Probing oxygen reduction and oxygen evolution reactions on bifunctional non-precious metal catalysts for metal–air batteries

Abstract

Non-precious metal (NPM) catalysts comprising cobalt and N-doped multiwalled carbon nanotubes (N-MWCNTs-Co) were synthesized by the solid-state pyrolysis (SSP) of melamine with Co₃O₄. N-MWCNTs-Co acts as an electrocatalyst for both the oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER), and hence can be used in secondary metal–air batteries and in unitized regenerative fuel cells. It is important to study the OER and ORR at high concentrations of KOH as most of the metal–air batteries employ KOH concentrations > 4 M. The OER potential of N-MWCNTs-Co at a current density of 10 mA cm⁻² was 0.6 V vs. Hg/HgO (corresponding to an overpotential of 0.364 V) in 6 M KOH. The activity towards the ORR of N-MWCNTs-Co at −0.1 V vs. Hg/HgO was 2.7 and 0.2 mA cm⁻² in 0.1 M and 6 M KOH respectively. Based on our ORR, OER and impedance studies, we postulate that the non-covalent interactions between the hydrated alkali metal cations and the adsorbed oxygen-based species on the catalyst surface are responsible for the reduction in ORR activity and enhancement in OER activity with increasing KOH concentration. Finally, we provide our initial results using this bifunctional catalyst in a zinc–air battery.

---

1. Introduction

The oxygen reduction reaction (ORR) and oxygen evolution reaction (OER)¹ are two fundamental electrochemical processes that occur in energy conversion/storage systems such as unitized regenerative fuel cells and metal–air batteries. In regenerative fuel cells and metal–air batteries, oxygen is reduced to water in the galvanic mode and water is oxidized back to oxygen in the electrolytic mode. Hence, to design an effective energy system, a bifunctional catalyst with low overpotential towards both ORR and OER is essential.^2–9 Although the redox couple involved is O₂/H₂O in both the reactions, the mechanism of the ORR is quite different from that of the OER.^10,11 Hence, designing a single bifunctional catalyst that is active towards both the ORR and the OER is a challenging task. Platinum (Pt) and Pt-based alloys are the benchmark catalysts for the ORR, however they are poor OER catalysts.¹² In contrast, iridium and ruthenium oxide-based electrocatalysts demonstrate excellent OER activity, but relatively poor ORR activity.¹³ Furthermore, the high cost and scarcity of platinum group and other precious metals hamper the use of such electrocatalysts in energy conversion/storage systems.^14–17

As an alternative to expensive precious metal catalysts, non-precious-metal (NPM) based materials such as perovskites,^18–20 cobalt oxide, manganese oxide and its derivatives^1,21–23 N-doped carbon nanotubes²⁴ and nitrogen-metal-containing carbon materials (hereafter called MNC) have been proposed and investigated.²⁵ Recently, MNC catalysts have been reported to demonstrate high ORR activity, comparable to that of Pt/C catalysts.²⁶ However, the use of an amorphous, conducting carbon source²⁷ to host the ORR active sites within its micropores dilutes the active site density in the resultant MNC catalyst. To meet the required current density, thicker (>100 μm) MNC-catalyst-based electrodes are required,²⁸ which result in high mass-transfer overpotentials. Moreover, the amorphous carbon support used in Pt-based catalysts has been identified as being prone to corrosion at high potentials, typically seen during transient conditions.^29–31 In the case of MNC catalysts, in addition to carbon corrosion, the oxidative decay of the ORR active site by peroxide radicals produced as intermediates during the ORR also contributes to loss in catalytic activity.³² Hence, MNC catalysts ideally need to be designed with a corrosion resistant support and with the ability to reduce oxygen completely (4 electron reduction of oxygen). Nitrogen-doped carbon nanotubes^33–37 and graphene² based MNC catalysts have been explored towards this end. In the literature, Xia et al. reported metal–organic framework-derived N-doped carbon nanotubes as superior bifunctional oxygen electrocatalyst.³⁸

In this study, we have developed N-doped multi-walled carbon nanotubes N-MWCNTs-Co via the solid-state pyrolysis (SSP) of melamine and Co₃O₄. Using this catalyst, we have investigated the effect of non-covalent interactions existing between the hydrated alkali metal cations and the adsorbed oxygen based species generated on the electrode surface during the ORR/OER. Electrochemical impedance spectroscopy (EIS) and ORR polarization analysis were employed for the electrochemical characterization of N-MWCNTs-Co. In the literature, the effect of non-covalent interactions on ORR kinetics has been reported for Pt catalysts.^39,40 Since the metal–air batteries are operated with an electrolyte having a high concentration of KOH, it is essential to study the ORR/OER kinetics of the catalyst in high concentration alkaline media, as we have done in this study.

2. Experimental

2.1 Materials

Cobaltous nitrate was procured from Fisher Scientific. Melamine, 5 wt% Nafion® solution and carbon felt were purchased from Alfa Aesar (Kolkata, India). KOH and LiOH were procured from Rankem, India. High purity oxygen and nitrogen cylinders were acquired from Indo Gas Agency, Chennai, India.

2.2 Synthesis

Co₃O₄ was prepared in-house by oxidative decomposition of cobaltous nitrate in air at 500 °C for 1 h. The obtained Co₃O₄ was in the Fd3m space group and its powder X-ray diffraction pattern matched with JCPDS card no. 76-1802. 140 mg of melamine was vacuum-sealed in a 7 mL quartz ampule along with 30 mg of Co₃O₄ and pyrolyzed at 800 °C for 20, 60 or 120 minutes. As we have found 120 minutes to be the optimum pyrolysis time at 800 °C, catalysts were synthesized by pyrolyzing for 120 minutes at 600, 700 or 900 °C in order to identify the optimum pyrolysis temperature. After pyrolysis, the ampule was cut open and the contents inside were washed with 0.1 M HNO₃ for 6 h, followed by rinsing with copious amount of water. The catalysts thus obtained were labeled as “N-MWCNTs-Co-T-x” with the suffix ‘T’ and ‘x’ being the temperature and duration of the pyrolysis.

2.3 Characterization

The catalyst samples were characterized using powder X-ray diffraction (XRD, Bruker 8 fitted with Cu Kα X-ray source), scanning electron microscopy (FEI Quanta FEG 450), transmission electron microscopy (TEM, JEOL JEM 2100), Fourier transform infrared spectroscopy (FT-IR, JASCO-4100 Type A) and Raman spectroscopy (Bruker FT-Multi Ram). X-ray photoelectron spectroscopy (XPS) of samples N-MWCNTs-Co-800-20 min, N-MWCNTs-Co-800-60 min and N-MWCNTs-Co-800-120 min were recorded on a Kratos AXIS-165 surface analysis system using Al Kα radiation. The analyzer was set at a pass energy of 20 eV for obtaining high-resolution spectra of all the individual elements in each sample tested. The background was subtracted using the Shirley-type background correction and the curves were fitted with Gaussian product functions. The obtained binding energies were calibrated using the C 1s peak at 284.8 eV as the reference.

2.4 Electrochemical evaluation

5.7 mg of the catalyst (N-MWCNTs-Co-800-20 min, N-MWCNTs-Co-800-60 min, N-MWCNTs-Co-800-120 min, N-MWCNTs-Co-600-120 min, N-MWCNTs-Co-700-120 min or N-MWCNTs-Co-900-120 min) was dispersed by sonication in 1.6 mL of water–isopropyl alcohol (5 : 1 v/v) solution to obtain a homogenous, ink-like dispersion. 12.5 μL of Nafion® ionomer (5 wt%) solution was added to the catalyst dispersion to act as binder. The glassy carbon disk of a rotating disk electrode (RDE, 5 mm diameter, Pine Research Instrumentation) was coated with this ink to yield a thin-film electrode with a catalyst loading of 0.2 mg cm⁻². The ORR activity of the catalyst was evaluated using linear sweep voltammetry (LSV) in O₂-saturated KOH solutions (0.1, 1, 4, and 6 M), at a rotation rate of 1600 rpm. For background correction, LSV was performed in N₂-saturated KOH solutions (0.1, 1, 4, and 6 M), at a rotation rate of 1600 rpm. The ORR polarization of N-MWCNTs-Co-800-120 min catalyst was also measured using LSV in 4 M KOH solution containing 0.2 M zinc acetate to assess the interference of zinc on the ORR. An RDE coated with commercial 20 wt% Pt/C catalyst (50 μg_Pt cm⁻²) and 20 wt% IrO₂/C (0.2 mg cm⁻²) were used as the reference. A Biologic VSP-300 Model potentiostat was used to control the electrode potential with respect to the Hg/HgO, (1 M) OH⁻ reference electrode (Hg/HgO).

An accelerated stability test (AST) comprising potential cycling between −0.13 and 0.73 V vs. Hg/HgO (OER region) at 500 mV s⁻¹ was conducted on the N-MWCNTs-Co-800-120 min coated RDE in O₂-saturated 6 M KOH solution to understand the durability and ORR activity of the catalyst after it was employed for the OER.

Electrochemical impedance spectroscopy (EIS) was performed between 100 kHz and 30 mHz with a voltage perturbation of 10 mV_rms at base DC potentials of −0.08, −0.1, −0.12 and −0.14 V vs. Hg/HgO (in the ORR regime). Similarly, EIS was performed at DC potentials of 0.72, 0.74, 0.76, 0.78 V vs. Hg/HgO (in the OER regime).

An electrochemical quartz crystal microbalance (EQCM7290, Metrohm, Autolab) wherein the quartz crystal was coated with a 100 nm polished gold layer was used to study the ORR. The resonance frequency and the sensitivity coefficient of the quartz crystal were 6 MHz and 0.0815 Hz ng⁻¹ cm⁻² respectively. This quartz crystal was coated with the catalyst ink (catalyst loading = 200 μg cm⁻²) and used in the EQCM to follow the change in mass of the electrode during the ORR.

For testing the performance of the N-MWCNTs-Co-800-120 min catalyst in a zinc–air battery, a bipolar air electrode was made by coating a 4 cm² carbon felt (5 mm thick) with appropriate catalyst ink. The catalyst loading on the felt electrode was 3 mg cm⁻². A carbon felt coated with 0.25 g cm⁻² of zinc was used as the anode. 0.2 M zinc acetate dissolved in 6 M KOH was used as the electrolyte. Synthetic air was bubbled near the cathode at 50 mL min⁻¹ during the discharge of the battery.

3. Results and discussions

3.1 Physical and chemical characterizations

Fig. 1(a) and (b) show the SEM micrographs of N-MWCNTs-Co-800-120 min catalyst indicating the formation of fibrous material with a foam-like structure at the tip of the fibers. TEM micrographs (Fig. 1(c)–(e)) indicated the presence of graphene and multiwalled carbon nanotubes (MWCNTs). The diameter of the MWCNTs ranged between 20 nm and 50 nm and its wall structure was composed of many discontinuous graphene sheets, indicating the presence of both short-range and long-range order in the same material. Carbon–hydrogen–nitrogen (CHN) analysis indicated a nitrogen content of about 2 ± 0.3 wt% in the catalyst. We believe that both the graphene and the MWCNTs were doped with various nitrogen functionalities such as primary amine, secondary amine, and pyridinic and pyrolic nitrogen.²⁵ EDAX measured loading of Co, N, and C were 2.3 ± 0.4, 2.0 ± 0.5, 95.6 ± 0.05 respectively. Fig. S1† shows the TEM and STEM images and the corresponding EDAX mapping showing elemental distribution on N-MWCNTs-Co-800-120 min. Cobalt and nitrogen were uniformly present on the MWCNTs in the form of aggregated particles. Fig. 1(c) shows the presence of encapsulated cobalt nanoparticles within a graphene structure. In the literature, pyridinic nitrogen and quaternary nitrogen that are covalently attached to the edges of the conducting carbon matrix has been identified as the ORR active site in alkaline media.⁴¹ However, there are reports that suggest that the transition metal coordinated to pyridinic type of nitrogen also serves as the catalytic site for ORR.^42–44 Fig. S2 in the ESI† presents the active site structures hypothesized in the literature.

Fig. 1 (a) and (b) SEM images, (c)–(e) TEM images and (f) power XRD patterns of N-MWCNT-Co-800-120 min catalyst. (f) Includes the XRD of pyrolyzed melamine.

Fig. 1(f) shows the powder XRD pattern of the catalyst and the pyrolyzed melamine. The features at around 10° and 26.5° indicated the presence of graphene and MWCNTs respectively in the catalyst. Pyrolyzed melamine exhibited many diffraction peaks related to its fragments.⁴⁵ On comparing the XRD of the pyrolyzed melamine and the catalyst, it was understood that the presence of Co₃O₄ promoted the formation MWCNTs and graphene. The FT-IR spectrum⁴⁶ of the catalyst featured the peaks due to N–H, aromatic C=C and C=N, C–H and C=O stretches, confirming the functionalization of the catalyst (Fig. 2(a)). The Raman spectrum of the catalyst exhibited two strong peaks at around 1302 and 1600 cm⁻¹, commonly known as the D and G bands of carbon (Fig. 2(b)). The presence of the D band was associated with the defects/edges⁴⁷ that were likely developed due to the nitrogen/oxygen functionalization on the MWCNTs. The ratio of the intensity of D and G bands was around 1.0, indicating the presence of defects in the catalysts. Fig. S3† provides the XRD and Raman spectra of the N-MWCNTs-Co-800-20 min and N-MWCNTs-Co-800-60 min catalysts. The ratio of the intensity of the XRD features at 26.5° and 10° was used to qualitatively analyze the growing phase with the duration of pyrolysis. The intensity ratio was 1.31, 1.55 and 1.75 for N-MWCNTs-Co-800-20 min, N-MWCNTs-Co-800-60 min and N-MWCNTs-Co-800-120 min respectively, indicating the continuous conversion of N-doped graphene into N-MWCNTs with time. The Raman spectra also showed a change in I_D/I_G value from 1.16 to 1.03 as the pyrolysis duration increased from 20 min to 120 min. Based on CHN analysis, the nitrogen content of the catalyst obtained after 20 min and 60 min of pyrolysis was estimated to be 5.0 ± 0.2 and 3.8 ± 0.2 wt% respectively. In summary, from TEM, XRD, CHN and Raman spectroscopy studies, it was understood that Co₃O₄ catalyzed the formation of functionalized graphene and N-MWCNTs.

Fig. 2 (a) FT-IR spectra (b) Raman spectra of N-MWCNT-Co-800-120 min catalyst.

Fig. S4† shows the TGA profile of the N-MWCNTs-Co-800-120 min catalyst measured in air at 20 °C min⁻¹ ramp rate. About 9.5 wt% of the mass was left. Employing the XRD analysis (inset to Fig. S4†), the left over mass was identified as Co₃O₄. Hence, the Co content of the N-MWCNTs-Co-800-120 min is about 6.9 wt%.

Fig. 3 shows the XPS spectra of N-MWCNTs-Co-800-120 min measured in the C 1s, N 1s and Co 2p regions. The XPS spectra of the other two catalysts are provided in Fig. S5 and S6.† As shown in Fig. 3(a), the C 1s peak can be deconvoluted into four peaks. The peaks at 284.8 eV, 285.9 eV, 286.8 eV and 289.2 eV represent C=C, C=N or C–O, C–N or C–O–C and O–C=O respectively. Fig. 3(b) presents the Co 2p region, which was deconvoluted into three peaks. The peaks at 778.6 eV, 780.1 eV and 781.8 eV are attributed to metallic cobalt, Co₃O₄ and to a satellite peak of Co. The N 1s peak could be clearly deconvoluted into two peaks (Fig. 3(c)). The peaks at 398.8 eV, 401.3 eV are associated with pyridinic N and primary or secondary amines respectively. The XPS N 1s results corroborates well with FT-IR results, confirming the presence of primary/secondary amines (Fig. 2(a)) in N-MWCNTs-Co-800-120 min. Table S1 in the ESI† lists the elemental composition of the catalysts obtained through XPS analysis. The surface nitrogen content decreased by a factor of about 2.5 as the pyrolysis duration of catalyst increased from 20 min to 120 min. This related well with the CHN data, wherein the bulk nitrogen content was reported to decrease from 5 wt% to 2 wt% for a corresponding increase in pyrolysis time. As the duration of pyrolysis is increased, the nitrogen is continuously removed from the catalyst due to the high temperature involved during synthesis of the catalysts. Please note that the XRD predicted the growth of graphitic materials at the cost of graphene and Raman spectroscopy predicted increase in the intensity of I_G in comparison to that of I_D. Possibly, these nitrogen defect sites are removed to facilitate conversion of graphene layers into MWCNTs.

Fig. 3 XPS spectra of N-MWCNT-Co-800-120 min catalyst.

3.2 Electrochemical characterization

The catalyst obtained after 120 min of pyrolysis at 800 °C (N-MWCNTs-Co-800-120 min) exhibited the highest ORR activity in O₂-saturated 0.1 M KOH solution (Fig. S7(a)†). To study the effect of pyrolysis temperature on ORR activity, the ORR polarization curves pertaining to N-MWCNTs-Co-600-120 min, N-MWCNTs-Co-700-120 min, N-MWCNTs-Co-800-120 min and N-MWCNTs-Co-900-120 min were recorded (Fig. S7(b)†) in O₂ saturated 0.1 M KOH solution and N-MWCNTs-Co-800-120 min was found to exhibit highest ORR activity. In the literature, while using high-pressure pyrolysis method, mostly non-precious metal catalyst synthesized at 800 °C have been identified as the best suited catalyst for ORR.^25,48 Hereafter only N-MWCNTs-Co-800-120 min is used for further discussion. Fig. 4(a) shows the ORR polarization curves for N-MWCNTs-Co-800-120 min in various oxygen saturated alkali solutions (0.1, 1, 4, 6 M KOH). With increasing KOH concentration, the ORR onset potential shifted to more negative values, from −0.045 V vs. Hg/HgO in 0.1 M KOH to −0.08 V vs. Hg/HgO in 6 M KOH. The negative shift in potential with increasing concentration of KOH has been reported in the literature⁴⁹ for NPM catalysts. Fig. S8† shows the ORR polarization curves for 20 wt% Pt/C recorded at various KOH solutions and 20 wt% IrO₂/C recorded at 0.1 M KOH solution for comparison. The ORR onset potential of IrO₂/C was much lower in comparison to that of Pt/C. Table 1 lists the activity coefficients and kinematic viscosity for four KOH solutions, and the concentration, diffusion coefficient and reduction potential of oxygen in these solutions. These values were all collected from the literature.^50–52 The oxygen solubility decreases significantly with increase in KOH concentration. The observed limiting current also decreased with KOH concentration, as expected. The ORR can proceed via a four-electron or a two-electron pathway.⁵³ While the four-electron pathway generates OH⁻ ions, the two-electron pathway leads to the formation of peroxide ions. Direct transformation of oxygen to OH⁻ ions by the four-electron pathway is desirable due to its high energy efficiency. The peroxide species produced by the two-electron pathway is corrosive and can lead to premature degradation of the electrode and/or electrolyte. Fig. S9† shows the Koutecky–Levich (KL) plots for the catalyst at all the four KOH concentrations. All the KL lines were linear and parallel at all the potentials studied in the ORR regime. In addition, these lines were parallel to the theoretical line drawn assuming a four-electron process, indicating nearly complete reduction of oxygen over the catalyst surface. Fig. 4(b) shows the kinetic current (i_k), estimated from the KL plot at −0.1 V vs. Hg/HgO, and plotted against the concentration of KOH. Since, the number of electrons transferred was nearly 4, indicating complete reduction of oxygen.

Fig. 4 (a) ORR polarization plots of N-MWCNT-Co-800-120 min catalyst measured in O₂-saturated KOH solutions (0.1 M, 1 M, 4 M and 6 M), (b) plot of kinetic current (i_k) estimated at −0.1 V vs. Hg/HgO versus KOH concentration (c) Nyquist plots obtained using N-MWCNT-Co-800-120 min coated RDE in O₂-saturated 0.1 M KOH solution (d) plot of R_ct1 versus potential in 0.1 M, 1 M, 4 M and 6 M KOH solutions. (e) Plot of R_ct2 versus potential in various KOH solutions, (f) plot of R_ct1 versus potential in various KOH solutions for 20% Pt/C catalyst (g) comparison of ORR polarization plots of N-MWCNT-Co-800-120 min catalyst measured in O₂ saturated 0.1 M KOH and 0.1 M LiOH solutions and (h) comparison of R_ct1 and R_ct2 versus potential in O₂ saturated 0.1 M KOH and 0.1 M LiOH solutions.

Table 1 Activity coefficient and kinematic viscosity (ν) of the four KOH solutions and concentration (C_O2), diffusion coefficient (D_O2) and reduction potential (E_O2/H2O) of oxygen in those solutions

KOH concentration (M)	CO2 (mol cm^−3)	DO2 (cm^2 s^−1)	ν (cm^2 s^−1)	EO2/H2O (V) vs. Hg/HgO	Activity coefficient
0.1	1.1 × 10^−6	1.9 × 10^−5	0.95 × 10^−2	0.360	0.8
1	0.83 × 10^−6	1.65 × 10^−5	0.95 × 10^−2	0.308	0.75
4	0.33 × 10^−6	1.05 × 10^−5	1.23 × 10^−2	0.257	1.35
6	1.7 × 10^−7	7.5 × 10^−6	1.6 × 10^−2	0.234	2.2

---

To obtain more clarity on the mechanism of the ORR, we performed EIS analysis on the catalyst coated RDE (rotating at 1600 rpm) at various DC potentials in the ORR regime. In a nitrogen-saturated solution, only a Warburg-capacitive behavior was observed indicating the absence of any charge-transfer reactions (Fig. S10(a)†). Fig. 4(c) shows the representative Nyquist plots and corresponding Bode phase plots obtained in O₂ saturated 0.1 M KOH solution as a function of potential. The Nyquist plots and corresponding Bode phase plots for the remaining KOH concentrations are provided in Fig. S10(b)–(d).† The equivalent circuit used is shown as inset to Fig. S10(b).† Fig. 4(c) shows the presence of two semicircles at all potentials, indicating two possible charge-transfer steps for the ORR on the catalyst. With increasing overpotential (in the cathodic direction), the second semicircle grows and the first semicircle decreases in diameter. Hence, we believe that the second semicircle was due to a mass-transfer influenced charge transfer step. The best fit to this data was obtained with an impedance circuit corresponding to a transmissive-finite diffusion model.^54–56 We believe that this behavior is due to species diffusion through the attached film of partly reduced oxygen intermediate species ((O₂H)_ads⁻) adsorbed on the surface of the catalyst, which is stabilized by hydrated alkali metal cations (Scheme 1). The reduction process associated with the conversion of oxygen to intermediate species corresponds to the first semicircle in the Nyquist plots. Constant phase element (CPE, Q) is employed to fit the experimental impedance data instead of pure capacitor, as the goodness of it is poor with the later. The physical reason for replacing capacitor with Q is well-established in the literature and it is due to the frequency dependence of the measured electrode–electrolyte capacitor, which is arising due to surface inhomogeneity, roughness and current and potential distribution associated with the electrode geometry.^57–59 The expected phase angle for the pure capacitor is 90°, anything lower than that indicates the presence of CPE. The Bode phase plot shown in the right side of Fig. 4(c) clearly indicates presence of two time constants. The phase angle associated with the high-frequency side peak decreases slightly with overpotential, whereas the phase angle associated with low-frequency side peak is increasing with increase in overpotential. This again confirms the dominance of capacitive contribution arise due to the increase in the surface concentration of (O₂H)_ads⁻ intermediate species adsorbed on the surface of the catalyst with increase in overpotential.

Scheme 1 The growth of (O₂H)_ads⁻–M⁺(H₂O)_x cluster layer with overpotential on the electrode–electrolyte interface.

The results of EQCM measurements performed in oxygen and nitrogen saturated 0.1 M KOH are shown in Fig. S11.† In oxygen saturated electrolyte, as the potential was scanned from 0.015 V to −0.28 V vs. Hg/HgO, a negative change in the frequency was observed at potentials lower than −0.1 V, indicating increase in the mass of the catalyst coated quartz crystal. This increase in mass was attributed to adsorbed intermediate species, such as (O₂H)_ads⁻, produced during the ORR. The change in frequency at −0.275 V vs. Hg/HgO was equivalent to 877 ng cm⁻². In nitrogen-saturated electrolyte, a slight negative change in the frequency was observed at potentials below −0.1 V, which was attributed to the potential induced change in the interfacial region, such as formation of hydroquinone from quinone etc. This result supports the interpretation of the EIS data. Strmcnik et al.³⁹ and Katsounaros et al.⁴⁰ reported the existence of non-covalent binding between OH_ads and hydrated cation clusters (OH_ads–M⁺(H₂O)_x) located at the outer Helmholtz plane (OHP) of the double layer during electrocatalytic reactions over platinum in alkaline media. The presence of this layer of cation clusters is reported to increase the overpotential towards O₂ reduction and to decrease the overpotential towards H₂O₂ oxidation. Since the hydration energy of Li⁺ is higher than that of K⁺, the concentration of OH_ads–Li⁺(H₂O)_x cluster was higher on the catalyst surface at a given overpotential. Hence, the ORR was reported to have higher overpotential in a LiOH solution compared to a KOH solution. Fig. S12† shows the Nyquist plots of Pt/C catalysts in various KOH solutions for comparison. Platinum exhibited only one semicircle⁶⁰ across the ORR regime and the charge transfer resistance decreased with increase in overpotential.

In alkaline media, the ORR on NPM catalyst proceeds via a multistep mechanism comprising adsorption and desorption phenomena, and it is well documented in the literature.⁴³ In Fig. 4(d) and (e), R_ct1 and R_ct2 correspond to the first and second semicircle respectively, and were determined at various KOH concentrations. There was not much change in the R_ct values when the concentration of KOH changed from 0.1 M to 1 M. The inset of Fig. 4(e) shows the change in i_k values as a function of potential for 0.1 M and 1 M KOH solutions. From the R_ct values and the i_k values, no change was observed in the ORR mechanism or the order of the ORR reaction with respect to KOH concentration (zero order) while varying the KOH concentration from 0.1 M to 1 M. However, the observed R_ct values for 4 M and 6 M KOH solutions greatly vary. For comparison, Fig. 4(f) shows the change in R_ct values of Pt/C as a function of potential in the four KOH solutions. The R_ct values of Pt/C were nearly one order magnitude higher in comparison to the R_ct1 values of N-MWCNTs-Co-800-120 min catalyst at potentials close to equilibrium. Similar R_ct values have been reported for Pt/C in the literature.⁵⁵

Employing the reduction potential and activity coefficients given in Table 1 and the i_k values obtained from the K–L analysis, exchange current density (i⁰) values and transfer coefficient (α) values were calculated (see Table 2) using the Nernst and Tafel equations given below.

Table 2 Kinetic parameters for the ORR in various KOH solutions

KOH concentration (M)	Transfer coefficient (α)		Exchange current density (i^0/A cm^−2)	Exchange current density (i^0/A cm^−2)
	N-MWCNTs-Co-800-120 min	20% Pt/C	N-MWCNTs-Co-800-120 min	20% Pt/C
0.1	0.80	0.44	1.56 × 10^−9	3.6 × 10^−6
1	0.86	0.78	2.74 × 10^−9	9.51 × 10^−8
4	1.16	0.93	1.70 × 10^−10	1.38 × 10^−8
6	1.30	0.94	1.58 × 10^−11	1.41 × 10^−8

---

The i⁰ values for 0.1 and 1 M KOH were in the order of 10⁻⁹ A cm⁻² and decreased to 10⁻¹⁰ and 10⁻¹¹A cm⁻² in case of 4 M and 6 M KOH respectively, indicating a change in surface coverage by the species present in aqueous KOH solution (water, hydrated potassium ion and hydroxide). From Fig. 4(a), a negative shift in the onset of ORR with KOH concentration could be observed. For comparison, the i⁰ and α values of 20 wt% Pt/C catalyst are also reported in Table 2. The i⁰ values of the Pt/C catalyst were on the order of 10⁻⁸ A cm⁻² in the concentration range of 1 M to 6 M KOH. Although the R_ct values of Pt/C are higher than the corresponding values for the N-MWCNTs-Co-800-120 min, the observed ORR activity of Pt/C catalyst was higher than that of the N-MWCNTs-Co-800-120 min catalyst in 4 M and 6 M solutions as the exchange current densities on Pt/C were nearly 2 to 3 orders of magnitude higher to those on N-MWCNTs-Co-800-120 min. However, the difference between the E_1/2 values of Pt/C and N-MWCNTs-Co-800-120 min decreased from about 100 mV to 60 mV as the concentration of KOH was increased from 4 M to 6 M. Hence, at the high KOH concentrations required for zinc–air batteries, the N-MWCNTs-Co-800-120 min catalyst can adequately substitute for Pt/C, especially given its much higher activity for the OER.

To study the effect of cation on activity, we performed ORR polarization (Fig. 4(g)) and impedance analysis for the N-MWCNTs-Co-800-120 min catalyst in 0.1 M LiOH solution. The overpotential for the ORR reaction in this solution was higher in comparison to 0.1 M KOH and this effect was more pronounced in the mixed control and limiting current regions of the ORR curve. The R_ct1 and R_ct2 values measured in 0.1 M KOH and 0.1 M LiOH are shown in Fig. 4(h). Across the potentials studied, the R_ct1 value for LiOH was about 5 ohm-cm² higher than in KOH. Although the activity coefficient of 0.1 M KOH and 0.1 M LiOH are almost the same,⁵⁰ the difference in R_ct1 values indicated cation specific interactions with the catalyst surface even near equilibrium conditions. The difference between the R_ct2 values in LiOH and KOH solutions increased with overpotential, confirming the high level of interaction between the ORR intermediates adsorbed on the catalyst surface and hydrated Li⁺, compared to hydrated K⁺. Based on the above analysis, we suggest the following inner-sphere electron transfer mechanism consisting of multi-step elementary electrochemical reactions for the ORR on the N-MWCNTs-Co-800-120 min catalyst surface.

O₂ + active site ↔ O_2ads (I)

O_2ads + e⁻ → O_2ads⁻ (II)

2O_2ads⁻ + H₂O → (O₂H)_ads⁻ + O₂ + OH⁻ (III)

(O₂H)_ads⁻ + H₂O + e⁻ → (OH)_ads + 2OH⁻ (IV)

(OH)_ads + e⁻ → OH⁻ (V)

At potentials close to onset potential, the first electron transfer could be rate limiting as confirmed by the impedance analysis, wherein R_ct1 was higher than R_ct2. Near the onset potential, the surface coverage due to (O₂H)_ads⁻ species would be lower. As the overpotential increased (closed circuit conditions) more and more (O₂H)_ads⁻ species would form and would gradually begin to cover the electrode surface if the reduction of (O₂H)_ads⁻ species into (OH)_ads turns into the rate-determining step. The layer of (O₂H)_ads⁻ can be stabilized by the hydrated alkali metal cations, via electrostatic interactions, forming a (O₂H)_ads⁻–M⁺(H₂O)_x type of cluster layer (Scheme 1). A layer of such clusters would impede the oxygen and electrolyte mass-transport to the electrode surface. This cluster is in a quasi-specifically adsorbed state as the alkali metal ion contacts the electrode surface via adsorbed (O₂H)⁻. Hence, at potentials lower than −0.08 V vs. Hg/HgO, step (IV) could be the rate-limiting step as the tunneling of electron and insertion of H₂O into the activated complex present in the cluster layer could be an energy intensive process. Moreover, the availability of water at the catalyst–electrolyte interface could be impacted by the competitive H₂O absorption by the alkali cation to form its hydration shell.

In the literature two kinds of active sites have been reported namely (i) transition metal ion coordinated to the pyridinic nitrogen covalently attached to the edges of the graphitic crystallites⁴¹ and (ii) metal free catalysts having pyridinic and quaternary type of nitrogen tethered to the edges of graphitic crystallites.⁴⁰ To identify the N-MWCNTs-Co-800-120 min catalyst's active site, the ORR activity was measured in presence of 10 mM 2,2′-bipyridine (Bpy) or disodium salt of ethylene diamine tetraacetic acid (EDTA) (Fig. 5). Ligation of cobalt ion, present in the active site, with Bpy/EDTA was expected to modify the ORR activity of the catalyst. Both in presence of Bpy and EDTA, the ORR activity did not change appreciably at the onset potential and in the kinetic region, whereas the limiting current was lowered. This indicated that in the presence of Bpy/EDTA, oxygen diffusion towards the ORR active site was impeded, most likely due to the diffusion barrier offered by adsorbed EDTA/Bpy or EDTA/Bpy co-ordinated to the exposed cobalt site if any. Hence, we inferred that the active site comprised of only carbon and nitrogen, wherein the role of cobalt is to catalyze the formation of carbon nanotubes functionalized with catalytic sites constituted of carbon and nitrogen. TEM results (Fig. 1(c)) indicated the presence of cobalt particles encapsulated by carbon (confirmed by the EDAX and electron diffraction measurements), and no visible cobalt particles were spotted outside the nanotubes.

Fig. 5 ORR activity of N-MWCNT-Co-800-120 min catalyst measured in O₂ saturated 0.1 M KOH solution with and without Bpy and EDTA.

Fig. 6 shows the ORR activity of N-MWCNTs-Co-800-120 min catalyst and Pt in presence and absence of 0.2 M zinc acetate in 4 M KOH solution. It is important (from the perspective of application in a zinc air battery) to check the effect of zinc containing electrolyte on the ORR activity of a bifunctional electrocatalyst. The ORR activity of N-MWCNTs-Co-800-120 min catalyst was not affected by the presence of zinc, however a 50 mV negative shift was observed in case of platinum, indicating a detrimental interference of zinc on the ORR activity of Pt/C due to adsorption effects.

Fig. 6 ORR activity of N-MWCNT-Co-800-120 min catalyst and Pt/C in the presence (open triangle) and absence (open circle) of 0.2 M zinc acetate in 4 M KOH solution.

Fig. 7 shows the OER polarization curves of N-MWCNTs-Co-800-120 min catalyst at four different KOH concentrations and the OER activity measured in the presence of zinc acetate in 4 M KOH solution. The highest OER activity of 10 mA cm⁻² at 0.6 V vs. Hg/HgO (1.59 V vs. RHE) was obtained in 6 M KOH solution. This value is comparable to the reported OER activity values reported in the MNC literature.⁶¹ As the potentials in the OER region are more positive relative to the potentials in the ORR region, the electrostatic attraction between the electrode surface and hydrated metal ion would be minimized, thereby enhancing the OER activity with increase in KOH concentration in the solution. The OER activity of N-MWCNTs-Co-800-120 min catalyst was not affected by the presence of zinc across the potential range studied. Fig. S13(a)–(b)† shows the OER polarization of 20 wt% Pt/C and 20 wt% IrO₂/C respectively, in various KOH solutions. At all KOH concentrations, the OER activity of Pt/C was lower than that of N-MWCNTs-Co-800-120 min. At 0.6 V, in 6 M KOH, an OER current density of 1.35 mA cm⁻² was observed on the Pt/C catalyst, which was about 8 times lower than that of N-MWCNTs-Co-120 min catalyst. At all KOH concentrations, the OER activity on 20 wt% IrO₂/C was comparable to that of N-MWCNTs-Co-800-120 min. For instance, at 0.6 V and 0.7 V, in 6 M KOH, the OER current densities were 10 mA cm⁻² and 64 mA cm⁻² respectively; the corresponding values were 10 mA cm⁻² and 59 mA cm⁻² on N-MWCNTs-Co-800-120 min. Fig. S14† shows the OER activity of 20 wt% Pt/C and 20 wt% IrO₂/C in presence of 0.2 M zinc acetate; due to the interference of zinc, the OER activities of Pt/C and IrO₂/C were further reduced. Fig. S15 and S16† show the Nyquist plots obtained in various KOH solution as a function of potential, relevant to the OER region. At any given potential, the R_ct observed with N-MWCNTs-800-120 min was lower than the Pt/C counterpart. Fig. S17(a) and (b)† presents the summary of R_ct values obtained as a function of potential in various KOH solutions for N-MWCNTs-800-120 min and Pt/C. With increasing KOH concentration, R_ct decreased for both N-MWCNTs-800-120 min and Pt/C.

Fig. 7 OER polarization of N-MWCNT-Co-800-120 min catalyst measured in N₂ saturated KOH solutions (0.1 M, 1 M, 4 M and 6 M) and in 4 M KOH solution containing 0.2 M zinc acetate.

To identify the N-MWCNTs-Co-800-120 min catalyst's active site, the OER activity was measured in presence of 10 mM 2,2′-bipyridine (Bpy) or disodium salt of ethylene diamine tetraacetic acid (EDTA) in 0.1 M KOH (Fig. S18†). Ligation of cobalt ion, present in the active site, with Bpy/EDTA was expected to modify the OER activity of the catalyst. Both in presence of Bpy and EDTA, the OER activity did not change appreciably at the onset potential and in the kinetic region, which confirms that Co is not part of the OER active site.

3.2.1 Discriminating between OER activity and carbon support corrosion on N-MWCNTs-Co-800-120 min. As the potential in the OER region is sufficiently positive to trigger carbon oxidation, it is important to unequivocally show that the current observed at potentials > 0.6 V vs. Hg/HgO is due to the OER and not due to carbon corrosion. To this end, an RRDE experiment was setup, wherein the disk coated with catalyst was subjected a linear sweep experiment in the OER domain in 0.1 M KOH solution, and the Pt ring was maintained at −0.22 V vs. Hg/HgO to reduce the evolved oxygen at its mass-transfer limiting rate. Fig. S19† shows the OER and ORR currents measured at disk and ring respectively, confirming the ability of N-MWCNTs-Co-800-120 min to catalyze the OER reaction. As the oxygen produced at the disk could partially dissolve in the electrolyte, its detection in the Pt ring is feasible but may not be quantitative. Additionally, this experiment does not rule out the concomitant oxidation of carbon alongside the OER. Please note that the reduction of CO₂ at the ring while applying −0.22 V is not feasible and hence only oxygen will be detected.

To resolve this issue, 50 polarization cycles in the OER regime (between 0.35 and 0.8 V at 20 mV s⁻¹) was carried out using Ketjenblack carbon (reference, independent run) and N-MWCNTs-Co-800-120 min (sample) in 100 mL of 0.1 M KOH solution (Fig. S18†). Fig. S20(a)† shows the data obtained on N-MWCNTs-Co-800-120 min; the current obtained increased with the number of cycles. Fig. S20(b)† shows the results of the same experiment performed with Ketjenblack carbon, wherein a decrease in oxidation current observed with the number of cycles. This provides partial evidence for the OER on the N-MWCNTs-Co-800-120 min catalyst and for carbon corrosion on the Ketjenblack carbon, but again does not rule out carbon corrosion in the former. A similar cycling experiment was repeated in 100 mL of 0.1 M KOH solution containing 15 mg of Ca(OH)₂. Upon cycling, the electrode was washed in distilled water, dried and photographed. Photographs of RDE electrodes coated with Ketjenblack and N-MWCNTs-Co-800-120 min upon cycling are shown as insets to Fig. S20.† The Ketjenblack coated RDE electrode turned white due to the precipitation of CaCO₃ as a result of the reaction between electrochemically evolved CO₂ from the carbon surface and Ca(OH)₂ in the solution. Whereas, there was no visible white precipitate found on the N-MWCNTs-Co-800-120 min catalyst coated RDE, supporting the premise of a high selectivity towards the OER over carbon corrosion.

3.2.2 Corrosion behavior of the Ketjenblack carbon and N-MWCNTs-Co-120 min catalyst. Fig. S21† shows the potentiodynamic polarization curves for Ketjenblack carbon and N-MWCNTs-Co-800-120 min catalyst recorded in N₂ saturated 0.1 N KOH solution at 20 mV s⁻¹ san rate. The corrosion potential (E_corr) and corrosion current (I_corr) for the Ketjenblack and N-MWCNTs-Co-800-120 min materials are reported in Table S2.† The E_corr of N-MWCNTs-gvbCo-800-120 min was about 40 mV positive to that of Ketjenblack carbon. Edward et al. reported a polarization resistance (R_p) of 1570.61 ohm cm² for Co–W–CNT in 3% NaCl electrolyte.⁶² In our study, the R_p value (calculated based on the Stern Geary equation⁶³) of Ketjenblack carbon and N-MWCNTs-Co-800-120 min were about 90 and 1048 ohm cm², indicating that the N-MWCNTs-Co-800-120 min was almost 12 times more corrosion resistant than Ketjenblack carbon.

3.2.3 AST testing and demonstration of zinc–air battery. Fig. 8 compares the ORR activity of N-MWCNTs-Co-800-120 min at the beginning of life and after 1000 AST cycles into the OER region in 6 M KOH solution. A 17 mV gain in ORR activity was observed in the case of N-MWCNTs-Co-800-120 min catalyst subsequent to the AST. As the ORR activity was retained even after one thousand AST cycles into the OER region, this catalyst can be regarded as a promising bi-functional catalyst for use in the oxygen electrode of rechargeable metal–air batteries. To demonstrate this promise, a zinc–air battery was assembled and tested at a current density of 0.5 A g⁻¹ for charge–discharge 100 cycles (Fig. S22†).

Fig. 8 ORR polarization curve of the N-MWCNT-Co-800-120 min catalyst before and after 1000 AST cycles in the OER region in 6 M KOH solution.

Fig. S23(a)† shows the photograph of the N-MWCNTs-Co-800-120 min and zinc coated carbon felt. Fig. S23(b)† shows the experimental zinc–air battery setup. Fig. S24† compares the SEM micrograph of the carbon felt and zinc-coated carbon felt. A uniform coating of zinc on the carbon fibers of the felt was observed and EDAX confirms the presence of zinc (Fig. S23(c)†). The average charging voltage of the battery was 2.0 V and corresponding discharge voltage was 1.15 V. Fig. 9(a) compares the first 16 cycles of the zinc–air battery while using N-MWCNTs-Co-800-120 min, 20 wt% Pt/C or 20 wt% IrO₂/C as air electrodes. In the ORR regime (the discharge profile of the zinc–air battery), the cell voltage exhibited by N-MWCNTs-Co-800-120 min was 44 and 83 mV higher (based on 5^th cycle) in comparison to of 20 wt% Pt/C and 20 wt% IrO₂/C respectively, which is in line with the half-cell studies performed in RDE mode. In the OER regime (charge profile of zinc–air battery), the IrO₂/C exhibited about 45 mV lower overpotential in comparison to that of the N-MWCNTs-Co-800C-120 min, which corroborates with the half-cell study (S13(b) and S14†). Fig. 9(b) shows the change in average charge and discharge voltage of the zinc–air battery, employing N-MWCNTs-Co-800C-120 min as air electrode, as a function of the current density. This demonstrates the ability of N-MWCNTs-Co-800C-120 min to work at higher current densities.⁶⁴

Fig. 9 (a) Charge–discharge curves of a zinc–air battery, with air electrodes N-MWCNTs-Co-800-120 min catalyst, 20 wt% Pt/C and 20 wt% IrO₂/C, measured at 0.5 A g⁻¹ current density (one cycle includes 15 min of charge followed by 15 min of discharge) and (b) charge and discharge voltages vs. current density for zinc–air battery using N-MWCNTs-Co-800-120 min.

4. Conclusions

A series of NPM catalysts were synthesized by the SSP route, of which the N-MWCNTs-Co-800-120 min catalyst was investigated in detail. From XRD, CHN and Raman spectroscopy studies, it was understood that the cobalt metal derived from Co₃O₄ catalyzed the formation of N-MWCNTs. As the duration of the pyrolysis was increased from 20 min to 120 min, the proportion of MWCNTs increased at the expense of graphene. The mechanism of ORR on the catalyst involved two different rate limiting steps, depending on the potential. This was attributed to the existence of non-covalent interactions between the adsorbed oxygen species and hydrated metal ions ((O₂H)_ads⁻–M⁺(H₂O)_x cluster layer). This cluster layer improved the OER performance of the catalyst from 0.8 V to 0.6 V vs. Hg/HgO (at 10 mA cm⁻²) as the concentration of KOH was increased from 0.1 M to 6 M. As a comparison, the OER activity of a Pt/C benchmark in 6 M KOH was about 8-fold lower than the N-MWCNTs-Co catalyst at 0.6 V vs. Hg/HgO, while the ORR activity of Pt/C was approximately 2-fold higher at −0.1 V vs. Hg/HgO (∼0.334 V overpotential) in 6 M KOH. Even though the ORR activity of Pt/C catalyst was higher than that of N-MWCNTs-Co-800-120 min, its sensitivity towards the presence of zinc in the electrolyte and poor OER performance in comparison to N-MWCNTs-800-120 min suggested that Pt/C would not be an efficient bifunctional catalyst for zinc–air battery applications (and presumably for other metal–air batteries as well). However, the N-MWCNTs-Co-800-120 min catalyst demonstrated excellent bifunctional performance and robustness; its ORR and OER activity were not affected by the presence of Zn. To demonstrate the suitability of the N-MWCNTs-800-120 min catalyst for both ORR and OER, we have demonstrated 100 cycles of charge–discharge in a zinc–air battery with minimal loss in performance.

Acknowledgements

We thank CSIR-New Delhi for financial support (CHY/1415/326/CSIR-KOTH). We thank Professor R. Dhamodharan for useful discussions on CO₂ detection by using Ca(OH)₂. Author TT thanks the UGC-CSIR for fellowship. The authors thank TEM facility provided by NCCR-IITM. Vijay Ramani would like to acknowledge the Hyosung S. R. Cho Professorship at Illinois Institute of Technology for partially funding the collaboration with IIT-Madras.

References

1. Y. Gorlin and T. F. Jaramillo, J. Am. Chem. Soc., 2010, 132, 13612–13614.
2. Y. Liang, Y. Li, H. Wang, J. Zhou, J. Wang, T. Regier and H. Dai, Nat. Mater., 2011, 10, 780–786.
3. Z. Chen, A. Yu, D. Higgins, H. Li, H. Wang and Z. Chen, Nano Lett., 2012, 12, 1946–1952.
4. E. M. Benbow, S. P. Kelly, L. Zhao, J. W. Reutenauer and S. L. Suib, J. Phys. Chem. C, 2011, 115, 22009–22017.
5. W. G. Hardin, D. A. Slanac, X. Wang, S. Dai, K. P. Johnston and K. J. AStevenson, J. Phys. Chem. C, 2013, 4, 1254–1259.
6. L. Sun, L. Cong, X. Gao, C. Yao, X. Guo, L. Tai, P. Mei, Y. Zeng, H. Xie and R. Wang, J. Phys. Chem. C, 2013, 117, 25890–25897.
7. D. A. Slanac, A. Lie, J. A. Paulson, K. J. Stevenson and K. P. Johnston, J. Phys. Chem. C, 2012, 116, 11032–11039.
8. K. Zhang, L. Zhang, X. Chen, X. He, X. Wang, S. Dong, P. Han, C. Zhang, S. Wang, L. Gu and G. Cui, J. Phys. Chem. C, 2013, 117, 858–865.
9. X. Zhang, Q. Xiao, Y. Zhang, X. Jiang, Z. Yang, Y. Xue, Y.-M. Yan and K. Sun, J. Phys. Chem. C, 2014, 118, 20229–20237.
10. B. S. Yeo and A. T. Bell, J. Am. Chem. Soc., 2011, 133, 5587–5593.
11. Z. Duan and G. Wang, Phys. Chem. Chem. Phys., 2011, 13, 20178.
12. F. D. Kong, S. Zhang, G. P. Yin, N. Zhang, Z. B. Wang and C. Y. Du, Electrochem. Commun., 2012, 14, 63–66.
13. Y. Shao, S. Park, J. Xiao, J. Zhang, Y. Wang and J. Liu, ACS Catal., 2012, 2, 844–857.
14. Y. Zhang, H. Zhang, Y. Ma, J. Cheng, H. Zhong, S. Song and H. Ma, J. Power Sources, 2010, 195, 142–145.
15. G. Chen, S. R. Bare and T. E. Mallouk, J. Electrochem. Soc., 2002, 149, A1092.
16. S. Y. Huang, P. Ganesan, H. Y. Jung and B. N. Popov, J. Power Sources, 2012, 198, 23–29.
17. G. Chen, D. A. Delafuente, S. Sarangapani and T. E. Mallouk, Catal. Today, 2001, 67, 341–355.
18. S. Malkhandi, B. Yang, A. K. Manohar, A. Manivannan, G. K. S. Prakash and S. R. Narayanan, J. Phys. Chem. Lett., 2012, 3, 967–972.
19. J. Suntivich, K. J. May, H. a Gasteiger, J. B. Goodenough and Y. Shao-horn, Science, 2011, 334, 2010–2012.
20. J. Suntivich, H. Gasteiger, N. Yabuuchi, H. Nakanishi, J. B. Goodenough and Y. Shao-Horn, Nat. Chem., 2011, 3, 546–550.
21. K. L. Pickrahn, S. W. Park, Y. Gorlin, H. Lee, T. F. Jaramillo and S. F. Bent, Adv. Energy Mater., 2012, 2, 1269–1277.
22. X. Wu and K. Scott, J. Power Sources, 2012, 206, 14–19.
23. Y. Gorlin and T. F. Jaramillo, J. Electrochem. Soc., 2012, 159, H782–H786.
24. M. Yang, D. Yang, H. Chen, Y. Gao and H. Li, J. Power Sources, 2015, 279, 28–35.
25. R. Kothandaraman, V. Nallathambi, K. Artyushkova and S. C. Barton, Appl. Catal., B, 2009, 92, 209–216.
26. F. Jaouen, E. Proietti, M. Lefevre, R. Chenitz, J.-P. Dodelet, G. Wu, H. T. Chung, C. M. Johnston and P. Zelenay, Energy Environ. Sci., 2011, 4, 114–130.
27. N. Ramaswamy, U. Tylus, Q. Jia and S. Mukerjee, J. Am. Chem. Soc., 2013, 135, 15443–15449.
28. D. Banham, S. Ye, K. Pei, J. Ozaki, T. Kishimoto and Y. Imashiro, J. Power Sources, 2015, 285, 334–348.
29. A. A. Franco, M. Guinard, B. Barthe and O. Lemaire, Electrochim. Acta, 2009, 54, 5267–5279.
30. D. S. Cameron, Platinum Met. Rev., 2009, 53, 147–154.
31. F. N. Büchi, M. Inaba and T. J. Schmidt, Polymer electrolyte fuel cell durability, Springer, 2009.
32. V. Goellner, V. Armel, A. Zitolo, E. Fonda and F. Jaouen, J. Electrochem. Soc., 2015, 162, H403–H414.
33. J. Shui, M. Wang, F. Du and L. Dai, Sci. Adv., 2015, 1, e1400129.
34. Y. Li, M. Gong, Y. Liang, J. Feng, J. E. Kim, H. Wang, G. Hong, B. Zhang and H. Dai, Nat. Commun., 2013, 4, 1805.
35. J. Zhang, Z. Zhao, Z. Xia and L. Dai, Nat. Nanotechnol., 2015, 10, 444–452.
36. F. Cheng and J. Chen, Chem. Soc. Rev., 2012, 41, 2172.
37. A. Sumboja, X. Ge, F. W. T. Goh, B. Li, D. Geng, T. S. A. Hor, Y. Zong and Z. Liu, ChemPlusChem, 2015, 80, 1341–1346.
38. B. Y. Xia, Y. Yan, N. Li, H. Bin Wu, X. Wen, D. Lou and X. Wang, Nat. Energy, 2016, 1, 1–8.
39. D. Strmcnik, K. Kodama, D. van der Vliet, J. Greeley, V. R. Stamenkovic and N. M. Marković, Nat. Chem., 2009, 1, 466–472.
40. I. Katsounaros and K. J. J. Mayrhofer, Chem. Commun., 2012, 48, 6660–6662.
41. G. Liu, X. Li, P. Ganesan and B. N. Popov, Electrochim. Acta, 2010, 55, 2853–2858.
42. M. Lefèvre, E. Proietti, F. Jaouen and J. P. Dodelet, Science, 2009, 324, 71–74.
43. K. M. Palanivelu, V. Prabhakaran, V. K. Ramani and K. Ramanujam, J. Electrochem. Soc., 2015, 162, 475–482.
44. N. Ramaswamy and S. Mukerjee, Adv. Phys. Chem., 2012, 2012.
45. X. Bai, C. Cao and X. Xu, Mater. Sci. Eng., B, 2010, 175, 95–99.
46. L. M. Ombaka, P. G. Ndungu and V. O. Nyamori, RSC Adv., 2015, 5, 109–122.
47. S. Shanmugam and T. Osaka, Chem. Commun., 2011, 47, 4463–4465.
48. N. Leonard, V. Nallathambi and S. C. Barton, J. Electrochem. Soc., 2013, 160, 788–792.
49. X. Li, W. Qu, J. Zhang and H. Wang, J. Electrochem. Soc., 2011, 158, A597–A604.
50. R. E. Davis, G. L. Horvath and C. W. Tobias, Electrochim. Acta, 1967, 12, 287–297.
51. K. E. Gubbins and R. D. Walker, J. Electrochem. Soc., 1965, 112, 469–471.
52. R. C. Weast and R. C. Weast, Handbook of chemistry and physics: a ready-reference book of chemical and physical data, 1969.
53. B. Šljukić, C. E. Banks and R. G. Compton, J. Iran. Chem. Soc., 2005, 2, 1–25.
54. J. S. Kim and S. I. Pyun, J. Solid State Electrochem., 2011, 15, 2447–2452.
55. F. Kuang, D. Zhang, Y. Li, Y. Wan and B. Hou, J. Solid State Electrochem., 2009, 13, 385–390.
56. M. E. Orazem and B. Tribollet, Electrochemical impedance spectroscopy, John Wiley & Sons, 2011, vol. 48.
57. B. Hirschorn, M. E. Orazem, B. Tribollet, V. Vivier, I. Frateur and M. Musiani, Electrochim. Acta, 2010, 55, 6218–6227.
58. A. S. Bondarenko, I. E. L. Stephens, H. A. Hansen, J. P. Francisco, V. Tripkovic, T. P. Johansson, J. Rossmeisl, J. K. Nørskov and I. Chorkendorff, Langmuir, 2011, 27, 2058–2066.
59. P. M. Gomadam and J. W. Weidner, Int. J. Energy Res., 2005, 29, 1133–1151.
60. J. Perez, J. Electrochem. Soc., 1998, 145, 2307–2313.
61. Y. Zhao, K. Kamiya, K. Hashimoto and S. Nakanishi, J. Phys. Chem. C, 2015, 119, 2583–2588.
62. E. Edward Anand and S. Natarajan, J. Mater. Eng. Perform., 2015, 24, 128–135.
63. M. Stern and A. L. Geary, J. Electrochem. Soc., 1957, 104, 559–563.
64. M. Prabu, P. Ramakrishnan and P. Ganesan, Nano Energy, 2015, 15, 92–103.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra13414a

---