Cross Laboratory Experimental Study of N

Cross-Laboratory Experimental Study of
Non-Noble-Metal Electrocatalysts for the
SPOTLIGHT
Oxygen Reduction Reaction
Frédéric Jaouen,*,† Juan Herranz,† Michel Lefèvre,† Jean-Pol Dodelet,† Ulrike I. Kramm,‡
Iris Herrmann,‡ Peter Bogdanoff,‡ Jun Maruyama,§ Toru Nagaoka,| Arnd Garsuch,⊥
Jeff R. Dahn,⊥ Tim Olson,# Svitlana Pylypenko,# Plamen Atanassov,# and Eugene A. Ustinov∇
Institut National de la Recherche Scientifique, Énergie, Matériaux & Télécommunications, 1650 Bd Lionel Boulet,
Varennes, Québec J3X 1S2, Canada, Helmholtz-Zentrum Berlin GmbH, Lise-Meitner-Campus, Glienicker Strasse 100,
14 109 Berlin, Germany, Environmental Technology Research Division and Processing Technology Research
Division, Osaka Municipal Technical Research Institute, 1-6-50 Morinomiya, Joto-Ku, Osaka 536-8553, Japan,
Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada,
Chemical & Nuclear Engineering Department, University of New Mexico, Albuquerque, New Mexico 87131, and
Ioffe Physical Technical Institute, Polytechnicheskaya 26, St. Petersburg 194021, Russian Federation
ABSTRACT Nine non-noble-metal catalysts (NNMCs) from five different laboratories were investigated for the catalysis of O2
electroreduction in an acidic medium. The catalyst precursors were synthesized by wet impregnation, planetary ball milling, a foaming-
agent technique, or a templating method. All catalyst precursors were subjected to one or more heat treatments at 700-1050 °C in
an inert or reactive atmosphere. These catalysts underwent an identical set of electrochemical characterizations, including rotating-
disk-electrode and polymer-electrolyte membrane fuel cell (PEMFC) tests and voltammetry under N2. Ex situ characterization was
comprised of X-ray photoelectron spectroscopy, neutron activation analysis, scanning electron microscopy, and N2 adsorption and
its analysis with an advanced model for carbonaceous powders. In PEMFC, several NNMCs display mass activities of 10-20 A g-1 at
0.8 V versus a reversible hydrogen electrode, and one shows 80 A g-1. The latter value corresponds to a volumetric activity of 19 A
cm-3 under reference conditions and represents one-seventh of the target defined by the U.S. Department of Energy for 2010 (130
A cm-3). The activity of all NNMCs is mainly governed by the microporous surface area, and active sites seem to be hosted in pore
sizes of 5-15 Å. The nitrogen and metal (iron or cobalt) seem to be present in sufficient amounts in the NNMCs and do not limit
activity. The paper discusses probable directions for synthesizing more active NNMCs. This could be achieved through multiple pyrolysis
steps, ball-milling steps, and control of the powder morphology by the addition of foaming agents and/or sulfur.
KEYWORDS: non-platinum group metal • catalyst • oxygen electroreduction • fuel cell • polymer electrolyte
I. INTRODUCTION tion is fast (2), the oxygen reduction reaction (ORR) is
S ince the advent in the 1960s of perfluorinated poly- sluggish at any pH. Therefore, the ORR requires the best
mers with proton conductivity of ∼10 S m-1 (1), the possible catalysts. However, while at high pH ORR electro-
research and development in low-temperature fuel catalysts can be chosen from a large repertoire (Pt, Pd, Ag,
cells (<100 °C) has progressively become dominated by Au, Ni-Co spinel oxides, manganese oxides, and iron or
polymer-electrolyte membrane fuel cells (PEMFCs). Rein- cobalt phthalocyanines or porphyrins) (3-12) the acidic
forced 20-30-µm-thick polymer membranes with resis- character of PEM precludes the use of most of these catalysts
tances as low as 20-30 mΩ cm2 are now available. This for reasons of either poor stability or activity. Today, Pt is
great advancement in acidic PEMs has, however, placed the the only viable catalyst for ORR in PEMFC, but still the
burden on the electrocatalysis side. While hydrogen oxida- cathode is the least efficient component. For reasons of cost,
the meaningful ORR activity of a Pt catalyst is the current
* Phone: +1 929 8176. Fax: +1 929 8102. E-mail: jaouen@emt.inrs.ca. obtained per Pt mass at a fixed cathode potential. This ratio
Received for review March 31, 2009 and accepted July 8, 2009
†
is called the mass activity.
Institut National de la Recherche Scientifique, Énergie, Matériaux & Télécom-
munications. The experimental conditions prevailing for reporting the
‡
Helmholtz-Zentrum Berlin GmbH. Pt mass activity for the ORR in a PEMFC are a cell voltage of
§
Environmental Technology Research Division, Osaka Municipal Technical
Research Institute.
0.9 V, O2 and H2 pressures of 1 bar, 100% relative humidity
|
Processing Technology Research Division, Osaka Municipal Technical Research (RH), and a temperature of 80 °C (13, 14). The Pt mass
Institute. activity has been dramatically improved by tailoring Pt in
⊥
Dalhousie University.
#
particles of 2-5 nm supported on carbon (Pt/C), alloying Pt
University of New Mexico.
∇
Ioffe Physical Technical Institute. with Ti, V, Cr, Mn, Fe, Co, or Ni (13, 15-20), optimizing the
DOI: 10.1021/am900219g catalyst-polymer electrolyte contact in porous cathodes
© 2009 American Chemical Society (21-23), or depositing a Pt monolayer on Au or Pd nano-
www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1623

Published on Web 07/20/2009
particles (24-26). Thus, the Pt loadings could be reduced the kinetic performance of today’s Pt-based cathodes, an
from 28 mg cm-2 with unsupported Pt (27) to 0.8 mg cm-2 NNMC should display no less than one-tenth of the ORR
today (0.4 per electrode) (13). The targeted Pt loading for activity of Pt/C, with the activity being defined by the current
SPOTLIGHT
automotive application is 0.1 mg cm-2 per electrode in 2015 per volume of porous cathode. This ratio is called the volu-
(14). While a negligible performance decay occurs when the metric activity. The potential chosen to report the NNMC
anode Pt loading is reduced from 0.4 to 0.1 mg cm-2 (28), activity was downscaled to 0.8 V (13) because the activity
an equal reduction at the cathode without adverse effects was too low to be measurable at 0.9 V. Under reference
on the performance requires that the mass activity of Pt conditions (O2 and H2 pressures of 1 bar; 100% RH; 80 °C),
catalysts be increased by 4-fold. the U.S. DOE targets for the NNMC activity were set at 130
Paradoxically, recent improvements in the PEMFC per- (year 2010) and 300 A cm-3 (year 2015) (14). The 2010
formance and its probable emergence as a consumer good target represents one-tenth of the volumetric activity at 0.8
have revived the problems of Pt availability and cost, which V of a Pt/C cathode in year 2005: 1300 A cm-3 (Supporting
are the fundamental incentives to search for non-noble metal Information, section A).
catalysts (NNMCs) (29). Even if the U.S. Department of NNMCs for the ORR in an acidic medium can be divided
Energy (DOE) target for year 2015 of 0.2 g of Pt kW-1 is in (i) inorganic catalysts and (ii) molecular or molecular-
achieved, a scenario of an all-PEMFC car production of, say, derived catalysts. Only a few inorganic structures with non-
100 million cars a year (in 2007, 73 million cars and trucks noble metals catalyze the ORR in an acidic medium, and
were produced worldwide (30)) with engines rated at 50 kW their activities have been, up to now, quite poor. Chalco-
would require 1000 tons of Pt a year. In 2007, the Pt world genides (35-38), nitrides (39, 40), and oxides (41, 42) have
production amounted to only 204 tons, 60% of it being been investigated. The second class of NNMCs, which is the
installed in catalytic converters for exhaust cleaning of focus of this paper, was introduced when Jasinski discovered
internal combustion engines (ICEs) (31). While an average-
that cobalt phthalocyanine catalyzes the ORR in an alkaline
size gasoline ICE vehicle requires only 1 g of Pt, 10 g is
medium (3). Later, the same molecule as well as other
expected for a PEMFC vehicle. On the other hand, diesel ICE
metal-N4 chelates was found to catalyze the ORR in an
vehicles contain on average, in Europe, 5 g of Pt (in 2007,
acidic medium as well (43). Since then, metal-N4 chelates
59 tons of Pt (31) were used to produce ∼11 million
have served as biomimetic analogues of cytochromes that
European diesel vehicles (30)). Increased Pt demand and
are key molecules for the respiratory chain (44-49). An
stagnating Pt production triggered the Pt price increase from
important step was the discovery that a high-temperature
600 to 2100 $/oz from 2003 to mid-2008 (32). In the long
treatment (400-1000 °C) of metal-N4 chelates in an inert
term, however, there could be enough Pt on earth to pro-
gas increased the stability and activity (50-57). Because
duce 3 billion PEMFC cars over the next 100-150 years at
these molecules decompose at such temperatures, the
the actual Pt mining rate (33, 34). However, is it sound to
nature of the active sites following the heat treatment has
deplete the reserves of that, in many applications, irreplace-
ever since been a subject of controversy (52-56, 58-60).
able metal? It is also questionable whether a recycling rate
of >95% of the Pt to be contained in PEMFC is feasible. This A second important step came with the high-temperature
is necessary for a sustainable worldwide PEMFC car fleet. synthesis of an NNMC free of metal-N4 chelate. Polyacry-
lonitrile mixed with CoII or FeII salt was heat treated at 800
A renewed interest in NNMCs for the ORR occurred a few
°C under argon (61). This work concluded that the high-
years ago, and targets of activity were defined (14). The
temperature synthesis of an NNMC requires the simulta-
practical activity for NNMCs is defined on a different ground
neous presence of a transition metal, N, and C. In that work,
than that for Pt-on-C catalysts (Pt/C). Because NNMCs have
the N and C sources were still joint (polyacrylonitrile). In a
a negligible cost against other PEMFC components, the
third step, it was shown that all three sources of a transition
concept of mass activity, so important for Pt for reasons of
metal, N, and C could be separately introduced in the oven.
cost, could at first sight appear to be out of place for NNMCs.
Gaseous sources for N were introduced in 1998-1999 in the
One might consider increasing the loading of poorly active
NNMCs by a factor of 50-100 versus the actual Pt loading form of CH3CN or NH3 (62, 63).
in order to match the performance of a Pt-based cathode. Since 1990, many groups have synthesized NNMCs using
Unfortunately, O2 diffusion and proton and electron con- a single or several heat-treatment steps; metal-N4 chelates
ductivity across the porous cathode are impervious to the or metal salts as a metal source; metal-N4 chelates, mol-
possible cost advantage of NNMCs. The maximum NNMC ecules, polymers, or NH3 or CH3CN gases as a N source;
loading that can be efficiently utilized under a practical metal-N4 chelates, molecules, polymers, or carbon powders
current density of >1 A cm-2 is restricted by the mass- and as a C source (64-98). The heat treatment of unsupported
charge-transport characteristics of the cathode. Today’s Pt- molecules may result in a low-porosity catalyst. To increase
based cathodes (0.4 mg of Pt cm-2; 45-50 wt % Pt on the porosity, a foaming agent can be added. This technique
carbon) are about 10 µm thick. In a first attempt to define has been applied on iron(III) chloride 5,10,15,20-tetrakis(4-
the target of activity for NNMCs, the maximum thickness methoxyphenyl)porphyrin(FeClTMPP)andcobalt(II)5,10,15,20-
envisaged for an efficient NNMC cathode was set to 100 µm tetrakis(4-methoxyphenyl)porphyrin (CoTMPP) with iron
(13). It follows that, in order for an NNMC cathode to equal oxalate as the foaming agent (71, 73, 84, 91).
1624 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

The present spotlight aims to answer two important concurring with the results found in ref 68. Pt nanoparticles (46
questions: wt %) on Vulcan XC-72 (labeled Pt/C) from Tanaka Kikinzoku
What is the present activity for the ORR of NNMCs serve as state-of-the-art Pt catalyst. They have a specific area
of 80 mPt2 gPt-1 (13).
SPOTLIGHT
obtained from a heat treatment and how does it compare II.2.2. Ink Formulation. NNMC. The ink formulation is
to U.S. DOE targets? (except for CoTMPP700) 10 mg of catalyst, 95 µL of 5 wt %
Does one physicochemical characteristic of the NNMC or Nafion in alcohol (Aldrich), and 350 µL of ethanol. The ink is
a combination of such characteristics explain their ORR prepared in a 5-mL glass vial, shaken for 15 min, and then
activity, regardless of the sources of metal, N, and C used ultrasonicated for 15 min. This cycle is then repeated twice. A
7-µL aliquot of ink is dropped onto the glassy carbon support
for their synthesis? (0.196 cm2). The 7-µL aliquot results in a catalyst loading of 800
NNMCs obtained through a heat treatment were synthe- µg cm-2. For catalysts UK63 and UK65 only, a 4-µL aliquot is
sized according to different schemes (diverse metal, N, and used for reasons explained in section III.
C precursors) at different laboratories and shipped to the Other ink formulations were tested on a case-by-case basis
Institut National de la Recherche Scientifique (INRS) between and, except for CoTMPP700, resulted in equal or lower mass
activity than with the above ink (Supporting Information, sec-
September 2007 and May 2008. Rotating-disk-electrode tion C). For CoTMPP700, the ink formulation (defined by the
(RDE) and PEMFC tests were performed. For each catalyst, University of New Mexico) consists of preparing two solutions:
PEMFC tests with different ink recipes (Nafion/catalyst ratio) solution 1 (10 mg of NNMC and 1 mL of deionized H2O) and
or NNMC loadings were run. Ex situ characterization was solution 2 (100 µL of a 5 wt % Nafion solution and 1 mL of
comprised of X-ray photoelectron spectroscopy (XPS), neu- deionized H2O). The final ink is prepared by mixing 400 µL
of solution 1 with 80 µL of solution 2 and adding 1520 µL of
tron activation analysis (NAA), N2 adsorption and its analysis deionized H2O. A 30-µL aliquot is dropped on the disk, resulting
with an advanced model for carbonaceous powders, and in a loading of 306 µg cm-2.
scanning electron microscopy (SEM). Pt/C. The ink formulation is 5 mg of catalyst (mass Pt and
All reported experimental results are based on new C), 40 µL of 5 wt % Nafion in alcohol, 222 µL of ethanol, and
10 µL of distilled H2O. The ink is then mixed as described above
measurements, not on previously published measurements,
for the usual NNMC ink. Then, a 7-µL aliquot is dropped on the
except for some data points in Figures 9-11 (see the figure glassy carbon, resulting in a catalyst loading of 657 µg (mass
captions). Pt and C) cm-2.
II.2.3. Voltammetry. For Pt/C, Pt is cleaned by cycling 20
II. EXPERIMENTAL METHODS times from -0.2 to +0.8 V vs SCE at 50 mV s-1 while the
II.1. Catalyst Synthesis. For space reasons, the fully de- electrode is idle. Then, for NNMC and Pt/C alike, one cycle in
tailed synthesis of all catalysts is given in the Supporting an O2-saturated electrolyte is recorded starting with an idle
Information, section B. For all catalysts, at least one heat- electrode at -0.25 V vs SCE and up to +0.75 V vs SCE, at which
treatment step is involved and there is at least one source of N, point the electrode is rotated at 1500 rpm. The scan rate is 10
C, and Fe (or Co). The acronyms of the catalysts are UK63, mV s-1. The scan from +0.75 to -0.25 V vs SCE under 1500
UK65,CHb200900,CoTMPP700,GAdFeCu,DAL900A,DAL900C, rpm is the raw data from which the ORR activity is extracted.
FC280, and M786. The catalysts can be classified according to N2 is then bubbled into the solution. The capacitive current ICAP
four synthesis approaches: is measured between -0.25 and + 0.75 V vs SCE at 10 mV s-1.
Approach i uses a metal-N4 chelate as the exclusive or main II.2.4. Corrections for Capacitive Current and O2
precursor for metal, N, and C. Pyrolysis is made in an inert gas. Diffusion Limitation. The current density, I (A cm-2), during
In two instances (UK63 and UK65), iron oxalate is added to the downward scan in the O2-saturated electrolyte is corrected
obtain high surface areas. Catalysts made according to approach by subtracting ICAP to yield the Faradaic current density, IF ) I
i are UK63, UK65, CHb200900, and CoTMPP700. - ICAP, defined as negative for a reduction reaction. Next, it is
The second approach (ii) uses a metal salt as the exclusive possible to calculate from IF the current that would be measured
metal precursor and a N-containing molecule as the exclusive if O2 diffusion were infinitely fast. That current is the kinetic
N source. No preexisting C support is used. Pyrolysis is carried current, IK, and is, for a given loading, controlled only by the
out under an inert atmosphere. Catalysts investigated in the ORR kinetics of the catalyst. The relationship between IK and IF
present paper that correspond to approach ii are GAdFeCu and is the Koutecky-Levich equation IK ) -IFIlim/(IF - Ilim) (ref (99),
DAL900A. p 290). IK is in A cm-2 and defined as <0 for a reduction while
A third approach (iii) consists of performing a second heat Ilim is the limiting current density. The Ilim value was taken from
treatment under a reactive atmosphere to catalysts that have the IF value at -0.25 V vs SCE.
undergone a first heat treatment under an inert atmosphere II.2.5. Definitions of NNMC Mass Activity and Pt/C
(approaches i and ii). The reactive atmosphere could be oxida- Mass Activity. In order to allow for a comparison with previ-
tive (O2 and CO2) or reductive (NH3 and N2/H2). Catalysts ously reported mass activities of NNMCs, the mass activity
synthesized according to that approach are UK63, UK65, and defined by eq I is reported in section III. Volumetric activities
DAL900C. The first two are already listed under approach i. as defined in the Introduction are reported in the Discussion
The fourth approach (iv) uses a metal salt as the exclusive section. The ORR mass activity of a C-based NNMC is defined
metal precursor and NH3 under pyrolysis as the exclusive N by
precursor. The C precursors are carbon black. The catalysts
made with this approach are FC280 and M786. IM(NNMC) ) -IK/mcatalyst [A g-1] (I)
II.2. RDE. II.2.1. Materials. Measurements are made at
room temperature and atmospheric pressure. The reference
electrode is a saturated calomel electrode (SCE) and the counter where IM > 0 and mcatalyst is the NNMC loading on the glassy
electrode a Pt wire. For the Pt/C catalyst, the SCE has a double carbon (g cm-2). The potential at which the IM value is reported
junction. The electrolyte is an aqueous solution of pH 1 of H2SO4 is 0.8 V vs a reversible hydrogen electrode (RHE), as chosen in
or HClO4 for NNMC or Pt/C, respectively. It was checked for ref 13. The mass activity of the Pt/C catalyst is similarly defined
some NNMCs that both electrolytes resulted in the same activity, by

IM(Pt/C) ) -IK/mcarbon [A g-1] (II) 2. The corresponding ink formula is 10 mg of Pt/C, 235 µL of
Nafion, 28 µL of ethanol, and 74 µL of deionized H2O. The
targeted Pt loading was 0.33 mg cm-2 (28-µL aliquot of ink).
SPOTLIGHT
where mcarbon is the carbon loading contained in the Pt/C film. The ink was mixed in the same way as that of the NNMC.
Defining the mass activity of Pt/C by eq II allows a direct II.3.3. Experimental Conditions. Teflon gaskets of thick-
comparison with NNMC mass activities because both activities ness 100, 250, or 500 µm are chosen for each MEA so that the
are only one step away from the volumetric activity defined in average compression of the GDL + active layers is 18-25%.
the Introduction, namely, multiplication by the effective density The adjustment of the gasket thickness is important when
of C in the porous cathode. increasing the cathode loading from 1 to 4 mg cm-2. The
II.2.6. Reference Electrode and Reference Potential. RDE thickness of the free-standing MEA (tMEA) and of the membrane
measurements in the present paper report the potential versus (tmemb) is measured with a micrometer. The former minus the
RHE. The change from the SCE to RHE scale is made by latter gives the sum of the anode and cathode thicknesses (GDLs
measuring the voltage ∆Eref between the SCE and a Pt foil + active layers), under no compression. The average compres-
immersed in the same electrolyte saturated in H2. ∆Eref was sion of the anode and cathode in the PEMFC is then
about 304 ( 2 mV at pH 1.
II.3. PEM Fuel Cell. II.3.1. MEA Preparation. The anode % compression ) 100(tMEA - tmemb - tgasket)/
catalyst is 40 wt % Pt/C from E-TEK with a loading of 0.35 mg
(tMEA - tmemb) (III)
of Pt cm-2 predeposited as a film on top of an ELAT gas diffusion
layer (GDL). A circular piece of 1.14 cm2 is cut out with a punch.
A thin coat (0.4-0.5 mg of Nafion cm-2) of a 5 wt % Nafion
solution is brushed onto the anode active side. The cathode is where tgasket is the sum of the thickness of the anode and cathode
prepared by pipetting the required aliquot of catalyst ink on the gaskets.
hydrophobic side of a 1.14 cm2 uncatalyzed ELAT GDL (thick- The temperature of the humidifiers and of the cell is raised
ness and mass of 360-400 µm and 20-25 mg cm-2, respec- under N2 flow to 100 and 80 °C, respectively. Once the
tively). The catalyst loading is measured by weighing the dry temperatures are stable, O2 and H2 flows are switched on (208
GDL before and after application of the ink. The mass gain is and 416 sccm, respectively, corresponding to ∼90 times the
due to the dry mass of a Nafion ionomer and the mass of the stoichiometric flow for 1 A), backpressures (gauge pressures)
catalyst. Because the ratio of Nafion to catalyst is known, it is are set to +1 bar (the gas pressure at either the anode or cathode
possible to deduce the catalyst mass from the mass change. The is thus 1 atm + 1 bar, and the O2 and H2 partial pressures are
anode and cathode are dried first on a heat plate and then at 1.5 bar because PH2O,sat at 80 °C is 0.5 bar), and the cell voltage
80 °C under vacuum, then weighed, and then hot-pressed at is set to 0 V for about 5 min (for measurement with NNMC only)
140 °C against a Nafion 117 membrane (thickness 160-170 to completely humidify the membrane. Then, the cell is allowed
µm, a 5 × 5 cm piece). The anode, cathode, and membrane to set at open-circuit potential (OCP) until the OCP stabilized.
are placed in a compression set made of two copper plates (11 Then the polarization curve of the NNMC is recorded by scan-
× 11 cm) and of two masks (thickness 280 µm), with each mask ning the cell voltage from 0.9-0.85 V down to 0.1 V at a scan
having a 1.14 cm2 hole to host the electrodes. One additional rate of 0.5 mV s-1. Immediately after, an electrochemical
mask of thickness 100 µm was added for the cathodes loaded impedance spectroscopy (EIS) measurement is taken at the
with 4 mg cm-2 of NNMC. The press is a Carver laboratory press OCP. Then, cyclic voltammetry under N2 at the cathode is
made of two metal plates (15 × 15 cm) heated by cartridges. recorded for all NNMC MEAs: the cell is allowed to cool at room
The compression test is installed in the press, the two heated temperature, and the humidifier temperatures are set to 60-70
plates of the press are brought in contact with each face of the °C; the anode and cathode backpressures are both +1 bar. At
set without any load for 1 min, and then a load of 500 lbs (0.4 the anode, H2 is diluted with N2, and at the cathode, pure N2 is
metric ton) is applied for 40 s. The membrane electrode used. Cyclic voltammetry is performed between 0.0 and 1.2 V
assembly (MEA) is installed in a fuel-cell fixture of Electrochem with a scan rate of 20 mV s-1.
Inc., and the experiments are controlled with a potentiostat For the Pt/C catalyst, the cathode is activated by a breakin at
PARSTAT 2273 (Princeton Applied Research). The MEA is 0.6 V for 5 h, followed by repeating 5 times the following cycle:
installed so that its 1.14 cm2 geometric active area is placed in 0.85 V for 15 min/OCP for 15 min/0.65 V for 15 min. The
the middle of the 5 cm2 area of serpentine gas channels, and polarization curve is recorded with a scan rate of 0.5 mV s-1
the rest of the MEA (Nafion 117, 5 × 5 cm piece) is clamped immediately after the last step at 0.65 V.
between two Teflon gaskets of thickness 250 µm. An additional II.3.4. iR Correction with EIS. Because the H2 oxidation
gasket of 100 µm was added at the cathode side for cathodes reaction is fast over Pt, the anode overpotential is less than 10
with a loading of 4 mg cm-2 of NNMC. mV (28) and the cell voltage is about the cathode potential, EC
II.3.2. Ink Formulation. NNMC. Two ink formulations are (V vs RHE), minus the ohmic drop, |I|R. The value R is measured
defined: one in which the Nafion-to-catalyst mass ratio (Naf/Cat) with EIS. For cathodes loaded with 1 mg cm-2 of NNMC, the R
is 1 (ink a) and the other in which it is 2 (ink b). The formulation values were 180-230 mΩ cm2 except for cathodes made with
of ink a is 10 mg of catalyst, 217 µL of Nafion (5 wt % Nafion the catalysts DAL900A and DAL900C (300 mΩ cm2) and with
solution, Aldrich, EW 1000), 272 µL of ethanol, and 136 µL of CoTMPP700 (390 mΩ cm2).
deionized H2O. The formulation of ink b is 10 mg of catalyst, II.4. Catalyst ex Situ Characterization. II.4.1. NAA and
435 µL of Nafion, 54 µL of ethanol, and 136 µL of deionized XPS. NAA is used to assess the bulk concentration of the metals
H2O. After the ink was prepared in a 5-mL glass vial, it was and is performed at Ecole Polytechnique de Montreal. Surface
mixed by alternating between 15 min of shaking and 15 min elemental analysis is done by XPS using a VG Escalab 200i
of ultrasonication and repeating this cycle twice. For NNMC instrument and a Al KR line (1486.6 eV) as the X-ray source.
loadings of 1 and 4 mg cm-2, aliquots of 71 and 284 µL were The N 1s narrow scan is deconvoluted in five peaks (Table 1),
deposited on 1.14 cm2 GDL (the same aliquots for ink a or ink one each for a N type except peaks II and III. Peaks II and III
b). may each be a lump response of two N functionalities. All peak
Pt/C. For this catalyst (46 wt % Pt on Vulcan, Tanaka), INRS functions are 30% Gaussian and 70% Lorentzian. Constraints
has studied the effect of various mass ratios of Nafion-to-carbon, on the peak position and width are applied (Table 1).
Naf/C (unpublished). For our method of preparation and for a The various N coordinations with the C support have been
1.14 cm2 MEA, the maximum activity was found for Naf/C ) presented in refs 100-102. Except for graphitic N in the

Table 1. Constraints Applied to the Centers and Width of the Gaussian-Lorentzian Functions for Fitting of
the Experimental N XPS Narrow-Scan Spectrum
SPOTLIGHT
position N3 (see the scheme corresponding to peak IV in Table been done with a new concept that considers the solid and
1), all N types are found at the edge of graphene layers (Table adsorbed gas as two components of a binary mixture (112-115).
1). Graphitic N bonded to three C atoms has a binding energy The gas-solid molecular parameters were determined by fitting
(BE) of 400.5-402.7 eV, while when bonded to two C atoms a N2 adsorption isotherm of a nongraphitized carbon black BP
and one H atom (position N1, Table 1), it has a BE of 400.1-401.2 280 Cabot (111). The fitting error was <2% over 6 orders of
eV (101). This lower BE overlaps with the BE ascribed to pyrrolic magnitude of pressure, proving that the new model accounts
N. Thus, peak III must be regarded as the sum of pyrrolic N and for the energetic heterogeneity of amorphous surfaces. In the
graphitic N of type N1 (101). Peak II must also be regarded as a present work, the newly developed NLDFT is relied on to
sum, namely, that of nitrile N and N bonded to a metal in an analyze the PSD of the catalysts. If one would have used the
Me-Nx center. The assignment to the Me-Nx type of N is based classical NLDFT model found in commercial softwares instead,
on studies of metal-less and metal-containing macrocycles. The the conclusion of the importance of micropores for the activity
replacement of protons by a metal ion introduces an electron- of such catalysts would not change. However, the fine subdivi-
withdrawing effect, leading to an increase in the N 1s BE sion of micropores made in Figure 11 would have been
(103, 104). Studies of iron and cobalt porphyrins have reported impossible.
a N 1s BE of ∼399.2 eV (93-96).
II.4.2. N2 Isotherm and Its Analysis. Measurements are III. RESULTS OF ORR MASS ACTIVITY IN RDE
made with a Quantachrome Instruments Autosorb-1. The pore- Figure 1A presents the Faradaic current density, IF, of the
size distribution (PSD) is determined from the adsorption
isotherm using nonlocal density functional theory (NLDFT). In various NNMCs and of the Pt/C catalyst, measured at a
view of the importance of micropores for the present NNMC rotation of 1500 rpm. The target NNMC curves, curves b in
(Discussion section), the correct quantification of micropores Figure 1A,B, are derived from the Pt/C Tafel plot (curve a in
is paramount. An NLDFT model specific for disordered carbon- Figure 1B), as explained in the Supporting Information,
aceous materials has been used. Its advantage against the section D. The target NNMC curve has the same Tafel slope
classical slit pore model is now discussed.
The NLDFT was developed for micro-mesoporous materials as Pt/C and one-tenth of the Pt/C mass activity, IM (eq II).
by Tarazona and co-workers (105, 106) and applies molecular The mass activity measured for Pt at 0.9 V (Figure 1B) is
simulations to bulk fluids or fluids confined in pores. The PSD about 65 A (g of C + Pt)-1, i.e., about 140 A gPt-1 at 20 °C.
analysis (107-109) is based on a set of local adsorption This value is close to values reported for the very same
isotherms (kernel), with each local isotherm being calculated catalyst in Table 4 of ref 13 (160-220 A gPt-1 at 60 °C).
for a given pore width. The task is to determine the contribution
of each kernel to the experimental isotherm. This deconvolution For UK63 and UK65, the loading was decreased from 800
task is usually solved by the Tikhonov regularization procedure to 460 µg cm-2. The loading of 800 µg cm-2 resulted in two
(110). Nowadays, this NLDFT-based PSD analysis is found in drawbacks: (i) a large spike at 0.75 V vs RHE and (ii) the
softwares of automatic analyzers produced by Micromeritics reading of the mass activity at 0.8 V vs RHE required
Instruments and Quantachrome Instruments. However, the
extrapolation of the Tafel slope observed at E > 0.8 V vs RHE.
theoretical basis was developed for crystalline surfaces like
graphite while activated carbon or carbon black shows a dif- For a 4e ORR reduction at 1500 rpm, a limiting current
ferent type of pore surface (111). The model proved imperfect of ∼6 mA cm-2 is expected. A flat plateau is observed for
for carbon powders, which appeared in poor fitting of the the catalysts Pt/C, UK63, UK65, FC280, and M786 (Figure
experimental adsorption isotherms. As a result, the PSD often 1A). The other catalysts show a tilted plateau and a transition
shows a gap around the pore size of 1 nm, which is a model-
engendered artefact (112). To improve the situation, it was of control from kinetics to diffusion that is not as sharp as
necessary to extend the NLDFT to amorphous solids, accounting that for the catalysts previously mentioned (except DAL900C,
for surface roughness and energetic heterogeneity. This has with a sharp transition but tilted plateau). This could be due

SPOTLIGHT
FIGURE 2. (A) N2 cyclic voltammetry in RDE. N2-saturated electrolyte

of pH 1. Scan rate 10 mV s-1. Curve labels are 1 (UK63), 2 (UK65), 3
(FC 280), 4 (M786), 5 (DAL900C), 6 (CoTMPP700), 7 (GAdFeCu), 8
(DAL900A), and 9 (CHb200900). (B) N2 cyclic voltammetry of the
NNMC cathodes in PEMFC. Scan rate 20 mV s-1, Naf/Cat ratio 2, and
NNMC loading 1 mg cm-2.
peak. The nature of the redox peaks is discussed in the
Supporting Information, section E. The N2-CVs measured
FIGURE 1. (A) Polarization curves in RDE at 1500 rpm. O2-saturated
electrolyte of pH 1 at 20 °C. NNMC loading of 800 µg cm-2 except under PEMFC conditions (Figure 2B) and at a scan rate of
UK63-65 (460 µg cm-2) and CoTMPP700 (306 µg cm-2). Curve a: Pt/ 20 mV s-1 are discussed in the next section.
C, 302 µg of Pt cm-2 (355 µg of C cm-2). Curve b: NNMC target for
year 2010 for a loading of 460 µg cm-2. Other curve labels are 1
(UK63), 2 (UK65), 3 (FC 280), 4 (M786), 5 (DAL900C), 6 (CoTMPP700), IV. PEMFC RESULTS
7 (GAdFeCu), 8 (DAL900A), and 9 (CHb200900). (B) Tafel plots (E vs IV.1. ORR Mass Activity. PEMFC polarization curves
log IM) in RDE. Same conditions as those in part A. Curve a: 46%
Pt/C. Curve b: NNMC target by U.S. DOE for year 2010. For curve a,
are shown only in the next section because graphs of the
the Pt activity is as defined by eq II [A (g of C)-1 contained in Pt/C]. potential versus the current in a linear scale cannot display
the ORR mass activity at 0.8 V vs RHE (small currents). The
to the restricted porosity inside the catalytic film, leading to first figure of this section presents the Tafel plots of the
O2 diffusion limitation not only in the quiescent electrolyte various NNMCs and of Pt/C (Figure 3). For each NNMC,
layer close to the electrode but also inside the porous PEMFC experiments were conducted with two possible
catalytic film. Possibly, the tilted plateau arises because O2 Nafion-to-catalyst mass ratios in the cathode ink (Naf/Cat of
is not fully reduced to H2O on these particular NNMCs. 1 or 2) and with two possible NNMC loadings at the cathode
Measurement of % H2O2 was not made because it is more (1 or 4 mg cm-2). Figure 3 presents only the Tafel plots
complex than previously believed. It has been recently obtained with Naf/Cat ) 2 and a loading of 1 mg cm-2. It
shown for an NNMC synthesized by Fe impregnation, like will be seen later that, for any NNMC, this combination
FC280, that % H2O2 decreases dramatically with an increase resulted in the highest ORR mass activity. The ORR mass
in the catalyst loading (116). This has also been observed activity at 0.8 V vs RHE (iR-corrected potential) is either
for a Ru-Se catalyst (117). directly read when the curve shows a straight Tafel slope at
Figure 1B presents the kinetically controlled current IM of 0.8 V or, for catalysts UK63, UK65, M786, and DAL900C, is
various NNMCs and of the Pt/C catalyst. Tafel slopes are obtained by extrapolation of the Tafel slope seen at poten-
observed for all NNMCs in the interval 0.9-0.75 V vs RHE tials >0.8 V. Such an extrapolation is shown for M786 and
with values in the range 54-65 mV decade-1 except for also for the Pt/C catalyst (dash-dotted lines in Figure 3). The
M786 (77 mV decade-1). The mass activities at 0.8 V vs RHE ORR mass activities thus obtained are reported later in this
obtained from Figure 1B are reported later. section.
Figure 2A presents the N2 cyclic voltammograms The targeted Tafel plot for NNMC by year 2010 according
(N2-CVs) in RDE recorded at a scan rate of 10 mV s-1. The to the U.S. DOE (one-tenth of the Pt/C activity) is represented
signal is expected to depend on the specific surface area of by curve b and was calculated from the experimental Pt/C
the catalysts but might also depend on the quality of the curve a in the same manner as that in section III. A Tafel
electrolyte-catalyst interface. Catalyst UK65 shows a redox slope of 65 mV decade-1 is assumed, and the targeted mass

SPOTLIGHT
FIGURE 3. Tafel plots of the various NNMC cathodes in a PEMFC.
NNMC loading 1 mg cm-2. Naf/Cat ) 2. Curve a: 46% Pt/C; 0.33 mg
of Pt cm-2; Naf/Cat ) 2. For curve a, the Pt activity is as defined by
eq II [A (g of C)-1 contained in Pt/C]. Curve b: estimated target
activity for NNMC for year 2010 (see the Supporting information,
section D).
activity is one-tenth of that measured on Pt/C in the same

PEMFC and under the same conditions.
Figure 2B reports the N2-CVs (scan rate 20 mV s-1)
measured in situ in a PEMFC on the same MEAs whose Tafel
plots are reported in Figure 3. N2-CVs in a PEMFC on
cathodes having other combinations of Naf/Cat ratios and
NNMC loadings were measured as well but are not shown. FIGURE 4. (A) ORR mass activity at 0.8 V vs RHE in RDE and PEMFC.
The catalysts UK63, UK65, and CHb200900 show a clear (B) Capacitance in RDE and PEMFC.
redox peak.
Generally, the NNMCs show more intense redox peaks In order to explain why, for a given NNMC, the mass
in a PEMFC than in a RDE (e.g., UK63). This is particularly activity may show large changes from RDE to PEMFC
true at a loading of 4 mg cm-2 (not shown). This does not (remark i above), one might call upon two phenomena: (a)
seem to be due to a better covering of the catalyst surface the extent of the interface Nafion ionomer/NNMC or (b) the
by the Nafion ionomer in the PEMFC cathodes than in the activation energy that governs the increase in activity with
RDE ones because the capacitance per mass of catalyst temperature (60 °C increase from RDE to PEMFC).
(values reported later) is generally not much larger in a The temperature effect was not experimentally investi-
PEMFC than in a RDE and sometimes even smaller. Maybe gated for all catalysts at intermediate temperatures and will
the pH of the Nafion ionomer is <1. This could lead to more only be examined in the Discussion section. Point a is now
intense redox peaks in PEMFC vs RDE because the Support- investigated by analyzing the N2-CVs measured in RDE and
ing Information, section E, shows that the redox signal PEMFC. The capacitance is averaged over one full cycle
increased when the pH decreased from 4 to 1. according to
Next, Figure 4A reports, for each catalyst, the ORR mass
activities at 0.8 V vs RHE in RDE and in PEMFC. The x axis ∫ ICAP dE
lists the NNMCs according to increasing RDE mass activity. CM ) [F g-1] (IV)
mcatalyst × 2v∆E
Considering all catalysts, the activity in a RDE spans from
0.4 to 18 A g-1 while the activity in a PEMFC spans from
0.7 to 80 A g-1. Three remarks are made: where CM is the average capacitance per mass of catalyst,
(i) For each catalyst, the mass activity is usually higher in ICAP is the capacitive current density, E is the potential,
a PEMFC than in a RDE. For example, M786 shows a factor mcatalyst is the catalyst loading, ν is scan rate, and ∆E is the
of 7.5-28 higher activity in PEMFC than in RDE. Only scanned potential window.
FC280, UK65, and UK63 show similar activities in PEMFC Figure 4B presents the CM values under RDE and PEMFC
and RDE. conditions. For each catalyst, we now investigate whether
(ii) For each catalyst, the mass activity is highest with a changes in the CM values could explain some of the large
ratio Naf/Cat of 2. changes observed in the IM values when switching from RDE
(iii) At a fixed ratio Naf/Cat of 2, an increase in the loading to PEMFC. When parts A and B of Figure 4 are compared, it
from 1 to 4 mg cm-2 resulted in slightly improved mass is obvious that the increased activity in PEMFC vs RDE
activity (FC280), slightly decreased mass activity (CHb200900, activity (e.g., GAdFeCu, DAL900C, and M786) cannot be
DAL900C, UK65, and UK63), or a decrease by a factor of explained by an improved Nafion-catalyst interface be-
about 2 (DAL900A, GAdFeCu, CoTMPP700, and M786). cause the average capacitance in a PEMFC is only negligibly

SPOTLIGHT
FIGURE 5. Cathode polarization curves of all NNMCs in the PEMFC. Each color corresponds to a given set of mass ratio Nafion/NNMC (Naf/Cat)
in the cathode and of NNMC loading at the cathode. Black: Naf/Cat mass ratio ) 1 and NNMC loading 1 mg cm-2. Blue: Naf/Cat ) 1 and NNMC
loading 4 mg cm-2. Green: Naf/Cat ) 2 and NNMC loading 1 mg cm-2. Red: Naf/Cat ) 2 and NNMC loading 4 mg cm-2.
higher in PEMFC versus that in RDE (M786) or even smaller

(GAdFeCu and DAL900C).
The other alternative to account for this effect of im-
proved activity in PEMFC versus RDE for some catalysts
would be a temperature effect (see the Discussion section).
IV.2. Fuel Cell Performance. While the previous
section focused on the mass activity of NNMCs under kinetic
control, here we are interested in the current density (mA
cm-2) of the porous cathode and how it can be maximized
by either increasing the catalyst loading (costless) or opti-
mizing the ink recipe. In the present work, NNMC cathodes
are fabricated onto a GDL, which is then hot pressed against
the membrane. We recognize that this procedure, although
adapted for measurement on 1 cm2 MEA, is probably not
the best fabrication method especially for MEAs >5 cm2. For
example, for the catalyst CoTMPP700, an alternative MEA
fabrication yielded an equal or slightly better performance
than that reported here (92, 93). This being pointed out, the FIGURE 6. Current density at 0.6 V (iR-corrected) in PEMFC. The
performance of the various catalysts with the present fab- catalysts are ranked from left to right by increasing mass activity at
0.8 V in PEMFC.
rication method is now compared.
Figure 5 presents all PEMFC polarization curves. At a inside any porous electrode yields such an effect. Water
fixed loading of 1 mg cm-2, it is seen that the Naf/Cat ratio flooding combined with another transport limitation might
of 2 (green curves) resulted for all NNMCs in an equal or also result in such an effect. However, such a modeling
higher performance than that of the Naf/Cat ratio of 1 (black example is not known by the authors to have been reported
curves) at all potentials, except for DAL900C, for which these in the literature.
two curves cross each other at 0.6 V. Next, at a fixed Naf/ For practical applications of the PEMFC, the operating
Cat ratio of 2, increasing the loading from 1 to 4 mg cm-2 voltage would likely not be <0.6 V because otherwise the
usually resulted in a higher performance at high potential, energy efficiency is too low and the heat production too high.
while at low potential, the performance was poorer (green Thus, the iR-corrected cell voltage of 0.6 V was chosen in
vs red curves). The intersection point of these two curves order to compare the current density of all NNMC cathodes
depends greatly on the NNMC. For a given NNMC, the fact (Figure 6). The x axis ranks the NNMCs according to mass
that the PEMFC polarization curve with high loading crosses activity at 0.8 V in the PEMFC. Thus, if all NNMC cathodes
that with low loading can be explained on theoretical had identical mass- and charge-transport properties, the
grounds by modeling of the porous cathode. For example, current density at 0.6 V should increase from left to right.
simultaneous limitation by proton and electron conduction Obviously, this is not necessarily the case, especially not for

Table 2. Elemental Surface (XPS) and Metal Bulk (NAA) Concentrations of the NNMCsa
Ntotal Npyridinic Nnitrile or Me-Nx Npyrrolic O metal Si S metal bulk
SPOTLIGHT
UK63 3.06 1.03 0.51 0.47 3.50 0.36 Fe 0.45 0.71 Fe
UK65 2.55 0.84 0.47 0.38 2.64 0.19 0.08Co
0.30 Fe 0.62 Fe
FC280 1.90 0.45 0.36 0.35 0.49 <0.1 Fe 0.06 Fe
M786 1.91 0.57 0.38 0.28 1.31 0.17 Fe
DAL900C 2.93 0.63 0.40 0.61 1.30 0.12 Fe 0.31 Fe
CoTMPP700 5.37 1.80 1.37 0.83 10.30 0.74 Co 0.85 1.29Co
0.30 Fe 0.14 Fe
GAdFeCu 1.02 0.11 0.11 0.27 2.74 0.48 Cu 0.02Cu
0.15 Fe
DAL900A 3.26 0.74 0.47 0.71 2.46 <0.1 Fe 0.13 0.10 Fe
CHb200900 1.30 0.23 0.17 0.32 5.75 0.26 Fe
a
Catalysts are listed according to the RDE activity (highest activity on the upper line). All concentrations are given in atom %. Bulk metal
content on the last column (atom % bulk).
FIGURE 7. N 1s narrow scan spectra of the NNMCs and their deconvolution.
DAL900C. All NNMC cathodes are strongly limited by mass V. XPS RESULTS
transport at 0.6 V because none of them display an increase The elemental surface concentrations of the NNMCs are
by a factor of 4 in the current density when the loading is found in Table 2. For N, the contributions of pyridinic, nitrile,
increased from 1 to 4 mg cm-2. Today’s harsh transport and pyrrolic N atoms are detailed because some of these N
limitation within the NNMC cathodes is a main barrier to the types are likely to be involved in the active sites. Figure 7
efficient implementation of NNMC into PEMFC. The cath- shows N 1s spectra of the various NNMCs and their decon-
odes with least poor transport properties are those made volution by five peaks (see section II.4.1).
with M786, UK63, UK65, and FC280. Two of these catalysts The N surface content for all NNMCs is 1.30-5.37 atom
are made from a fine carbon black support, and the two %. The N functionalities are not divided in the same way
others have high porosity and fine particles (see the Sup- for all catalysts. Some show a predominance of pyridinic N
porting Information, section F) because of the addition of a (UK63, UK65, FC280, M786, and CoTMPP700) and others
foaming agent in their synthesis. The poorest mass transport an equal repartition between pyridinic and pyrrolic N
is observed on cathodes made with CHb200900, DAL900A, (DAL900C and DAL900A), while still others show a predomi-
CoTMPP700, and DAL900C. The last three catalysts involve nance of pyrrolic N (GAdFeCu and CHb200900).
a silica template in their synthesis, and SEM micrographs The O content is in the range 0.49-5.75 atom % except
show large particles or flakes. Possibly the tortuosity of the for CoTMPP700 (10.3 atom %). O is chemically or physically
electron and/or proton path is higher in these NNMC adsorbed on the NNMCs upon exposure to air. For some
cathodes. NNMCs, O might also originate from the precursor mol-

ecules. There is no correlation between the O content and
NNMC activity.
Next, the surface Fe could be quantified on several
SPOTLIGHT
catalysts. One interesting point is that the surface Fe is

detected only in traces (<0.1) on DAL900A but quantified to
0.12 atom % on DAL900C. The latter is obtained from the
former by doing a second heat treatment under NH3, and
we saw that this resulted in much improved activity in
PEMFC (factor 8).
Next, Si is present on DAL900A and CoTMPP700 and
comes from the Si template used in their synthesis. Fluoride
is present only on DAL900A and comes from the HF leaching
used to remove the Si template.
S is present on UK65 and UK63 and comes from their
synthesis (Supporting Information, section B). S prevents
excess Fe from forming iron carbide during the heat treat-
ment because excess Fe will preferentially form iron sulfide,
Fe1-xS. The formation of iron carbide during the heat treat-
ment of the iron porphyrin would catalyze a long-range
graphitization of the C and would negatively affect the
surface area of the powder (84, 91).
VI. NEUTRON ACTIVATION RESULTS

FIGURE 8. (A) N2 adsorption isotherms of the various NNMCs.
The metal bulk contents of the NNMCs are found in the Experiment (squares) and fitting by the present NLDFT model (lines).
last column of Table 2. The Fe bulk content is in the range (B) Surface area distribution obtained from analysis by the present
of 0.06-0.71 atom % (0.29-3.3 wt %). CoTMPP700 is NLDFT model.
mainly Co-based, while catalyst UK65 contains a minor
amount of Co. Because the mass activity of Co-based particles of about 400 nm. For the other NNMCs (except
catalysts has been shown for one type of catalyst to be about M786), the carbon structure is entirely formed during the
10 times lower than that of Fe (78) and under the assump- heat treatment because of the partial graphitization of the
tion that this order of magnitude is valid for the other NNMCs carbon precursors. These catalysts show a much coarser
containing both Co and Fe (CoTMPP700 and UK65), it can structure than FC280. The particle size ranges from 0.5 to
be stated that all of the presently investigated catalysts owe 20-30 µm. The catalysts UK63 and UK65 have, within that
their activity to Fe, except CoTMPP700. The activity of the family, a finer structure with average particle sizes of 7 and
latter could equally come from Co and Fe. 4 µm, respectively. This is due to the presence of a foaming
Evolution of the catalytic activity with the metal bulk agent (iron oxalate) and of S in the preparation of these
content is analyzed in the Discussion section. catalysts. The particle size of such catalysts has been found
to be controlled by the crystal size of the iron oxalate used
VII. POROSIMETRY RESULTS (91, 119). The catalyst M786 is a cross between a carbon
Figure 8A shows that the amount of pores differs widely black support and a carbon support obtained from graphi-
among the various NNMCs. The shapes of the isotherms are, tization of carbon molecules because its preparation involves
however, grossly similar and typical of carbonaceous ma- 50% of a high-surface-area carbon black and 50% of PTCDA
terials comprising all types of pores. The isotherms are of (C24H8O6). This catalyst shows particles of 3 µm on average.
type I at P/P0 < 0.05, where filling of the micropores by the
The existence of large particles (>1 µm) is probably not
adsorbate occurs, and of type II at P/P0 values >0.05, where
desirable in a catalytic powder used to form a porous
the slope is indicative of the amount of mesopores (118).
The isotherms were fitted (solid lines) with a model for cathode of 10-50 µm thickness. Also, it can be expected
disordered carbonaceous materials (section II.4.2) and yielded that the larger the particle, the more difficult it is for the
the PSD (Figure 8B). The importance of the pore size for the Nafion ionomer to connect the active sites existing in it. It
kinetic activity of such NNMCs is outlined in the Discussion is also probably more difficult for O2 to reach these sites
section. because the pores that exist inside the particles are narrow.
Diffusion is much slower in narrow pores because the
VIII. SEM RESULTS mechanism changes from molecular (gas-gas collision) to
SEM micrographs of all catalysts are shown in the Sup- Knudsen (gas-wall collision) diffusion, which can decrease
porting Information, section F. A quick review of these the effective diffusion coefficients by decades. Larger par-
reveals a chasm between FC280 and all other NNMCs of the ticles might also induce fewer points of contact between the
present study. FC280 is based on a carbon black that shows particles and thus might result in a decreased electronic
a particle size of 20-50 nm and agglomerates of such conductivity of the porous cathode.

IX. DISCUSSION
IX.1. Volumetric Activity of the Present NNMC
Cathodes: Comparison to U.S. DOE Targets. Equa-
SPOTLIGHT
tion V converts mass activities at 0.8 V iR-free cell voltage,
IM (defined by eq I) measured under pressures PO2 and PH2
into volumetric activities at 0.8 V iR-free cell voltage, IV*,
under reference conditions of 1 bar of O2 and H2 pressures.
( )( )
PO2/ 0.79 PH2/ αc/2
IV* ) IMFeff (V)
PO2 PH2
where PO2* and PH2* are the reference pressures (1 bar) and
Feff is the effective density of a C-based NNMC in the porous
cathode. In eq V, the exponents for O2 and H2 pressures have
been determined by Neyerlin et al. (eq 12 in ref 120). The
coefficient Rc is the cathodic transfer coefficient and is
related to the Tafel slope through Rc ) RT ln 10/(F × Tafel
slope). It is assumed in eq V that the exponent 0.79,
determined experimentally for Pt/C, also describes the PO2
dependence of the activity of NNMCs. Next, the effective
density of a C-based NNMC in a porous cathode was previ-
ously assumed to be 0.4 g cm-3 (13). This assumption is
retained here. The value of 0.4 g cm-3 corresponds roughly FIGURE 9. (A) Mass activity in RDE vs Fe bulk content (measured by
to 50% porosity in the cathode. Applying eq V, the mass NAA) of NNMC. The red squares are literature data obtained by the
impregnation of different amounts of Fe on a non-microporous
activities at 0.8 V (in A g-1; measured under PO2 of 1.5 bar, carbon and heat treatment at 950 °C in NH3 for the duration
100% RH, and 80 °C) multiplied by 0.237 (if Rc ) 1, i.e., 70 necessary to obtain 30-35 wt % loss of C. Red squares reprinted
mV decade-1) give the volumetric activities (A cm-3) under with permission from ref 78. Copyright 2008 Elsevier. (B) Mass
activity in RDE vs N surface content (XPS) of NNMC. Red squares
the reference conditions (PO2 and PH2 of 1 bar, 100% RH, reprinted with permission from ref 78. Copyright 2008 Elsevier.
and 80 °C) defined in refs 13 and 14.
The estimation of the volumetric activities tells us that the microporous carbon black of low surface area followed by
two NNMCs presently investigated (UK63, volumetric activ- heat treatment with the dried powder at 950 °C in pure NH3,
ity of 5.6 A cm-3 with Naf/Cat ) 2; DAL900C, 6.0 A cm-3 that the activity was controlled by the microporous surface
with Naf/Cat ) 2) are a factor of 21-23 below the U.S. DOE area (pore size < 20 Å) produced during the heat treatment
target of year 2010 (130 A cm-3 at 0.8 V) and one catalyst (77). Thus, three factors are of particular interest in trying
(M786, 18.7 A cm-3 with Naf/Cat ) 2) is only a factor of 7 to explain the activity of NNMC: the metal content, the N
below the target. This is a large improvement compared to content, and the microporous area created during the heat
the volumetric activities of 0.7-1.5 A cm-3 under reference treatment.
conditions (mass activity 3-5 A g-1 under 1.5 bar of O2 Figure 9A presents, as a function of the bulk Fe content,
pressure) reported for the previous best INRS catalysts the ORR mass activity measured in RDE for the present
(78, 90) that were obtained by impregnation of iron acetate NNMC (circles) as well as data measured previously (squares)
on carbon black followed by heat treatment in NH3. This by changing on purpose the Fe content of catalysts synthe-
catalyst type is represented in the present study by FC280 sized otherwise under fixed conditions (78). In that work, it
(1.1-1.3 A cm-3). was observed that the activity increased linearly only up to
Comparing the present NNMC activities to recent litera- a Fe content of about 0.1 wt %, where the activity departed
ture data, high activity has also been reached by Wood (88). from the linear relationship (solid line in Figure 9A). For the
At 80 °C and under O2/H2 backpressures of 50 psig/30 psig, squares, the leveling off and decrease of the activity at Fe >
a volumetric activity of 19 A cm-3 was reported. When the 0.1 wt % was explained by the concomitant decrease of the
correction for PO2 and PH2 (eq V with Rc ) 1.03 experimen- microporous area of the catalysts (78). With regard to the
tally determined) is applied, this transforms into 2.7 A cm-3 very different ways of producing the NNMC of the present
under reference conditions (PO2 and PH2 of 1 bar). data (circles), the superimposition of the new data to the
IX.2. What Characteristics of NNMC Set Their former one (squares) fits surprisingly well. The only catalysts
Activity? It has been shown that the activity for the ORR with Fe content > 0.1 wt % that show increased activity are
is obtained when the simultaneous presence of a metal (Fe UK63 and UK65. Their activity is, however, less than pre-
and Co), N, and C is ensured during a heat treatment at T > dicted by the solid line by a factor 2-4, and this is inter-
600 °C. Recently, it has been shown at INRS, for catalysts preted as an Fe utilization of 25-50%. A detailed study of
synthesized by impregnation of an iron salt onto a non- similar catalysts is found in ref 84. Their activities were

found to correlate with one structure found in Fe Mössbauer
spectroscopy (D1). Only 30-60% of Fe was found to have
this structure. Moreover, only Fe with the right structure and
SPOTLIGHT
found at the surface can be an active site for the ORR.

Catalyst FC280 is found right on the solid line (Figure 9A).
This was expected because FC280 is synthesized in the same
way as the catalysts represented by the squares. For the six
other NNMCs of the present study, increased Fe content
resulted in decreased activity.
An explanation of the descending part of the bell-shaped
curve in Figure 9A was given in ref 78. Excess Fe can
catalyze NH3 decomposition into N2 and H2. Hydrogen is less
reactive than NH3, and consequently both the microporous
area and N content decreased with increasing Fe content,
resulting in a negative effect of excess Fe on the activity (78). FIGURE 10. Mass activity in RDE vs microporous surface area of
For the present NNMC catalysts, that explanation cannot NNMCs. The red squares are literature data (reanalyzed with the
present NLDFT model for the micropore surface) obtained on
hold because heat treatment was done under an inert catalysts made by impregnation of the same amount of Fe on the
atmosphere (except FC280, M786, and DAL900C). Among same non-microporous C and heat treatment at 950 °C in NH3 for
the presently investigated catalysts (except FC280 and times of 2-62 min (77).
M786), no C support is used in the synthesis and the
morphology of the catalyst depends on the carbonization- The total microporous surface area was subdivided into
graphitization process of precursor molecules. One possible three areas defined by pore sizes (i) <5 Å, (ii) 5-7.5 Å, and
explanation to account for the descending part of Figure 9A (iii) 7.5-20 Å. The recurrent feedback of plots of activity
is the catalysis of graphitization of the C precursor molecules versus different pore size intervals helped us to choose the
above meaningful subdivisions. Figure 11 shows the rela-
by iron or iron carbide at 800-1000 °C. A higher graphitiza-
tionship between the activity measured in RDE and each of
tion would most probably be detrimental to the ORR activity
these surface areas found in the catalysts. Quadrant A in
of the resulting catalysts. Iron carbide is believed to be the
Figure 11 shows two separate branches, one each for
catalyst responsible for carbon nanotube formation at tem-
catalysts made with a preexisting C support (red squares and
peratures of 700-1000 °C (121-123) or graphitization of
data point 3) and for catalysts made without a preexisting C
C at temperatures <800 °C (124). This explanation goes well
support (all black circles, except points 3 and 4). Interest-
along with the peculiar behavior of catalysts UK63 and UK65
ingly, point 4 found between the two trends corresponds to
in Figure 9A: the micropore surface area and activity of these
catalyst M786, which is made from a 50/50 wt % mix of a
catalysts is high in spite of 3 wt % Fe because the excess Fe
C support and of organic molecules. Quadrants B and C show
cannot form Fe3C because of the presence of S in these
a single branch for all data. The only data apart from the
catalysts (84), with Fe forming Fe1-xS preferentially to iron
trend are those of UK63 and UK65 (points 1 and 2) in
carbide (91).
quadrant C. Quadrant D confirms for the present NNMCs
The N content of the catalysts is now scrutinized. Figure that the mesoporous surface area (20-100 Å) does not
9B shows that the total surface content in N does not correlate with the activity.
determine the activity of the present NNMC or that of the In conclusion, analysis of the isotherms with the model
previous data (squares). A similar cloud of points is obtained from Ustinov confirms the importance of micropores for
if the N content is restricted to pyridinic or pyrrolic N atoms. NNMC synthesized in any of the ways described in the
This means that, even though N atoms are known to be Experimental Methods section. The pores hosting active sites
present in the active sites, the number of available N atoms seem to be 5-20 Å wide. This is a more refined analysis
is, in general, well in excess compared to other factors compared to size <20 Å previously proposed (77). Possibly
restricting the density of the active sites. pores <5 Å may host a few active sites too, but their
Next, the microporous surface area of the various NNMCs accessibility to O2 or H+ is very low for NNMCs synthesized
is looked at. Figure 10 shows that, despite some scattering, without a preexisting C support, explaining the trend of the
the data obtained on the present NNMC (black circles) show black circles in Figure 11A. The minimum pore size envis-
a trend of increasing activity with increased micropore agable to host FeN4 active sites is the projected length of two
surface area. By comparison with the trend observed previ- Fe-N bonds at a 45° angle, 2.77 Å [1.96 Å cos(π/4)]. If one,
ously (squares (77)), there seems, on the one hand, that a alternatively two, benzene rings (width 2.46 Å) are added
mismatch arises between the two groups of data. With the on either side of the FeN4 central moiety, the width of the
model especially developed for disordered carbonaceous active site is 7.7 and 12.6 Å, respectively. The latter corre-
materials (112-115) that does not show the artifacts en- sponds to the site structure envisaged by Figure 15 in ref
gendered by the slit pore model at pore sizes of 8-9 and 22 89.
Å (Figure 4 in ref 77), it is possible to investigate in even Last, the slope in quadrants B and C (Figure 11) is about
greater detail what exact pore sizes are important. 2 in a log-log plane, meaning that the activity is proportional

SPOTLIGHT
FIGURE 11. Mass activity in RDE vs surface area comprised of pores of specific sizes. The numbers correspond to the legend of Figure 10. The
red squares are literature data (reanalyzed with the present NLDFT model for the micropore surface) obtained on catalysts made by impregnation
of the same amount of Fe on the same non-microporous C and heat treatment at 950 °C in NH3 for times of 2-62 min (77).
to the square of the surface areas defined by pores of 5-7.5

and 7.5-20 Å, respectively. No explanation for that fact can
be proposed at this time.
IX.3. Comparison of PEMFC vs RDE Mass
Activities at 0.8 V. For the second effect (ii) introduced
in section IV, the improvement factor when changing from
the RDE system to the PEMFC system might be represented
by the ratio
[IM(PEMFC)/IM(RDE)]/[CM(PEMFC)/CM(RDE)] (VI)
where CM is the mass capacitance. The division by the factor

CM(PEMFC)/CM(RDE) is to take into account the sometimes
observed change in CM from the RDE system to the PEMFC
system. FIGURE 12. ORR mass activity normalized by mass capacitance:
effect of a switch from the RDE system (20 °C) to the PEMFC system
If the improvement factors (Figure 12) are solely due to (80 °C).
the temperature activation of each single site, then we can
estimate the values of activation energies, Ea, that would M786 of the present study, we find an improvement factor
explain such improvements. Ea of 10, 20, 40, and 50 kJ of only 2-3 in RDE when the temperature is increased from
mol-1 would result in improvement factors from 20 to 80 20 to 60 °C.
°C of 2, 4, 16, and 32, respectively. Zhang et al. reported New studies to gain knowledge on Ea of NNMCs are
an improvement of the ORR activity of a factor 2 from 20 important. Also, the optimization of the RDE ink formulation
to 70 °C (Ea ) 10 kJ mol-1) for a non-heat-treated iron might reveal itself as important as the optimization of the
porphyrin in 0.5 M H2SO4 (47). For Pt/C, Neyerlin et al. also PEMFC ink formulation in order to measure the true activity
found Ea ) 10 kJ mol-1 (Table III in ref 120). Song et al. of such catalysts.
reported an improvement of the ORR activity by a factor of IX.4. Prospectives for Making More Active
7 from 20 to 70 °C (Ea ) 35 kJ mol-1) for a non-heat-treated NNMC. Compared to Pt/C, the fundamentals for the ORR
cobalt phthalocyanine at pH 6 (49). On the other hand, non- mass activity of NNMCs obtained from the heat treatment
heat-treated iron phthalocyanines showed decreased activity of Fe (Co), N, and C are less understood. This stems (i) from
with increased temperature from 20 to 80 °C in 0.1 M H2SO4 the incompletely resolved structure of the active sites of heat-
(125). From measurements on two catalysts, like FC280 and treated NNMCs and (ii) from the incomplete knowledge of

the characteristics limiting their site density. The latter has activity is the micropore area created during pyrolysis, not
impeded the practical efforts to improve the activity of such its absolute value (129).
NNMCs. The present paper shows that micropores seem to Interestingly, disordered C can also be deliberately intro-
SPOTLIGHT
be important for any NNMCs obtained from the heat treat- duced in the micropores of a high-surface-area C in the form
ment of metal (Fe and Co), N, and C. of molecules. This is the approach that was adopted in the
This section reviews the information gathered mainly synthesis of M786. Black pearls was mixed with PTCDA in
during the last 5 years on NNMCs obtained at INRS by the a 1:1 mass ratio. In order to be sure that the PTCDA
impregnation of iron acetate on carbon black followed by molecules went into the micropores, the ball-milling tech-
heat treatment in NH3. The most recent views on what nique was used. This catalyst yields the best activity of this
factors limit the activity of “impregnation NNMC” are then study in PEMFC (Figure 3).
exerted on heat-treated NNMCs obtained from alternative
Now, our focus moves from “impregnation NNMC” to
synthesis schemes that were investigated in the present
“bulk NNMC”. We call “bulk NNMC” catalysts those that
study.
are obtained without a preexisting C support. This implies
The most accurate information on the structure of the
that the Fe and C precursors can be homogeneously
active sites has been up to now obtained from EXAFS,
mixed in contrast to “impregnation NNMC”. Thus, in bulk
which shows that Fe is coordinated by four N atoms.
NNMCs, Fe can be found at the surface but also in the
However, the remaining part of the site is important too
bulk. Bulk NNMCs could potentially lead to a high density
because many non-heat-treated Fe-N4 chelates do not
of active sites. However, one possible drawback is that a
show a high activity, such as that observed with heat-
treated NNMCs. Because the NNMC active sites are widely large proportion of the active sites may not be found at
accepted to comprise Fe (Co), N, and C coordinated the surface. Such sites will not participate in the ORR.
together in a particular way, all three may control the site Incomplete Fe utilization is evidenced for one such par-
density. Quite early, it was demonstrated for impregnated ticular catalyst by Figure 4 in ref 84. Another drawback of
NNMCs that increasing the Fe content increased the bulk NNMC is the lack of control on the organization of the
activity up to a content of 0.2-0.5 wt %, beyond which precursor molecules into a partially graphitized carbon-
the activity leveled off or decreased (63). The leveling off aceous material.
of the activity with Fe content >0.5 wt % remained To alleviate the problems inherent to the “bulk NNMC”
unexplained until recently (78). So, a piece of the puzzle approach, several actions are possible. One is to avoid large
was lacking. On the side of the C specific area, many agglomerates and low surface area by adding a foaming
different carbon powders with a wide range of specific ar- agent. Another is to ballmill bulk NNMC after pyrolysis. A
eas were tested but neither the Brunauer-Emmett-Teller third possible action is to create or to extend the porous
area or the resulting NNMCs correlated with the activity network of bulk NNMC by subjecting it to a second heat
(126). Some high-surface-area C supports resulted in treatment under a reactive atmosphere (84, 85, 88). The
poorly active catalysts. Why? A new piece of the puzzle increase in the activity due to the second heat treatment can
came into place when a connection was made between be of a decade. The exact conditions of the second heat
the amount of disordered C present in carbon black, the treatment (temperature, time, and etching gas) need to be
weight loss during pyrolysis in NH3, and the N content controlled precisely in order to connect the maximum
after pyrolysis in NH3 (76). It was understood that the number of active sites with the porous network extended
presence of disordered C in the initial C support is a during the second pyrolysis.
paramount requirement to obtain a good activity after
pyrolysis in NH3. Disordered C reacts faster with NH3 than In summary, the focus during the preparation of NNMC
graphitic C, fixes the N from NH3 in the NNMC, serves as from heat treatment should be on the control and optimiza-
a highly transformable C source to build the active sites tion of the heat-treatment conditions, not so much on the
with, and moreover, its gasification by NH3 creates mi- chemicals used to start from. It is seen in the present paper
cropores during pyrolysis. When using non-microporous that highly active NNMCs can be obtained from almost any
carbon blacks, micropore creation during pyrolysis in NH3 source of Fe, N, and C. Thus, the proper conditions of heat
seems to be a requirement to obtain high activity treatment must be found for the various raw materials. The
(77, 89, 127, 128). The maximization of the micropore time of heat treatment has a paramount importance, espe-
area requires minute control of the duration of pyrolysis, cially in a reactive atmosphere such as NH3, O2, or CO2. Also,
temperature of pyrolysis, and gas flow. In a next step, we while up to now, NNMC synthesis has consisted of usually a
started to work again on carbon blacks having a high single heat treatment followed by leaching of excess metal,
microporous area before pyrolysis but having concomi- reaching even more active NNMC will probably require
tantly a low content of disordered C. The most recent several synthesis steps that may include two or more heat
study on impregnation catalysts shows that the ORR treatments (80, 81, 83-85, 88, 130), one or two ball-milling
activity obtained with as-received microporous carbon steps (88, 131) and acid washing. These multisteps must be
blacks is lower than that obtained with as-received non- performed in an appropriate sequence in order to achieve
microporous carbon blacks because what steers the ORR the highest possible site density.

X. CONCLUSIONS (7) Meng, H.; Shen, P. K. Electrochem. Commun. 2006, 8, 588.
The mass activity for the ORR of the present NNMC is (8) Blizanac, B. B.; Ross, P. N.; Markovic, N. M. Electrochim. Acta
2007, 52, 2264.
governed mainly by the surface area of micropores having (9) Demarconnay, L.; Coutanceau, C.; Léger, J. M. Electrochim. Acta
SPOTLIGHT
a pore size of 5-20 Å (Figures 10 and 11) and might, as a 2004, 49, 4513.
(10) Rashkova, V.; Kitova, S.; Konstantinov, I.; Vitanov, T. Electro-
secondary factor, be limited by their N content (Figure 9). chim. Acta 2002, 47, 1555.
The activity is not limited by the metal content (Figure 9). (11) Matsuki, K.; Kamada, H. Electrochim. Acta 1986, 31, 13.
(12) Roche, I.; Chaı̂net, E.; Chatenet, M.; Vondrak, J. J. Phys. Chem. C
The mass activity for the ORR of the present NNMC in a 2007, 111, 1434.
RDE spans from 0.4 to 18 A g-1 at 0.8 V vs RHE and at 20 (13) Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl.
°C (Figure 1B), and that in PEMFC spans from 0.7 to 80 A Catal., B 2005, 56, 9.
(14) U.S. Department of Energy, Multi-Year Research, Development and
g-1 at 0.8 V vs RHE and at 80 °C (Figure 3). The ORR mass Demonstration plan: Planned Program Activities for 2005-2015
activity of the present NNMC is usually higher in the fuel cell (2007), http://www1.eere.energy.gov/hydrogenandfuelcells/mypp/
than in RDE, up to a multiplication factor >20 (Figure 12). pdfs/fuel_cells.pdf. Section 3.4, p 24.
(15) Mukerjee, S.; Srinivasan, S. J. Electoanal. Chem. 1993, 357, 201.
The reason for this huge difference is unresolved. Pt-based (16) Mukerjee, S.; Srinivasan, S.; Soriaga, M. P.; McBreen, J. J.
catalysts do not show such an effect. To explain this differ- Electrochem. Soc. 1995, 142, 1409.
ence with such NNMCs, new comprehensive studies are (17) Mukerjee, S.; Srinivasan, S.; Soriaga, M. P.; McBreen, J. J. Phys.
Chem. 1995, 99, 4577.
needed on (i) the effect of the temperature on the mass (18) Toda, T.; Igarashi, H.; Uchida, H.; Watanabe, M. J. Electrochem.
activity of NNMC and on (ii) the optimization of the catalyst Soc. 1999, 146, 3750.
(19) Min, M.; Cho, J.; Cho, K.; Kim, H. Electrochim. Acta 2000, 45,
ink with respect to the Nafion ionomer and catalyst ratio for 4211.
a correct measurement of the mass activity both in a RDE (20) Thompsett, D. Pt alloys as oxygen reduction catalysts. In Hand-
and a fuel cell. book of fuel cellssFundamentals, technology and applications;
Vielstich, W., Gasteiger, H. A., Lamm, A., Eds.; John Wiley and
Remaining tasks in the field of NNMCs are (i) further Sons: Chichester, U.K., 2003; Vol. 3; p 467.
increasing the activity of such catalysts, (ii) improving the (21) Ticianelli, E. A.; Derouin, C. R.; Redondo, A.; Srinivasan, S. J.
Electrochem. Soc. 1988, 135, 2209.
electrode fabrication and mass-transport properties, and (iii) (22) Wilson, M. S.; Gottesfeld, S. J. Appl. Electrochem. 1992, 22, 1.
developing a fundamental understanding of the factors (23) Wilson, M. S.; Gottesfeld, S. J. Electrochem. Soc. 1992, 139, L28.
affecting their stability or instability. The first task should be (24) Zhang, J.; Lima, F. H. B.; Shao, M. H.; Sasaki, K.; Wang, J. X.;
Hanson, J.; Adzic, R. R. J. Phys. Chem. B 2005, 109, 22701.
facilitated by the conclusions of this work and by the fact (25) Zhang, J.; Vukmirovic, M. B.; Sasaki, K.; Uribe, F.; Adzic, R. R. J.
that some catalysts in the present work were found to be Serb. Chem. Soc. 2005, 70, 513.
(26) Vukmirovic, M. B.; Zhang, J.; Sasaki, K.; Nilekar, A. U.; Uribe,
very active. The third task, not dealt with in this work, is now
F.; Mavrikakis, M.; Adzic, R. R. Electrochim. Acta 2007, 52, 2257.
becoming the most important as the activity reaches accept- (27) Larminie, J.; Dicks, A. L. Fuel Cell Systems Explained, 2nd ed.;
able values. John Wiley and Sons Inc.: Hoboken, NJ, 2003.
(28) Gasteiger, H. A.; Panels, J. E.; Yan, S. G. J. Power Sources 2004,
127, 162.
Acknowledgment. INRS acknowledges the support of (29) Dodelet, J. P. Oxygen reduction in PEM fuel cell conditions: heat-
NSERC and General Motors of Canada. E.A.U. acknowledges treated non-precious metal-N4 macrocyles and beyond. In N4-
the Russian Foundation for Basic Research (Project 06-03- Macrocyclic Metal Complexes; Zagal, J. H., Bedioui, F., Dodelet,
J. P., Eds.; Springer Science + Business Media Inc.: New York,
32268-a). 2006; p 83.
(30) Organisation internationale des constructeurs automobiles, http://
Supporting Information Available: Calculation of the oica.net/category/production-statistics, 2007.
volumetric activity for a Pt/C catalyst, full description of the (31) Platinum 2008; Johnson Matthey: Hertfordshire, England, 2008;
catalyst’s synthesis, summary of the RDE results with other p 60; http://www.platinum.matthey.com/publications/pgmre-
view.html.
ink recipes, definition of the target mass activity for NNMCs (32) http://www.platinum.matthey.com, 2008.
in RDE, discussion on the nature of the redox peaks seen (33) Cawthorn, R. G. South. Afr. J. Sci. 1999, 95, 481.
in the N2-CV, and SEM pictures of all NNMCs. This ma- (34) Jaffray, C.; Hards, G. Precious metal supply requirements. In
Handbook of fuel cellssFundamentals, technology and applica-
terial is available free of charge via the Internet at http://
tions; Vielstich, W., Gasteiger, H. A., Lamm, A., Eds.; John Wiley
pubs.acs.org. and Sons: Chichester, U.K., 2003; Vol. 3; p 509.
(35) Pattabi, M.; Castellanos, R. H.; Castillo, R.; Ocampo, A. L.;
REFERENCES AND NOTES Moreira, J.; Sebastian, P. J.; McClure, J. C.; Mathew, X. Int. J.
Hydrogen Energy 2001, 26, 171.
(1) Banerjee, S.; Curtin, D. E. J. Fluorine Chem. 2004, 125, 1211.
(36) Susac, D.; Sode, A.; Zhu, L.; Wong, P. C.; Teo, M.; Bizzotto, D.;
(2) Markovic, N. The hydrogen electrode reaction and the elec-
Mitchell, K. A. R.; Parsons, R. R.; Campbell, S. A. J. Phys. Chem.
trooxidation of CO and H2/CO mixtures on well-characterized B 2006, 110, 10762.
Pt and Pt-bimetallic surfaces. In Handbook of fuel cellss (37) Feng, Y.; He, T.; Alonso-Vante, N. Chem. Mater. 2008, 20, 26.
Fundamentals,technology andapplications;Vielstich,W.,Gasteiger, (38) Lee, K.; Zhang, L.; Zhang, J. Electrochem. Commun. 2007, 9,
H. A., Lamm, A., Eds.; John Wiley and Sons: Chichester, U.K., 1704.
2003; Vol. 2; p 368. (39) Zhong, H.; Zhang, H.; Liu, G.; Liang, Y.; Hu, J.; Yi, B. Electrochem.
(3) Jasinski, R. Nature 1964, 201, 1212. Commun. 2006, 8, 707.
(4) Markovic, N. M.; Gasteiger, H. A.; Ross, P. N. J. Phys. Chem. 1996, (40) Zhong, H.; Zhang, H.; Liang, Y.; Zhang, J.; Wang, M.; Wang, X.
100, 6715. J. Power Sources 2007, 164, 572.
(5) Yang, Y. F.; Zhou, Y. H.; Cha, C. S. Electrochim. Acta 1995, 40, (41) Liu, G.; Zhang, H. M.; Wang, M. R.; Zhong, H. X.; Chen, J. J. Power
2579. Sources 2007, 172, 503.
(6) Lee, H. K.; Shim, J. P.; Shim, M. J.; Kim, S. W.; Lee, J. S. Mater. (42) Ohnishi, R.; Takahashi, Y.; Takagaki, A.; Kubota, J.; Domen, K.
Chem. Phys. 1996, 45, 238. Chem. Lett. 2008, 37, 838.

(43) Jahnke, H.; Schönborn, M.; Zimmermann, G. Top. Curr. Chem. (81) Maruyama, J.; Okamura, J.; Miyazaki, K.; Abe, I. J. Phys. Chem.
1976, 61, 133. C 2007, 111, 6597.
(44) Boulatov, R. Billion-year-old oxygen cathode that actually works: (82) Wang, P.; Ma, Z.; Zhao, Z.; Jia, L. J. Electroanal. Chem. 2007, 611,
Respiratory oxygen reduction and its biomimetic analogs. In N4- 87.
SPOTLIGHT
Macrocyclic Metal Complexes; Zagal, H., Bedioui, F., Dodelet, (83) Maruyama, J.; Okamura, J.; Miyazaki, K.; Uchimoto, Y.; Abe, I.
J. P., Eds.; Springer Science + Business Media: New York, 2006; J. Phys. Chem. C 2008, 112, 2784.
p 1. (84) Koslowski, U. I.; Abs-Wurmbach, I.; Fiechter, S.; Bogdanoff, P.
(45) Song, E.; Shi, C.; Anson, F. C. Langmuir 1998, 14, 4315. J. Phys. Chem. C 2008, 112, 15356.
(46) Shi, C.; Anson, F. C. Inorg. Chem. 1998, 37, 4294. (85) Koslowski, U. I.; Herrmann, I.; Bogdanoff, P.; Barkschat, C.;
(47) Zhang, L.; Song, C.; Zhang, J.; Wang, H.; Wilkinson, D. P. J. Fiechter, S.; Iwata, N.; Takahashi, H.; Nishikoro, H. Electrochem.
Electrochem. Soc. 2005, A2421. Soc. Trans. 2008, 13, 125.
(48) Baranton, S.; Coutanceau, C.; Roux, C.; Hahn, F.; Leger, J. M. J. (86) Sirk, A. H. C.; Campbell, S. A.; Birss, V. I. J. Electrochem. Soc.
Electroanal. Chem. 2005, 577, 223. 2008, 155, B592.
(49) Song, C.; Zhang, L.; Zhang, J.; Wilkinson, D. P.; Baker, R. Fuel (87) Garsuch, A.; d’Eon, R.; Dahn, T.; Klepel, O.; Garsuch, R. R.;
Cells 2007, 7, 9. Dahn, J. R. J. Electrochem. Soc. 2008, 155, B236.
(50) Bagotzky, V. S.; Tarasevich, M. R.; Radyushkina, K. A.; Levina, (88) Wood, T. E.; Tan, Z.; Schmoekel, A. K.; O’Neill, D.; Atanasoski,
O. E.; Andrusyova, S. I. J. Power Sources 1977, 2, 233. R. J. Power Sources 2008, 178, 510.
(51) van Veen, J. A. R.; van Baar, J. F.; Kroese, K. J. J. Chem. Soc., (89) Charreteur, F.; Jaouen, F.; Ruggeri, S.; Dodelet, J. P. Electrochim.
Faraday Trans. 1981, 1, 2827. Acta 2008, 53, 2925.
(52) Scherson, D. A.; Gupta, S. L.; Fierro, C.; Yeager, E. B.; Kordesch, (90) Charreteur, F.; Ruggeri, S.; Jaouen, F.; Dodelet, J. P. Electrochim.
M. E.; Eldridge, J.; Hoffman, R. W.; Blue, J. Electrochim. Acta Acta 2008, 53, 6881.
1983, 28, 1205. (91) Herrmann, I.; Koslowski, U. I.; Radnik, J.; Fiechter, S.; Bogdanoff,
(53) Scherson, D.; Tanaka, A.; Gupta, S. L.; Tryk, D.; Fierro, C.; Holze, P. Electrochem. Soc. Trans. 2008, 13, 143.
R.; Yeager, E. B. Electrochim. Acta 1986, 31, 1247. (92) Olson, T. S.; Chapman, K.; Atanassov, P. J. Power Sources 2008,
(54) Van der Putten, A.; Elzing, A.; Visscher, W.; Barendrecht, E. J. 183, 557.
Electroanal. Chem. 1986, 205, 233. (93) Pylypenko, S.; Mukerjee, S.; Olson, T. S.; Atanassov, P. Electro-
(55) Wiesener, K. Electrochim. Acta 1986, 31, 1073. chim. Acta 2008, 53, 7875.
(56) Tarasevich, M. R.; Radyushkina, K. A. Mater. Chem. Phys. 1989, (94) Artyushkova, K.; Levendosky, S.; Atanassov, P.; Fulghum, J. Top.
22, 477. Catal. 2007, 46, 263.
(57) Franke, R.; Ohms, D.; Wiesener, K. J. Electroanal. Chem. 1989, (95) Artyushkova, K.; Pylypemko, S.; Olson, T. S.; Fulghum, J. E.;
260, 63. Atanassov, P. Langmuir 2008, 24, 9082.
(58) Yeager, E. B. Electrochim. Acta 1984, 29, 1527. (96) Ziegelbauer, J. M.; Olson, T. S.; Pylypenko, S.; Alamgir, F.; Jaye,
(59) van Veen, J. A. R.; Colijn, H. A.; van Baar, J. F. Electrochim. Acta C.; Atanassov, P.; Mukerjee, S. J. Phys. Chem. C 2008, 112, 8839.
(97) Garsuch, A.; McIntyre, K.; Michaud, X.; Stevens, D. A.; Dahn,
1988, 33, 801.
J. R. J. Electrochem. Soc. 2008, 155, B953.
(60) Wiesener, K.; Ohms, D.; Neumann, V.; Franke, R. Mater. Chem.
(98) Garsuch, A.; Sattler, R. R.; Witt, S.; Klepel, O. Microporous
Phys. 1989, 22, 457.
Mesoporous Mater. 2006, 89, 164.
(61) Gupta, S. L.; Tryk, D.; Bae, I.; Aldred, W.; Yeager, E. B. J. Appl.
(99) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals
Electrochem. 1989, 19, 19.
and Applications; John Wiley & Sons: New York, 1980.
(62) Côté, R.; Lalande, G.; Guay, D.; Dodelet, J. P.; Denes, G. J.
(100) Jimenez-Mateos, J. M.; Fierro, J. L. G. Surf. Interface Anal. 1996,
Electrochem. Soc. 1998, 145, 2411.
24, 223.
(63) Wang, H.; Côté, R.; Faubert, G.; Guay, D.; Dodelet, J. P. J. Phys.
(101) Casanovas, J.; Ricart, J. M.; Rubio, J.; Illas, F.; Jimenez-Mateos,
Chem. B 1999, 103, 2042.
J. M. J. Am. Chem. Soc. 1996, 118, 8071.
(64) Biloul, A.; Coowar, F.; Contamin, O.; Scarbeck, G.; Savy, M.; van (102) Faubert, G.; Côté, R.; Dodelet, J. P.; Lefèvre, M.; Bertrand, P.
den Ham, D.; Riga, J.; Verbist, J. J. J. Electroanal. Chem. 1990, Electrochim. Acta 1999, 44, 2589.
289, 189. (103) Karweik, D. H.; Winograd, N. Inorg. Chem. 1976, 15, 2336.
(65) Fabjan, C.; Frithum, G.; Hartl, H. Ber. Bunsenges. Phys. Chem. (104) Roe, S. P.; Hill, J. O.; Magee, R. J. Aust. J. Chem. 1986, 39, 1377.
1990, 94, 937. (105) Tarazona, P. Phys. Rev. A 1985, 31, 2672.
(66) Ohms, D.; Herzog, S.; Franke, R.; Neumann, V.; Wiesener, K.; (106) Tarazona, P.; Marconi, U. M. B.; Evans, R. Mol. Phys. 1987, 60,
Gamburcev, S.; Kaisheva, A.; Iliev, I. J. Power Sources 1992, 38, 573.
327. (107) Seaton, N. A.; Walton, J. P. R. B.; Quirke, N. Carbon 1989, 27,
(67) Bouwkamp-Wijnoltz, A. L.; Visscher, W.; van Veen, J. A. R.; Tang, 853.
S. C. Electrochim. Acta 1999, 45, 379. (108) Lastoskie, C.; Gubbins, K. T.; Quirke, N. J. Phys. Chem. 1993, 97,
(68) Gojkovic, S. L.; Gupta, S.; Savinell, R. F. J. Electroanal. Chem. 4786.
1999, 462, 63. (109) Ravikovitch, P. I.; Neimark, A. V. Stud. Surf. Sci. Catal. 2000, 129,
(69) Contamin, O.; Debiemme-chouvy, C.; Savy, M.; Scarbeck, G. J. 597.
New Mater. Electrochem. Syst. 2000, 3, 67. (110) Tikhonov, A. N.; Arsenin, V. Y. Solutions of ill-posed problems;
(70) Sun, G. Q.; Wang, J. T.; Gupta, S.; Savinell, R. F. J. Appl. Wiley: New York, 1977.
Electrochem. 2001, 31, 1025. (111) Gardner, L.; Kruk, M.; Jaroniec, M. J. Phys. Chem. B 2001, 105,
(71) Bogdanoff, P.; Herrmann, I.; Hilgendorff, M.; Dorbandt, I.; 12516.
Fiechter, S.; Tributsch, H. J. New Mater. Electrochem. Syst. 2004, (112) Ustinov, E. A.; Do, D. D.; Fenelonov, V. B. Carbon 2006, 44, 653.
7, 85. (113) Ustinov, E. A.; Do, D. D.; Jaroniec, M. Appl. Surf. Sci. 2005, 252,
(72) Ye, S.; Vijh, A. K. J. Solid State Electrochem. 2005, 9, 146. 548.
(73) Herrmann, I.; Bogdanoff, P.; Schmithals, G.; Fiechter, S. Elec- (114) Ustinov, E. A.; Do, D. D.; Fenelonov, V. B. Appl. Surf. Sci. 2007,
trochem. Soc. Trans. 2006, 3, 211. 253, 5610.
(74) Medard, C.; Lefèvre, M.; Jaouen, F.; Dodelet, J. P.; Lindbergh, (115) Ustinov, E. A.; Kukushkina, J.; Sokolov, V.; Kravchik, A.; Ter-
G. Electrochim. Acta 2006, 51, 3202. eshchenko, G.; Melgunov, M.; Eletski, P.; Yakovlev, V.; Fenelonov,
(75) Subramanian, N. P.; Kumaragaru, S. P.; Colon-Mercado, H.; Kim, V. A refined method of porous structure analysis of activated
H.; Popov, B. N.; Black, T.; Chen, D. A. J. Power Sources 2006, carbons with N2 adsorption isotherms. International Conference
157, 56. on Carbon (CARBON′08), Nagano, Japan, 2008.
(76) Jaouen, F.; Charreteur, F.; Dodelet, J. P. J. Electrochem. Soc. 2006, (116) Bonakdarpour, A.; Lefevre, M.; Yang, R.; Jaouen, F.; Dahn, T.;
153, A689. Dodelet, J. P.; Dahn, J. R. Electrochem. Solid-State Lett. 2008, 11,
(77) Jaouen, F.; Lefèvre, M.; Dodelet, J. P.; Cai, M. J. Phys. Chem. B B105.
2006, 110, 5553. (117) Bonakdarpour, A.; Delacote, C.; Yang, R.; Wieckowski, A.; Dahn,
(78) Jaouen, F.; Dodelet, J. P. Electrochim. Acta 2007, 52, 5975. J. R. Electrochem. Commun. 2008, 10, 611.
(79) Maruyama, J.; Abe, I. Chem. Commun. 2007, 27, 2879. (118) Lowell, S. Introduction to powder surface area; John Wiley & Sons:
(80) Maruyama, J.; Abe, I. J. Electrochem. Soc. 2007, 154, B297. New York, 1979.

(119) Herrmann, I. In Innovative Elektrokatalyse: Platinfreie Kathoden- (126) Jaouen, F.; Marcotte, S.; Dodelet, J. P.; Lindbergh, G. J. Phys.
katalysatoren für brennstoffzellen; Vdm Verlag: Saarbrücken, Chem. B 2003, 107, 1376.
2008; p 153. (127) Jaouen, F.; Dodelet, J. P. J. Phys. Chem. C 2007, 111, 5963.
(120) Neyerlin, K. C.; Gu, W.; Jorne, J.; Gasteiger, H. A. J. Electrochem.
SPOTLIGHT
(128) Jaouen, F.; Serventi, A. M.; Lefevre, M.; Dodelet, J. P.; Bertrand,
Soc. 2006, 153, A1955. P. J. Phys. Chem. C 2007, 111, 5971.
(121) Hou, H.; Schaper, A. K.; Weller, F.; Greiner, A. Chem. Mater.
(129) Lefevre, M.; Dodelet, J. P. Electrochim. Acta 2008, 53, 8269.
2002, 14, 3990.
(122) Perez-Cabero, M.; Rodriguez-Ramos, I.; Guerrero-Ruiz, A. J. (130) Herranz, J.; Lefevre, M.; Larouche, N.; Stansfield, B.; Dodelet,
Catal. 2003, 215, 305. J. P. J. Phys. Chem. C 2007, 111, 19033.
(123) Schaper, A. K.; Hou, H.; Greiner, A.; Phillipp, F. J. Catal. 2004, (131) Proietti, E.; Ruggeri, S.; Dodelet, J. P. J. Electrochem. Soc. 2008,
222, 250. 155, B340.
(124) Sinclair, R.; Itoh, T.; Chin, R. Microsc. Microanal. 2002, 8, 288.
(125) Baker, R.; Wilkinson, D. P.; Zhang, J. Electrochim. Acta 2008,
53, 6906. AM900219G

Cross Laboratory Experimental Study of N

Uploaded by

Copyright:

Available Formats

Cross Laboratory Experimental Study of N

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cross Laboratory Experimental Study of N

Uploaded by

Copyright:

Available Formats

Cross-Laboratory Experimental Study of

Non-Noble-Metal Electrocatalysts for the

I. INTRODUCTION tion is fast (2), the oxygen reduction reaction (ORR) is

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1623

1624 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1625

1626 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1627

FIGURE 2. (A) N2 cyclic voltammetry in RDE. N2-saturated electrolyte

1628 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

activity is one-tenth of that measured on Pt/C in the same

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1629

higher in PEMFC versus that in RDE (M786) or even smaller

1630 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

FIGURE 7. N 1s narrow scan spectra of the NNMCs and their deconvolution.

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1631

catalysts. One interesting point is that the surface Fe is

VI. NEUTRON ACTIVATION RESULTS

1632 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1633

found at the surface can be an active site for the ORR.

1634 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

to the square of the surface areas defined by pores of 5-7.5

where CM is the mass capacitance. The division by the factor

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1635

1636 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1637

1638 VOL. 1 • NO. 8 • 1623–1639 • 2009 Jaouen et al. www.acsami.org

www.acsami.org VOL. 1 • NO. 8 • 1623–1639 • 2009 1639

You might also like