the concentration of the active sites,
which is governed by the choice of synthesis temperature. Based
on XRD, TG and Raman data, the studied samples become more
active as they have more structural defects. A beneficial role of
structural disorder in N-doped carbon materials for ORR, in
particular the exposure of the edge plane sites, was previously
reported by others [27,37–40].
3 4
Co O
-1
and 679 cm ) [34].
Inorganic species are not freely exposed on the surface of the
acid-treated materials. They are likely to be covered by carbon
deposit, particularly metallic cobalt is encased in CNTs. Similar
observations are reported for other carbon nanostructures
synthesized in the presence of cobalt [2,35,36].
Number of defects (the density of graphene layer edges) may be
3
.2. Electrocatalytic activity/selectivity of the N-doped carbon
expressed by the ID1/I
measurements (Fig. S4 (a,b,c) in Supplementary materials) [41].
The ID1/I ratio increases from 1.6 via 2.4 up to 3.5 for C-800, C-700
G
ratio estimated by Raman spectroscopic
materials
G
The electrocatalytic performance of all the samples for ORR was
assessed by rotating ring disk (RDE) and cyclic voltammetry (CV)
measurements.
and C-600, respectively. The fraction of edge plane sites increases
with decreasing synthesis temperature as a consequence of
changes in structural ordering and morphology of the carbona-
ceous component.
N-doping (Table 1, Table 2) is another important factor that
determines the electrocatalytic activity of the studied samples. In
particular, increasing concentration of pyridinic nitrogen (revealed
by XPS) seems to correlate with enhancing electrocatalytic
performance. The order of activity for ORR is the following: C-
800 < C-700 < C-600.
The current-potential curves were recorded in 0.1 M KOH
ꢁ
1
saturated with oxygen at a scan rate of 5 mV s with rotation rate
changing from 400 to 1600 rpm for GC coated with C-600, C-700
and C-800 in Fig. 3 (a,b,c). The comparison of RDE results obtained
for the samples at the rotation rate of 1600 rpm (Fig. S5 in
Supplementary materials) demonstrates a gradual improvement
in ORR activity for C–800 < C–700 < C-600. The onset potential, i.e.
ꢁ
0.03, ꢁ 0.07 and ꢁ 0.08 V, the half-wave potential, i.e. ꢁ 0.17, ꢁ
Similar dependence between activity for ORR and morphologi-
cal/textural features and contribution o pyridinic nitrogen located
on edge plane was previously evidenced for other N-doped carbon
0
0
.19 and ꢁ 0.25 V and the limiting current (at the potential of ꢁ
.6 V), i.e. 0.64, 0.60 and 0.57 mA for C-600, C-700 and C-800,
respectively, indicate that the behavior of the electrocatalysts
prepared at 600, 700 and 800 C is affected by synthesis
temperature. Similar behaviour was noticed by Kundu et al. for
3
materials derived from CH CN [10–12,14,15,19].
It is worth to mention that the performance (the onset
potential, limiting current and number of electrons) for ORR (in
ꢀ
ꢁ
1
Fig. 3. The current-potential relations recorded in oxygen-saturated 0.1 M KOH solution by sweeping the potential at 5 mV s for a GC coated with the catalysts denoted as C-
00 (a), C-700 (b) and C-800 (c) at various rotation rates.
6

A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
53
alkaline medium) of C-800 is comparable to that of N-CNTs
ꢀ
prepared at 800 C with the use of CH
42].
In order to determine the number of electrons transferred in
3
CN and Co-based catalyst
[
ORR for the studied samples the RDE data in the limiting current
regime are analyzed using the Levich equation [43]:
i = 0.62 n F A D2/3
C
y
ꢁ1/6
v1/2 = B v1/2
where i is the measured limiting current, n is the number of
electrons transferred per oxygen molecule, F is the Faraday’s
ꢁ1
constant (96485C mol ), A is the geometric electrode area
2
(
0.20 cm ), D and C are the diffusion coefficient of dissolved
ꢁ
5
2 ꢁ1
oxygen (1.9 ꢅ10 cm s ) and the concentration of dissolved
ꢁ
6
ꢁ3
oxygen (1.2 ꢅ10 mol cm ) in 0.1 M KOH, respectively;
y
is the
2
ꢁ1
kinematic viscosity of the electrolyte solution (0.01 cm s ), and
v
ꢁ1
is the rotation rate (rad s ) [44].
Fig. 4 shows the Levich plots (i versus v1/2), which represent a
linear relationship between the measured limiting current (at the
potential of ꢁ 0.6 V) and the square root of the rotation rate. The
plots obtained from experimental data are between the theoretical
lines characteristic of 2- and 4-electron reduction of oxygen. It
implies that in the presence of C-600, C-700 and C-800 molecular
oxygen undergoes simultaneously the 2- and 4-electron reduction
reactions.
In alkaline electrolytes, oxygen may be reduced through a direct
ꢁ
four electron reduction process to OH following reaction (a), or
ꢁ
through a series of two electron reduction processes via HO
2
as an
ꢁ
intermediate, reactions (b) and (c). Alternatively, the HO
2
formed
according
ꢁ
in the reaction (b) may disproportionate to HO and O
2
to reaction (d). Reactions (c) and (d) may proceed concurrently so
that the O formed in (d) is re-reduced through reaction (b)
2
followed again by (c) and (d) repeatedly until all the O
2
is
ꢁ
converted to OH . If reactions (c) and (d) are very fast, it would
ꢁ
then appear as if oxygen is reduced directly to OH through the
four electron transfer process [45].
O + 4e ! 4OHꢁ
ꢁ
(
a) O
2
+ 2H
2
ꢁ
2ꢁ
ꢁ
(
b) O
2
+ H
2
O + 2e ! HO + OH
2ꢁ
2
O + 2e ! 3OHꢁ
ꢁ
HO + H
(c)
(d)
2ꢁ
ꢁ
2
2
HO ! 2OH + O
Similar number of electrons in ORR (in alkaline medium) equal
to 3 was reported for N-doped CNTs derived from CH CN and Co/
3
GC, and other N-doped carbon materials, e.g. N-doped graphene,
N-doped CNTs, N-doped CNFs [20,24,37,46–48].
Fig. S6 (a,b,c) in Supplementary materials show the CV curves
recorded in 0.1 M KOH saturated with oxygen for GC coated with C-
6
00, C-700 and C-800. The reduction of oxygen to water or/and
hydrogen peroxide occurs at the potentials of ꢁ 0.16, ꢁ 0.17 and ꢁ
.20 V for C-600, C-700 and C-800, respectively. Similar electro-
chemical response in alkaline medium was noticed for N-doped
carbon nanostructures, e.g. nanotubes, nanofibers
22,28,37,44,49,50].
To check the electrocatalytic behaviour of the studied samples
towards hydrogen peroxide reduction the CV curves were recorded
in 0.1 M KOH with the addition of H (Fig. S7 (a,b,c) in
0
[
2 2
O
Supplementary materials). The reduction of hydrogen peroxide
occurs at the potentials of ꢁ 0.15, ꢁ 0.16 and ꢁ 0.17 V for C-600, C-
Fig. 4. The Levich plots for oxygen reduction reaction over a GC coated with the
carbon materials denoted as C-600 (circles) C-700 (squares) and C-800 (triangles).
7
00 and C-800, respectively. It directly confirms that the studied
samples are able to catalyze not only oxygen reduction but also

5
4
A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
hydrogen peroxide reduction. Both electrocatalytic reactions are
observed at the similar potentials.
The analysis based on Koutecky-Levich (K.-L.) equation is
applied to confirm that for C-600, C-700 and C-800, the reduction
process involves 2- and 4-electron reactions
ꢁ
1
kꢁ1
+ (0.62 n F A D2/3
ꢁ1/6 1/2 ꢁ1
kꢁ1
ꢁ1
ꢁ1/2
i
= i
C
y
v
)
= i
+ B
v
where i is the measured current and i
electron transfer kinetics.
k
is the current controlled by
The K.-L. plots at various potentials for GC coated with C-600, C-
00 and C-800 are shown in Fig. S8 (a,b,c) in Supplementary
7
k
materials. The kinetic current (i ) of oxygen reduction can be
obtained from the intercept of the K.-L. plots for various potentials
and then the kinetic rate constant (k) of ORR (shown in Table 3) is
obtained from the following equation [51]:
i
k
= n F A k C
Fig. 5. The Tafel plots for oxygen reduction reaction in 0.1 M KOH solution at a GC
coated with the N-doped carbon materials denoted as C-600 (circles), C-700
(squares) and C-800 (triangles).
The order of the kinetic current is the following: C-800 < C-
00 < C-700. However, there is no significant difference in the
6
values of kinetic current for the samples prepared at 600 and
ꢀ
7
00 C. The activity of C-600 and C-700 is comparable and both
both oxygen and hydrogen peroxide reduction reactions [36,58].
ꢀ
samples are more active than that prepared at 800 C.
Fig. 5 shows Tafel plots of mass-transport-corrected currents at
GC electrode coated with C-600, C-700 and C-800. The kinetic
currents, free of mass transfer effects, are obtained from the
intercepts of the K.-L. plots.
In the low-overpotential region (between ꢁ 0.10 and ꢁ 0.15 V),
where the current is not dependent on the rotation rate and the
ORR rate is kinetically controlled, the Tafel slope for C-600, C-700
ꢀ
According to the literature, Co-N
x
species are stable until 700 C
[1,2,4,7]. However, other papers report the presence of Co-N
x
species in the samples prepared at higher temperatures such as
ꢀ
8
00 or 900 C [53,59–61].
According to the literature [45,53,62], the presence of Co-N
bonds may be reflected in FT-IR spectra by the band at low
ꢁ1
ꢁ1
wavenumbers, i.e. ca. 860 cm or 418 cm . Such spectroscopic
response is not observed for C-600, C-700 and C-800 (Fig. S9).
Moreover, the presence of Co-N bonds may be manifested in Co
ꢁ1
and C-800 is ꢁ 69, ꢁ 50, ꢁ 46 mV dec , respectively. Similar Tafel
ꢁ1
slopes (the values between ꢁ 48 and ꢁ 66 mV dec ) for oxygen
reduction in 0.1 M KOH were reported for Co-containing N-doped
carbon materials [52–54].
2
p core level spectra by the signal at around 779–782 eV
[
(
45,53,62]. Analysis of the XPS spectrum recorded for C-600
shown in Fig. S3 (d)) reveals that Co 2p3/2 signal consists of four
In the high-overpotential region (between ꢁ 0.15 and ꢁ 0.25 V),
components corresponding to cobalt carbide (Co
metallic cobalt (Co) (at ꢄ 780 eV), cobalt in oxide form, i.e. Co
x
C) (at ꢄ 779 eV),
where the current is dependent on the rotation rate, the Tafel slope
3 4
O ,
ꢁ1
is ꢁ 142, ꢁ 107, ꢁ 114 mV dec . The Tafel behavior of C-600, C-700
(
(
at ꢄ 782 eV) and cobalt ions in an electronegative surrounding
and C-800 towards oxygen reduction in alkaline solution is similar
at ꢄ 784 eV). We are unable to assign fourth component strictly.
ꢁ
ꢁ
1
1
to that reported for polycrystalline Pt, e.g. ꢁ 60 and ꢁ 120 mV dec
The component at around 784 eV may be related to cobalt-chlorine
groups, which are formed during acid treatment. Co 2p3/2 signal for
cobalt carbide and metallic cobalt does not exhibit satellites lines,
whereas for the remaining compounds displays intense satellites
lines located at about 5 eV above the main XPS lines. The existence
[
[
ꢁ
55], N-doped carbon nanotubes, e.g. ꢁ 62 and ꢁ 123 mV dec
56], and Co-containing N-doped graphene, e.g. ꢁ 59 and
ꢁ1
152 mV dec [57].
The electrocatalytic performance of the studied samples may be
related to the presence of Co-N species claimed to be active for
x
of Co
x
C, Co and Co
3
O
4
in C-600 and also in C-700 and C-800 is
Table 3
ꢀ
Comparison of the Koutecky-Levich data (mean ꢃ standard deviation for two determinations) of the N-doped carbon materials prepared at 600, 700 and 800 C deposited on
GC.
Rate constant (cm s 1) x 103
ꢁ
Potential (V)
C-600/GC
Electron number
Kinetic current (mA)
ꢁ
ꢁ
ꢁ
ꢁ
0.25
0.30
0.35
0.40
2.8
2.8
2.8
2.9
1.71 ꢃ 0.01
2.19 ꢃ 0.02
2.38 ꢃ 0.02
2.46 ꢃ 0.01
26.0 ꢃ 0.1
33.8 ꢃ 0.3
36.1 ꢃ 0.1
36.4 ꢃ 0.1
C-700/GC
ꢁ
ꢁ
ꢁ
ꢁ
0.25
0.30
0.35
0.40
2.4
2.4
2.5
2.6
1.74 ꢃ 0.03
2.57 ꢃ 0.04
2.77 ꢃ 0.09
2.88 ꢃ 0.11
31.2 ꢃ 0.5
46.8 ꢃ 0.6
48.7 ꢃ 1.8
48.6 ꢃ 2.2
C-800/GC
ꢁ
ꢁ
ꢁ
ꢁ
0.25
0.30
0.35
0.40
2.6
2.6
2.7
2.8
0.68 ꢃ 0.02
1.05 ꢃ 0.01
1.29 ꢃ 0.02
1.52 ꢃ 0.01
11.2 ꢃ 0.3
17.4 ꢃ 0.1
20.7 ꢃ 0.3
23.9 ꢃ 0.3

A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
55
proved by other methods such as XRD or Raman spectroscopy. We
cannot provide any evidences to confirm the existence of Co ꢁꢁ N
bonds in the studied samples. More advanced techniques such as
ToF-SIMS (time-of-flight secondary ion mass spectroscopy), EXAFS
C-800 with increasing synthesis temperature. Such correlation
confirms that both nitrogen forms are active for ORR, in particular
N-6 is responsible for 4-electron reaction pathway whereas N-4
favor 2-electron reaction pathway. Our observations are similar to
those published by Wong et al. [67].
(
extended X-ray absorption fine structure) and XAS (X-ray
ꢀ
absorption spectroscopy) should be employed to detect Co ꢁꢁ N
For comparison, we carbonized acetonitrile at 700 C (for 30 or
bonds [1–3].
180 minutes) with the use of Co free Mg-Al LDHs. Such samples,
denoted as C-MgAl-700-30 or C-MgAl-700-180 respectively, after
acid treatment have higher nitrogen concentration (6.1 or 6.7 wt. %,
respectively) than C-700 sample (3.1 wt. %) prepared with the use
of Co-containing LDH. C-MgAl-700-30 and C-MgAl-700-180 show
similar activity (Fig. S14) and selectivity (Fig. S15) in ORR. However,
both samples are less active than C-700 (Fig. S16). It implies that
ORR activity is not proportional to the total amount of incorporated
nitrogen as previously reported in the literature [46,49] and cobalt
plays the role in the creation of the active centers.
N 1 s XPS signal recorded for C-MgAl-700-30 (Fig. S17) consists
of four components: N-6 (27.9 %), N-5 (59.1 %), N-4 (8.0 %) and N-O
(5.0 %). It reveals that the sample prepared with cobalt (C-700)
shows larger percentage of N-6 (34.2 vs. 27.9 %) and N-4 (11.5 vs.
8.0 %) but lower percentage of N-5 (48.8 vs. 59.1 %) than the sample
prepared without cobalt (C-MgAl-700-30). As C-700 is more active
than C-MgAl-700-30 it suggests that N-6 and/or N-4 are more
important for ORR activity than N-5 as reported in the literature
[24,46], or another reason may affect better performance of C-700
as compared to C-MgAl-700-30.
The concentration of cobalt determined by XPS is 2.08, 0.02 and
0
.03 wt. % for C-600, C-700 and C-800, respectively (Table 2). We
cannot confirm the relation between the electrocatalytic perfor-
mance of C-600, C-700 and C-800 and the content of Co-containing
species estimated by XPS. Indeed, C-600 is more active than C-800.
However, the activities of C-600 and C-700 are comparable despite
the concentration of cobalt estimated by XPS is significantly higher
for C-600 than for C-700. It is worth to underline that the kinetic
currents calculated for C-700 are even slightly higher in compari-
son to those calculated for C-600 (Table 3). The concentration of
ꢀ
cobalt evaluated by EDS for the samples prepared at 600 and 800 C
is 2.9 and 1.4 wt. %, respectively. Considering the difference in the
analysis depth between XPS (up to 10 nm) and EDS (up to 1 mm)
methods it suggests that Co-containing species do not appear at
the surface of the samples. The distinct difference in the
concentration of cobalt measured by XPS and EDS for C-800
(
0.03 and 1.4 wt. %, respectively), suggests that Co-containing
species are deeply covered by carbon deposit.
We assume that the performance of C-600, C-700 and C-800 for
ORR is determined by the coexistence of structural defects and N-
and O-containing functionalities.
XPS confirms the presence of N- and O-containing groups
incorporated in the carbon framework of carbonaceous constitu-
ent of C-600, C-700 and C-800 (Table 2). The N 1 s spectra of C-600,
C-700 and C-800 are shown in Fig. S3 (b) in Supplementary
materials. N 1 s signals consist of four components identified as
pyridinic nitrogen (denoted as N-6) (398–399 eV), pyrrolic
nitrogen (denoted as N-5) (400–401 eV), quaternary nitrogen
According to SEM images (Fig. S18), C-MgAl-700-30 does not
contain carbon nanotubes but only plate-like carbon particles. The
presence of carbon platelets in C-MgAl-700-30 results in higher
2
ꢁ1
specific surface area (SBET = 881 m g ) and higher fraction of edge
plane sites (ID1/I = 4.5), while the presence of CNTs in C-700 lead
to lower specific surface area (SBET = 225 m g ) and lower fraction
of edge plane sites (ID1/I = 2.4). The concentration of the active
G
2
ꢁ1
G
sites located on edge planes (N-6/N-5) is higher in C-MgAl-700-30
than in C-700, however the sample prepared without cobalt is less
active than that prepared with cobalt. Therefore, the higher activity
of C-700 may be caused by the existence of nitrogen-coordinated
Co centers (Co-N-C), which presence cannot be however confirmed
by applied techniques.
(
denoted as N-4) (402–403 eV) and oxidized nitrogen (denoted as
N-O) (404–405 eV) [63]. The contribution of N-6 component
decreases from 37.4 via 34.2 down to 24.4 % whereas the
contribution of N-4 component increases from 9.2 via 11.5 up to
2
0.4 % with increasing synthesis temperature for C-600, C-700 and
The proportion of nitrogen atoms located on graphitic edge
plane (N-6) to nitrogen atoms located on graphitic basal plane (N-
4) is roughly equal to 3.0 : 1 and 3.5 : 1 in C-700 and C-MgAl-700-
30, respectively. The number of electrons (according to the Levich
approach) involved in ORR is 2.2 and 2.9 for C-700 and C-MgAl-
700-30, respectively (Fig. S15). It shows that the presence of N-6 is
more advantageous for ORR selectivity than the presence of N-4.
Not only N-containing groups but also O-containing groups
being present in C-600, C-700 and C-800 affect the ORR
performance. It is claimed that oxygen molecule much easily
adsorbs on carbon atom located near O-containing groups, which
facilitates 2-electron reduction [28]. In addition, oxygen atoms
located in the carbon framework may be involved in the ORR as
mediators [68,69].
C-800, respectively. It is consistent with previous paper [64]
showing that N-6 groups are not thermally stable and they convert
into N-4 groups. The contribution of N-5 does not vary for C-600, C-
700 and C-800, i.e. 48.0, 48.8 and 48.7 %, respectively. Among N-
containing species, N-5 is dominating whereas the contribution of
N-O is very low. Faubert et al. observed a similar trend with
increasing carbonization temperature (i.e. decrease in N-6 content,
increase in N-4 content, no change in N-5 content) for the samples
prepared at 600, 800 and 1000 C with the use of CH
ꢀ
3
CN and Fe/C
[10].
It has been reported that pyridinic (N-6), pyrrolic (N-5) and
quaternary (N-4) nitrogens are responsible for ORR activity of N-
doped carbon materials [24,46,65]. It is not clearly defined, which
nitrogen forms are associated with 2-electron and which ones with
The surface oxygen concentration determined by XPS is 7.0, 6.7
and 7.6 wt. % for C-600, C-700 and C-800, respectively. The O 1 s
spectra for C-600, C-700 and C-800 samples are shown in Fig. S3 (c)
4-electron reduction [66]. It is claimed that adsorption of oxygen
molecule on carbon atom adjacent to nitrogen atom by the side-
one mode (known as the Yeager model) facilitates 4-electron
reduction leading to OH [28].
in Supplementary materials.
O 1 s signals consist of five
ꢁ
components characteristic of O-containing groups such as oxygen
Based on the literature [48,67], N-5 and N-6 are responsible for
atoms connected to metal atoms (530–531 eV), oxygen atoms in
4
-electron reduction, whereas N-4 is responsible for 2-electron
ꢁꢁ OH groups (531–532 eV), oxygen atoms in C
¼
O and C ꢁꢁ O ꢁꢁ C
reduction. The studied samples contain the sites, which could be
responsible for both reaction pathways. RDE data (Fig. 4) demon-
strate that number of electrons involved in ORR approaches 3 (for
C-600), oscillates between 3 and 2 (for C-700) or approaches 2 (for
C-800). It agrees well with XPS results showing that the
contribution of N-6 and N-4 gradually changes for C-600, C-700,
groups (532–533 eV), oxygen atoms in carboxyl (COOH) and
2ꢁ
carbonate (CO
3
) groups and in adsorbed water (H O) (533–
2
534 eV) and oxygen atoms in oxygen (O ) molecules or bonded to
2
aromatic carbon compounds (535–536 eV) [28,63]. Analysis of XPS
spectra reveals that only a small amount of oxygen (less than 3 %) is
attributed to metal oxides/hydroxides, i.e. MgAl O , Al(OH) and
2 4 3

5
6
A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
Co
3
O
4
, whereas the major part of oxygen (97 %) is associated with
data from Fig. S12 (a,b) in Supplementary materials and [22]),
which indicates that the former are more active for ORR than the
latter. In particular, higher kinetic currents (> 1 mA) (Table 3) are
observed for the acid-treated catalysts (C-700 and C-800) than
those (< 1 mA) for the corresponding as-prepared samples [22].
In addition, Fig. S13 (a,b) in Supplementary materials show that
oxygen incorporated in the carbon framework. According to the
literature [16,28,42,70,71], carbonyl (including carbonyl oxygen of
quinones) and carboxyl groups present on the surface of the carbon
materials, e.g. N-doped CNTs, are responsible for their activity for
ORR. The oxygen molecule is adsorbed on carbon atom in carbonyl
ꢁ1
group by the end-one mode (known as the Pauling model), which
facilitates 2-electron reduction leading to HO
the apparent activity (mA mgcat ) of the acid-treated materials is
higher than that of the corresponding as-prepared materials.
Limiting current normalized to the mass of the catalyst is 12.0 and
ꢁ
2
[28]. More precise
analysis of XPS spectra reveals the contribution of carbonyl (24.0,
.8 and 7.2 %) and carboxyl (12.5, 42.1 and 43.8 %) groups for C-600,
ꢁ1
7
6.1 mA mgcat
for the acid-treated sample (C-700) and the
C-700 and C-800, respectively. The total contribution of carbonyl
and carboxyl groups, i.e. 36.5 to 49.9 and 51.0 for C-600, C-700 and
C-800, respectively, increases with rising synthesis temperature.
Thus, a gradual decrease in number of electrons from 3 down to 2
with rising synthesis temperature, illustrated in the Levich plots
corresponding as-prepared sample, respectively. The values for
another acid-treated sample (C-800) and its corresponding as-
ꢁ
1
prepared sample are 11.4 and 7.7 mA mgcat , respectively.
The acid-treated materials appear to be more active than the
corresponding as-prepared materials a consequence of higher
exposure of carbonaceous component and better access to the
active sites. Similar observations for washed and unwashed
(
Fig. 4), is not only a result of the conversion of N-6 into N-4 but
also consequence of increasing amount of O-containing
a
functionalities responsible for 2-electron reaction pathway.
Metallic cobalt or cobalt in the form of the oxide in the studied
samples (evidenced by XRD, XPS) is not accessible by the
electrolyte (dissolved oxygen) and therefore it does not participate
in ORR. The presence of metallic cobalt or cobalt in the form of the
oxide exposed on the surface of the electrode materials should be
manifested in CV curves recorded in alkaline medium by a pair of
peaks associated with redox transformation between cobalt
catalysts derived from CH
reported by Singh et al. [18].
3
CN and Co-based catalysts were
The number of electrons involved in electrode process for the
acid-treated samples (n ꢄ 3) and for the as-prepared samples
(n ꢄ 2) is different. It indicates that washing the as-prepared
materials with acid removes freely exposed cobalt and as a
consequence the acid-treated materials become less selective for
2-electron reduction than the corresponding as-prepared ones as
they do not contain cobalt which is active for reduction of oxygen
to hydrogen peroxide.
hydroxide (Co(OH)
electrochemical response is not observed for C-600, C-700 and C-
00 (Fig. S6 (a,b,c)), which confirms that metallic cobalt or cobalt in
2
) and cobalt oxyhydroxide (CoOOH) [72]. Such
8
Our results are consistent with those reported by Matter et al.
[15]. The authors also observed that washing the catalyst derived
from CH CN and Co/SiO with acid resulted in improving
3 2
the form of the oxide existing in the studied samples is not
electrocatalytically active. Our results agree with the published
data demonstrating that metallic cobalt wrapped with carbona-
ceous layers is not responsible for ORR activity [2,3,7,24,73,74]. In
our previous publication [19] we showed that cobalt enclosed
selectivity for 4-electron reduction.
4. Conclusions
3
insight N-CNTs derived from CH CN and Mg-Co-Al LDHs or Co-Al
LDHs was electrochemically silent. Kundu et al. also demonstrated
Novel electrocatalysts derived from CH
were prepared at 600, 700 and 800 C. The choice of synthesis
temperature determines their activity/selectivity for ORR evaluat-
3
CN and Mg-Co-Al LDHs
ꢀ
that cobalt encapsulated in N-CNTs derived from CH
electrocatalytically inactive [19].
3
CN was
ꢀ
Treating the as-prepared catalysts with acid eliminates
inorganic compounds (metal/metal oxides) and creates cavities.
The comparison of pore size distribution for the acid-treated (C-
ed in alkaline medium. The samples prepared at 600 and 700 C are
ꢀ
more active than that prepared at 800 C. Higher activity of C-600
and C-700 is associated with higher contribution of carbon
nanostructures with plate-like morphology providing higher
specific surface area and offering higher population of the active
sites. Our results show that carbon nanotubes are not advanta-
geous for ORR performance.
The proportion of nitrogen atoms located on graphitic edge
plane (N-6) to nitrogen atoms located on graphitic basal plane (N-
4) is roughly equal to 4.4 : 1, 3.0 : 1 and 1.2 : 1 in C-600, C-700 and
C-800, respectively. It reveals that the presence of N-6 is more
beneficial for ORR activity/selectivity than the presence of N-4. It
also indicates that not all morphological surfaces of N-doped
graphite-like carbons are populated with the sites equally active/
selective for ORR.
600, C-700 and C-800) and corresponding as-prepared samples is
shown in Fig. S10 (a,b,c) in Supplementary materials.
Based on nitrogen sorption data (Fig. S11 in Supplementary
materials), C-600, C-700 and C-800 contain predominantly slit-
shaped mesopores and/or mesopores occurring in aggregates of
platy particles [75]. The average pore diameter is 7, 11 and 9 nm for
C-600, C-700 and C-800, respectively. The total pore volume is
3
ꢁ1
0
.60, 0.63 and 0.32 cm g
for C-600, C-700 and C-800,
3
ꢁ1
respectively, whereas it is only 0.14, 0.07 and 0.08 cm g
for
the corresponding as-prepared materials. The acid-treated sam-
ples have also higher specific surface areas, i.e. 342, 225 and
2
ꢁ1
1
33 m g
for C-600, C-700 and C-800, respectively, than the
corresponding as-prepared samples. For comparison, the specific
The N-doped carbon materials are able to catalyze the oxygen
reduction reaction by 2 and 4 electrons. The order of selectivity for
4-electron reduction is the following: C-800 < C-700 < C-600.
Number of transferred electrons for the most selective sample
approaches 3. Among studied samples, C-600 has the highest
concentration of N-containing groups, in particular N-6 being
responsible for 4-electron reduction. Moreover, C-600 has the
lowest concentration of O-containing groups, especially carbonyl
and carboxyl being responsible for 2-electron reduction.
The electrocatalytic performance of C-600, C-700 and C-800
cannot be related only to the coexistence of structural defects and
N- and O-containing functionalities. The presence of nitrogen-
coordinated Co centers (Co-N-C) in the studied samples is not
confirmed by FT-IR and XPS. However, the existence of Co-N-C
surface area of the as-prepared samples obtained at 600, 700 and
ꢀ
2 ꢁ1
8
00 C is 116, 9 and 13 m g , respectively [22].
The onset potential for the acid-treated samples and corre-
ꢀ
sponding as-prepared materials synthesized at 700 and 800 C, i.e.
ꢁ
0.07 vs. ꢁ 0.07 V and ꢁ 0.08 vs. ꢁ 0.10 V, respectively [22], is
similar. In contrast, the half-wave potential for the acid-treated
samples and corresponding as-prepared materials synthesized at
ꢀ
7
00 and 800 C, i.e. ꢁ 0.19 vs. ꢁ 0.23 V and ꢁ 0.25 vs. ꢁ 0.31 V,
respectively, is shifted by 40 and 60 mV.
The Tafel slopes (in the low-overpotential region) for the acid-
treated samples are lower than those for corresponding as-
ꢀ
prepared materials synthesized at 700 and 800 C, i.e. ꢁ 50 vs.
ꢁ1
ꢁ1
ꢁ
64 mV dec and ꢁ 46 vs. ꢁ 65 mV dec , respectively (compared

A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
57
cannot be excluded especially because the sample prepared with
Mg-Co-Al LDHs appears to be more active than that prepared with
Mg-Al LDHs, which suggests that Co-containing centers are likely
to be involved in ORR.
The acid-treated materials appear to be more active for ORR
than the corresponding as-prepared materials as a consequence of
higher exposure of carbonaceous component and better access to
the active sites. In addition, washing the as-prepared materials
with acid removes freely exposed cobalt, as a result the acid-
treated materials become less selective for 2-electron reduction
than the corresponding as-prepared ones.
[19] S. Kundu, T. Chikka Nagaiah, W. Xia, Y. Wang, S. van Dommele, J.H. Bitter, M.
Santa, G. Grundmeier, M. Bron, W. Schuhmann, M. Muhler, J. Phys. Chem. C 113
(
2009) 14302–14310.
20] C.-M. Chen, Y.-M. Dai, J.G. Huang, J.-M. Jehng, Carbon 44 (2006) 1808–1820.
[21] A. Pacuła, R. Mokaya, Microporous and Mesoporous Mater. 106 (2007) 147–
54.
[
1
[
[
22] A. Pacuła, K. Ikeda, T. Masuda, K. Uosaki, J. Power Sources 220 (2012) 20–30.
23] G. Liu, X. Li, P. Ganesan, B.N. Popov, Applied Catalysis B: Environmental 93
(2009) 156–165.
[24] D. Shin, B. Jeong, B.S. Mun, H. Jeon, H.-J. Shin, J. Baik, J. Lee, J. Phys. Chem. C. 117
(
2013) 11619–11624.
[25] H.-J. Zhang, H.-Ch. Kong, X. Yuan, Q.-Z. Jiang, J. Yang, Z.-F. Ma, Int. J. Hydrogen
Energ. (2012) 1–8.
[26] A. Pacuła, P. Nowak, W. Makowski, R.P. Socha, Microporous Mesoporous Mater.
01 (2015) 1–9.
2
[
[
27] S. Maldonado, S. Morin, K.J. Stevenson, Carbon 44 (2006) 1429–1437.
28] R.-S. Zhong, Y.-H. Qin, D.-F. Niu, J.-W. Tian, X.-S. Zhang, X.-G. Zhou, S.-G. Sun,
W.-K. Yuan, J. Power Sources 225 (2013) 192–199.
Acknowledgments
[
[
29] A. Sadezky, H. Muckenhuber, H. Grothe, R. Niessner, U. Poschl, Carbon 43
(2005) 1731–1742.
30] S. Maldonado, K.J. Stevenson, J. Phys. Chem. B 109 (2005) 4707–4716.
This work was financially supported by World Premier
International Research Center (WPI) Initiative on Materials Nano-
architectonics from the Ministry of Education, Culture, Sports,
Science and Technology (MEXT), Japan.
Electrochemical measurements were conducted using the
equipment purchased thanks to the financial support provided
by National Science Centre in Poland within decision number DEC-
[
31] Y. Wu, J. Zang, L. Dong, Y. Zhang, Y. Wang, Y. Zhang, J. Power Sources 305 (2016)
6
4–71.
[32] J. Chakkamadathil Mohandas, M. Kumaran Gnanamani, G. Jacobs, W. Ma, Y. Ji,
S. Khalid, B.H. Davis, ACS Catal. 1 (2011) 1581–1588.
[
[
33] E. Wang, J. Chen, X. Hu, K.-Ch. Chou, X. Hou, Ceram. Int. 42 (2016) 11429–11434.
34] C.-W. Tang, C.-B. Wang, S.-H. Chien, Thermochimica Acta 473 (2008) 68–73.
[35] G. Wei, J.S. Wainright, R.F. Savinell, J. New Mat. Electrochem. Systems 3 (2000)
21–129.
1
2013/11/B/ST5/01417.
[36] Z. Wang, S. Xiao, Z. Zhu, X. Long, X. Zheng, X. Lu, S. Yang, ACS Appl. Mater.
Raman spectra were recorded by means of the equipment
Interfaces 7 (2015) 4048–4055.
purchased thanks to the financial support of the European Regional
Development Fund in the framework of the Polish Innovation
Economy Operational Program (contract no. POIG.02.01.00-12-
[37] D. Higgins, Z. Chen, Z. Chen, Electrochim. Acta 56 (2011) 1570–1575.
[38] G. Zhong, H. Wang, H. Yu, F. Peng, Electrochemistry Commun. 40 (2014) 5–8.
[
[
39] S. Yasuda, L. Yu, J. Kim, K. Murakoshi, Chem. Commun. 49 (2013) 9627–9629.
40] Y. Jiang, L. Yang, T. Sun, J. Zhao, Z. Lyu, O. Zhuo, X. Wang, Q. Wu, J. Ma, Z. Hu, ACS
Catalysis 5 (2015) 6707–6712.
0
23/08).
Prof. Katsuaki Shimazu and Dr. Toshikazu Kawaguchi (Hokkaido
[41] Y. Wang, D.C. Alsmeyer, R.L. McCreery, Chem. Mater. 2 (1990) 557–563.
[
[
[
[
[
42] N. Alexeyeva, E. Shulga, V. Kisand, I. Kink, K. Tammeveski, J. Electroanalytical
University) are gratefully acknowledged for XPS measurements. Dr
Paweł Nowak (Jerzy Haber Institute of Catalysis and Surface
Chemistry, Polish Academy of Sciences) is gratefully acknowledged
for fruitful discussions.
Chem. 648 (2010) 169–175.
43] V.S. Bagotsky, Fundamentals of Electrochemistry, second ed., John Wiley &
Sons, Inc, Hoboken, New Jersey, 2006.
44] M. Vikkisk, I. Kruusenberg, U. Joost, E. Shulga, K. Tammeveski, Electrochim.
Acta 87 (2013) 709–716.
45] J. Masa, A. Zhao, W. Xia, M. Muhler, W. Schuhmann, Electrochim. Acta 128
(
2014) 271–278.
Appendix A. Supplementary data
46] L. Lai, J.R. Potts, D. Zhan, L. Wang, Ch. Kok Poh, Ch. Tang, H. Gong, Z. Shen, J. Lin,
R.S. Ruoff, Energy Environ. Sci. 5 (2012) 7936–7942.
Supplementary data associated with this article can be found, in
[47] M. Vikkisk, I. Kruusenberg, U. Joost, E. Shulga, I. Kink, K. Tammeveski, Applied
Catalysis B: Environmental 147 (2014) 369–376.
the
online
version,
at
http://dx.doi.org/10.1016/j.
[48] H. Li, W. Kang, L. Wang, Q. Yue, S. Xu, H. Wang, J. Liu, Carbon 54 (2013) 249–257.
electacta.2016.06.150.
[49] T.C. Nagaiah, S. Kundu, M. Boron, M. Muhler, W. Schuhmann, Electrochemistry
Commun. 12 (2010) 338–341.
[
[
50] Z. Mo, S. Liao, Y. Zheng, Z. Fu, Carbon 50 (2012) 2620–2627.
51] R. Jiang, D. Chu, J. Power Sources 245 (2014) 352–361.
References
[
[
52] X. Li, G. Liu, B.N. Popov, J. Power Sources 195 (2010) 6373–6378.
53] M. Li, X. Bo, Y. Zhang, C. Han, A. Nsabimana, L. Guo, J. Mater. Chem. A 2 (2014)
[
[
[
[
[
[
1] G. Faubert, G. Lalande, R. Côté, D. Guay, J.P. Dodelet, L.T. Weng, P. Bertrand, G.
Dénès, Electrochim. Acta 41 (1996) 1689–1701.
2] V. Nallathambi, J.-W. Lee, S.P. Kumaraguru, G. Wu, B.N. Popov, J. Power Sources
11672–11682.
[
[
[
54] A. Sarapuu, L. Samolberg, K. Kreek, M. Koel, L. Matisen, K. Tammaveski, J.
183 (2008) 34–42.
Electroanal. Chem. 746 (2015) 9–17.
3] A.L. Bouwkamp-Wijnoltz, W. Visscher, J.A.R. van Veen, S.C. Tang, Electrochim.
Acta 45 (1999) 379–386.
4] S. Li, L. Zhang, H. Liu, M. Pan, L. Zan, J. Zhang, Electrochim. Acta 55 (2010) 4403–
55] N. Alexeyeva, K. Tammaveski, A. Lopez-Cudero, J. Solla-Gullón, J.M. Feliu,
Electrochimica Acta 55 (2010) 794–803.
56] Z. Chen, D. Higgins, Z. Chen, Carbon 48 (2010) 3057–3065.
4
411.
[
57] X. Fu, J. Jin, Y. Liu, Q. Liu, Q. Niu, J. Zhang, X. Cao, Electrochem. Commun. 28
5] U.I. Kramm, I. Herrmann-Geppert, S. Fiechter, G. Zehl, I. Zizak, I. Dorbandt, D.
Schmeißer, P. Bogdanoff, J. Mater. Chem. A 2 (2014) 2663–2670.
6] C.W.B. Bezerra, L. Zhang, K. Lee, H. Liu, A.L.B. Marques, E.P. Marques, H. Wang, J.
Zhang, Electrochim. Acta 53 (2008) 4937–4951.
(
2013) 5–8.
[58] Z.X. Liu, B.H. Liu, Z.P. Li, Int. J. Hydrogen Energ. 39 (2014) 5689–5695.
[
59] Z. Luo, B. Zhao, Y. Lin, H. Zhang, Z. Tang, J. Li, J. Yang, Electrochim. Acta 161
(
2015) 72–79.
[
[
7] H.-S. Oh, H. Kim, J. Power Sources 212 (2012) 220–225.
8] U.I. Kramm, I. Herrmann-Geppert, J. Behrends, K. Lips, S. Fiechter, P. j.
Bogdanoff, Am. Chem. Soc. 138 (2016) 635–640.
[
[
60] L. Wang, L. Zhang, L. Bai, L. Han, S. Dong, Electrochem. Commun. 34 (2013) 68–
7
2.
61] H. Wang, X. Bo, A. Wang, L. Guo, Electrochem. Commun. 36 (2013) 75–79.
62] W.M. Millán, T.T. Thompson, L.G. Arriaga, M.A. Smit, Int. J. Hydrogen Energ. 34
2009) 694–702.
[
9] G. Lalande, R. Côté, D. Guay, J.P. Dodelet, L.T. Wenig, P. Bertrand, Electrochim.
Acta 42 (1997) 1379–1388.
[
(
[
10] G. Faubert, R. Côté, D. Guay, J.P. Dodelet, G. Dénès, C. Poleunis, P. Bertrand,
[63] A. Pacuła, R.P. Socha, M. Zimowska, M. Ruggiero-Mikołajczyk, D. Mucha, P.
Electrochim. Acta 43 (1998) 1969–1984.
Nowak Applied Clay Science 72 (2013) 163–174.
[
[
[
11] P.H. Matter, L. Zhang, U.S. Ozkan, J. Catal. 239 (2006) 83–96.
12] P.H. Matter, U.S. Ozkan, Catal. Lett. 109 (2006) 115–123.
13] P.H. Matter, E. Wang, M. Arias, E.J. Biddinger, U.S. Ozkan, J. Phys. Chem. B 110
2006) 18374–18384.
14] P.H. Matter, E. Wang, U.S. Ozkan, J. Catal. 243 (2006) 395–403.
[
[
64] G. Liu, X. Li, P. Ganesan, B.N. Popov, Electrochim. Acta 55 (2010) 2853–2858.
65] H. Shi, Y. Shen, F. He, Y. Li, A. Li, S. Liu, Y. Zhang, J. Mater. Chem. A 2 (2014)
1
5704–15716.
66] Z. Yang, H. Nie, X. Chen, X. Chen, S. Huang, J. Power Sources 236 (2013) 238–
49.
(
[
[
[
[
2
15] P.H. Matter, E. Wang, M. Arias, E.J. Biddinger, U.S. Ozkan, A, J. Mol. Catal. 264
67] W.Y. Wong, W.R.W. Daud, A.B. Mohamad, A.A.H. Kadhum, E.H. Majlan, K.S. Loh,
(
2007) 73–81.
16] E.J. Biddinger, D. von Deak, U.S. Ozkan, Top. Catal. 52 (2009) 1566–1574.
17] P.H. Matter, E. Wang, M. Arias, J.-M.M. Millet, U.S. Ozkan, J. Phys. Chem. C 111
2007) 1444–1450.
18] D. Singh, I.I. Soykal, J. Tian, D. von Deak, J. King, J.T. Miller, U.S. Ozkan, J. Catal.
04 (2013) 100–111.
Diamond & Related Materials 22 (2012) 12–22.
[
[
[
[
68] J. Wu, D. Zhang, Y. Wang, Y. Wan, B. Hou, J. Power Sources 198 (2012) 122–126.
69] I. Kruusenberg, J. Mondal, L. Matisen, V. Sammelselg, K. Tammeveski,
Electrochem. Commun. 33 (2013) 18–22.
(
[
[
70] M. Zhang, Y. Yan, K. Gong, L. Mao, Z. Guo, Y. Chen, Langmuir 20 (2004) 8781–
3
8
785.

5
8
A. Pacuła et al. / Electrochimica Acta 212 (2016) 47–58
[
[
[
71] I. Kruusenberg, J. Mondal, L. Matisen, V. Sammelselg, K. Tammeveski,
Electrochemistry Commun. 33 (2013) 18–22.
72] H. Huang, Q. Wang, Q. Wei, Y. Huang, Int. J. Hydrogen Energ. 40 (2015) 6072–
[74] H. Niwa, M. Kobayashi, K. Horiba, Y. Harada, M. Oshima, K. Terakura, T. Ikeda, Y.
Koshigoe, J. Ozaki, S. Miyata, S. Ueda, Y. Yamashita, H. Yoshikawa, K. Kobayashi,
J. Power Sources 196 (2011) 1006–1011.
[75] F. Rouquerol, J. Rouquerol, K. Sing, Adsorption by powders and porous solids,
Principles, methodology and applications, Academic Press, London, 1999.
6
084.
73] N.P. Subramanian, X. Li, V. Nallathambi, S.P. Kumaraguru, H. Colon-Mercado, G.
Wu, J.-W. Lee, B.N. Popov, J. Power Sources 188 (2009) 38–44.