Hydrogen peroxide formation rates in a PEMFC anode and cathode

Article abstract of DOI:10.1149/1.2801980

Hydrogen peroxide (H2 O2) formation rates in a proton exchange membrane fuel cell (PEMFC) anode and cathode were estimated as a function of humidity and temperature by studying the oxygen reduction reaction (ORR) on a rotating ring disk electrode. Fuel cell conditions were replicated by depositing a film of Pt/Vulcan XC-72 catalyst onto the disk and by varying the temperature, dissolved O2 concentration, and the acidity levels in hydrochloric acid (HCl O4). The HCl O4 acidity was correlated to ionomer water activity and hence fuel cell humidity. The H2 O2 formation rates showed a linear dependence on oxygen concentration and square dependence on water activity. The H2 O2 selectivity in ORR was independent of oxygen concentration but increased with the decrease in water activity (i.e., decreased humidity). Potential dependent activation energy for the H2 O2 formation reaction was estimated from data obtained at different temperatures.

Full text of DOI:10.1149/1.2801980

B50
Journal of The Electrochemical Society, 155 ͑1͒ B50-B57 ͑2008͒
0013-4651/2007/155͑1͒/B50/8/$23.00 © The Electrochemical Society
Hydrogen Peroxide Formation Rates in a PEMFC Anode
and Cathode
Effect of Humidity and Temperature
Vijay A. Sethuraman,a,c, John W. Weidner,
*
a,**,z
b,d,**
Andrew T. Haug,
b,**
Sathya Motupally,b and Lesia V. Protsailo
aCenter for Electrochemical Engineering, Department of Chemical Engineering,
University of South Carolina, Columbia, South Carolina 29208, USA
bUTC Power, South Windsor, Connecticut 06074, USA
Hydrogen peroxide ͑H2O2͒ formation rates in a proton exchange membrane fuel cell ͑PEMFC͒ anode and cathode were estimated
as a function of humidity and temperature by studying the oxygen reduction reaction ͑ORR͒ on a rotating ring disk electrode. Fuel
cell conditions were replicated by depositing a film of Pt/Vulcan XC-72 catalyst onto the disk and by varying the temperature,
dissolved O2 concentration, and the acidity levels in hydrochloric acid ͑HClO4͒. The HClO4 acidity was correlated to ionomer
water activity and hence fuel cell humidity. The H2O2 formation rates showed a linear dependence on oxygen concentration and
square dependence on water activity. The H2O2 selectivity in ORR was independent of oxygen concentration but increased with
the decrease in water activity ͑i.e., decreased humidity͒. Potential dependent activation energy for the H2O2 formation reaction was
estimated from data obtained at different temperatures.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2801980͔ All rights reserved.
Manuscript submitted July 9, 2007; revised manuscript received September 26, 2007. Available electronically November 8, 2007.
Proton exchange membrane fuel cell ͑PEMFC͒ technology, ow-
These radicals react with the perfluorosulfonic acid type ionomer
in the electrode and the membrane to produce hydrofluoric acid
͑HF͒.14 The sequence is listed below
ing to its high efficiency, operational flexibility, and superior modu-
larity, has the capability to be the structural and fundamental unit of
an impending hydrogen economy. Two main issues that impede its
progress toward commercialization are cost and durability. The U.S.
Department of Energy’s ͑DOE͒ projected performance require-
ments1 for the year 2010 are 5000 h ͑with 20,000 start/stops͒ at
$45/kW for automotive stacks and upwards of 40,000 h at
$400–$750/kW for stationary power plants. In addition, current en-
gineering requirements demand stack operation at higher tempera-
tures ͑Ͼ100°C͒ and low relative humidities ͑Ͻ75% RH͒. Elevated
temperature operation offers better tolerance to CO, faster oxygen
reduction reaction ͑ORR͒ kinetics, and better water and thermal
management enabling easier system integration. However, elevated
temperatures and the desire to operate at ambient pressures means
the fuel cell needs to be operated at lower relative humidities. Since
much of ionomer and membrane technologies have evolved around
the water dependent perfluorinated systems such as Nafion, both
high temperature and low humidity conditions cause severe perfor-
mance degradation2-8 and remain an impediment toward achieving
DOE’s performance and durability targets.
Rp − CF2COOH + OH· → Rp − CF2 + CO2 + H2O
2Rp − CF2 + 2OH· → Rp − CF2OH + Rp − COF + HF ͓5͔
Rp − COF + H2O → Rp − COOH + HF ͓6͔
͓4͔
The fluoride emission rate is a measure of membrane degradation
given in Eq. 5 and 6. Since two-thirds of Nafion is fluorine ͑on a
mass basis͒, this chemical degradation results in mechanical insta-
bility in the membranes causing pinholes and eventual failure. Since
this degradation is initiated by the peroxide-radical attack, under-
standing H2O2 kinetics at the electrocatalyst/ionomer interface at
low humidities and elevated temperatures from a PEM fuel cell
context is vital toward explaining the increased degradation rate
observed under such conditions. Though Liu and Zuckerbrod15 have
reported a method for in situ detection of H2O2 formation, it only
served as a qualitative indicator of the existence of peroxide. In situ
quantification of peroxide kinetics is very difficult owing to its in-
stability.
Rotating ring disk electrode ͑RRDE͒ studies on supported Pt
catalysts have been successfully used16-21 as a technique to quantify
peroxide formation and for screening oxygen reduction catalysts.
Paulus et al.16,18 reported the use of a thin-film RRDE method for
characterizing oxygen reduction reaction ͑ORR͒ in supported high
surface area catalysts and were able to quantify the amount of H2O2
produced during the ORR. They decreased the film thickness and
improved the ionomer-catalyst film stability at higher rotation
speeds, which resulted in uniform collection efficiencies and better
peroxide measurements. Antoine and Durand19 reported a weak
platinum particle size effect on H2O2 production during ORR and
agreed with previously reported observations that H2O2 yields were
higher for potentials less than 0.4 V vs the standard hydrogen elec-
trode ͑SHE͒. Enayetullah et al.20 studied ORR electrocatalysis on
polycrystalline Pt microelectrodes in various concentrations of trif-
luoromethane sulfonic acid ͑TFMSA͒. They reported higher Tafel
slopes and lower activation energies for ORR in higher concentra-
tions of TFMSA, which was attributed to lower water activity.
Murthi et al.21 studied ORR in supported Pt and Pt alloy catalysts in
1 M and 6 M TFMSA as a way to study the effect of water activity.
They reported higher peroxide yields in 6 M TFMSA solution com-
pared to a 1 M solution. However, none of these studies correlate
One of the mechanisms for catalyst/ionomer chemical degrada-
tion in PEMFCs involves the formation of hydroxyl and hydroper-
oxyl ͑OH· and OOH·͒ radicals9,10 caused by hydrogen peroxide
͑H2O2͒ formation on the catalyst surface via Reaction 1
O2 + 2H+ + 2e− ↔ H2O2 E0 = 0.695 V vs SHE
͓1͔
and subsequent decomposition via Reactions 2 and 3. Using a novel
in situ spin trap electron paramagnetic resonance study,11 Panchenko
et al.12 reported no evidence of OH· and OOH· radicals in the an-
ode. They observed the presence of radicals in the cathode and near
the membrane-cathode interface. Therefore, the H2O2 diffuses into
the membrane and chemically breaks down to hydroxyl radicals and
ions on metal ions present13 in the membrane
H2O2 + M2+ → M3+ + OH· + OH−
OH· + H2O2 → OOH· + H2O
͓2͔
͓3͔
*
Electrochemical Society Student Member.
**
Electrochemical Society Active Member.
c Present address: Environmental Energy Technology Division, Lawrence Berkeley
National Laboratory, Berkeley, CA 94720, USA.
d Present address: 3M Fuel Cell Components, 3M Center, St. Paul, MN 55144.
z E-mail: weidner@engr.sc.edu
Downloaded on 2015-06-08 to IP 128.122.253.228 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 B50-B57 ͑2008͒
B51
the measured peroxide yields and selectivity to peroxide formation
rates in a PEM fuel cell as a function of cell operating conditions.
Therefore, the objective of this investigation was to predict H2O2
formation rates in a PEM fuel cell. This was accomplished by mea-
suring H2O2 formation rates at rotating ring disk electrode ͑RRDE͒.
Fuel cell conditions were replicated by depositing a film of Pt/
Vulcan XC-72 catalyst onto the disk and by varying the temperature,
dissolved O2 concentration, and the acidity levels in HClO4. The
HClO4 acidity was correlated to ionomer water activity and hence
fuel cell humidity. Peroxide formation rate in the anode was pre-
dicted using oxygen permeability measured across Nafion 112 mem-
brane and the fraction of oxygen that reduces the H2O2. Peroxide
formation rate in the cathode was predicted using the rate constants
measured from RRDE experiments and the local concentrations of
oxygen and protons.
NC͒ was scanned between 0 and 1.2 V vs SHE to characterize H2O2
formation within the potential range relevant to fuel cell operating
conditions. Potentials were determined using a mercury-mercurous
sulfate ͑Hg/Hg2SO4͒ reference electrode. All potentials in this
study, however, refer to that of the standard hydrogen electrode
͑SHE͒. A high-surface area Pt cylindrical mesh ͑5 mm diameter,
50 mm length͒ attached to a Pt wire ͑0.5 mm thick, 5 mm length͒
was used as the counter electrode.
Effect of oxygen concentration.— The effect of oxygen concen-
tration on ORR and H2O2 formation kinetics was studied by varying
the concentration of oxygen in the solution. The following three
gases were used: oxygen ͑UHP grade, Praxair͒, air ͑Industrial,
Praxair͒ and 10.01% oxygen in nitrogen ͑Airgas͒. A gas flow meter
͑0–500 mL, Dwyer Instruments Inc., Michigan City, IN͒ was used
to control the flow of the gas feed at ϳ100 mL min−1 into the elec-
trolyte. The electrochemical cell was sealed during the experiments
to keep air from affecting the concentration of dissolved oxygen in
the electrolyte. The concentration of dissolved oxygen in the elec-
trolyte was estimated using the solubility values for oxygen in pure
liquid water at 25°C and 101 kPa.26
Experimental
Rotating ring disk electrode (RRDE).— For the RRDE studies,
commercially available Pt/Vulcan catalyst ͑20% Pt on Vulcan XC-
72R carbon, Johnson Matthey Inc., West Chester, PA͒ was used.
Catalyst coated glassy carbon electrodes were prepared as described
by Schmidt et al.22 Aqueous suspensions of 1 mg catalyst mL−1
were obtained by pulse-sonicating 20 mg Pt/Vulcan catalyst with
20 mL triple-distilled, ultrapure water ͑Millipore Corp.͒ in an ice
bath ͑70% duty cycle, 60 W, 15 min͒. Sonication was done using a
Braun-Sonic U type 853973/1 sonicator. A glassy carbon disk served
as the substrate for the supported catalyst and was polished to a
mirror finish ͑0.05 ␮m deagglomerated alumina, Buehler͒ prior to
catalyst coating. An aliquot of 14 ␮L catalyst suspension was pipet-
ted onto the carbon substrate, which corresponded to a Pt loading of
ϳ14.1 ␮g Pt cm−2. After evaporation of water for 30 min in N2
atmosphere ͑381 mm Hg, vacuum͒, 14 ␮L of diluted Nafion solu-
tion ͑5% aqueous solution, 1100 EW; Solution Technology Inc.,
Mendenhall, PA͒ was pipetted on the electrode surface and further
evaporated for 30 min in N2 atmosphere ͑381 mm Hg, vacuum͒.
Nafion was used to adhere the Pt/Vulcan particles onto the glassy
carbon electrode ͑the ratio of H2O/Nafion solution used was
ϳ100/1͒. Previous work by Paulus et al. indicates that this proce-
dure yielded a Nafion film thickness of ϳ0.1 ␮m and that the utili-
zation of the Pt/Vulcan catalyst ͑based on H-adsorption charge͒ on
the electrode with this film was ϳ100%.
The catalyst-Nafion coated electrode was immersed in deaerated
͓ultrahigh purity ͑UHP͒ nitrogen, Praxair͔ perchloric acid ͑HClO4,
70% ULTREX II ultrapure reagent grade, J. T. Baker͒ of varying
concentrations for further synchronized chronoamperometric and
potentiodynamic experiments. Though a variety of supporting elec-
trolytes are reported in the literature, anion adsorption on Pt is mini-
mal for only a few electrolytes.23 ͓e.g., trifluoromethane sulfonic
acid ͑TFMSA͒ and HClO4͔. In addition, the ultrapure reagent grade
HClO4 used in this study is free of ionic impurities; especially since
Cl− ions, even in trace amounts ͑i.e., 1 ppm͒, are shown to drasti-
cally change both the activity and the reaction pathway of ORR on
Pt catalysts.23-25 All RRDE experiments were performed at atmo-
spheric pressure and all solutions were prepared from ultrapure wa-
ter ͑Millipore Inc., 18.2 M⍀ cm͒.
Effect of pH.— The effect of proton concentration on ORR and
H2O2 formation kinetics was studied by varying the acidity of
HClO4 in the 2.0–0.1 M concentration window ͑ϳ−0.301–1 pH,
assuming Ka ӷ 1 for HClO4͒. Between solution changes, the elec-
trochemical cell and its components were washed and boiled in
deionized water for 5 h to ensure accurate pH levels. The catalyst-
Nafion coated electrode was also cleaned in a sonicator before every
experiment with triple distilled ultrapure water.
Collection efficiency.— Standard procedure27 for the determina-
tion of collection efficiency of a ring-disk electrode was followed.
The electrodes were prepared as described above. The experiment
was carried out in an electrochemical cell in deaerated ͑UHP nitro-
gen, Praxair͒ 0.1 M H2SO4 ͑96.5%, J. T. Baker͒ with 10 mmol L−1
K3Fe͑CN͒ ͑99.7%, J. T. Baker͒. The disk electrode was swept at
6
1 mV s−1 ͓vs SHE͔ while the Pt ring was held at a constant potential
of 1.2 V ͓vs SHE͔. At this ring potential, the oxidation of
͓Fe͑CN͒ ͔4−, produced at the disk electrode, to ͓Fe͑CN͒ ͔3−, pro-
6
6
ceeds under pure diffusion control. The collection efficiency was
determined as N = Iring/Idisk = 0.20, which was independent of disk
potential and consistent with the theoretical collection efficiency
provided by the manufacturer of the ring-disk electrode.28
The two-electron transfer reaction of O2 reduction to H2O2, cap-
tured by the Pt ring, was analyzed in this work ͑see Table I for
parameters used in the analysis͒. At the ring, the H2O2 produced at
the disk is oxidized via the reverse of Reaction 1. The fraction of
H2O2 formation, ␹
, can be determined from the collection effi-
H O
2
2
ciency, ring and disk currents by the expression
2Iring/N
␹
=
2
͓7͔
H O
2
Idisk + Iring/N
The measured current density j corresponding to H2O2 formation
on a film covered RDE for the first-order ORR kinetics was previ-
ously reported to take the following expression,29 in terms of kinetic
and mass-transport dependent currents
The electrochemical measurements were conducted in a standard
electrochemical cell ͑RDE Cell, Pine Instrument Company, Ray-
leigh, NC͒ immersed in a custom-made jacketed vessel, the tem-
perature of which was controlled by a refrigerated/heating circulator
͑Julabo Labortechnik GMBH͒. A ring-disk electrode setup with a
bi-potentiostat ͑Bi-Stat, Princeton Applied Research Inc., Oak
Ridge, TN͒ was used in conjunction with rotation-control equipment
͑Pine Instrument Company, Rayleigh, NC͒. EC-Lab software ͑ver-
sion 8.60, Bio-logic Science Instruments, Claix, France͒ was used to
control the bi-potentiostat. The Pt ring electrode was held at 1.2 V
vs SHE where the oxidation of peroxide is diffusion limited. The
catalyst coated glassy carbon disk electrode ͑5 mm diameter,
0.1966 cm2 area, DT21 Series, Pine Instrument Company, Rayleigh,
1
1
␦
1
f
=
+
+
͓8͔
nFDfO CfO
j
jkin
jD
2
2
where j is
Iring
j =
͓9͔
NA
jkin is the current density in the absence of mass transfer effects and
jD is the diffusion current given by the Levich equation
jD = 0.62nFD*O2/3C*O
␯
2
␻
͓10͔
−1/6 1/2
2
Downloaded on 2015-06-08 to IP 128.122.253.228 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B52
Journal of The Electrochemical Society, 155 ͑1͒ B50-B57 ͑2008͒
Table I. Parameters used in the analysis of measured current at
the Pt ringa.
Parameter
Value
Comments
a
A
b
1
0.164025 cm2
2
Measured
Ref. 28
Measured
Ref. 26a
Ref. 48
1.274 mol cm−3
2.2 ϫ 10−6 cm2 s−1
0.695 V vs SHE
1100 g mol−1
96,485 C mol equiv.−1
0.2
CO*
2
DO*
2
E0
EW
F
Ref. 49
Ref. 50
Measured
Ref. 50
Measured
Assumed
Ref. 16
Ref. 50
Estimated
Measured
N
R
8.314 J mol−1 K−1
298 K
T0
␣
0.5
␦
10−5 cm
f
␳
␷
␻
1 g cm−3
0.009 cm2 s−1
2500 s−1
a The mole fraction solubility Xl of oxygen in water is given as ln Xl
= A2 + ͑B2/T*͒ + C2 lnT*, where T* = ͑T/100͒K. A2 = −66.7354,
B2 = 87.4755 and C2 = 24.4526.
The concentration of O2 in the solution was calculated from the
partial pressure of O2 in the inlet gas and O2 solubility data for pure
liquid water at corresponding temperature and 101 kPa.26 The dif-
ference in O2 solubility in pure liquid water and in HClO4 ͑up to
2 M͒ was assumed to be negligible. Combining Eq. 8 and 10 and
solving for jkin gives
jnFDfO COf DO*2/3CO*
␻
1/2
2
2
2
2
jkin
=
f
nFDfO COf DO*2/3CO*
␻
− ␦fjD*O2/3C*O
␻
− 1.6␯1/6jDO CfO
1/2
1/2
2
2
2
2
2
2
2
2
͓11͔
The purely kinetic portion of the H2O2 formation rate is
jkin
a
b
RH O
=
= kf͑CO ͒ ͑CH
+
͒
͓12͔
2
2
2
2F
where
␣F␩
RT0
k = k0 exp
͓13͔
ͫ ͬ
f
f
Figure 1. ͑a͒ Polarization curves for the oxygen reduction reaction on a
Pt/Vulcan XC-72R ͑14.1 ␮g Pt cm−2͒ thin-film RRDE ͓2500 rpm͔ in 2.0 M
HClO4 solution ͓pH −0.3͔ bubbled with 10% O2 ͓᭺͔, air ͓᭡͔ and O2 ͓b͔.
͑b͒ Ring currents corresponding to H2O2 oxidation ͓Ering = 1.2 V vs SHE͔.
͑c͒ % H2O2, given by Eq. 11, formed during the oxygen reduction reaction
on Pt/Vulcan XC-72R in 2.0 M HClO4 solution ͓pH −0.3͔ bubbled with
10% O2 ͓᭺͔, air ͓᭡͔ and O2 ͓᭝͔ at 25°C. Ering = 1.2 V vs SHE, 1 mV/s,
2500 rpm.
In Eq. 12, a and b are reaction orders with respect to O2 and H+,
respectively. Only the forward rate term is used in Eq. 12 because at
0.6 V vs SHE and below, the rate of oxidation of H2O2 ͑the reverse
reaction͒ is negligible. The kinetic rate constant kf was estimated for
different potentials by plotting H2O2 production rate as a function of
oxygen concentration for various potentials. Since the electrode re-
action rate was shown earlier by Damjanovic and Hudson30 to be
faster on an oxide-free Pt surface than on an oxide-covered surface,
both the forward and the reverse scans were used to estimate the
reaction rate constant. The potential dependence of this rate constant
is given in Eq. 13.
Results and Discussion
The activation energies for hydrogen peroxide formation reaction
were evaluated by using the Arrhenius equation31 shown below,
H2O2 kinetic studies using RRDE.— Figure 1a shows polariza-
tion curves for the oxygen reduction reaction and Fig. 1b shows ring
currents corresponding to H2O2 oxidation. It should be noted that
hydrogen evolution starts in the neighborhood of ϳ25–50 mV vs
SHE. Hence, both ring and disk currents in the 25–50 mV region
include a fractional contribution due to hydrogen evolution ͓2H+
+ 2e− → H2͔ and oxidation ͓H2 → 2H+ + 2e−͔ respectively. This is
noted by the decrease in the disk currents and an increase in the ring
currents in the 25–50 mV potential region. Data obtained below
25 mV were not used in this analysis. Figure 1c shows the fraction
Ea
k0f = k0f,0 exp
͓14͔
ͫ ͬ
RT
The activation energies for H2O2 formation on supported Pt cata-
lysts were compared to the computationally estimated activation en-
ergies reported in the literature. For example, using density func-
tional theory, Anderson and Albu,32 Sidik and Anderson,33 and Wang
and Balbuena34 have reported activation energies for H2O2 forma-
tion on Pt1, Pt2, and Pt3 sites, respectively.
Downloaded on 2015-06-08 to IP 128.122.253.228 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 B50-B57 ͑2008͒
B53
Figure 2. H2O2 formation rates ͓mol cm−2 s−1͔ on Pt/Vulcan XC-72R in
2.0 M HClO4 solution ͓pH − 0.3͔ as a function of dissolved oxygen con-
centration ͓mol L−1͔ for the following four overpotentials: 0.695 V ͓᭺͔,
0.495 V ͓ࡗ͔ , 0.295 V ͓᭝͔ and 0.095 V ͓b͔. The symbols represent data
and the lines represent linear fits with zero intercepts.
of peroxide produced in the ORR at the disk, as captured by the ring
and shows no dependence on oxygen concentration.
Figure 2 shows the purely kinetic portion of H2O2 formation
rates obtained from Eq. 12 as a function of oxygen concentration in
the 2 M HClO4 for different overpotentials. The data show a linear
dependence of the oxygen concentration at all potentials ͑i.e., a
= 1͒. In this figure, overpotential of 0.670 V represents a potential
of 0.025 V vs SHE because the equilibrium potential for H2O2 for-
mation is 0.695 V. Four representative overpotentials were chosen
for this plot. The anode experiences the highest overpotential for
peroxide formation during fuel cell operation. The cathode potential
is above the H2O2 equilibrium potential at open circuit but experi-
ences a significant drop during load conditions, and can go negative
of H2O2 equilibrium potential.
Figure 3 shows the potential dependence of the rate constants
estimated from Eq. 12. The 25°C data between ␩ values 0–0.3 V
and 0.3–0.65 V was fit with two separate linear equations in Fig. 3.
The respective intercepts represent k0f ͑Eq. 13͒ and are independent
of CH+. These values, as a function of T, were used to obtain acti-
vation energy.
Figure 4. ͑a͒ Rate and ͑b͒ % of H2O2 formed during the oxygen reduction
reaction on Pt/Vulcan XC-72R in 0.1 M ͓᭺͔, 1.0 M ͓᭡͔, 1.8 M ͓छ͔ and
2.0 M ͓b͔ HClO4 solution ͓pH −0.3͔ bubbled with O2 at 25°C. Ering
= 1.2 V vs SHE, 1 mV/s, 2500 rpm.
effect of proton concentration. Figure 4 shows an increased rate of
H2O2 formation with increasing proton concentration. Since the disk
currents were similar for all acid concentrations, the increased ring
currents meant that selectivity toward peroxide formation was a
function of proton concentration.
Figure 5 shows the dependence of H2O2 formation rate on proton
concentration. This is consistent with the RRDE results of Murthi et
al. in 1 and 6 M TFMSA. The points are measurements and lines are
predictions according to Eq. 12. The reaction order with respect to
H+ in the H2O2 formation reaction was found to best fit the data for
Figure 4 shows the ring currents and the fraction of H2O2 formed
with different acidities. Perchloric acid systems in the 0.1 and 2.0 M
concentration range equilibrated with pure O2 were used to study the
Figure 3. Electrochemical rate constant for H2O2 formation, kf, as a function
of overpotential, ␩ = Eapp − E0, E0 = 0.695 V vs SHE. The data between ␩
values 0–0.3 V and 0.3–0.65 V were fit with two separate linear equations.
The value of ␣ = 0.05, T = 25°C. Data was obtained in 2 M HClO4 bubbled
with pure O2 at 1 atm.
Figure 5. Mass transport corrected H2O2 formation rates ͓mol cm−2 s−1͔ on
Pt/Vulcan XC-72R in HClO4 solution as a function of acidity ͓M͔ for the
following three overpotentials: 0.695 V ͓᭺͔, 0.345 V ͓ࡗ͔ and 0.095 V ͓᭝͔.
The lines are predictions according to Eq. 13. The reaction order with respect
to H+ in the H2O2 formation reaction, b = 2.
Downloaded on 2015-06-08 to IP 128.122.253.228 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B54
Journal of The Electrochemical Society, 155 ͑1͒ B50-B57 ͑2008͒
b = 2. The reason for the change in selectivity shown by the Pt
catalyst in favor of the two-electron transfer at higher acid concen-
trations ͑or lower water activities͒ is beyond the scope of this article.
There is a possibility that the adsorption of ClO−4 anions could in-
fluence the reaction pathway in ORR on Pt catalysts similar to that
of Cl− anions as previously reported by Schmidt et al.,23 Stamenk-
ovic et al.,25 and Markovic and Ross.24 However, the bond strength
of ClO−4 adsorption on Pt is much weaker than Cl− and SO24− and its
influence on the reaction pathway of ORR is also very minimal.23,25
The H2O2 formation rates measured as a function of water activ-
ity, potential, and temperature using RRDE experiments was used to
predict H2O2 formation rates at the anode and cathode of PEM fuel
cell. Peroxide formation rate at the anode was predicted using oxy-
gen permeability from the cathode and ␹
. Peroxide formation
H O
2
2
rate at the cathode was predicted via Eq. 12, i.e., as a product of the
rate constant and the local reactant concentrations. Peroxide forma-
tion at the cathode occurs only for fuel cells operating under con-
siderable load ͑i.e., high cell current͒ such that the local potential
goes negative relative to the equilibrium potential for peroxide for-
mation. For estimation of peroxide rates, local potential at the cath-
ode was taken to be 0.6 V ͑i.e., ␩ = 0.095 V͒.
Nafion is a super-acid catalyst and hence the local acidity at the
catalyst-membrane interfaces was calculated from the local water
content and the fixed number of sulfonic acid groups. The water
sorption properties of Nafion as a function of temperature and water
activity had been studied by several laboratories.33-37
Using a tapered element oscillating microbalance technique,
Jalani et al.37 measured water uptake in Nafion as a function of
water activity in vapor phase between 30 and 110°C and reported
that the water uptake increased with temperature and was highest at
110°C. The difference in water uptake between 30°C and 110°C is
negligible for lower water activities ͑aw Ͻ 0.7͒. This is reported by
Jalani et al. ͑experimental͒ and discussed in detail by Motupally et
al. ͑simulations͒.40 For this work, the absorption isotherm of Nafion
117 membranes measured at 30°C by Zawodzinski et al.38,39 was
used. Between water activity values of 0 and 1, the experimentally
measured absorption isotherm was fit to the following polynomial41
Figure 6. ͑a͒ pH ͑-----͒ and ␭ ͑—͒ vs water activity, aw, plots for Nafion; ␭
is the amount of water per sulfonic acid group ͓mol basis͔ and aw is the
effective mole fraction of water and is equal to the equilibrium relative
humidity expressed as a fraction.
shown͒ is seen for HClO4, also a strong acid. The acid was assumed
to be completely dissociated i.e., Ka ӷ 1. This approach in relating
membrane electrode assembly acidity to water activity and hence to
the humidity of the incoming gases facilitates in computing peroxide
rates inside a fuel cell. Quantitatively, the measured peroxide rates
via the RRDE experiments at a particular oxygen concentration, pH
value, and temperature should equal the peroxide rates inside the
fuel cell at the same pH value and temperature. Since rate constants
were measured at 25°C using RRDE, they had to be estimated at
higher temperatures in order to be used to calculate peroxide forma-
tion rates in a fuel cell. Activation energies for H2O2 formation on
supported Pt catalysts were estimated from kinetic currents obtained
at 15, 25, 35, and 45°C. Figure 7 shows the Arrhenius plot of the
logarithmic pre-exponential term k0f vs the inverse of temperature
for the two overpotential regions ͑0–0.3 V and 0.3–0.65 V͒. The
activation energies were obtained from the slope ͑Ea/R͒ of a linear
fit according to Eq. 14. This procedure is analogous to those de-
scribed by Neyerlin et al.43 and Bard and Faulkner.44
2
3
␭ = 0.043 + 17.81͓aw͔ − 39.85͓aw͔ + 36.0͓aw͔
͓15͔
Oxygen flux across Nafion at higher temperatures.— Oxygen
permeability through Nafion depends greatly on the water content of
the membrane. It has been shown by Sakai et al.45 that O2 diffusion
rates in a completely dry Nafion membrane have values similar to
that in polytetrafluorethylene and approaches the limit of liquid wa-
ter with increasing water content. Figure 8 shows experimentally
measured O2 permeability, corrected for an O2 feed at 101 kPa,
across a Nafion 112 membrane as a function of temperature and
relative humidity. These permeation rates were estimated using elec-
trochemical monitoring technique and are comparable to those esti-
In this equation, ␭ represents the number of water molecules per
sulfonic acid group in the polymer and aw represents the activity of
water, which is the effective mole fraction of water given by p0/p*,
where p* is the vapor pressure of water, in bar; p* was calculated
from the Antoine correlation
B1
ln p* = A1 −
͓16͔
T + C1
The constants are A1 = 11.6832, B1 = 3816.44, C1 = −46.13.42
Inside a fuel cell, this water activity is essentially the equilibrium
relative humidity expressed as a fraction. The concentrations of H2O
and H+ in the polymer are, respectively, expressed as
␳␭
CH O
=
͓17͔
2
EW
CH O
2
CH =
+
͓18͔
␭
In these equations, EW is the equivalent weight of the polymer
͑taken to be 1100͒ and ␳ is the humidity-dependent density of the
polymer given by
1.98 + 0.0324␭
␳ =
͓19͔
1 + 0.0648␭
It was assumed that all sulfonic acid groups exist in a completely
dissociated form. Figure 6 shows the variation of ␭ and pH of
Nafion as a function of water activity. Even at vapor-saturated con-
ditions ͓␭ = 14͔, the pH of Nafion is below 0. This trend ͑not
Figure 7. Arrhenius plots for H2O2 formation reaction on Pt/Vulcan XC-72R
in 2 M HClO4 at overpotential of 0.195 V ͑-᭝-͒ and 0.695 V ͑-b-͒ vs NHE.
Downloaded on 2015-06-08 to IP 128.122.253.228 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 B50-B57 ͑2008͒
B55
Figure 8. Oxygen permeability ͓mol cm−1 s−1͔ in Nafion 112 measured by
electrochemical monitoring technique as a function of temperature for two
different relative humidities, 25% and 100% RH. These rates were normal-
ized to a 101 kPa O2 feed.
Figure 10. Rates of H2O2 formation ͓mol cm−2 s−1͔ in the anode side of a
PEM fuel cell for different relative humidities and temperatures. Local po-
tential at the anode was assumed to be ϳ0 V vs SHE, which translates to an
overpotential of 0.695 V for H2O2 formation.
mated by the gas chromatography method.46 Between 25% and
100% relative humidity of the feed gas, the permeabilities differ by
as much as an order of magnitude. Permeability for other tempera-
tures and water contents were estimated by the following equation
which was derived by fitting the measured permeability values
The potential profile across the membrane, measured in situ by
Liu and Zuckerbrod15 ͑Fig. 19͒ and modeled by Burlatsky et al.47 at
open circuit conditions, indicates that the potential at the anode-
membrane interface is ϳ0 V. For the purpose of calculating H2O2
rates at the anode/membrane interface, a potential of ϳ0 V ͑i.e.,
␩ = 0.695 V͒ was assumed to exist at the interface.
POm = ͑1.002 ϫ 10−14 − 9.985 ϫ 10−15aw͒exp͓͑0.0127 + 2.3467
2
ϫ 10−2aw͒T͔
͓20͔
The oxygen flux across the membrane from the cathode to the
anode is
Oxygen solubility at the membrane-cathode catalyst layer inter-
DOm
face, CcO , was estimated using the following relation
2
FO
=
͑CcO − CaO
͒
2
͓23͔
2
2
2
POm
DOm
␦
2
COc
=
͓21͔
The concentration of oxygen at the anode-membrane interface
approaches zero, assuming all of the oxygen crossing over the mem-
brane to the anode side is reduced to water or reacts chemically with
hydrogen
2
2
DmO values for different temperatures and relative humidities
2
were obtained from the work of Sakai et al.45 and was fit to the
following expression
POm
2
RaH O = ␹
͓24͔
DmO = 9.78 ϫ 10−8 + 3.5 ϫ 10−9T + 10−4aw
͓22͔
H O
2
2
2
2
␦
2
While the fraction of oxygen that is reduced to peroxide is a
strong function of water activity, it is not a function of oxygen
Figure 9 shows the peroxide rates at the cathode-membrane in-
terface when the local potential at the cathode is 0.6 V and the gas
feed is pure oxygen at 1 atm. Though the amount of peroxide
formed at the cathode is only of the order of few micromoles at fully
humidified conditions, the relative difference between dry and fully
humidified conditions at 95°C is significant.
concentration ͑see Fig. 1c͒. An expression for ␹
vs CH+ was
H O
2
2
obtained from measured values at room temperature
2
3
␹
= 0.2081 − 0.1208͑aw͒ − 0.072͑aw͒ − 2.132 ϫ 10−14͑aw͒
H O
2
2
͓25͔
Figure 10 shows the H2O2 formation rates at the anode-
membrane interface. This goes through a peak because oxygen per-
meability decreases with decreasing water activity, whereas H2O2
selectivity increases with decrease in water activity.
These anode and cathode H2O2 formation rates cannot be di-
rectly correlated with the fluoride emission rates because there are
several intermediate reactions between H2O2 formation and actual
membrane degradation. Though additional mechanisms involving
direct radical formation on Pt followed by their chemical attack on
the membrane were suggested,47 they need further experimental
validation and verification. However, quantification of H2O2 forma-
tion rates described in this work is important in the mechanistic
understanding of membrane degradation. This quantification would
help in validating a durability mechanism especially at elevated tem-
peratures and low relative humidities.
Conclusion
Figure 9. Rates of H2O2 formation ͓mol cm−2 s−1͔ in the cathode side of a
PEM fuel cell for different relative humidities and temperatures. Local po-
tential at the cathode was assumed to be ϳ0.6 V vs SHE, which translates to
an overpotential of 0.095 V for H2O2 formation.
H2O2 formation rates in a PEM fuel cell anode and cathode were
estimated by studying the ORR kinetics on a ring disk electrode.
Fuel cell conditions were replicated by depositing a film of Pt/
Downloaded on 2015-06-08 to IP 128.122.253.228 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B56
Journal of The Electrochemical Society, 155 ͑1͒ B50-B57 ͑2008͒
Superscript
Vulcan XC-72 catalyst onto the disk and by varying the temperature,
dissolved O2 concentration, and the acidity levels in HClO4. The
HClO4 acidity was correlated to ionomer water activity and hence
fuel cell humidity. H2O2 formation rates showed a linear depen-
dence on oxygen concentration and a square law dependence on
water activity. The H2O2 selectivity in ORR was independent of
oxygen concentration but increased with decrease in water activity
͑i.e., decreased humidity͒. Potential dependent activation energy for
the H2O2 formation reaction was estimated from data obtained at
different temperatures. Anode peroxide formation rates are propor-
tional to the oxygen flux from the cathode and were estimated to be
three orders of magnitude lower than cathode formation rates for a
cell operating under load conditions ͑i.e., V ഛ 0.6 V͒.
0
a
c
standard state or equilibrium
anode
cathode
Subscript
b
D
backward reaction
diffusion
disk Pt/Nafion coated disk electrode
f
Nafion film or forward reaction
kin kinetic
ring Pt ring electrode
References
1. http://www.eere.energy.gov/hydrogenandfuelcells/
2. H. A. Gasteiger, W. Gu, R. Makharia, M. F. Mathias, and B. Sompalli, in Hand-
book of Fuel Cells—Fundamentals, Technology and Applications, Vol. 3, W. Viel-
stich, A. Lamm, and H. A. Gasteiger, Editors, p. 593, Wiley, New York ͑2003͒.
3. D. P. Wilkinson and J. St.-Pierre, Handbook of Fuel Cells: Fundamentals, Tech-
nology and Applications, Vol. 3, W. Vielstich, A. Lamm and H. A. Gasteiger,
Editors, p. 611, Wiley, New York ͑2003͒.
4. C. Paik, T. Skiba, V. Mittal, S. Motupally, and T. Jarvi, Abstract 771, The Electro-
chemical Society Meeting Abstracts, Vol. 2005-1, Quebec City, Canada, May 15–
20, 2005.
5. V. Mittal, C. Paik, and S. Motupally, Abstract 772, The Electrochemical Society
Meeting Abstracts, Vol. 2005-1, Quebec City, Canada, May 15–20, 2005.
6. S. Hommura, K. Kawahara, and T. Shimohira, Abstract 806, The Electrochemical
Society Meeting Abstracts, Vol. 2005-1, Quebec City, Canada, May 15–20, 2005.
7. M. Inaba, H. Yamada, J. Tokunaga, and A. Tasaka, Abstract 1506, The Electro-
chemical Society Meeting Abstracts, Vol. 2005-1, Quebec City, Canada, May 15–
20, 2005.
Acknowledgments
The U.S. Department of Energy supported this work under con-
tract no. DOE-DE-FC36-02AL, for which the authors are grateful.
The authors thank Michael Fortin for his help with the experimental
setup for oxygen crossover measurements.
The University of South Carolina assisted in meeting the publication
costs of this article.
8. T. Takeshita, F. Miura, and Y. Morimoto, Abstract 1511, The Electrochemical So-
ciety Meeting Abstracts, Vol. 2005-1, Quebec City, Canada, May 15–20, 2005.
9. A. B. LaConti, M. Hamdan, and R. C. McDonald, in Handbook of Fuel Cells—
Fundamentals, Technology and Applications, Vol. 3, W. Vielstich, A. Lamm and H.
A. Gasteiger, Editors, p. 647, Wiley, New York ͑2003͒.
10. E. Endoh, S. Terazono, H. Widjaja, and Y. Takimoto, Electrochem. Solid-State
Lett., 7, A209 ͑2004͒.
List of Symbols
a
reaction order with respect to O2 in the H2O2 formation reaction
aw water activity
A
b
disk area, cm2
reaction order with respect to H+ in the H2O2 formation reaction
proton concentration, mol cm−3
water concentration in the membrane, mol cm−3
oxygen concentration in the bulk of the electrolyte, mol cm−3
oxygen concentration in Nafion film, mol cm−3
oxygen concentration in Nafion 112 membrane-anode catalyst layer inter-
face, mol cm−3
oxygen concentration in Nafion 112 membrane-cathode catalyst layer in-
11. A. Panchenko, Dipl.-Chem., Institute für Phzsikalische Chemie der Universität
CH
CH
+
Stuttgart ͑2004͒.
O
2
12. A. Panchenko, H. Dilger, J. Kerres, M. Hein, A. Ullrich, T. Kaz, and E. Roduner,
Phys. Chem. Chem. Phys., 6, 2891 ͑2004͒.
CO*
COf
COa
2
2
2
13. W. Bi, G. E. Gray, and T. Fuller, Electrochem. Solid-State Lett., 10, B101 ͑2007͒.
14. D. E. Curtin, R. D. Lousenberg, T. J. Henry, P. C. Tangeman, and M. E. Tisack, in
Proceedings of the Second European Polymer Electrolyte Fuel Cell Forum, Vol. 1,
p. 25 ͑2003͒; D. E. Curtin, R. D. Lousenberg, T. J. Henry, P. C. Tangeman, and M.
E. Tisack, J. Power Sources, 131, 41 ͑2004͒.
15. W. Liu and D. Zuckerbrod, J. Electrochem. Soc., 152, A1165 ͑2005͒.
16. U. A. Paulus, T. J. Schmidt, H. A. Gasteiger, and R. J. Behm, J. Electroanal.
Chem., 495, 134 ͑2001͒.
COc
2
terface, mol cm−3
DO*
DOf
oxygen diffusion coefficient in the electrolyte, cm2 s−1
oxygen diffusion coefficient in Nafion film, cm2 s−1
diffusion coefficient of O2 in Nafion 112 membrane, cm2 s−1
equilibrium potential, 0.695 V vs SHE at 25°C and 101 kPa
activation energy for H2O2 formation, J mol−1
2
2
DOm
17. E. Claude, T. Addou, J.-M. Latour, and P. Aldebert, J. Appl. Electrochem., 28, 57
͑1998͒.
2
E0
Ea*
18. U. A. Paulus, A. A. Wokaun, G. G. Scherer, T. J. Schmidt, V. Stamenkovic, N. M.
Markovic, and P. N. Ross, Electrochim. Acta, 47, 3787 ͑2002͒.
19. O. Antoine and R. Durand, J. Appl. Electrochem., 30, 839 ͑2000͒.
20. M. A. Enayetullah, T. D. DeVilbiss, and J. O’M. Bockris, J. Electrochem. Soc.,
136, 3369 ͑1989͒.
21. V. S. Murthi, R. C. Urian, and S. Mukerjee, J. Phys. Chem. B, 108, 11011 ͑2004͒.
22. T. J. Schmidt, H. A. Gasteiger, G. D. Stäb, P. M. Urban, D. M. Kolb, and R. J.
Behm, J. Electrochem. Soc., 145, 2354 ͑1998͒.
Eapp applied potential, V vs SHE
equivalent weight of Nafion polymer, 1100 g equiv−1
Faraday constant, 96,485 C mol−1
EW
F
Iring ring current, mA
Idisk disk current, mA
j
jdisk
jD
jkin
kb
kf
total peroxide current density, mA cm−2
disk current density, mA cm−2
23. T. J. Schmidt, U. A. Paulus, H. A. Gasteiger, and R. J. Behm, J. Electroanal.
diffusion-limited current density, mA cm−2
kinetic current density, mA cm−2
Chem., 508, 41 ͑2002͒.
24. N. Markovic and P. N. Ross, J. Electroanal. Chem., 330, 499 ͑1992͒.
25. V. Stamenkovic, N. M. Markovic, and P. N. Ross, Jr., J. Electroanal. Chem., 500,
44 ͑2001͒.
26. L. H. Gevantman, in CRC Handbook of Chemistry and Physics, 79th ed., D. R.
Lide, Editor, p. 8, CRC, New York ͑1998–99͒.
27. W. J. Albery and M. L. Hitchman, Ring-Disc Electrodes Clarendon Press, Oxford,
United Kingdom, ͑1971͒.
28. Rotated Ring-Disk Electrodes: DT21 Series, Pine Instrument Company, Raleigh,
NC 27617.
rate constant for H2O2 electro-oxidation, s−1
rate constant for H2O2 formation, mol2 cm−5 s−1
collection efficiency
N
n
number of electrons transferred per O2 molecule in H2O2 formation, 2
POm
permeability of O2 in Nafion 112 membrane, mol cm−1 s−1
2
universal gas constant, 8.314 J mol−1 K−1
R
T
t
temperature, K
time, s
29. S. K. Zečević, J. S. Wainright, M. H. Litt, S. Lj. Gojković, and R. F. Savinell, J.
Electrochem. Soc., 144, 2973 ͑1997͒.
Greek
30. A. Damjanovic and G. Hudson, J. Electrochem. Soc., 135, 2269 ͑1988͒.
31. U. A. Paulus, A. Wokaun, G. G. Scherer, T. J. Schmidt, V. Stamenkovic, V. Rad-
milovic, N. M. Markovic, and P. N. Ross, J. Phys. Chem. B, 106, 4181 ͑2002͒.
32. A. B. Anderson and T. V. Albu, J. Electrochem. Soc., 147, 4229 ͑2000͒.
33. R. A. Sidik and A. B. Anderson, J. Electroanal. Chem., 528, 69 ͑2002͒.
34. Y. Wang and P. B. Balbuena, J. Chem. Theory Comput., 1, 935 ͑2005͒.
35. D. R. Morris and X. Sun, J. Appl. Polym. Sci., 50, 1445 ͑1993͒.
36. D. Rivin, C. E. Kendrick, P. W. Gibson, and N. S. Schneider, Polymer, 42, 623
͑2001͒.
␣
␦
transfer coefficient
Pt/C electrode thickness, cm
Nafion film thickness, cm
␦
f
density of Nafion, g cm−3
␳
kinematic viscosity, cm2 s−1
overpotential, V vs SHE
␯
␩
␭
moles of water per sulfonic acid group in Nafion
fraction of O2 reducing to H2O2
electrode rotation rate, s−1
␹
H
O
37. N. H. Jalani, P. Choi, and R. Datta, J. Membr. Sci., 254, 31 ͑2005͒.
38. T. Zawodzinski, M. Neeman, L. Sillerud, and S. Gottesfeld, J. Phys. Chem., 95,
2
2
␻
Downloaded on 2015-06-08 to IP 128.122.253.228 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 B50-B57 ͑2008͒
B57
6040 ͑1991͒.
44. A. J. Bard and L. R. Faulkner, Electrochemical Methods, Wiley, New York ͑1980͒.
45. T. Sakai, H. Takenaka, and E. Torikai, J. Electrochem. Soc., 133, 88 ͑1986͒.
46. K. Broka and P. Ekdunge, J. Appl. Electrochem., 27, 117 ͑1997͒.
47. S. Burlatsky, T. Jarvi, and V. Atrazhev, Abstract No. 502, The Electrochemical
Society Meeting Abstracts, Vol. 2005-2, Los Angeles, CA, Oct 16–21, 2005.
48. Q. Guo, V. A. Sethuraman, and R. E. White, J. Electrochem. Soc., 151, A983
͑2004͒.
49. K. A. Mauritz and R. B. Moore, Chem. Rev. (Washington, D.C.), 104, 4535 ͑2004͒.
50. R. H. Perry and D. W. Green, Perry’s Chemical Engineer’s Handbook, 7th ed.,
McGraw-Hill, New York, ͑1997͒.
39. J. T. Hinatsu, M. Mizuhata, and H. Takenaka, J. Electrochem. Soc., 141, 1493
͑1994͒.
40. S. Motupally, A. J. Becker, and J. W. Weidner, J. Electrochem. Soc., 147, 3171
͑2000͒.
41. T. Fuller, Ph.D. Thesis, University of California, Berkeley ͑1989͒.
42. R. C. Reid, J. M. Prausnitz, and T. K. Sherwood, The Properties of Gases and
Liquids, McGraw-Hill, New York ͑1977͒.
43. K. C. Neyerlin, W. Gu, J. Jorne, and H. A. Gasteiger, J. Electrochem. Soc., 153,
A1955 ͑2006͒.
Downloaded on 2015-06-08 to IP 128.122.253.228 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).