Mechanistic Discussion of the Oxygen Reduction Reaction at Nitrogen

ARTICLE pubs.acs.org/JPCC

Mechanistic Discussion of the Oxygen Reduction Reaction at Nitrogen-Doped Carbon Nanotubes Jaclyn D. Wiggins-Camacho and Keith J. Stevenson* Department of Chemistry and Biochemistry, Center for Nano- and Molecular Science and Technology, Center for Electrochemistry, University of Texas at Austin, Austin, Texas 78712, United States

bS Supporting Information ABSTRACT: The oxygen reduction reaction (ORR) at undoped and nitrogen-doped carbon nanotubes (CNTs and N-CNTs, respectively) was studied by cyclic voltammety, rotating disk electrode voltammetry, and gasometric analysis in neutral and alkaline aqueous solutions. At undoped CNTs, the ORR proceeds by two successive two-electron processes with hydroperoxide (HO2
) as the intermediate. At N-CNTs, the ORR occurs through a “pseudo”-four-electron pathway involving a catalytic regenerative process in which hydroperoxide is chemically disproportionated to form hydroxide (OH
) and molecular oxygen (O2). The ORR mechanism at both undoped and N-doped varieties is supported by steady state polarization and gasometric measurements of hydroperoxide disproportionation rates. An enhancement of over 1000-fold for hydroperoxide disproportionation is observed for N-CNTs, with rates comparable to the best known peroxide decomposition catalysts. A positive correlation between nitrogen content and ORR activities is observed where the ORR potential shifts by up to 11.6 mV per at. % N incorporated into the N-CNTs and exhibits an oxygen reduction potential, Ep, of
0.23 V vs Hg/Hg2SO4 (+0.640 V vs NHE) in 1 M Na2HPO4 for N-CNTs containing 7.4 at. % N. A detailed mechanism is proposed that involves a dual site reduction in which O2 is initially reduced at a N
C type site in a 2-electron process to form HO2
, which then can undergo either further electrochemical reduction to form OH
species or chemical disproportionation to form OH
species and molecular O2 at a decorating FexOy/Fe surface phase.

’ INTRODUCTION The oxygen reduction reaction (ORR) mechanism at nitrogendoped carbons (N-carbons) has been a widely discussed topic in the scientific literature. Currently, there is no consensus regarding how the role of defects, edge plane sites, heteroatom doping, and the presence of metal oxides (i.e., Fe, Co, Mn) affect the ORR mechanism. Understanding the specifics of this mechanism will allow researchers to tailor more efficient and stable support materials for a wide range of applications, including the improvement of fuel cells,1
5 biological sensors,6,7 and oxygen cathodes.8 Previous research in our group has shown that adding nitrogen into carbon nanotubes (CNTs) increases the edge plane structure, and edge planes are known to be reactive sites showing an increased electrochemical reactivity.9
11 In addition, several other groups have shown that incorporating nitrogen into carbon materials affects the electrocatalytic behavior, promoting the ORR by facilitating the growth of a more favorable structure (i.e., incorporation of disorder and edge plane exposure).12
17 Previous work in our group has shown that the presence of residual iron growth catalyst (ferrocene) does not significantly alter the electrochemical response of the undoped CNTs at neutral pH11,18 and provides evidence in support of the ORR proceeding by two successive two-electron processes to form OH
, with HO2
as the intermediate (eqs 1, 2).10,19 O2 þ H2 O þ 2e
T HO2
þ OH


E0 ¼
0:065 V vs NHE

ð1Þ r 2011 American Chemical Society

HO2
þ H2 O þ 2e
T 3OH


HO2
T ð1=2ÞO2 þ OH


E0 ¼ 0:867 V vs NHE

ð2Þ ð3Þ

At nitrogen-doped CNTs (N-CNTs), the ORR proceeds through a “pseudo” 4-electron pathway invoving a catalytic regenerative process in which the HO2
intermediate is formed electrochemically (eq 1) then chemically disproportionated to form OH
and O2 (eq 3) in what is known as a catalytic electrochemical
chemical (EC0 ) reduction mechanism.10 It has been proposed that this rapid heterogeneous chemical disproportionation step (eq 3) is what allows for the electrocatalytic enhancement observed at N-CNTs, as they are able to bypass the kinetically unfavorable second electrochemical reduction step, effectively lowering the overpotential of the overall reaction.10 Furthermore, researchers have demonstrated that this fast disproportionation is facilitated by a surface stabilized metal oxide particle (Fe, Co, Mn) similar to the Haber
Weiss mechanism for the catalytic decomposition of hydrogen peroxide.20 Other experimental and theoretical groups investigating the ORR at transition-metal-containing N-carbons have indicated that the ORR proceeds through a two-step mechanism.12,21
24 Specifically, strong evidence has been provided that a direct Received: June 7, 2011 Revised: August 25, 2011 Published: September 21, 2011 20002

dx.doi.org/10.1021/jp205336w | J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C

ARTICLE

4-electron transfer mechanism (eq 4) involving a molecular N4
Fe type active site, as has been proposed for heat treated N4
Me macrocycles,25
30 is unlikely by showing that the electronic properties of the N-CNTs (i.e., density of states at the Fermi level (D(EF)), conductivity, mobile charge carriers) depend most strongly on the concentration and coordination of nitrogen and are largely unaffected by removal of the residual Fe by acid washing.10,11,31 O2 þ 4Hþ þ 4e
f 2H2 O

E0 ¼ 0:401 V vs NHE

ð4Þ

Wang et al. also reported that the addition of Fe in the carbon catalyst is beneficial for the incorporation of nitrogen and that this N doping introduces electronic states around the Fermi level, enhancing electron transfer for O2 reduction.32 In addition, our group’s electrochemical studies have shown that residual surfacebound FexOy/Fe sites on the N-CNTs can be electrochemically passivated without compromising the observed electrocatalytic enhancement,18 and recently, we have shown that the addition of a competitive inhibitor (CN
) to the electrolyte solution at N-CNTs does not poison the ORR, like what is seen for the ORR mediated via a molecular heme-like N4
Fe active site.33 These findings are further supported by theoretical work suggesting that nitrogen atoms incorporated at specific sites within graphitic carbon can significantly enhance the ORR activity.24 Specifically, theory indicates that O2 preferentially adsorbs on carbon sites adjacent to pyridinic nitrogen sites, supporting a mechanism involving surface-adsorbed radical and peroxy intermediates following the formation of surface adsorbed HO2
.22,23 It has been shown that the initial reduction step (eq 1) possibly occurs via a three-step surface-mediated mechanism shown in eqs 5
7, in which O2 is first reduced to the surfacebound superoxide radical ((O2
•)ads, eq 5), which is subsequently converted to the adsorbed hydroperoxide radical ((HO2•)ads, eq 6), and then reduced to hydroperoxide (HO2
, eq 7).10,34,35 O2 þ e
T ðO2 •
Þads

ð5Þ

ðO2 •
Þads þ H2 O T ðHO2 • Þads þ OH


ð6Þ

ðHO2 • Þads þ e
T HO2


ð7Þ

The specific mechanism for the further disproportionation and reduction of HO2
(eqs 2, 3) is not as well understood, and there is much discussion throughout the literature as to how the following reduction proceeds. In this work, the ORR mechanism at N-CNTs is investigated specifically in terms of the chemical disproportionation of HO2
while being mindful of the roles of the nitrogen coordinations and presence of residual surfacebound iron oxide particles (FexOy/Fe surface sites). Mechanistic differences between the CNTs and N-CNTs are discussed, and a complex surface-mediated ORR mechanism involving a dual site reduction is proposed.

’ EXPERIMENTAL SECTION Synthesis of Carbon Nanotubes. Both undoped and N-doped carbon nanotubes (CNT/N-CNTs) were synthesized via our previously reported10,36 floating catalyst chemical vapor deposition (CVD) method using a ferrocene growth catalyst (Alfa Aesar) and m-xylene, anhydrous (Sigma-Aldrich) or pyridine (Fisher) as the carbon source, respectively. Briefly, 1.0 mL of a

20 mg mL
1 precursor solution was injected at a rate of 0.1 mL min
1 into a dual zone tube furnace, with a quartz tube spanning both zones. The precursor solution was heated to the vaporization temperature in the first zone (150/130 °C for m-xylene and pyridine, respectively) and carried into the second zone by carrier gases (Ar
H2 or Ar
NH3 for undoped and N-CNTs, respectively), at a total flow rate of 575 sccm, where the NH3 flow rate was systematically increased to change the doping level in the N-CNTs, as previously reported.9 The mixture was pyrolyzed upon reaching the second zone at the respective appropriate temperature (700 °C for undoped and 800 °C for N-CNTs), resulting in the growth of multiwalled CNTs (MWCNTs) along the inner wall of the quartz tube. The CNT/N-CNTs were collected after cooling the tube to room temperature in an argon environment and stored in an airtight vial prior to use. In all experiments, unless otherwise noted, five N-doping levels were investigated: 0.0, 4.0, 5.0, 6.3, and 7.4 at. % N (content determined by XPS). CNT Mat Assembly. CNT/N-CNT mat electrodes were made via a vacuum filtration method, modified from previously reported procedures.37
39 Briefly, a suspension of 0.40 mg mL
1 CNT/N-CNTs in ethanol was sonicated for 16 h to create a uniform suspension. One milliliter of the suspension was then vacuum-filtered through an Anodisc porous alumina membrane (Whatman, 0.2 μm pore size, 13 mm diameter) and allowed to dry over vacuum to evaporate any excess ethanol. The alumina membranes were then removed with 6 M NaOH, and the resulting mat was thoroughly rinsed with Nanopure water (18 MΩ cm) until the pH of the rinsate was neutral. The mat was then transferred onto an 18  18 mm glass coverslip and dried for 30 min in an oven at 60 °C. The resulting mats were stored in sealed containers prior to analysis. Acid-treated mats (Supporting Information) were “super-washed” in a 0.6 mg/mL suspension of 3 M HCl at 90 °C for 12 h to remove residual Fe catalyst. Fe removal (less than 0.2 at. % Fe by XPS) was confirmed with high-resolution XPS analysis as well as STEM imaging.11 The acid-treated N-CNTs were collected via vacuum filtration and rinsed with Nanopure water (18 MΩ cm) until the pH of the rinsate was neutral and dried for 20 min in an oven at 60 °C. Mats were then assembled according to the previously described procedure. X-ray Photoelectron Spectroscopy. XPS was performed using a Kratos Axis DLD spectrometer with an Al Kα source and an analyzer resolution of 0.1 eV. The nitrogen content is reported as at. % N as determined from high-resolution XPS measurements of the N 1s spectral line and is reported relative to the other elements present (carbon, oxygen, iron). Spectra were analyzed using Kratos Vision software, with Shirley background correction. The N 1s and Fe 2p functionalities were fitted using CasaXPS version 2.3.15 according to Artyushkova, et al.40,41 Thermogravimetric Analysis. Thermogravimetric analysis (TGA) was performed using a Mettler Toledo TGA/DSC 1 with STARe software. For all samples, 1
5 mg of CNT/N-CNTs were weighed out and placed in an alumina sample holder. The samples were heated from 25 to 900 °C at a ramp rate of 50 °C min
1 in a mixture of air (60 mL min
1) and N2 (40 mL min
1). Analysis of the data to calculate the remaining Fe2O3 product was performed using STARe software. Electrochemical Analysis. Electrochemical analyses were performed using either an Autolab PGSTAT30 potentiostat interfaced with GPES software (v. 4.9) or a CH 700 potentiostat (CH Instruments) running CHI software (v. 3.01). Unless 20003

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C otherwise indicated, an Hg/Hg2SO4 (saturated K2SO4) reference electrode (CH Instruments, E0 = +0.640 V vs NHE) was used for all experiments presented herein, and all potentials are reported versus this reference. Due to their hydrophobic nature, the CNT/N-CNT working electrodes were wetted in an O2-saturated environment by cycling 10 times from 0.0 to
1.0 V vs Hg/Hg2SO4. Wetting was ensured by the emergence of reproducible electrochemical activity and confirmed visually. All experiments were performed a minimum of three times, with the reported standard deviations resulting from these replicates. O2 reduction measurements were conducted in O2-saturated 1 M KNO3 and 1 M Na2HPO4, pH 6.40 ( 0.03, in a custom-built three-electrode cell employing a CNT/N-CNT mat as the working electrode and a Pt mesh counter electrode. Following electrochemical wetting, cyclic voltammetry (CV) was used to determine the ORR potentials (taken as the apex of the reduction peak) for each CNT/N-CNT mat by cycling from 0.0 V until the O2 reduction peak was observed (ranging from
0.4 to
1.0 V). The electrode was first cycled 30 times in saturated O2 to ensure reproducibility of the peak potential, then a single scan was taken at 20 mV/s. Peroxide disproportionation rates were determined electrochemically using a 0.5 cm diameter glassy carbon (GC) disk electrode (PINE Instruments AFE2M050GC) and rotator (PINE Instruments AFMSRX SN 1250) apparatus to construct polarization curves following the methods of Fu, et al.42 Before each experiment, the GC electrode was polished to a mirror finish with 0.3 and 0.05 μm alumina slurries successively on microcloth (Buehler) and sonicated in ultra pure H2O (18 MΩ cm) for 15 min. A 0.4 mg mL
1 suspension of CNT/N-CNTs was sonicated for 16 h to ensure homogeneity and consistency with previous methods, and 50 μL of this suspension was dropcast via pipet directly onto the GC electrode, which was then dried under an Ar stream and immediately immersed into the electrolyte solution. This suspension of CNT/N-CNTs adheres strongly to the GC surface without the use of another binder, such as Nafion, which is especially advantageous for kinetic experiments. In addition to the CNT/N-CNT working electrode and Hg/Hg2SO4 reference electrode, a gold wire counter electrode was used in all rotating disk electrode (RDE) experiments. A 125 mL, five-neck glass electrochemical cell containing 100 mL of 1 M electrolyte (KOH, KNO3, H2SO4; Fisher) was employed for these experiments, and the two necks not used for the electrodes were used as a gas inlet and a sealed injection port for the H2O2 injection. Following electrochemical wetting, the electrolyte solution was purged with Ar for 20 min to ensure that the reaction would proceed in an O2-free environment. The electrode was rotated at 1000 rpm, and 4 μL 10 M H2O2 (Fisher, standardized with KMnO4) was injected via microsyringe into the electrochemical cell to achieve a final concentration of 400 μM H2O2, which was determined to be an optimal concentration.43 Following injection, the solution was allowed to mix for 60 s, then a linear sweep from open circuit potential (OCP) spanning (0.4 V was taken at a scan rate of 0.17 mV/s. The reduction sweep (OCP
0.4 V) was taken first, then the electrolyte solution was replaced and purged, a fresh injection was performed, and the oxidation sweep (OCP + 0.4 V) was measured. Electrochemical detection of hydroxyl radicals was performed following the procedure of Hu et al.44 Due to the large capacitive background of the N-CNT electrodes, these experiments were performed only on the 0.0 and 4.0 at. % N-CNT electrode mats. These experiments employed our custom-built, flat-bottomed,

ARTICLE

three-electrode cell and a Pt mesh counter electrode. Detection of the hydroxyl radical is dependent on the conversion of 4-hydroxybenzoic acid (4-HBA) to the 3,4-dihydroxybenzoic acid (3,4-DHBA) in the presence of 3 OH. Control CVs were taken in the background electrolyte, 20 mM PBS buffer pH 6.28 (diluted from 10, Fisher), 2 mM 4-HBA (Sigma-Aldrich), and 2 mM 3,4-DHBA (Sigma-Aldrich) by sweeping from
0.4 to +0.1 V at 50 mV/s. The production of the hydroxyl radical was monitored by taking a CV every 5 min for a 30 min time period following the addition of 2 mM H2O2 to the electrolyte (2 mM 4-HBA in 20 mM PBS). The formation of 3 OH is determined by the presence of the reversible 3,4-DHBA redox peaks in the voltammogram. Gasometric Analysis. Gasometric analysis was performed using a home-built apparatus consisting of an inverted buret submerged in a 500 mL beaker and connected to a sealed 100 mL round-bottom flask via a syringe and tubing. A designated amount of N-CNTs and 50 mL of O2-saturated 1 M electrolyte (KOH, KNO3, H2SO4, Fisher) was placed in the round-bottom flask and stirred continuously to create a homogeneous suspension and ensure that the reaction is mass transport limited. These measurements were performed by normalizing the N-CNT content to 200 μM Fe (3
9 mg N-CNTs), calculated from the TGA results (Supporting Information Figure SF1, Table ST1). The flask was sealed with a rubber stopper and held in a constant 23 °C water bath throughout the analysis. The connection to the buret was made with a 20 G needle attached to PTFE tubing, with a syringe inserted into the flask stopper and the other end inserted into the open end of the inverted buret. The buret and beaker were filled with the same electrolyte used in the reaction flask. A 0.5 mL portion of 5 M H2O2 (0.05 M total H2O2) was injected into the flask via syringe, and the volume displacement in the buret was monitored for 30 min, with measurements taken in 30 s intervals. The rate of H2O2 disproportionation was calculated by measuring the volume of O2 evolved as a function of time, and the concentration of HO2
was determined as [HO2
] = [H2O2]0
(2pO2VO2/RTVsol).8

’ RESULTS AND DISCUSSION As shown in Figure 1, the N-CNTs display an enhanced electrocatalytic response toward O2 reduction with increased current densities and positive potential shifts with N-doping in comparison with undoped CNTs by as much as 11.6 mV/at. % N, indicating that the N-CNTs are more efficiently catalyzing the ORR by decreasing the overpotential. This inherent catalysis by the N-CNTs is consistent with other reports of catalytic enhancements toward ORR of other N-doped nonprecious metal catalysts (NPMCs).45
48 Table 1 lists the O2 reduction potentials (Ep), where Ep is taken as the apex of the reduction peak for undoped and N-CNT electrodes in both electrolytes investigated (1 M KNO3 and 1 M Na2HPO4). Interestingly, although this potential shift trend is observed in both electrolytes, we observe a 2 larger positive shift of the reduction potential in 1 M Na2HPO4 (Figure 1B) to
0.227 ( 0.007 V vs Hg/Hg2SO4 (0.413 V vs NHE), as compared with
0.451 ( 0.009 V vs Hg/Hg2SO4 (0.189 V vs NHE) in 1 M KNO3 (Figure 1A) for 7.4 at. % N-CNT electrodes. Speculatively, the larger Ep shift in the Na2HPO4 electrolyte may be due to the differing anions’ interactions with the FexOy/Fe particles, with the phosphate anion acting as a stabilizing agent.18 The decrease in the current in 1 M Na2HPO4 is explained by a decrease in the O2 solubility in this electrolyte, 1.16 versus 1.98 mM in 1 M 20004

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C

ARTICLE

Table 1. Oxygen Reduction Potentials for Undoped CNT and N-CNT Mats in Neutral Electrolyte, pH 6.40 ( 0.03 Ep (V vs Hg/Hg2SO4)

Ep (V vs NHE)

nitrogen content (at. % N)

1 M KNO3

1 M Na2HPO4
0.88 ( 0.01

1 M KNO3
0.14 ( 0.03

1 M Na2HPO4

0.0


0.78 ( 0.03

4.0


0.489 ( 0.007
0.26 ( 0.01

0.151 ( 0.007


0.24 ( 0.01

5.0


0.48 ( 0.01


0.24 ( 0.01

0.16 ( 0.01

0.40 ( 0.01

6.3


0.46 ( 0.01


0.23 ( 0.02

0.18 ( 0.01

0.41 ( 0.02

7.4


0.451 ( 0.009
0.227 ( 0.007

0.189 ( 0.009

0.413 ( 0.007

0.38 ( 0.01

Figure 2. N 1s peak ratios for nitrogen functionalities (A, B) and Fe 2p peak ratios for iron functionalities (C, D) versus ORR potential in (A, C) 1 M KNO3 and (B, D) 1 M Na2HPO4, pH 6.40 ( 0.03.

Figure 1. Oxygen reduction reaction (ORR) cyclic voltammograms for N-CNT mat electrodes in (A) 1 M KNO3 and (B) 1 M Na2HPO4, pH 6.40 ( 0.03. Scan rate = 20 mV/s.

KNO3.33 In addition, subtle differences in wettability of the freestanding mat electrodes could possibly influence the electroactive surface area. As demonstrated in the Supporting Information (Figure SF2), studies with drop-cast CNT and N-CNTs still show corresponding positive potential shifts in Ep in Na2HPO4, NaC2H3O2, and Na3C6H8O7 in comparison with KNO3, yet with comparable reduction currents for ORR. The influence of supporting electrolyte composition on ORR has been detailed previously.18 To further understand this catalytic effect, the ORR reduction potentials are analyzed in relation to the various surface nitrogen and iron functionalities present in our N-CTS as determined from XPS analysis.9,11 Highly linear relationships (R2 > 0.9) are found between both the pyridinic type and pyrrolic/pyridonic nitrogen coordinations. As shown in Figure 2, the reduction potentials in both 1 M KNO3 (Figure 2A) and 1 M Na2HPO4 (Figure 2B) correlate to a simultaneous increase in pyridinic type nitrogen and decrease in pyrrolic/pyridonic type functionalities, and no significant relationships are observed with the other functionalities. Figure SF3 shows the ORR potentials plotted vs total nitrogen content (at. %) with linear regression showing rate of ORR peak potential shift per at. % N incorporated into N-CNTs for both 1 M KNO3and 1 M Na2HPO4, pH 6.3 ( 0.03. This observation is also in agreement with previous results reported by others45,47,48 and ourselves for ORR.10 Furthermore,

Figure SF4 shows that the ORR potentials do not change after acid washing in 3 M HCl. In fact, the ORR potentials actually shift slightly more positive by the removal of the residual surface Fe by acid washing for N-CNTS in 3 M HCl, and the trend for correlation of positive shift of Ep vs nitrogen content (at. % N) remains the same. The influence of acid washing is a subtle effect that we do not fully understand in terms of the mechanistic details, but the changes in catalytic activity are possibly related to changes to removal of pore blocking species and unstable phases associated with the active site for ORR, as also speculated by others.31,49,50 We have also demonstrated that the density of states at the Fermi level increases with pyridinic nitrogen,11 and other groups have also indicated that the pyridinic coordination is the primary functionality involved in the electrocatalytic enhancement.3,14,21,49
54 In addition, the reduction potential, Ep, is correlated with the Fe
C and Fe(NO) (Fe
C/Fe(NO)) functionalities (Figure 2C, D), suggesting that the Fe
C and Fe
N coordinations with respect to the active carbon and nitrogen sites may play a key role, possibly by increasing the number of active sites and stabilizing surface bound FexOy/Fe species near nitrogen sites. These observations are supported by recent reports on the role of nitrogen functionalities and defects on the stability of Pt and Pt
Ru catalyst nanoparticles supported on N-doped highly ordered pyrolytic graphite3,54,55 as well as previous reports that indicate that the ORR activity can be closely correlated with pyridinic functionalities, metal oxide sites, and M
Nx centers, supporting a dual-site reduction mechanism.40,41,49,50 Both the effect of the addition of iron precursors and iron loading on ORR activity have been studied extensively by our ourselves9
11,18 and others31,56,57 in the preparation of NPMCs. Although ORR activity has been reported to increase when Fe content increases from 0.5 to 3 wt. %, the addition of more Fe above this amount does not appear to increase ORR activity. For the N-CNT mat electrodes used in this study, we have not observed any significant changes in ORR activity with increasing Fe content as measured by TGA (Figure SF5) or after acid washing in 3 M HCl (Figure SF4). This data is also supported by our previous studies that show no increase in ORR activity when adding more Fe either during CVD synthesis9 or intentionally during ORR experiments.18 Because HO2
is the key intermediate in the O2 reduction peroxide pathway, it is necessary to understand the kinetics of the 20005

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C

ARTICLE

Table 2. Heterogeneous Hydrogen Peroxide Disproportionation Rates Calculated from Gasometric Analysis at a Constant 200 μM Fe Content nitrogen

khet  10
5

khet  10
5

khet  10
7

content

(cm/s),

(cm/s),

(cm/s),

(at. % N)

1 M KNO3

1 M KOH

1 M H2SO4

4.0

1.1 ( 0.2

1.3 ( 0.7

5.0

1.20 ( 0.03

1.3 ( 0.8

1.2 ( 0.5 11 ( 0.6

6.3 7.4

1.9 ( 0.4 2.6 ( 0.7

1.8 ( 0.4 2.0 ( 0.3

6.4 ( 0.6 6.6 ( 0.9

Table 3. Heterogeneous HO2
Disproportionation Rates Calculated from Steady State Polarization Curves nitrogen

khet  10
5

khet  10
4

content

(cm/s),

(cm/s),

(cm/s),

(at. % N)

1 M KNO3

1 M KOH

1 M H2SO4

Figure 3. Gasometric analysis of N-CNTs in (A) 1 M KNO3, (B) 1 M KOH, and (C) 1 M H2SO4. [H2O2] = 0.05 M.


heterogeneous disproportionation of HO2 at the N-CNTs. Gasometric analysis was performed to calculate the heterogeneous chemical disproportionation rate of HO2
at the N-CNTs; however, the rates at undoped CNTs were too slow to measure via this method. Because of the question regarding the participation of the trace metals used in the growth process, it is especially important to investigate the HO2
disproportionation kinetics in relation to the residual Fe content in the N-CNTs. To control this, the mass of N-CNTs used in each analysis was normalized to 200 μM Fe, ensuring that any measured differences in the forward rate constant would be due to the N-CNTs themselves rather than the presence of excess residual Fe catalyst. Residual Fe in the N-CNTs (Fe2O3 above 750 °C)10 ranges from 6.17 to 16.13% for the undoped to 7.4 at. % N-CNTs, as determined by TGA (Supporting Information Figure SF1, Table ST1). The increase in Fe content with N-doping is consistent with our previous results that indicate that the residual Fe content in the N-CNTs increases with the amount of NH3 carrier gas used during the synthesis.9 The experiments presented herein measure the forward rate constants for the chemical step shown in eq 3 of the peroxide pathway mechanism by monitoring the production of O2 over time in a closed system, where the O2 is produced as a product of the direct disproportionation of HO2
. This is done according eq 8, where the volume of evolved O2 is proportional to the concentration of HO2
in the solution.8   
 2pO2 VO2 HO2 ¼ ½H2 O2 0
ð8Þ RTVsol As indicated by eq 9, the homogeneous rate constant is equal to the slope of the ln[HO2
] versus time plots shown in Figure 3 for 20 min of analysis in neutral (1 M KNO3, Figure 3A), basic

khet  10
5

0.0

0.61 ( 0.09

0.041 ( 0.003

1.2 ( 0.2

4.0 5.0

3.2 ( 0.8 5.1 ( 0.8

2.5 ( 0.1 6.7 ( 0.9

1.5 ( 0.4 1.9 ( 0.9

6.3

7.4 ( 0.9

9.4 ( 0.3

6.2 ( 0.6

7.4

11 ( 2

9( 1

6.1 ( 0.7

(1 M KOH, Figure 3B), and acidic (1 M H2SO4, Figure 3C) environments.
ð9Þ khom ¼ d ln cHO2
=dt These values are easily converted to the heterogeneous rate constant through eq 10, where (A/m)cat is the BET surface area of the N-CNTs.   ð10Þ khet ¼ ½Vsol khom = mcat ðA=mÞcat The calculated rate constants for N-CNTs in all three media are shown in Table 2. From Figure 3, it is apparent that HO2
is disproportionated much faster at N-CNTs containing more nitrogen, regardless of the amount of residual Fe content. This trend holds true for all three electrolytes investigated. Among the three electrolytes, it is apparent that although the rates in 1 M KNO3 (Figure 3A) and 1 M KOH (Figure 3B) are on the same order of magnitude, the rates in 1 M H2SO4 (Figure 3C) are decreased by as much as 2 orders of magnitude relative to the others. In addition, some anomalous behavior is observed for the 5.0 at. % N-CNT in the acidic electrolyte (Figure 3C, inset), which may be attributed to the complexities of the surface-mediated mechanism, perhaps indicating the binding of a surface iron oxide was stronger for these N-CNTs. To further investigate the kinetics of the HO2
disproportionation step, including the rates for the undoped CNTs, the disproportionation rates are measured electrochemically by constructing polarization curves following the analysis methods of Fu et al.42 The heterogeneous rate is calculated from eq 11, where jd is the y-intercept between the oxidation and reduction sides of the polarization curves (Tafel plots) and [mH2O2] is the 20006

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C

ARTICLE

Figure 4. Tafel plots from polarization curves of N-CNTs in (A) 1 M KNO3, (B) 1 M KOH, and (C) 1 M H2SO4. Scan rate = 0.17 mV/s, 1000 rpm, [H2O2] = 400 μM.

starting concentration of peroxide (400 μM).   d½H2 O2  jjd j ¼ kobs m H 2 O2 ¼
dt nF

ð11Þ

surface

The polarization curves were taken while rotating the electrode at 1000 rpm to ensure that the system was mass-transfercontrolled (as with the stirring in the gasometric analysis); however, because no binder was used, the electrode could not be rotated any faster without the CNT/N-CNTs disengaging from the electrode surface. Figure 4 shows the Tafel plots constructed for CNT/N-CNTs in 1 M KNO3 (Figure 4A), KOH (Figure 4B), and H2SO4 (Figure 4C), and the corresponding calculated rate constants are shown in Table 3. Interestingly, the electrochemically measured rates in all three electrolytes are faster than those measured at open circuit (chemically), which can be attributed to the fact that when this rate is measured electrochemically, the electrode is poised at a potential rather than measuring the pure chemical step at open circuit potential. Comparing the rates across the three different electrolytes also yields interesting analysis. Whereas in the gasometric experiments, the rates in neutral (1 M KNO3) and basic (1 M KOH) media were comparable, when applying a potential, we observe an order of magnitude enhancement in the disproportionation rate in 1 M KOH, consistent with previous simulations.10 In addition, although we still observe that the rates in 1 M H2SO4 are slower than in the neutral and basic electrolytes, this difference is considerably less defined than the 2 orders of magnitude difference observed gasometrically. Importantly, this calculation (eq 11) relies on knowing the number of electrons transferred in the reaction. Although many groups assume n = 2 for this step (as indicated in eq 2), it is important to determine this value electrochemically from the Tafel slopes, since the number of electrons transferred yields important information about the rate-determining step of the ORR mechanism.42 We used the experimentally determined slopes (Figure 4) to calculate the empirical Tafel constants, α and b, where for a Nernstian system in which n = 1, we expect b =
0.118 V/dec.58 By calculating these kinetic parameters for all of our data, we are able to determine an average value of α for each N-doping level in each electrolyte, which is then used to calculate b. From these calculations, we determined that n = 2 for the undoped CNTs, and n = 1 for all N-CNTs. This supports the conclusion that the undoped CNTs are undergoing a different reduction process for ORR, that is, two successive two-electron electrochemical reductions, whereas at N-CNTs, ORR involves a more complex chemical disproportionation step as part of an EC0 mechanism. The disproportionation mechanism of the N-CNTs

may be a more complex surface-mediated mechanism involving a surface-bound FexOy/Fe particle stabilized near a N-coordinated site, similar to those mechanisms proposed for metal oxide films or surface-bound metal oxide particles.42,59
61 This hypothesis is further supported by the slower disproportionation rate observed in 1 M H2SO4, where some of the surface-bound FexOy/Fe particle could be dissolved into solution in such strong acid, effectively reducing the number of adsorption sites and slowing the overall disproportionation process (i.e., further slowing the rate-determining step). In a previous report, we determined, using electrochemical simulations, that for the disproportionation of HO2
to effectively “outrun” the second electrochemical reduction step, the effective rate constant from the disproportionation of HO2
must be on the order of 10
3 cm/s or greater.10 This value is in agreement with Jaouen and Dodelet, who also determined that the reaction rate must be on the order of 10
3 cm/s, assuming a rotation rate of 1500 rpm.28 Noting that our rotation speed is only 1000 rpm, which would lead to an overall underestimation of the rate, the heterogeneous disproportionation rates of HO2
reported herein range from 1.5 ( 0.4  10
5 cm/s (4.0 at. % N-CNTs in H2SO4) to 9 ( 1  10
4 cm/s (7.4 at. % N-CNTs in KOH). We note that our values are 2 orders of magnitude faster than those reported by Jaouen and Dodelet for Fe/N/C catalysts, but are still 2 orders of magnitude slower than the predicted threshold of 10
3 cm/s by simulation. This discrepancy may be due in part to the limitations of the simulation, as explained previously,10 in particular with regard to the involvement of adsorbed reactive intermediates such as HO2•, O2•-, and •OH.62 In addition, because the rate-limiting step is likely the adsorption of HO2
, the method used to determine the disproportionation rates yields a further underestimation of the disproportionation rates. As HO2
is generated in situ during the ORR, adsorption is no longer the limiting step; instead, the reaction is limited by the slowest subsequent step, which is likely several orders of magnitude faster than the initial adsorption of HO2
. Furthermore, the experiments used to determine the HO2
disproportionation rates are constructed in such a way that bulk peroxide is added into the cell rather than generated in situ. Therefore, the observed rate will be limited by the slow adsorption step, resulting in a slower estimated rate constant than that predicted by simulation. Taking these two sources of bias into consideration, we can reasonably conclude that the rate constants measured herein are fast enough to produce the observed single reduction peak response for the peroxide pathway mechanism, as seen in Figure 1 and in the Supporting Information (Figure SF2). This represents what is referred to as a “pseudo”-four-electron transfer 20007

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C mechanism, in which, when calculating the number of electrons transferred using traditional methods (i.e., rotating ring disk electrode, RRDE), it would appear as if four electrons are transferred directly at the electrode, an argument that is commonly used for the presence of an N4
Me active site.26 We predict that the rate-limiting step of this surface-mediated O2 reduction mechanism is a one electron reduction of the HO2
intermediate on a surface-bound metal oxide particle (FexOy/Fe site) stabilized near an N-coordination site, similar to those proposed for peroxide disproportionation at an Fe
OOH surface.59,63

Figure 5. Cyclic voltammograms for (A) undoped CNT (0.0 at. % N) and (B) 4.0 at. % N-CNT mats in 20 mM PBS buffer, pH 6.40 ( 0.03, showing detection of hydroxyl radical. Scan rate = 50 mV/s.

ARTICLE

Such mechanisms involve the production of radical intermediates due to the interaction of the iron oxide surface with peroxide (hydroperoxide). To determine if such a surfacemediated reaction occurs at the N-CNTs, we sought to detect the production of the •OH intermediate. This was accomplished by exploiting the electrochemical conversion of 4-hydroxybenzoic acid (4-HBA) to 3,4-dihydroxybenzoic acid (3,4-DHBA), which occurs only in the presence of •OH and can easily be detected using CV. The 3,4-DHBA species is formed in a solution of 4-HBA only if •OH is produced during the chemical disproportionation of the H2O2 at the N-CNT electrode. Due to the high capacitance of our electrodes, this experiment could be performed on only the 4.0 at. % N-CNT mats because in the higher doping level N-CNTs, the background capacitance dominates the electrochemical response. Regardless, as evident from Figure 5, the addition of the H2O2 to the undoped mat electrode (Figure 5A) yields no observable redox activity associated with the formation of 3,4-DHBA; however, the 4.0 at. % N-CNT mats show evidence of the 3,4-DHBA redox couple after the addition of the H2O2 to the electrolyte, indicating that •OH is being produced. This result indicates that HO2
is disproportionated at the surface of the N-CNTs as radical intermediates such as • OH, strongly suggesting that the disproportionation mechanism involves a surface-stabilized FexOy/Fe species. In consideration of the experimental results, we propose that the heterogeneous disproportionation of peroxide (eq 3) at N-CNTs proceeds via a surface-mediated mechanism in which HO2
is adsorbed on a surface-bound iron oxide particle stabilized near a N-containing carbon site, (FeIII
OH)s to form [FeIII
O(H2O2)ads
]s (eq 12), which is then converted to [FeII(O2•
)ads]s (eq 13). Evidence for this step is supported by our calculation of the disproportionation rates and the determination that n = 1 at the N-CNTs, suggesting that the rate-limiting step is, in fact, the surface-mediated adsorption of HO2
onto

Scheme 1. Graphical Representation of Proposed ORR Mechanism at CNT versus N-CNT Electrodes with Representative Cyclic Voltammograms for CNTs and 6.3 at. % N-CNTs in 1 M Na2HPO4 (V vs NHE)

20008

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C

ARTICLE

[FeIII
OH]s, which is then converted to [FeII]s and O2•
via a one-electron transfer step (eqs 12, 13). [FeII(O2•
)ads]s then reacts to form [FeIII
OH]s, •OH, and [FeIII(O2•
)ads]s species (eqs 14
16). The hydroxyl radical (•OH) can react with )ads (eq 17), which then can undergo H2O2 to form (HO2••
deprotonation to (O2 )ads; however, in basic solutions, the hydroperoxide radical is favored (eq 18). Finally, HO2• and O2•
can combine at [FeIII
OH]s to form OH
and O2, completing the disproportionation process (eq 19). Scheme 1 shows a simplified graphical version of our proposed mechanism for peroxide disproportionation.     2 FeIII
OH s þ 2HO2
T 2 FeIII
OðH2 O2 Þ
ads s

ð12Þ

    2 FeIII
OðH2 O2 Þ
ads s f 2 FeII ðO2 •
Þads s þ 2H2 O ð13Þ    II  Fe þ H2 O2 f FeIII
OH s þ • OH

ð14Þ

 II    Fe þ O2 þ OH
f FeIII
OHðO2 •
Þads s

ð15Þ

   II  Fe þ • OH f FeIII -OH s

ð16Þ

•

OH þ H2 O2 f H2 O þ ðHO2 • Þads

ðHO2 • Þads T ðO2 •
Þads þ Hþ

pKa ¼ 4:8

ð18Þ

    2 FeIII -OHðHO2 • Þads =ðO2 •
Þads s f 2 FeII s þ 2H2 O=OH
þ 2O2

’ ASSOCIATED CONTENT

bS

Supporting Information. TGA analysis for determining the amount of Fe loading, cyclic voltammograms showing the ORR response in different electrolytes, and plots of the ORR peak potentials correlated to both N content (at. %) and Fe content (wt %) are provided in the Supporting Information. In addition, a comparison of the ORR peak potentials between acidtreated and untreated N-CNT mats is provided. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

ð17Þ


presented herein indicates that the inherent catalytic activity of the N-CNTs (as observed as a significant positive shift in oxygen reduction potential) arises from a dual site mechanism in which both the N-doping and residual FexOy/Fe surface phase particles play key roles.

’ ACKNOWLEDGMENT We thank Dr. Jennifer L. Lyon for assistance with the Supporting Information and helpful discussions. Financial support of this work was provided by the R. A. Welch Foundation (Grant F-1529). The Kratos XPS was funded by the National Science Foundation under Grant CHE-0618242. ’ REFERENCES

ð19Þ

As shown in eqs 12, 13, and 19, this reduction process occurs in two equivalents, ultimately resulting in a total of four electrons transferred during the ORR, although not all four electrons result from the direct electrochemical reduction O2 to OH
(eq 4). While four electrons are in fact being transferred in this reaction, two of them are coming from the initial reduction of oxygen, and the second two result from the interaction with the surface-bound iron oxide particle, as shown.

’ CONCLUSION We propose that the ORR at N-CNTs involves a dual site reduction in which O2 is initially electrochemically reduced in a 2-electron process to form HO2
. This step is then followed by a rapid chemical disproportionation step at a FexOy/Fe surface site stabilized by the presence of pyridinic type surface functionalities. This mechanism is supported by the confirmed presence of the radical intermediate •OH and the HO2
decomposition rates measured by both steady state polarization curves and gasometric analysis. The confirmation of the presence of •OH lends additional support to a surface-mediated mechanism at an iron oxide particle involving peroxy intermediates. The kinetics of the heterogeneous disproportionation of HO2
to OH
and O2 at the N-CNTs is shown to be sufficiently fast in neutral and basic solutions (10
4
10
5 cm/s) to allow this chemical step to overtake the slower electrochemical reduction of HO2
that dominates the ORR at undoped CNTs. Overall, the data

(1) Qu, L.; Liu, Y.; Baek, J.-B.; Dai, L. ACS Nano 2010, 4 (3), 1321–1326. (2) Chen, Z.; Higgins, D.; Chen, Z. Carbon 2010, 48 (11), 3057–3065. (3) Zhou, Y.; Neyerlin, K.; Olson, T. S.; Pylypenko, S.; Bult, J.; Dinh, H. N.; Gennett, T.; Shao, Z.; Hayre, R. O. Energy Environ. Sci. 2010, 3 (10), 1437–1446. (4) Gong, K.; Du, F.; Xia, Z.; Durstock, M.; Dai, L. Science 2009, 323 (5915), 760–764. (5) Lyth, S. M.; Nabae, Y.; Moriya, S.; Kuroki, S.; Kakimoto, M.-a.; Ozaki, J.-i.; Miyata, S. J. Phys. Chem. C 2009, 113 (47), 20148–20151. (6) Tang, Y.; Allen, B. L.; Kauffman, D. R.; Star, A. J. Am. Chem. Soc. 2009, 131 (37), 13200–13201. (7) Lyon, J. L.; Stevenson, K. J. Electrochim. Acta 2008, 53 (23), 6714–6721. (8) Falcon, H.; Carbonio, R. E. J. Electroanal. Chem. 1992, 339 (1
2), 69–83. (9) Maldonado, S.; Morin, S.; Stevenson, K. J. Carbon 2006, 44 (8), 1429–1437. (10) Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B 2005, 109 (10), 4707–4716. (11) Wiggins-Camacho, J. D.; Stevenson, K. J. J. Phys. Chem. C 2009, 113 (44), 19082–19090. (12) Matter, P. H.; Wang, E.; Arias, M.; Biddinger, E. J.; Ozkan, U. S. J. Phys. Chem. B 2006, 110 (37), 18374–18384. (13) Biddinger, E. J.; Deak, D.; Ozkan, U. S. Top. Catal. 2009, 52 (11), 1566–1574. (14) Liu, G.; Li, X.; Ganesan, P.; Popov, B. N. Electrochim. Acta 2010, 55 (8), 2853–2858. (15) Subramanian, N. P.; Kumaraguru, S. P.; Colon-Mercado, H.; Kim, H.; Popov, B. N.; Black, T.; Chen, D. A. J. Power Sources 2006, 157 (1), 56–63. (16) Liu, G.; Li, X.; Lee, J.-W.; Popov, B. N. Catal. Sci. Technol. 2011, 1 (2), 207–217. 20009

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010

The Journal of Physical Chemistry C (17) Tominaga, H.; Ikeda, W.; Nagai, M. Phys. Chem. Chem. Phys. 2011, 13 (7), 2659–2662. (18) Lyon, J. L.; Stevenson, K. J. Langmuir 2007, 23 (22), 11311–11318. (19) Kinoshita, K. Carbon: Electrochemical and Physiochemical Properties; John Wiley & Sons: New York, 1988. (20) Chlistunoff, J. J. Phys. Chem. C 2011, 115 (14), 6496–6507. (21) Kundu, S.; Nagaiah, T. C.; Xia, W.; Wang, Y.; Van Dommele, S.; Bitter, J. H.; Santa, M.; Grundmeier, G.; Bron, M.; Schuhmann, W.; Muhler, M. J. Phys. Chem. C 2009, 113 (32), 14302–14310. (22) Sidik, R. A.; Anderson, A. B.; Subramanian, N. P.; Kumaraguru, S. P.; Popov, B. N. J. Phys. Chem. B 2006, 110 (4), 1787–1793. (23) Kurak, K. A.; Anderson, A. B. J. Phys. Chem. C 2009, 113 (16), 6730–6734. (24) Ikeda, T.; Boero, M.; Huang, S.-F.; Terakura, K.; Oshima, M.; Ozaki, J.-i. J. Phys. Chem. C 2008, 112 (38), 14706–14709. (25) Lefevre, M.; Dodelet, J. P.; Bertrand, P. J. Phys. Chem. B 2002, 106 (34), 8705–8713. (26) Koslowski, U. I.; Abs-Wurmbach, I.; Fiechter, S.; Bogdanoff, P. J. Phys. Chem. C 2008, 112 (39), 15356–15366. (27) Pumera, M.; Miyahara, Y. Nanoscale 2009, 1 (2), 260–265. (28) Jaouen, F.; Dodelet, J.-P. J. Phys. Chem. C 2009, 113 (34), 15422–15432. (29) Bron, M.; Radnik, J.; Fieber-Erdmann, M.; Bogdanoff, P.; Fiechter, S. J. Electroanal. Chem. 2002, 535 (1
2), 113–119. (30) Maruyama, J.; Abe, I. J. Electrochem. Soc. 2007, 154 (3), B297–B304. (31) Wu, G.; Johnston, C. M.; Mack, N. H.; Artyushkova, K.; Ferrandon, M.; Nelson, M.; Lezama-Pacheco, J. S.; Conradson, S. D.; More, K. L.; Myers, D. J.; Zelenay, P. J. Mater. Chem. 2011, 21 (30), 11392–11405. (32) Wang, P.; Wang, Z.; Jia, L.; Xiao, Z. Phys. Chem. Chem. Phys. 2009, 11 (15), 2730–2740. (33) Wiggins-Camacho, J. D.; Stevenson, K. J. Environ. Sci. Technol. 2011, 45 (8), 3650–3656. (34) Yeager, E. Electrochim. Acta 1984, 29 (11), 1527–37. (35) Yang, H.-H.; McCreery, R. L. J. Electrochem. Soc. 2000, 147 (9), 3420–3428. (36) Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B 2004, 108 (31), 11375–11383. (37) Zhou, Y.; Hu, L.; Gruner, G. Appl. Phys. Lett. 2006, 88 (12), 123109/1–123109/3. (38) Trancik, J. E.; Calabrese Barton, S.; Hone, J. Nano Lett. 2008, 8 (4), 982–987. (39) Ng, S. H.; Wang, J.; Guo, Z. P.; Chen, J.; Wang, G. X.; Liu, H. K. Electrochim. Acta 2005, 51 (1), 23–28. (40) Artyushkova, K.; Levendosky, S.; Atanassov, P.; Fulghum, J. Top. Catal. 2007, 46 (3
4), 263–275. (41) Artyushkova, K.; Pylypenko, S.; Olson, T. S.; Fulghum, J. E.; Atanassov, P. Langmuir 2008, 24 (16), 9082–9088. (42) Fu, D.; Zhang, X.; Keech, P. G.; Shoesmith, D. W.; Wren, J. C. Electrochim. Acta 2010, 55 (11), 3787. (43) Lyon, J. L.; Stevenson, K. J. ECS Trans. 2009, 16 (23, Bioelectroanalysis), 1
12. (44) Hu, Y.-L.; Lu, Y.; Zhou, G.-J.; Xia, X.-H. Talanta 2008, 74 (4), 760–765. (45) Wu, G.; More, K. L.; Johnston, C. M.; Zelenay, P. Science 2011, 332 (6028), 443–447. (46) Arbizzani, C.; Righi, S.; Soavi, F.; Mastragostino, M. Int. J. Hydrogen Energy 2011, 36 (8), 5038–5046. (47) Shanmugam, S.; Osaka, T. Chem. Commun. 2011, 47 (15), 4463–4465. (48) Li, H.; Liu, H.; Jong, Z.; Qu, W.; Geng, D.; Sun, X.; Wang, H. Int. J. Hydrogen Energy 2011, 36 (3), 2258–2265. (49) Olson, T. S.; Pylypenko, S.; Atanassov, P.; Asazawa, K.; Yamada, K.; Tanaka, H. J. Phys. Chem. C 2010, 114 (11), 5049–5059. (50) Olson, T. S.; Pylypenko, S.; Fulghum, J. E.; Atanassov, P. J. Electrochem. Soc. 2010, 157 (1), B54–B63. (51) Kundu, S.; Xia, W.; Busser, W.; Becker, M.; Schmidt, D.; Havenith, M.; Muhler, M. Phys. Chem. Chem. Phys. 2010, 12 (17), 4351–4359.

ARTICLE

(52) Rao, C. V.; Cabrera, C. R.; Ishikawa, Y. J. Phys. Chem. Lett. 2010, 1 (18), 2622–2627. (53) Higgins, D.; Chen, Z.; Chen, Z. Electrochim. Acta 2011, 56 (3), 1570–1575. (54) Holme, T.; Zhou, Y.; Pasquarelli, R.; O’Hayre, R. Phys. Chem. Chem. Phys. 2010, 12 (32), 9461–9468. (55) Pylypenko, S.; Queen, A.; Olson, T. S.; Dameron, A.; O’Neill, K.; Neyerlin, K. C.; Pivovar, B.; Dinh, H. N.; Ginley, D. S.; Gennett, T.; O’Hayre, R. J. Phys. Chem. C 2011, 115 (28), 13667–13675. (56) Lefevre, M.; Proietti, E.; Jaouen, F.; Dodelet, J.-P. Science 2009, 324 (5923), 71–74. (57) Jaouen, F.; Dodelet, J.-P. Electrochim. Acta 2007, 52 (19), 5975–5984. (58) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; John Wiley & Sons, Inc.: New York, 2001. (59) Lin, S.-S.; Gurol, M. D. Environ. Sci. Technol. 1998, 32 (10), 1417–1423. (60) Perez-Benito, J. F. J. Phys. Chem. A 2004, 108 (22), 4853–4858. (61) Roche, I.; Chainet, E.; Chatenet, M.; Vondrak, J. J. Phys. Chem. C 2007, 111 (3), 1434–1443. (62) Appleby, A. J.; Savy, M. J. Electroanal. Chem. Interfacial Electrochem. 1978, 92 (1), 15–30. (63) Haber, F.; Weiss, J. Proc. R. Soc. London, Ser. A 1934, 147, 332–351.

20010

dx.doi.org/10.1021/jp205336w |J. Phys. Chem. C 2011, 115, 20002–20010