Investigation of activity of sputtered transition-metal (TM)-C-N (TM = V, Cr, Mn, Co, Ni) Catalysts for Oxygen Reduction Reaction

Article abstract of DOI:10.1149/1.2803519

Variations with heat-treatment temperature of the structure and the electrocatalytic activity of sputtered transition-metal (TM)-C-N (TM=V, Cr, Mn, Co, and Ni) films for the oxygen reduction reaction (ORR) in acid and alkaline electrolytes were studied. The films prepared were all originally amorphous. At a critical heat-treatment temperature (Tc, different for each composition) substantial nitrogen release occurred and the films transformed to a heterogeneous N-containing carbon structure with either Cr3 C2, Co, Ni, V8 C7, or Mn7 C3, depending on the TM used. In acid electrolyte, heat-treated Crx C1-x-y Ny, Cox C1-x-y Ny, and Nix C1-x-y Ny films began to show oxygen reduction activity at Tc, while Vx C1-x-y Ny and Mnx C1-x-y Ny films showed little or no activity for ORR in acid electrolyte for all temperatures studied. However, all of the heat-treated TMx C1-x-y Ny (TM=V, Cr, Mn, Co, and Ni) films showed oxygen reduction activity in alkaline electrolyte even when heated to temperatures below Tc. Because all the TMx C1-x-y Ny films heated above Tc contained N-containing carbon, the activity of heat-treated sputtered C1-x Nx and C films, as well as graphite powder, were studied for comparison. Their activities were much lower in acid electrolyte than those of the Crx C1-x-y Ny, Cox C1-x-y Ny, and Nix C1-x-y Ny films heat-treated at the same temperatures. C1-x Nx films, C films, and graphite powder all showed activity in alkaline electrolyte but were less active than some heated Cox C1-x-y Ny and Nix C1-x-y Ny films. The choice of TM and the heat-treatment temperature played important roles in determining the activity of the sputtered TMx C1-x-y Ny films for ORR in acid and alkaline electrolytes. The corrosion stability of the heat-treated TMx C1-x-y Ny libraries was studied as well. Heat-treated Crx C1-x-y Ny libraries showed good passivation against corrosion.

Full text of DOI:10.1149/1.2803519

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B79
0013-4651/2007/155͑1͒/B79/13/$23.00 © The Electrochemical Society
Investigation of Activity of Sputtered Transition-Metal
(TM)–C–N (TM ؍ V, Cr, Mn, Co, Ni) Catalysts for Oxygen
Reduction Reaction
,z
*
*
Ruizhi Yang, Krystal Stevens, and J. R. Dahn
Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Nova Scotia,
Canada B3H 3J5
Variations with heat-treatment temperature of the structure and the electrocatalytic activity of sputtered transition-metal ͑TM͒–
C–N ͑TM = V, Cr, Mn, Co, and Ni͒ films for the oxygen reduction reaction ͑ORR͒ in acid and alkaline electrolytes were studied.
The films prepared were all originally amorphous. At a critical heat-treatment temperature ͑Tc, different for each composition͒
substantial nitrogen release occurred and the films transformed to a heterogeneous N-containing carbon structure with either
Cr3C2, Co, Ni, V8C7, or Mn7C3, depending on the TM used. In acid electrolyte, heat-treated CrxC1−x−yNy, CoxC1−x−yNy, and
NixC1−x−yNy films began to show oxygen reduction activity at Tc, while VxC1−x−yNy and MnxC1−x−yNy films showed little or no
activity for ORR in acid electrolyte for all temperatures studied. However, all of the heat-treated TMxC1−x−yNy ͑TM = V, Cr, Mn,
Co, and Ni͒ films showed oxygen reduction activity in alkaline electrolyte even when heated to temperatures below Tc. Because
all the TMxC1−x−yNy films heated above Tc contained N-containing carbon, the activity of heat-treated sputtered C1−xNx and C
films, as well as graphite powder, were studied for comparison. Their activities were much lower in acid electrolyte than those of
the CrxC1−x−yNy, CoxC1−x−yNy, and NixC1−x−yNy films heat-treated at the same temperatures. C1−xNx films, C films, and graphite
powder all showed activity in alkaline electrolyte but were less active than some heated CoxC1−x−yNy and NixC1−x−yNy films. The
choice of TM and the heat-treatment temperature played important roles in determining the activity of the sputtered TMxC1−x−yNy
films for ORR in acid and alkaline electrolytes. The corrosion stability of the heat-treated TMxC1−x−yNy libraries was studied as
well. Heat-treated CrxC1−x−yNy libraries showed good passivation against corrosion.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2803519͔ All rights reserved.
Manuscript submitted July 16, 2007; revised manuscript received October 1, 2007. Available electronically November 14, 2007.
The increasing need for polymer electrolyte membrane fuel cells
͑PEMFCs͒ as power sources for residential and transportation appli-
cations has intensified research on the oxygen reduction catalysts
used in these cells. So far, Pt and Pt–alloy catalysts are the best
electrocatalysts for the oxygen reduction reaction ͑ORR͒ due to their
ability to reduce oxygen efficiently to water in a four-electron re-
duction. However, Pt is expensive and of limited availability. The
replacement of Pt by non-noble metal catalysts could lead to signifi-
cant cost reduction in PEMFCs and is desirable, provided that simi-
lar area-specific catalytic activity can be obtained.1,2
Comprehensive reviews on the development of non-noble cata-
lysts for PEMFCs have been published recently.2-4 Heat-treating
carbon-supported transition-metal ͑TM͒ and nitrogen-containing
complexes5-13 or TM salts14-16 in inert gas, ammonia, or acetonitrile
vapor has led to promising ORR electrocatalysts. Dodelet and co-
workers proposed that FeN2/C and FeN4/C or CoN4/C active sites
exist in Fe-based17 and Co-based catalysts,18 based on time-of-flight
secondary ion mass spectrometry ͑TOF SIMS͒ results. Recently, Oz-
kan et al.19-21 and Stevenson et al.22 proposed that a metal site is not
needed for ORR and that the active sites are located on the edge
planes of nitrogen-containing carbon. Popov et al.23 proposed that
active sites with pyridinic nitrogen are responsible for the catalytic
ORR activity. Although the nature of the active sites for ORR is not
yet clear, the choice of TM, the nitrogen content, the carbon type,
and the preparation method are all believed to influence the forma-
tion of the active sites and thus the activity of the catalysts.
pyrolysis in acetonitrile vapor or ammonia at elevated temperature.
The activity of Cr, Mn, Fe, and Co-based catalysts decreased ac-
cording to the following sequence: Fe Ͼ Co Ͼ Cr Ͼ Mn. The other
metal-based materials showed little activity.26,27 Recently, carbon-
supported Ru-chelate and Ru-based bimetallic electrocatalysts that
produced only 2–3% H2O2 were reported by Popov and co-workers,
which suggests that the central metal plays an important role in the
catalytic activity.28,29
Active Fe–C–N and Co–C–N thin-film electrocatalysts prepared
by combinatorial magnetron sputter deposition method in our lab
showed excellent ORR activity in acid electrolyte after heat-
treatment in inert gas.30-33 In our previous work,30-33 the effect of
heat-treatment temperature on the composition, structure, and ORR
catalytic activity of sputtered Fe–C–N ͑acid electrolyte͒ and Co–
C–N ͑acid and alkaline electrolyte͒ was studied in detail. In this
paper, the effect of TM choice and heat-treatment temperature on the
catalytic activity in acid and alkaline electrolytes and corrosion sta-
bility of sputtered TM–C–N ͑TM = V, Cr, Mn, Co, and Ni͒ films are
reported. Results for TM = Fe are not included here because the
sputtered Fe–C–N and Co–C–N catalysts showed similar trends in
the variation of the ORR activity with the metal ͑Fe or Co͒ content
and the heat-treatment temperature, which was reported in a previ-
ous paper.31 Comparisons to the activities of sputtered Co, C1−xNx,
and C films, as well as powdered graphite, are made as well in this
paper.
Experimental
The influence of the central metal ion of N4-phthalocyanines on
ORR activity was reported to vary as Fe Ͼ Co Ͼ Ni Ͼ Cu
Ϸ Mn.24 Chu and Jiang investigated the ORR catalytic activity of a
series of heat-treated TM ͑V, Mn, Fe, Co, Ni, Cu, and Zn͒ tetraphe-
nylporphyrin ͑TPP͒ and binary TM TPP ͑V/Fe, Co/Fe, Ni/Fe, and
Cu/Fe͒. All of these heat-treated TM TPP showed high activity ex-
cept for metal-free TPP and ZnTPP. Heat-treated FeTPP/CoTPP was
shown to be the best ORR catalyst.25 Dodelet et al. studied the ORR
activity of Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Mo-based catalysts
prepared by adsorbing various TM hydroxides or acetates on carbon
blacks or perylenetetracarboxylic diahydride ͑PTCDA͒ followed by
Deposition of thin-film samples.— The V, Cr, Mn, and Co sput-
tering targets ͑2.00 in. diameter, 0.250 in. thick, and 99.95% pure͒
were obtained from Williams Advanced Materials ͑Buffalo, NY͒.
The Ni target ͑2.00 in. diameter, 0.039 in. thick, and 99.99% pure͒
was obtained from Advent Research Materials ͑Eyansham, Oxford,
U.K.͒. The C sputtering target ͑2.00 in. diameter, 0.250 in. thick,
and 99.999% pure͒ was obtained from Kurt J. Lesker Co. ͑Clairton,
PA͒. All targets, with the exception of Ni, were mounted on
0.125 in. thick copper backing plates using SilverTech PT-1 silver
epoxy from Williams Advanced Materials. The Ni target was held in
place on the copper backing plate by the magnetic field from the
magnetron itself. V, Cr, Mn, Co, Ni, and C were sputtered using
MDX-1K DC ͑Advanced Energy͒ power supplies with powers of
41, 33, 15, 15, 15, and 150 W, respectively. Specially designed
*
Electrochemical Society Active Member.
z E-mail: jeff.dahn@dal.ca
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B80
Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
masks were positioned in front of each target to produce the desired
composition variations, while the substrate table was rotated. More
details about the production of binary and pseudo-binary composi-
tion spreads using our equipment can be found elsewhere.34 For the
constant composition samples, “constant” masks34 were used in
front of both C and TM targets. For the composition spread libraries
produced, a constant mask was used for the carbon target and a
“linear out” mask was used for the TM target.34
Nitrogen was incorporated into the libraries by sputtering in an
Ar/N2 gas mixture.30-33 Gas flows were independently controlled
using mass-flow controllers ͑MKS Instruments, Inc.͒. Ultrahigh-
purity Ar ͑99.999%͒ and N2 ͑99.999%͒ were supplied at constant
flow rates during the sputtering. One composition spread and one
constant composition sputtering run was completed for each TM–
C–N system. The composition-spread run for each TM–C–N system
was sputtered under 7.5 sccm Ar and 2.0 sccm N2, which corre-
sponds to a total pressure of 0.277 Pa ͑2.10 mTorr͒ with N2/Ar
= 20:80. The constant composition run for each TM–C–N system
was sputtered under 4.5 sccm Ar and 3.0 sccm N2, which corre-
sponds to a total pressure of 0.224 Pa ͑1.70 mTorr͒ with N2/Ar
= 40:60.
The constant composition C1−xNx film was sputtered under
4.5 sccm Ar and 3.0 sccm N2. The constant composition C film was
sputtered under 7.2 sccm Ar, which corresponds to a total pressure
of 0.224 Pa ͑1.70 mTorr͒. For the purpose of comparison of activi-
ties in alkaline solution, a constant composition Co film was also
sputtered under 6.6 sccm Ar, which corresponds to a total pressure
of 0.216 Pa ͑1.64 mTorr͒.
The composition-spread and constant composition samples for
each TM–C–N system were deposited onto a number of 80
ϫ 15 mm SiO2-coated Si͑100͒ wafer strips ͑500 nm thick layer of
SiO2͒ for heat-treatment, grazing incidence X-ray diffraction ͑XRD͒
measurements ͑constant composition sample only͒, electron micro-
probe analysis, and film thickness determination and a number of
5 mm diameter glassy carbon ͑GC͒ disks ͑Tokai Carbon Co.͒ for
electrochemical measurements using the rotating ring-disk electrode
͑RRDE͒ technique. Similar sputtering-table layouts were used for
the C1−xNx, C, and Co films. The TM–C–N composition spread was
also deposited onto 80 ϫ 10 mm quartz substrates for heat-
treatment and corrosion testing. More details about the various sub-
strates mounted on the substrate table as well as a projection of the
targets and their associated masks on the substrate table can be
found elsewhere.32
The TMxC1−x−yNy, C1−xNx, and C films deposited on the
SiO2-coated Si͑100͒ wafer strips, GC disks, and quartz substrates
were heat-treated at different temperatures in Ar using a Modular
Process Technology RTP-600S rapid thermal processor ͑RTP͒. The
samples were heated at 5°C s−1 in flowing Ar ͑2 L min−1͒ and held
at the target temperature for 5 min before cooling.
Composition of the sputtered samples.— The composition of the
sputtered samples was measured using a JEOL JXA-8200 super-
probe electron microprobe system equipped with one energy disper-
sive ͑EDS͒ spectrometer and five wavelength dispersive ͑WDS͒
spectrometers. The standards used were: $ilicon nitride ͑Si3N4͒ for
N; V, Cr, and Ni metals for V, Cr, and Ni, respectively; MnO2 for
Mn; a Mn–Co alloy ͑Mn:Co = 30:70 by weight͒ for Co; and pyroli-
tic graphte for C. Overlap of the N͑K␣͒ and Co͓L␤͑2͔͒ lines was
carefully considered and accommodated. TM ͑V, Cr, Mn, Co, and
Ni͒, N, and C atomic compositions were all determined by WDS
detectors. An accelerating voltage of 3 kV was employed to elimi-
nate the effect of the substrate on the measurement. The analysis
spot size was 50 ␮m. The composition was measured every 2 mm
along the 76 mm long composition-spread samples before and after
various heat-treatments. The composition of the as-sputtered
composition-spread and constant composition samples are reported
in Table I.
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B81
Thickness of the sputtered samples.— The thickness of the com-
position-spread and constant composition samples sputtered on Si
wafers was determined using a Dektak 8 Advanced Development
Profilometer ͑Veeco Instruments Inc.͒. The film thickness was found
to increase linearly with increasing TM content across the
TMxC1−x−yNy ͑TM = V, Cr, Mn, Co, and Ni͒ films. The thicknesses
of the composition-spread and constant composition samples are
reported in Table I.
drawn parallel to the baseline ͑i.e., from 0.9 V to approximately
0.8 V͒ and the second parallel to the increasing current signal. For
all the measurements, the ring potential was held at 1.2 V vs RHE in
35
36
order to oxidize any H2O2 produced in acid solution and HO2−
produced in alkaline solution. The % H2O2 produced in acid solu-
tion ͑or % HO−2 produced in alkaline solution͒ was calculated using
the following equation11,35,37
2IR/N
% H2O2 = 100
͓1͔
Grazing incidence XRD for the constant composition sample.—
The crystallographic structure of the constant composition films af-
ter heat-treatment to various temperatures was determined by graz-
ing incidence XRD using a Siemens D5000 diffractometer equipped
with a grazing incidence attachment. Cu K␣ radiation was used at
an incidence angle of 0.5° and with divergence slits set to 0.1°. Such
conditions lead to an illuminated spot of 40 ϫ 12 mm in the center
of the 80 ϫ 15 mm constant-composition specimen. A Soller slit
box in front of the detector was used to ensure that the scattered
X-rays from this large illuminated sample area that entered the de-
tector all had the same scattering angle. Data were collected over a
scattering angle range of 10–80° at 0.05° intervals. Total scan times
varied from 8 to 12 h.
ID + ͑IR/N͒
where ID is the faradaic current at the disk, IR is the faradaic current
at the ring, and N = 0.2 is the RRDE collection efficiency.
Corrosion testing.— Thin-film composition-spread samples of
TMxC1−x−yNy ͑TM = V, Cr, Mn, Co, and Ni͒ on quartz strips heated
at 700, 800, and 1000°C were immersed in 0.5 M H2SO4 ͑reagent,
analytical grade͒ at 25°C for 1 week. Separate CrxC1−x−yNy
composition-spread films heated at 700, 800, and 1000°C were im-
mersed in 0.5 M H2SO4 solution at 80°C for 1 week. After treat-
ment, the samples were rinsed with NANOpure water and dried in
air for further study. The composition as a function of position was
measured and compared to the composition measured before acid
treatment.
Electrochemical characterization
.— The TMxC1−x−yNy ͑TM ϭ
V, Cr, Mn, Co, and Ni͒, C1−xNx, C, and Co films were sputtered
directly on GC disks as described in the Deposition of thin-film
samples section. The loading of VxC1−x−yNy, CrxC1−x−yNy,
MnxC1−x−yNy, CoxC1−x−yNy, NixC1−x−yNy, C1−xNx, C, and Co films
Results
Composition of the thin-film TMxC1ϪxϪyNy (TM ϭ V, Cr, Mn,
Co, and Ni) libraries.— The N content vs TM content in the as-
sputtered TMxC1−x−yNy films and those heat-treated subsequently at
700, 800, or 1000°C is shown in Fig. 1. For the as-sputtered
TMxC1−x−yNy films, the atomic fraction of N varies smoothly from
0.186 to 0.406 for VxC1−x−yNy, from 0.225 to 0.461 for
CrxC1−x−yNy, from 0.254 to 0.445 for MnxC1−x−yNy, from 0.184 to
0.320 for CoxC1−x−yNy, and from 0.189 to 0.362 for NixC1−x−yNy.
The variation of N content with heat-treatment temperature in these
TMxC1−x−yNy films shows one striking feature: the N content de-
creases as heat-treatment temperature increases. However, the
amount of nitrogen lost depends on both the choice of TM and the
heat-treatment temperature.
VxC1−x−yNy, CrxC1−x−yNy, and MnxC1−x−yNy do not lose much
nitrogen when heated to 700°C. The N content decreases sharply at
about 1.2% with respect to Co for CoxC1−x−yNy and at about 1.5%
with respect to Ni for NixC1−x−yNy heated to 700°C. VxC1−x−yNy
and MnxC1−x−yNy libraries still retain significant nitrogen when
heated to 800°C, with MnxC1−x−yNy retaining more nitrogen than
VxC1−x−yNy. CrxC1−x−yNy, CoxC1−x−yNy, and NixC1−x−yNy lose nitro-
gen, in a “knee-shaped” transition when heated to 800°C, at about
3.8% with respect to Cr for the CrxC1−x−yNy film, 0.7% with respect
to Co for the CoxC1−x−yNy film, and about 1.5% with respect to Ni
for the NixC1−x−yNy film. The thermal stability of the sputtered
TMxC1−x−yNy libraries decreases in the order of Co Ͻ ͑Ϸ͒Ni
Ͻ Cr Ͻ V Ͻ Mn.
on the GC disks were 79, 75, 67, 56, 63, 50, 42, and 24 ␮g cm−2
,
respectively. The electrochemical properties of a graphite powder
sample ͑Fluka, 50870 graphite͒ were also measured for comparison.
An ink of graphite was prepared by mixing 10 mg of powder, 95 ␮L
of Nafion solution ͑5% wt from Aldrich͒, and 350 ␮L of ethanol,
followed by ultrasonicating for 40 min. Seven ␮L of ink was pipet-
ted onto a GC disk, resulting in a graphite powder loading of
780 ␮g cm−2
.
The electrocatalytic activity for the ORR of the samples on the
GC disks was studied with the RRDE technique using a Pine Elec-
trochemical system ͑AFMSRX rotator and AFCBP1 bipotentio-
stat͒. The RRDE electrode consisted of a catalyst film-coated GC
disk ͑5 mm diameter͒ surrounded by a Pt ring ͑7.5 mm inside diam-
eter and 8.5 mm outer diameter͒. The electrochemical measurements
were made in a one-compartment cell. The electrolyte was either a
0.1 M HClO4 solution or a 0.1 M KOH solution prepared from
double-distilled HClO4 ͑GFS Chemicals, Inc.͒ or KOH ͑GFS
Chemicals, Inc.͒ and 18 M⍀ Barnstead NANOpure water. A Pt-wire
counter electrode and Koslow Hg/Hg2SO4 reference electrode ͑used
in 0.1 M HClO4 solution͒ or Ag/AgCl reference electrode ͑Bioana-
lytical Systems, Inc., used in 0.1 M KOH solution͒ were used for
the measurements. The potential of the Hg/Hg2SO4 and Ag/AgCl
reference electrodes were calibrated vs the reversible hydrogen elec-
trode ͑RHE͒. The potentials reported in this study all refer to that of
the RHE potential.
The samples on the GC disks were first electrochemically
cleaned by sweeping the potential between 0 and 1.0 V ͑vs RHE͒ at
50 mV s−1 in an Ar-saturated 0.1 M HClO4 solution or 0.1 M KOH
solution until steady-state cyclic voltammograms ͑CVs͒ were ob-
tained. For each catalyst tested, a CV was first collected in Ar-
saturated 0.1 M HClO4 solution or 0.1 M KOH solution from
1.0 to 0.05 V at 5 mV s−1 to determine the nonfaradaic current. The
CVs during oxygen reduction in 0.1 M HClO4 or 0.1 M KOH so-
lution were performed between 1.0 and 0.05 V at 5 mV s−1 with the
electrode stationary and with the electrode rotated at 900 rpm. O2
gas was purged into the solution at a rate of 1 L min−1. The faradaic
current was determined by subtracting the CV in Ar-saturated 0.1 M
HClO4 or 0.1 M KOH solution from that obtained in O2-saturated
0.1 M HClO4 solution or 0.1 M KOH solution. The onset potential,
at which the current for oxygen reduction is first observed,4 was
determined by the point of intersection of two tangent lines, one
Electrocatalytic activity of the sample that showed the highest
activity in each TMxC1ϪxϪyNy library after heat-treatment.— Fif-
teen GC disks, as three repeats for five different compositions, were
included in each preparation of the composition-spread
TMxC1−x−yNy libraries. One group of five disks was heated to
700°C, one group of five disks was heated to 800°C, and the final
group of five disks was heated to 1000°C. The activity of the sput-
tered films on all 15 disks ͑15 disks for each of the five choices of
TM = a total of 75 disks͒ was measured in acid solution. The com-
position of the most active GC disk sample for a given TM is indi-
cated by circles in Fig. 1.
Figure 2 shows the CVs of the sample on the GC disk that
showed the highest activity in each TMxC1−x−yNy film after heat-
treatment measured in O2-saturated 0.1 M HClO4 solution at 25°C.
No other data is shown in order to limit the amount of data pre-
sented in this paper to a manageable level. The TM content ͑shown
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B82
Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
on each panel in Fig. 2͒ for the most active compositions in the
samples varied only between 10.0 and 11.8%. The heat-treatment
temperature for the most active sample was 800°C for CrxC1−x−yNy,
CoxC1−x−yNy, and NixC1−x−yNy and 1000°C for VxC1−x−yNy and
MnxC1−x−yNy. Notice from Fig. 1 that VxC1−x−yNy and MnxC1−x−yNy
only lose substantial N when heated above 800°C, whereas the Cr,
Co, and Ni-based libraries had already lost substantial N when
heated to 800°C. For all TMs tested, the most active sample was
found to be that with the highest metal content of the disks tested
and after heating to a temperature at which substantial N loss had
occurred. The activity of the samples toward ORR decreases in the
order of Co Ͼ Cr Ͼ Ni Ͼ Mn Ͼ V. V11.3C88.6N0.1 shows little or
no activity. The onset potential of Co10.0C87.3N2.7 is about 0.76 V vs
RHE.
In order to investigate the dependence of the activity of the
TMxC1−x−yNy films on the heat-treatment temperature in detail,
constant-composition TMxC1−x−yNy films with nominally the same
value of x ͑x = 0.10͒ were prepared and heat-treated at a closely
spaced series of temperatures. The atomic TM content, shown in
Table I, in the as-sputtered films was measured to lie between 9.2
and 11.4%. Constant-composition C1−xNx ͑composition included in
Table I͒ and C films were also prepared for comparison.
Electrocatalytic
activity
of
the
constant-composition
TMxC1ϪxϪyNy, C1ϪxNx and C samples heat-treated at different tem-
peratures.— Figures 3-7 show the variation with heat-treatment
temperature of the N content, onset potentials, and the current den-
sities at fixed potentials ͑vs RHE͒ for the ORR measured in
O2-saturated 0.1 M HClO4 and/or O2-saturated 0.1 M KOH solution
at 25°C for the constant composition TM–C–N films. The potentials
for comparing the current densities vs heat-treatment temperature
were selected to be about 0.1 V below the onset potential of the best
sample for each TM.
The data for the constant-composition V–C–N film presented in
Fig. 3 shows that there is a dramatic decrease in the N content at
Tc = 825°C. The onset potential of the film in alkaline electrolyte
does not change significantly with heat-treatment temperature be-
tween 650 and 1000°C, although the maximum occurs for the
sample heat-treated at 900°C. The current density in alkaline solu-
tion at 0.6 V increases slightly between 650 and 825°C ͑Tc͒ and
then increases significantly above Tc, with the maximum around
900°C. The significant increase in activity above Tc could be caused
by the porosity induced by the N loss at Tc that would increase the
surface area. The constant-composition V–C–N film shows little or
no activity in acid solution before or after heat-treatment.
There is a dramatic decrease in N content at Tc = 850°C for the
Mn–C–N film ͑Fig. 4͒. The onset potential in alkaline electrolyte
increases with heat-treatment temperature between 600 and 900°C,
with the maximum occurring for the sample heat-treated at 900°C.
The current density at 0.6 V increases slightly between 600 and
850°C ͑Tc͒ and then increases significantly above Tc with the maxi-
mum occurring at 900°C. The constant-composition Mn–C–N film
shows little or no activity in acid solution before and after heat-
treatment, which is similar to the V–C–N film.
Figure 1. ͑Color online͒ Atomic % N vs atomic % TM for TMxC1−x−yNy
library before and after heat-treatment at 700, 800, and 1000°C. TM = ͑a͒ V,
͑b͒ Cr, ͑c͒ Mn, ͑d͒ Co, and ͑e͒ Ni. The balance of the composition is carbon.
The circle-shaped data points indicate the compositions of the sample that
showed the highest activity for ORR in acidic electrolyte in each
TMxC1−x−yNy library.
For the constant-composition Cr–C–N film ͑Fig. 5͒, there is a
dramatic decrease in N content at Tc = 800°C. The onset potential
Figure 2. Disk current densities for the
samples on GC disks that showed the
highest activity in an acidic environment
for ORR in the TMxC1−x−yNy libraries.
TM = ͑a͒ V, ͑b͒ Cr, ͑c͒ Mn, ͑d͒ Co, and
͑e͒ Ni. The measurement was performed
in O2-saturated 0.1 M HClO4 at 25°C.
The sweep rate was 5 mV s−1 and the ro-
tation speed was 900 rpm. The average
composition of the film on each disk is
indicated in each panel.
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B83
Figure 3. ͑a͒ N content, ͑b͒ onset potential, and ͑c͒ current density at 0.6 V
for ORR in 0.1 M KOH solution vs heat-treatment temperature of the
VxC1−x−yNy samples. The samples were derived from the constant-
composition sample described in Table I.
Figure 4. ͑a͒ N content, ͑b͒ onset potential, and ͑c͒ current density at 0.6 V
for ORR in 0.1 M KOH solution vs heat-treatment temperature of the
MnxC1−x−yNy samples. The samples were derived from the constant-
composition sample described in Table I.
decrease in N content at Tc = 725°C. The onset potential for ORR in
acid increases with the heat-treatment temperature between 700 and
850°C, as the catalyst becomes active above Tc = 725°C. The maxi-
mum onset potential and current density at 0.5 V in acid solution
occurs for the film heat-treated at 850°C. The maximum onset po-
tential and the current density at 0.6 V in alkaline solution occurs
for the film heat-treated at 900°C; however, the film begins to show
activity at a lower temperature ͑650°C͒, i.e., below Tc, in 0.1 M
KOH solution than in 0.1 M HClO4 solution. In addition, the ORR
activity of the film in alkaline solution is much higher than that in
acid solution. This is similar to the behavior of the Co–C–N and
Cr–C–N films. In our previous work on the activity of sputtered
Co–C–N films in acid and alkaline solutions, we noted33 that there
were samples heated to just below Tc that were active in alkaline
solution but not in acid solution. We speculated that small pores
created by N loss above Tc are too small for the adsorption of acidic
species that block catalytic sites, so the films become active above
Tc in acid. We further speculated that the sites are not blocked by
ions in the alkaline solution regardless of whether or not the samples
have been heated through Tc.
for ORR in acid solution increases with the heat-treatment tempera-
ture between 775 and 850°C, as the catalyst becomes active above
Tc = 800°C. The onset potential decreases slightly with heat-
treatment temperature above 850°C. The film begins to show activ-
ity at a lower temperature ͑700°C͒ in 0.1 M KOH solution than in
0.1 M HClO4 solution. The onset potential measured in 0.1 M KOH
solution increases with heat-treatment temperature between 700 and
800°C and then does not change much, although the maximum oc-
curs for the sample heat-treated at 900°C. The current density at
0.6 V in acid electrolyte is strongly dependent on the heat-treatment
temperature, reaching a maximum for the sample heat-treated at
850°C. The current density at 0.6 V in the alkaline electrolyte in-
creases between 700 and 800°C and then does not change signifi-
cantly, although, as with the testing in acid, the maximum occurs for
the sample heat-treated at 900°C.
There is a dramatic decrease in N content at Tc = 725°C for the
constant-composition Co–C–N film ͑Fig. 6͒. The onset potential for
ORR in acid increases with the heat-treatment temperature between
700 and 800°C, as the catalyst becomes active above Tc = 725°C.
The maximum onset potential occurs for the film heat-treated at
800°C. The samples begin to show activity at a lower heat-treatment
temperature ͑650°C͒ in 0.1 M KOH solution than in 0.1 M HClO4
solution. The maximum onset potential in alkaline solution occurs
for the sample heat-treated at 850°C. The current density at 0.7 V in
acid electrolyte is strongly dependent on the heat-treatment tempera-
ture, reaching a maximum for the sample heat-treated at 800°C. The
maximum current density at 0.7 V in alkaline electrolyte occurs for
the sample heat-treated at 850°C. The observation that the Co–C–N
film starts to show activity at Tc in acid solution and below Tc in
alkaline solution is similar to that seen for the Cr–C–N film.
For the constant-composition Ni–C–N film, there is a dramatic
The activity for the ORR, the optimum heat-treatment tempera-
ture, and the composition of the most active sample in the constant-
composition TM–C–N ͑TM = V, Cr, Mn, Co, and Ni͒ films in acid
and alkaline solution are summarized in Table II. In 0.1 M HClO4
solution, the activity of the most active heat-treated sample in TM–
C–N ͑TM = Cr, Co, and Ni͒ follows the order Co Ͼ Cr Ͼ Ni. In
0.1 M KOH solution, the activity of the most active heat-treated
sample in TM–C–N ͑TM = V, Cr, Mn, Co, and Ni͒ follows the
order Co Ͼ Ni Ͼ Mn Ͼ V Ͼ Cr.
Figure 8 shows the CV and the percentage of HO−2 ͑% HO2−͒
produced at the disk electrode vs the disk potential for the most
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B84
Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
Figure 6. ͑a͒ N content, ͑b͒ onset potential, and ͑c͒ current density at 0.7 V
for ORR in 0.1 M HClO4 and 0.1 M KOH solution vs heat-treatment tem-
perature of the CoxC1−x−yNy samples. The samples were derived from the
constant-composition sample described in Table I.
Figure 5. ͑a͒ N content, ͑b͒ onset potential, and ͑c͒ current density at 0.6 V
for ORR in 0.1 M HClO4 and 0.1 M KOH solution vs heat-treatment tem-
perature of the CrxC1−x−yNy samples. The samples were derived from the
constant-composition sample described in Table I.
active heat-treated V–C–N and Mn–C–N samples, whose composi-
tions are given in Table II, measured in O2-saturated 0.1 M KOH
solution at 25°C. No well-defined diffusion-limited current is
reached, although there is a plateau between 0.4 and 0.6 V in the
CV curves in both cases. For V–C–N, the percentage of HO2−
͑% HO−2͒ first increases as the disk potential decreases and reaches
95.0% at 0.60 V and then decreases with the disk potential. For
Mn–C–N, the percentage of HO−2 ͑% HO−2͒ first increases as the
disk potential decreases and reaches about 55% at 0.50 V, and then
does not change much with the disk potential. The fraction of per-
oxide produced is very large.
increases as the disk potential decreases, reaching 55% at 0.60 V,
and then decreases slightly for lower disk potentials. The % HO−2 is
between 40 and 55% between 0.70 and 0.05 V, which is slightly
higher than the % H2O2 produced in acid solution. For Ni–C–N, the
activity in alkaline solution is much higher than in acid solution. The
onset potentials measured in acid and alkaline solution are 0.60 and
0.84 V, respectively. The % HO−2 ͑about 60–90% between 0.80 and
0.05 V͒ produced in alkaline solution is higher than the % H2O2
͑about 40–55% between 0.60 and 0.05 V͒ produced in acid solution.
All of these thin-film catalysts produce large amounts of peroxide.
Figure 9 shows the CVs and the percentage of H2O2 ͑% H2O2͒
͑in acid solution͒ and HO−2 ͑% HO−2͒ ͑in alkaline solution͒ produced
at the disk electrode vs the disk potential for the most active heat-
treated Cr–C–N, Co–C–N, and Ni–C–N samples, whose composi-
tions are given in Table II, measured in O2-saturated 0.1 M HClO4
solution and in O2-saturated 0.1 M KOH solution at 25°C. No well-
defined diffusion-limited current is reached for ORR over the poten-
tial range measured in the CV curves in acid and alkaline solution
for the Cr–C–N, Co–C–N, and Ni–C–N samples. For Cr–C–N, the
activity in acid and alkaline solution are similar. The % HO−2 ͑about
85% between 0.6 and 0.3 V͒ produced in alkaline solution is higher
than the % H2O2 ͑about 65% between 0.6 and 0.3 V͒ produced in
acid solution. For Co–C–N, the CV curve in acid solution shows two
clear reduction waves. The % H2O2 in acid is between 22 and 50%
between 0.80 and 0.05 V and increases as the disk potential de-
creases, which correlates to the two steep reduction waves in the CV
curve. The Co–C–N CV curve in alkaline solution shows a relatively
well-defined current plateau, which indicates that the activity in al-
kaline solution is higher than that in acid solution. The % HO2−
Grazing incidence XRD of the constant-composition
TMxC1ϪxϪyNy samples heat-treated at different temperatures.—
Figures 10-14 show grazing incidence XRD spectra for the constant-
composition TM–C–N films heat-treated to a range of different tem-
peratures between 650 and 1000°C. In all the cases, at a critical
temperature ͑Tc͒, the film transforms to a heterogeneous mixture of
N-containing carbon and V8C7 ͑Fig. 10͒, Cr3C2 ͑Fig. 11͒, Mn7C3
͑Fig. 12͒, Co ͑Fig. 13͒, or Ni ͑Fig. 14͒. The observed phases in the
XRD pattern of the heat-treated TM–C–N ͑TM = V, Cr, Mn, Co,
and Ni͒ films and the equilibrium phases in the TM–C ͑TM = V, Cr,
Mn, Co, and Ni͒ binary-phase diagrams38 are summarized in Table
III. Previous studies on Co–C–N samples33 have shown that Tc de-
creases with TM content and increases with N content in the films.
Even though the constant-composition samples were designed to
have similar TM and N contents in the as-sputtered state, the varia-
tions that exist between the samples ͑see Table I͒ make it impossible
to rank the five TMs for their effectiveness at catalyzing the trans-
formation at Tc. The observed phases in the heat-treated TM–C–N
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B85
Figure 8. Disk current density and % HO−2 produced for the heat-treated
VxC1−x−yNy and MnxC1−x−yNy constant-composition samples on the GC disks
that showed the highest activity for ORR measured in O2-saturated 0.1 M
KOH. The measurements were performed at 25°C with a sweep rate of
5 mV s−1 and rotation speed of 900 rpm. The average compositions of the
films on the disks are indicated.
before and after heat-treatment. RRDE of graphite powder was also
measured for comparison in alkaline solution. First, the data for the
C1−xNx and C samples are presented and later compared to the
heated TM–C–N and Co films ͑in alkaline solution͒ in the Discus-
sion section.
Figure 7. ͑a͒ N content, ͑b͒ onset potential, and ͑c͒ current density for ORR
at 0.6 V in 0.1 M KOH solution or 0.5 V in 0.1 M HClO4 vs heat-treatment
temperature of the NixC1−x−yNy samples. The samples were derived from the
constant-composition sample described in Table I.
Figure 15 shows the variation with heat-treatment temperature of
the N content, the onset potential, and the current densities at 0.3 V
for the ORR measured in O2-saturated 0.1 M HClO4 and the current
densities at 0.6 V ͑vs RHE͒ in O2-saturated 0.1 M KOH solution at
25°C for the constant-composition C61.1N38.9 film ͑the composition
shown here is for the as-sputtered sample͒. The N content decreases
as the heat-treatment temperature increases. The onset potential for
ORR in acid increases with the heat-treatment temperature in the
range between 700 and 900°C and then decreases as it is heated to
1000 and 1100°C. The maximum onset potential and current density
at 0.3 V in acid solution occurs for the film heat-treated to 900°C.
The film begins to show activity at 400°C in 0.1 M KOH solution.
The maximum onset potential and the current density at 0.6 V in
alkaline solution occurs for the film heat-treated at 1000°C. The
ORR activity of the film in alkaline solution is much higher than that
in acid solution.
samples are in good agreement with equilibrium-phase diagrams,
which indicates that heating above Tc causes a transformation to a
nanocomposite of expected equilibrium phases.
In all the cases, the transformation at Tc coincides with a signifi-
cant loss of N as shown in Fig. 3-7, respectively. The thermal sta-
bility of the amorphous constant-composition TMxC1−x−yNy ͑x
= 0.10͒ films increases in the order of Co Ϸ Ni Ͻ Cr Ͻ V Ͻ Mn
based on the value of Tc. This is in agreement with the results for
thermal stability obtained for the TMxC1−x−yNy libraries ͓i.e., Co
Ͻ ͑Ϸ͒Ni Ͻ Cr Ͻ V Ͻ Mn͔ that can be inferred from Fig. 1.
Comparison of the electrocatalytic activity of heated TM–C–N
films to heated C1ϪxNx and C films.— The results above show that
each of the TM–C–N systems transforms to N-containing graphitic
carbon and either TM or a TM-carbide above Tc. Therefore, it is
natural to consider the role that the TM plays in the formation of the
active sites for ORR in both acid and alkaline solution. Experiments
focusing on the case of TM = Co were made. In this case, films of
C1−xNx, C, and Co were prepared and studied by XRD and RRDE
Figure 16 shows the CVs and the percentages of H2O2
͑% H2O2͒ ͑in acid solution͒ and HO−2 ͑% HO2−͒ ͑in alkaline solu-
tion͒ produced at the disk electrode vs the disk potential for the
constant-composition C61.1N38.9 sample heat-treated at 900°C ͑i.e.,
the most active one in acid solution, the composition is C78.6N21.4
Table II. The activity for ORR, the optimum heat-treatment temperature, and the compositions of the most active sample in the constant-
composition TM–C–N (TM = V, Cr, Mn, Co, and Ni) samples in acid and alkaline solutions.
Composition of
the most active sample
after heat-treatment
Onset
potential
͑V vs RHE͒
Composition of the
as-sputtered sample
Tc
͑°C͒
Optimum heat-treatment
Electrolyte
temperature ͑°C͒
V–C–N
Cr–C–N
V
9.4C47.4N43.2
825
800
0.1 M KOH
0.1 M HClO4
0.1 M KOH
0.1 M KOH
0.1 M HClO4
0.1 M KOH
0.1 M HClO4
0.1 M KOH
900
850
900
900
800
850
850
900
V
13.2C83.1N3.7
0.79
0.80
0.78
0.82
0.81
0.87
0.60
0.84
Cr11.4C47.2N41.3
Cr14.2C80.1N5.7
Cr14.6C82.0N3.4
Mn13.3C83.7N3.0
Co11.5C82.5N6.0
Co11.9C82.8N5.3
Ni13.2C82.7N4.1
Ni13.4C83.2N3.4
Mn–C–N
Co–C–N
Mn9.7C52.7N37.6
Co9.2C57.3N33.5
850
725
Ni–C–N
Ni10.1C50.5N39.4
725
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B86
Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
Figure 11. ͑Color online͒ Grazing incidence XRD measurements showing
intensity vs scattering angle for the heat-treated constant-composition sample
of Cr–C–N. The heat-treatment was done in flowing N2 and the heat-
treatment temperature is indicated next to each curve. The major Bragg
peaks of the phases present are indicated.
after heat-treatment at 900°C͒ measured in O2-saturated 0.1 M
HClO4 solution at 25°C and heat-treated at 1000°C ͑i.e., the most
active one in alkaline solution, the composition is C88.7N11.3 after
heat-treatment at 1000°C͒ measured in O2-saturated 0.1 M KOH
solution at 25°C. The onset potentials measured for the sample
heated at 900°C in acid solution and for the sample heated at
1000°C in alkaline solution are 0.40 and 0.83 V, respectively. The
% HO−2 ͑about 30–60% between 0.80 and 0.05 V͒ produced in al-
kaline solution for the sample heated at 1000°C is lower than the %
H2O2 ͑about 60–77% between 0.40 and 0.05 V͒ produced in acid
solution for the sample heated at 900°C.
Figure 17 shows the variation with heat-treatment temperature of
the onset potential and the current densities at 0.6 V ͑vs RHE͒ for
the ORR measured in O2-saturated 0.1 M KOH solution at 25°C for
the sputtered C film. The onset potential ͑0.71–0.73 V͒ does not
change much with heat-treatment temperature in the range between
25 and 900°C and then decreases to 0.62 V when it is heated to
1000°C. The current density at 0.6 V increases slightly over the
range between 25 and 800°C and then decreases at 900 and 1000°C.
The C film shows little or no activity in acid solution before and
Figure 9. Disk current density and % H2O2 ͑% HO−2 in 0.1 M KOH͒ pro-
duced for the heat-treated CrxC1−x−yNy, CoxC1−x−yNy, and NixC1−x−yNy
constant-composition samples on the GC disks that showed the highest ac-
tivity for ORR measured in O2-saturated 0.1 M HClO4 and in O2-saturated
0.1 M KOH. The measurements were performed at 25°C with a sweep rate
of 5 mV s−1 and rotation speed of 900 rpm. The average compositions of the
films on the disks are indicated.
Figure 10. Grazing incidence XRD measurements showing intensity vs scat-
tering angle for the heat-treated constant-composition sample of V–C–N.
The heat-treatment was done in flowing N2 and the heat-treatment tempera-
ture is indicated next to each curve. The major Bragg peaks of the phases
present are indicated.
Figure 12. Grazing incidence XRD measurements showing intensity vs scat-
tering angle for the heat-treated constant-composition sample of Mn–C–N.
The heat-treatment was done in flowing N2 and the heat-treatment tempera-
ture is indicated next to each curve. The major Bragg peaks of the phases
present are indicated.
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B87
Figure 13. Grazing incidence XRD measurements showing intensity vs scat-
tering angle for the heat-treated constant-composition sample of Co–C–N.
The heat-treatment was done in flowing N2 and the heat-treatment tempera-
ture is indicated next to each curve. The major Bragg peaks of the phases
present are indicated.
after heat-treatment. Recall that the C1−xNx films heated between
700 and 1100°C are somewhat active for ORR in acid solution as
shown in Fig. 16, confirming the importance of N doping on the
activity of C in acid solution as reported in the literature.39,40
Figure 15. ͑a͒ N content, ͑b͒ onset potential for ORR, and ͑c͒ current density
for ORR at 0.6 V in 0.1 M KOH solution or 0.3 V in 0.1 M HClO4 vs
heat-treatment temperature of the C1−xNx samples. The samples were derived
from the constant-composition sample described in Table I.
Figure 18 shows the CV and the percentages of HO−2 ͑% HO2−͒
produced at the disk electrode vs the disk potential for the C film
heat-treated at 800°C ͑i.e., the most active one͒ measured in
O2-saturated 0.1 M KOH solution at 25°C. There are two waves in
the CV curve, possibly corresponding to two active sites for the
ORR, which is consistent with literature experiments on GC,41,42 on
the roughened basal plane of graphite43 and on the edge of pyrolytic
graphite.44,45 The two waves were also observed in the CV curves of
the as-sputtered C film and the C films heated at all temperatures up
to 1000°C ͑not shown here͒. These results suggest that the as-
sputtered C film and the C films heated up to 1000°C are
disordered,42,46 which agrees with the XRD measurements on these
Figure 14. Grazing incidence XRD measurements showing intensity vs scat-
tering angle for the heat-treated constant-composition sample of Ni–C–N.
The heat-treatment was done in flowing N2 and the heat-treatment tempera-
ture is indicated next to each curve. The major Bragg peaks of the phases
present are indicated.
Table III. The critical temperature for the phase transformation,
the observed phases (from the grazing incidence XRD patterns)
in the heat-treated constant-composition samples of TM–C–N
(TM = V, Cr, Mn, Co, and Ni), and the equilibrium phases in the
TM–C (TM = V, Cr, Mn, Co, and Ni) binary-phase diagrams.36
Critical
temperature
Equilibrium
phases
for TM–C
͑Tc͒
͑°C͒
As-sputtered
sample
Observed phases
Figure 16. Disk current density and % H2O2 ͑% HO−2 in 0.1 M KOH͒
produced for the heat-treated C1−xNx sample on the GC disk that showed the
highest activity for ORR measured in O2-saturated 0.1 M HClO4 and in
O2-saturated 0.1 M KOH at 25°C. The sweep rate was 5 mV s−1 and the
rotation speed was 900 rpm. The average compositions of the films on the
disks are indicated.
V9.4C47.4N43.2
Cr11.4C47.2N41.3
Mn9.7C52.7N37.6
Co9.2C57.3N33.5
Ni10.1C50.5N39.4
825
800
850
725
725
V8C7, N-carbon
Cr3C2, N-carbon
Mn7C3, N-carbon
Co, N-carbon
V8C7, carbon
Cr3C2, carbon
Mn7C3, carbon
Co, carbon
Ni, N-carbon
Ni, carbon
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B88
Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
Figure 17. ͑a͒ Onset potential and ͑b͒ current density at 0.6 V for ORR in
0.1 M KOH solution vs heat-treatment temperature of the C samples. The
samples were derived from the constant-composition sample.
films that show no clear Bragg peaks ͑not shown͒. The percentage of
HO−2 ͑% HO−2͒ produced on the C film heat-treated at 800°C is
about 60% between 0.60 and 0.05 V.
Stability of the TMxC1ϪxϪyNy films heat-treated at different tem-
peratures against corrosion in acid.— Figure 19 shows the differ-
ence in the atomic content of TM ͑⌬%TM = % of TM lost͒ before
and after soaking in 0.5 M H2SO4 at 25°C for 1 week plotted vs the
initial ͑before acid͒ atomic content of TM ͑%TM͒ in the
TMxC1−x−yNy libraries heat-treated at 700, 800, and 1000°C. The
films with low TM content show relatively good stability with re-
spect to TM dissolution. For initial TM contents, %TM ͑TM
= V,Mn,Co,Ni͒ = 1.0–1.2% and % Cr = 6.5%, little or no TM is
dissolved from the films. More TM is dissolved from the films when
the TM concentration is relatively high in the TMxC1−x−yNy librar-
Figure 19. ͑Color online͒ The change in TM content ͑⌬%TM = % of tran-
sition metal lost͒ caused by soaking in 0.5 M H2SO4 at 25°C for 1 week vs
the initial atomic content of TM ͑%TM͒ in the TMxC1−x−yNy libraries heat-
treated at 700, 800, and 1000°C. TM = ͑a͒ V, ͑b͒ Cr, ͑c͒ Mn, ͑d͒ Co, and ͑e͒
Ni. The inset in ͑b͒ is ⌬%Cr before and after soaking in 0.5 M H2SO4 at
80°C for 1 week vs the initial % Cr in the CrxC1−x−yNy library.
ies. The amount of TM dissolved ͑⌬%TM͒ from the films increases
as the heat-treatment temperature increases. Especially for
CoxC1−x−yNy and NixC1−x−yNy films heat-treated at 1000°C, Co and
Ni are almost completely dissolved from the films at high Co and Ni
concentrations as shown in Fig. 19d and e. Interestingly, the
CrxC1−x−yNy composition spread is much more stable against corro-
sion than the VxC1−x−yNy, MnxC1−x−yNy, CoxC1−x−yNy, and
NixC1−x−yNy spreads. Only about 1.0% Cr was lost, even at high Cr
concentration ͑12.1% Cr͒, after acid treatment at 25°C for 1 week,
and ⌬%Cr is only about 1.5% at 12.1% Cr after acid treatment at
80°C for 1 week, as shown in Fig. 19b. This indicates that Cr com-
ponent in the heated CrxC1−x−yNy film provides passivation against
Figure 18. Disk current density and % HO−2 produced for the heat-treated C
sample on the GC disk that showed that highest activity for ORR measured
in O2-saturated 0.1 M KOH. The measurement was performed at 25°C. The
sweep rate was 5 mV s−1 and the rotation speed was 900 rpm.
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B89
Figure 21. ͑Color online͒ Disk current density of ͑a͒ sputtered Co heated at
900°C, ͑b͒ sputtered Co–C–N heated at 900°C and ͑c͒ sputtered Co–C–N
heated at 900°C and then treated by soaking in 1.0 M H2SO4 at 25°C for
one month and at 80°C for one month on the GC disks for ORR measured in
O2-saturated 0.1 M KOH. The measurement was performed at 25°C. The
sweep rate was 5 mV s−1 and the rotation speed was 900 rpm.
Figure 20. ͑Color online͒ Disk current density of ͑a͒ sputtered C heated at
900°C, ͑b͒ sputtered C–N heated at 900°C, ͑c͒ sputtered C–N heated at
1000°C, ͑d͒ sputtered C–N heated at 1100°C, ͑e͒ graphite, ͑f͒ sputtered
V–C–N heated at 900°C, ͑g͒ sputtered Cr–C–N heated at 900°C, ͑h͒ sput-
tered Mn–C–N heated at 900°C, ͑i͒ sputtered Co–C–N heated at 900°C, and
͑j͒ sputtered Ni–C–N heated at 900°C on the GC disks for ORR measured in
O2-saturated 0.1 M KOH. The measurement was performed at 25°C. The
sweep rate was 5 mV s−1 and the rotation speed was 900 rpm.
tions of sites͒ are active for ORR in alkaline solution in these
samples. Nevertheless, some comparison graphs of the data pre-
sented in the Results section are useful.
Figure 20 compares the activities in alkaline solution of sputtered
C heated at 900°C, sputtered C1−xNx heated at 900, 1000, and
1100°C, powdered graphite, and sputtered TMxC1−x−yNy ͑TM ϭ V,
Cr, Mn, Co, Ni͒ heated at 900°C. Results for graphite are included
because graphitic carbon appears in the XRD patterns of the sput-
tered TMxC1−x−yNy ͑TM ϭ V, Cr, Mn, Co, Ni͒ films after heating
above Tc ͑Fig. 10-14͒. The onset potentials of sputtered C1−xNx
heated at 900, 1000, and 1100°C ͑curves b, c, and d in Fig. 20͒ are
higher than that of sputtered C heated at 900°C ͑curve a in Fig. 20͒.
The onset potentials of C1−xNx heated at 1000 and 1100°C are also
higher than that of graphite ͑curve e in Fig. 20͒. These results sug-
gest that N doping can improve the activity of C in alkaline solution
as well as in acid solution.
The activity of activated carbon for ORR in alkaline solution
reported in Ref. 50 is better than the C1−xNx samples and compa-
rable to those of the TM–C–N samples. Note that when N is lost
from TMxC1−x−yNy and C1−xNx samples during heating, it is likely
that some porosity, similar to the porosity in activated carbon, may
form. Thus, the correlation of the increased activity with TM type or
N content in the heated films may be a result of increased porosity
or increased numbers of particular surface functional groups formed
on surfaces which become exposed through N loss.
The samples that show the highest activity in Fig. 20 are the
CoxC1−x−yNy and NixC1−x−yNy samples ͑curves i and j͒ heated to
900°C. These samples contain metallic Co and metallic Ni as shown
in Fig. 13 and 14. It is, therefore, extremely important to compare
their activity to the activity of the metals alone. Ni hydroxide51 and
Co hydroxide52 are electroactive for ORR in alkaline solution. Pour-
baix diagrams show53 that Ni and Co are passivated by an oxide or
hydroxide layer above ca. 0.20 and 0.15 V ͑vs RHE͒ in an alkaline
solution having pH Ͼ 9, respectively.
Figure 21 shows CVs in oxygen-saturated alkaline solution for a
sputtered Co film and for two samples of CoxC1−x−yNy. The
CoxC1−x−yNy samples were a film heated at 900°C ͑from the
constant-composition sample described in Table II͒ and an identical
film ͑also heated to 900°C͒ that was soaked in 1.0 M H2SO4 at
25°C for 1 month and then at 80°C for another month to dissolve
any exposed Co in the film. The activity of the heated CoxC1−x−yNy
film is higher than that of heated Co film and does not change
significantly after the Co is dissolved away, indicating that other
active sites account for the improved activity of the heated
CoxC1−x−yNy film ͑and the NixC1−x−yNy film as well͒.
corrosion in acid solution. The stability of heated CrxC1−x−yNy in
47-49
acid may arise from the presence of Cr3C2
in the film.
Discussion
Activity in acid solution.— The results from the previous section
are summarized in this section and comparisons between the differ-
ent samples are made. Heat-treated VxC1−x−yNy and MnxC1−x−yNy
films show little or no activity for ORR in acid electrolyte. The
heat-treated CrxC1−x−yNy, CoxC1−x−yNy, and NixC1−x−yNy films show
activity after heating to Tc and above in acid electrolyte. Heat-
treated C1−xNx films show much lower activity in acid solution than
the CrxC1−x−yNy, CoxC1−x−yNy, and NixC1−x−yNy films heat-treated
at the same temperatures. The as-sputtered and heated C films show
little or no activity in acid electrolyte. Therefore, it is clear that
activity in acid solutions appears above Tc and for TM–C–N mate-
rials with TM = Cr, Co, and Ni. It has also been shown that active
materials are produced when samples with TM = Fe are heated
30,31
above Tc
and the activity of the Fe–C–N is a bit higher than
that of Co–C–N. Therefore, the activity for ORR in acid solution
follows the order FexC1−x−yNy Ͼ CoxC1−x−yNy Ͼ CrxC1−x−yNy
Ͼ NixC1−x−yNy. The results show that TM plays an important role in
the formation of the active site for ORR in acid solution.
Activity in alkaline solution.— All of the heat-treated
TMxC1−x−yNy ͑TM = V, Cr, Mn, Co, and Ni͒ films show activity in
alkaline electrolyte when heated to some temperature below Tc.
Heat-treated C1−xNx and as-sputtered or heated C films all show
activity in alkaline electrolyte. What is the source of this activity?
The situation in alkaline solution is complicated, because
GC,41,42 graphite,42-44 and activated carbon50 all show significant
activity for ORR. In addition, metallic Co and Ni, which convert to
hydroxides or oxides in alkaline solution, also show significant ac-
tivity for ORR.51,52 Because the TM–C–N samples are carbonaceous
͑though amorphous͒ before heating and most show evidence of
N-containing graphite after heating above Tc, it is possible that their
activity in alkaline solution could arise due to their carbon-
containing character alone. Because the Co–C–N and Ni–C–N
samples clearly contain Co or Ni after heating above Tc, their activ-
ity in alkaline solution could arise from the metal content. We ac-
knowledge that it is difficult to be certain which sites ͑or combina-
The results in Fig. 20 show that the onset potentials for ORR in
alkaline solution follow the order CoxC1−x−yNy Ͼ NixC1−x−yNy
Ͼ MnxC1−x−yNy Ͼ C1−xNx heated at 1000°C Ͼ VxC1−x−yNy
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B90
Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
Ͼ CrxC1−x−yNy. The difference between the activity of C1−xNx and
TMxC1−x−yNy ͑TM = V, Cr, Mn, Co, and Ni͒ suggests that the TM
may play a role in the formation of the active site for ORR. Appar-
ently, Co, Ni, and Mn are beneficial while V and Cr are not useful in
the formation of active sites for ORR in alkaline solution.
To summarize, the results in this paper and in the literature sug-
gest that the choice of TM is important in the formation of the active
sites for ORR both in acid and alkaline solution. As mentioned
earlier, the trends in the activities of heat-treated CrxC1−x−yNy,
CoxC1−x−yNy, and NixC1−x−yNy films in acid and alkaline solutions
are different from the trends shown by heat-treated VxC1−x−yNy and
MnxC1−x−yNy films. Heat-treated CrxC1−x−yNy, CoxC1−x−yNy, and
NixC1−x−yNy films begin to show activity at Tc in acid electrolyte
and show activity in alkaline electrolyte when heated to a tempera-
ture below Tc. Heat-treated VxC1−x−yNy and MnxC1−x−yNy films
show little or no activity for ORR in acid electrolyte but show ac-
tivity in alkaline electrolyte when heated.
NixC1−x−yNy films heat-treated at the same temperatures. The as-
sputtered and heated C films show little or no activity in acid elec-
trolyte.
All the heat-treated sputtered TMxC1−x−yNy ͑TM = V, Cr, Mn,
Co, and Ni͒ films show activity in alkaline electrolyte when heated
to a temperature below Tc. Heat-treated C1−xNx and as-sputtered or
heated C films all show activity in alkaline electrolyte. Because
disordered carbons, graphite, and transition metals all show activity
for ORR in alkaline electrolyte, it is difficult to make conclusive
statements about the results in alkaline electrolyte. However, heat-
treated Co–C–N samples which were acid-etched for 2 months still
showed equivalent ORR to fresh heat-treated samples, suggesting
the Co metal, formed after heating above Tc, is not involved in the
most active sites.
Corrosion-stability experiments on the heat-treated TMxC1−x−yNy
͑TM = V, Cr, Mn, Co, and Ni͒ libraries showed that the heat-treated
CrxC1−x−yNy libraries show good passivation against corrosion,
which is believed to be provided by the Cr component. This sug-
gests that combining Cr into TMxC1−x−yNy ͑TM = Fe or Co͒ films
may improve the stability of the films against corrosion and could
increase catalyst lifetime in PEMFC applications.
For noble-metal catalysts, the pH effect for ORR on Pt54 and on
Ag55 is basically attributed to the adsorption of spectator species that
either block the active sites required for adsorption of ORR inter-
mediates or that lower the adsorption energy of the reaction inter-
mediates. In our previous work on the activity of sputtered Co–C–N
films in acid and alkaline solutions, we noted33 that there were
samples heated to just below Tc that were active in alkaline solution
but not in acid solution. We speculated that small pores created by N
loss above Tc are too small for the adsorption of acidic species that
block catalytic sites, so the films become active above Tc in acid and
the sites are not blocked by ions in the alkaline solution regardless
of whether the samples have been heated through Tc.
An alternative explanation for the trends observed could be
based on the onset of microporosity above Tc. The detailed studies
in this work show that carbon, graphite, and CNx are all active for
ORR in the alkaline solution. More active sites appear above Tc for
TM–C–N, which increase the activity in alkaline solution. Dodelet
and co-workers56 argue that microporosity is essential for the for-
mation of MNx catalytic sites within slit-shaped pores in carbon that
are active in acid solution. Therefore, the activity in acid solution
above Tc could result from the porosity created by N loss above Tc,
i.e., micropores are essential to obtain catalytic activity in acid
solution.56
Acknowledgments
This work was supported by the Natural Sciences and Engineer-
ing Research Council of Canada and by 3M Company. We also
thank David O’Neill, Alison Schmoeckel, George Vernstrom, Rado-
slav Atanasoski, Tom Wood, and Allen Siedle of 3M Company for
many useful discussions. This work was supported in part by the
U.S. Department of Energy ͑DOE͒, Cooperative Agreement no. DE-
FC36-03GO13106, under a subcontract from 3M Company. DOE
support does not constitute an endorsement by DOE of the views
expressed in this article.
Dalhousie University assisted in meeting the publication costs of this
article.
References
1. B. C. H. Steele and A. Heinzel, Nature (London), 414, 345 ͑2001͒.
2. H. A. Gasteiger, S. S. Kocha, B. Sompalli, and F. T. Wagner, Appl. Catal., B, 56,
9 ͑2005͒.
3. B. Wang, J. Power Sources, 152, 1 ͑2005͒.
4. L. Zhang, J. Zhang, D. P. Wilkinson, and H. Wang, J. Power Sources, 156, 171
͑2006͒.
5. P. Gouérec, A. Biloul, O. Contamin, G. Scarbeck, M. Savy, J. Riga, L. T. Weng,
and P. Bertrand, J. Electroanal. Chem., 422, 61 ͑1997͒.
6. P. Gouérec, M. Savy, and J. Riga, Electrochim. Acta, 43, 743 ͑1998͒.
7. S. L. Gojković, S. Gupta, and R. F. Savinell, J. Electrochem. Soc., 145, 3493
͑1998͒.
8. G. Faubert, R. Côté, D. Guay, J. P. Dodelet, G. Dénès, and P. Bertrand, Electro-
chim. Acta, 43, 341 ͑1998͒.
9. A. L. Bouwkamp-Wijnoltz, W. Visscher, J. A. R. van Veen, E. Boellaard, A. M. van
der Kraan, and S. C. Tang, J. Phys. Chem. B, 106, 12993 ͑2002͒.
10. S. Marcotte, D. Villers, N. Guillet, L. Roué, and J. P. Dodelet, Electrochim. Acta,
50, 179 ͑2004͒.
Heat-treated VxC1−x−yNy and MnxC1−x−yNy films show little or
no activity for ORR in acid electrolyte and show activity in alkaline
solution. Apparently, V and Mn seem to show little or no ability to
create the sites which are active in acid above Tc. These samples
behave similarly to the heated carbon films, and it is therefore be-
lieved that their activity arises primarily from the carbon in the
samples. For carbon catalysts, the pH effect is attributed to the ad-
sorption of ORR intermediates that block the active sites at low
pH.57
11. M. Lefèvre and J. P. Dodelet, Electrochim. Acta, 48, 2749 ͑2003͒.
12. M. Bron, J. Radnik, M. Fieber-Erdmann, P. Bogdanoff, and S. Fiechter, J. Elec-
troanal. Chem., 535, 113 ͑2002͒.
Conclusion
13. K. Sawai and N. Suzuki, J. Electrochem. Soc., 151, A682 ͑2004͒.
14. J. Fournier, G. Lalande, R. Côté, D. Guay, and J. P. Dodelet, J. Electrochem. Soc.,
144, 218 ͑1997͒.
15. F. Jaouen, S. Marcotte, J. P. Dodelet, and G. Lindbergh, J. Phys. Chem. B, 107,
1376 ͑2003͒.
16. C. Médard, M. Lefèvre, J. P. Dodelet, F. Jaouen, and G. Lindbergh, Electrochim.
Acta, 51, 3202 ͑2006͒.
17. M. Lefèvre, J. P. Dodelet, and P. Bertrand, J. Phys. Chem. B, 106, 8705 ͑2002͒.
18. M. Lefèvre, J. P. Dodelet, and P. Bertrand, J. Phys. Chem. B, 109, 16718 ͑2005͒.
19. P. H. Matter and U. S. Ozkan, Catal. Lett., 109, 115 ͑2006͒.
20. P. H. Matter, L. Zhang, and U. S. Ozkan, J. Catal., 239, 83 ͑2006͒.
21. P. H. Matter, E. Wang, and U. S. Ozkan, J. Catal., 243, 395 ͑2006͒.
22. S. Maldonado and K. J. Stevenson, J. Phys. Chem. B, 109, 4707 ͑2005͒.
23. B. Popov, in Abstract 495, The Electrochemical Society Meeting Abstracts, Vol.
2006-2, Cancún, Mexico, Oct 29–Nov 3, 2006.
24. K. Wiesener, D. Ohms, V. Neumann, and R. Franke, Mater. Chem. Phys., 22, 457
͑1989͒.
25. D. Chu and R. Jiang, Solid State Ionics, 148, 591 ͑2002͒.
26. R. Côté, G. Lalande, G. Faubert, D. Guay, J. P. Dodelet, and G. Dénès, J. New
Mater. Electrochem. Syst., 1, 7 ͑1998͒.
Sputtered TMxC1−x−yNy ͑TM = V, Cr, Mn, Co, and Ni͒ films
transform from an amorphous structure to a nanoscale mixture of
N-containing carbon and V8C7, Cr3C2, Mn7C3, Co, or Ni when
heated above Tc. The transformation temperature, Tc, of the sput-
tered TMxC1−x−yNy ͑x = 0.10͒ films increases in the order of Co
Ϸ Ni Ͻ Cr Ͻ V Ͻ Mn. No obvious structure changes were ob-
served in the XRD patterns measured on C1−xNx and C films after
heat-treatment.
The activity of the heat-treated sputtered TMxC1−x−yNy, C1−xNx,
and C films for ORR depends on the N doping, TM type, heat-
treatment temperature, and electrolyte. Heat-treated VxC1−x−yNy and
MnxC1−x−yNy films show little or no activity for ORR in acid elec-
trolyte. Heat-treated CrxC1−x−yNy, CoxC1−x−yNy, and NixC1−x−yNy
films begin to show activity in acid electrolyte at Tc. Heat-treated
C1−xNx films show some activity in acid solution but their activities
are much lower than that of the CrxC1−x−yNy, CoxC1−x−yNy, and
27. P. He, M. Lefèvre, G. Faubert, and J. P. Dodelet, J. New Mater. Electrochem. Syst.,
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑1͒ B79-B91 ͑2008͒
B91
2, 243 ͑1999͒.
64, 63 ͑1975͒.
28. L. Liu, H. Kim, J.-W. Lee, and B. N. Popov, J. Electrochem. Soc., 154, A123
͑2007͒.
42. J. Xu, W. Huang, and R. L. McCreery, J. Electroanal. Chem., 410, 235 ͑1996͒.
43. E. Yeager, Electrochim. Acta, 29, 1527 ͑1984͒.
29. L. Liu, J.-W. Lee, and B. N. Popov, J. Power Sources, 162, 1099 ͑2006͒.
30. E. B. Easton, A. Bonakdarpour, and J. R. Dahn, Electrochem. Solid-State Lett., 9,
A463 ͑2006͒.
31. E. B. Easton, R. Yang, A. Bonakdarpour, and J. R. Dahn, Electrochem. Solid-State
Lett., 10, B6 ͑2007͒.
44. I. Morcos and E. Yeager, Electrochim. Acta, 15, 953 ͑1970͒.
45. C. Paliteiro, A. Hamnett, and J. B. Goodenough, J. Electroanal. Chem. Interfacial
Electrochem., 233, 147 ͑1987͒.
46. B. Šljukić, C. E. Banks, and R. G. Compton, J. Iranian Chem. Soc., 2, 1 ͑2005͒.
47. T. S. Sidhu, S. Prakash, and R. D. Agrawal, J. Therm. Spray Technol., 15, 811
͑2006͒.
48. J. M. Guilemany, N. Espallargas, P. H. Suegama, and A. V. Benedetti, Corros. Sci.,
48, 2998 ͑2006͒.
49. T. S. Sidhu, S. Prakash, and R. D. Agrawal, Surf. Coat. Technol., 201, 792 ͑2006͒.
50. M. Appel and A. J. Appleby, Electrochim. Acta, 23, 1243 ͑1978͒.
51. S. Motupally, M. Jain, V. Srinivasan, and J. W. Weidner, J. Electrochem. Soc., 145,
34 ͑1998͒.
32. R. Yang, A. Bonakdarpour, E. B. Easton, P. Stoffyn-Egli, and J. R. Dahn, J. Elec-
trochem. Soc., 154, A275 ͑2007͒.
33. R. Yang, K. Stevens, A. Bonakdarpour, and J. R. Dahn, J. Electrochem. Soc., 154,
B893 ͑2007͒.
34. J. R. Dahn, S. Trussler, T. D. Hatchard, A. Bonakdarpour, J. R. Mueller-Neuhaus,
K. C. Hewitt, and M. Fleischauer, Chem. Mater., 14, 3519 ͑2002͒.
35. O. Antoine and R. Durand, J. Appl. Electrochem., 30, 839 ͑2000͒.
36. T. J. Schmidt, V. Stamenkovic, M. Arenz, N. M. Marković, and P. N. Ross, Jr.,
Electrochim. Acta, 47, 3765 ͑2002͒.
52. J. R. T. Johnsson Wass, I. Panas, J. Ásbjörnsson, and E. Ahlberg, J. Electroanal.
Chem., 599, 295 ͑2007͒.
37. U. A. Paulus, T. J. Schmidt, H. A. Gasteiger, and R. J. Behm, J. Electroanal.
53. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, p. 77,
National Association of Corrosion Engineers Press, Houston, TX ͑1974͒.
54. N. M. Marković and P. N. Ross, Jr., Surf. Sci. Rep., 45, 117 ͑2002͒.
55. B. B. Blizanac, P. N. Ross, and N. M. Marković, Electrochim. Acta, 52, 2264
͑2007͒.
Chem., 495, 134 ͑2001͒.
38. W. G. Moffatt, The Handbook of Binary Phase Diagrams, Genium Publishing
Corporation, New York ͑1990͒.
39. R. A. Sidik, A. B. Anderson, N. P. Subramanian, S. P. Kumaraguru, and B. N.
Popov, J. Phys. Chem. B, 110, 1787 ͑2006͒.
56. F. Jaouen, M. Lefèvre, J. P. Dodelet, and M. Cai, J. Phys. Chem. B, 110, 5553
͑2006͒.
40. P. H. Matter, E. Wang, M. Arias, E. J. Biddinger, and U. S. Ozkan, J. Mol. Catal.
A: Chem., 264, 73 ͑2007͒.
41. R. J. Taylor and A. A. Humffray, J. Electroanal. Chem. Interfacial Electrochem.,
57. H.-H. Yang and R. L. McCreery, J. Electrochem. Soc., 147, 3420 ͑2000͒.
Downloaded on 2014-11-15 to IP 129.130.252.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).