S1 DOI: 1.138/NMAT387 Supplementary Information Co 3 O 4 Nanocrystals on Graphene as a Synergistic Catalyst for Oxygen Reduction Reaction Yongye Liang, 1 Yanguang Li, 1 Hailiang Wang, 1 Jigang Zhou, 2 Jian Wang, 2 Tom Regier 2 and Hongjie Dai *1. 1 Department of Chemistry, Stanford University, Stanford, CA 9435, USA and 2 Canadian Light Source Inc., Saskatoon, SK, Canada. Synthesis of Mildly Oxidized Graphene Oxide (mgo) mgo was made by a modified Hummers method using a lower concentration of oxidizing agent. Graphite flakes (1 g, Superior Graphite Co.) were grounded with NaCl (2 g) for 1-15 minutes. Afterwards, the NaCl was washed away by repeatedly rinsing with water in a vacuum filtration apparatus. The remaining graphite was dried in an oven at 7 C for 3 minutes. The dried solid was transferred to a 25 ml round bottom flask. 23 ml of concentrated sulfuric acid was added and the mixture was stirred at room temperature for 24 hours. Next, the flask was heated in an oil bath at 4 C. 1 mg of NaNO 3 was added to the suspension and allowed to dissolve in 5 minutes. This step was followed by the slow addition of 5 mg of KMnO 4 (3 g for Hummers GO), keeping the reaction temperature below 45 C. The solution was allowed to stir for 3 minutes. Afterwards, 3 ml of water was added to the flask, followed by another 3 ml after 5 minutes. After another 5 minutes, 4 ml of water was added. 15 minutes later, the flask was removed from the oil bath and 14 ml of water NATURE MATERIALS www.nature.com/naturematerials 1
DOI: 1.138/NMAT387 S2 and 1 ml of 3% H 2 O 2 were added to end the reaction. This suspension was stirred at room temperature for 5 minutes. It was then repeatedly centrifuged and washed with 5% HCl solution twice, followed by copious amounts of water. The final precipitate was dispersed in 1 ml of water and bath sonicated for 3 min. Any indispensable solid was crushed down by a centrifugation at 5 rpm 5 minutes, and a brown homogeneous supernatant was collected. Synthesis of Co 3 O 4 /rmgo, Co 3 O 4 /N-rmGO Hybrids, rmgo, N-rmGO and free Co 3 O 4 nanoparticles. mgo was collected from the aqueous solution by centrifugation and redispersed in anhydrous ethanol (EtOH). The concentration of the final mgo EtOH suspension was ~.33 mg/ml (concentration of our mgo stock suspension was determined by measuring the mass of the mgo lyophilized from a certain volume of the suspension). For the first step synthesis of hybrid without NH 4 OH, 1.2 ml of.2 M Co(Ac) 2 aqueous solution was added to 24 ml of mgo EtOH suspension, followed by the addition of 1.2 ml of water at RT. The reaction was kept at 8 o C with stirring for 1 h. After that, the reaction mixture from the first step was transferred to a 4 ml autoclave for hydrothermal reaction at 15 o C for 3 h. This hydrothermal step also reduced mgo to rmgo. The resulted product was collected by centrifugation and washed with ethanol and water. The resulting Co 3 O 4 /rmgo hybrid was ~2 mg after lyophilization. To prepare Co 3 O 4 /N-rmGO hybrid with the addition of NH 4 OH, the first step 2 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S3 reaction mixture was prepared by adding 1.2 ml of.2 M Co(Ac) 2 aqueous solution to 24 ml of mgo EtOH suspension, followed by the addition of.5 ml of NH 4 OH (3% solution) and.7 ml of water at RT. The following steps were the same as above. The resulting Co 3 O 4 /N-rmGO hybrid was ~2 mg after lyophilization. The mass ratio of graphene in the hybrid was determined by thermal-gravimetric analysis, in which the hybrid material was heated in air at 5 for 2 hours and a weight loss of ~3 % was measured. This corresponded to the removal of graphene from the hybrid by oxidation. Co 3 O 4 was about 7% by mass (~2% by atom) in our hybrid. rmgo was made through the same steps as making Co 3 O 4 /rmgo without adding any Co salt in the first step. N-rmGO was made through the same steps as making Co 3 O 4 /N-rmGO without adding any Co salt in the first step. This produced N-doped reduced GO with N clearly resolved in the GO sample by XPS (Fig.S2b). Free Co 3 O 4 nanoparticle was made through the same steps as making Co 3 O 4 /N-rmGO without adding any mgo in the first step. Sample preparation for SEM, TEM and XRD SEM samples were prepared by drop-drying the samples from their aqueous suspensions onto silicon substrates. TEM samples were prepared by drop-drying the samples from their diluted aqueous suspensions onto copper grids. XRD samples were prepared by drop-drying the samples from their aqueous suspensions onto glass substrates. NATURE MATERIALS www.nature.com/naturematerials 3
DOI: 1.138/NMAT387 S4 XANES Measurements XANES were recorded in the surface sensitive total electron yield (TEY) with use of specimen current. Data were first normalized to the incident photon flux I measured with a refreshed gold mesh at SGM prior to the measurement. After background correction, the XANES are then normalized to the edge jump, the difference in absorption coefficient just below and at a flat region above the edge (3, 565 and 8 ev for C, O and Co respectively). Electrochemical Measurements 1. Cyclic voltammetry (CV). 5 mg of catalyst and 16-16 µl (16 µl for hybrids or Pt/C, 15% of Nafion to catalyst ratio; 16 µl for N-rmGO or rmgo, 1% of Nafion to catalyst ratio) of 5 wt% Nafion solution were dispersed in 1 ml of 3:1 v/v water/isopropanol mixed solvent by at least 3 min sonication to form a homogeneous ink. Then 2.4 μl of the catalyst ink (containing 12 μg of catalyst) was loaded onto a glassy carbon electrode of 3 mm in diameter (loading ~.17 mg/cm 2 ). Cyclic voltammetry (using the pontentiostat from CH Instruments) was conducted in a home-made electrochemical cell using saturated calomel electrode as the reference electrode, a graphite rod as the counter electrode and the sample modified glassy carbon electrode as the working electrode. Electrolyte was saturated with oxygen by bubbling O 2 prior to the start of each experiment. A flow of O 2 was maintained over the electrolyte during the recording of CVs in order to ensure its continued O 2 4 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S5 saturation. The working electrode was cycled at least 5 times before data were recorded at a scan rate of 5mVs 1. In control experiments, CV measurements were also performed in Ar by switching to Ar flow through the electrochemical cell. 2. Rotating disk electrode (RDE) measurement. For the RDE measurements, catalyst inks were prepared by the same method as CV s. 4 μl ink (containing 2 mg catalyst) was loaded on a glassy carbon rotating disk electrode of 5 mm in diameter (Pine Instruments) giving a loading of.1 mg/cm 2.The working electrode was scanned cathodically at a rate of 5 mvs 1 with varying rotating speed from 4 rpm to 225 rpm. Koutecky Levich plots (J -1 vs. ω -1/2 ) in the insets of Figure 2 of the main text were analyzed at various electrode potentials. The slopes of their best linear fit lines were used to calculate the number of electrons transferred (n) on the basis of the Koutecky-Levich equation 1 : 1 1 1 1 1 = + = + J J J Bω J 1/2 L K K B nfc D ν 2/3 1/6 =.62 o( o) K o J = nfkc where J is the measured current density, J K and J L are the kinetic- and diffusionlimiting current densities, ω is the angular velocity, n is transferred electron number, F is the Faraday constant, C o is the bulk concentration of O 2, v is the kinematic viscosity of the electrolyte, and k is the electron-transfer rate constant. For the Tafel plot, the kinetic current was calculated from the mass-transport correction of RDE by: J K J J L = ( J J ) L NATURE MATERIALS www.nature.com/naturematerials 5
DOI: 1.138/NMAT387 S6 3. Rotating ring-disk electrode (RRDE) measurement. For the RRDE measurements, catalyst inks and electrodes were prepared by the same method as RDE s. The ink was dried slowly in air and the drying condition was adjusted by trial and error until a uniform catalyst distribution across the electrode surface was obtained. The disk electrode was scanned cathodically at a rate of 5 mvs 1 and the ring potential was constant at 1.5 V vs RHE. The % HO - 2 and the electron transfer number (n) were determined by the followed equations 2 : - I /N % HO 2 = 2 I I /N I n = 4 I I /N where I d is disk current, I r is ring current and N is current collection efficiency (N) of the Pt ring. N was determined to be.4 from the reduction of K 3 Fe[CN] 6. 4. Oxygen electrode activities on carbon fiber paper. For measurements on carbon fiber paper, the working electrode was prepared by loading ~.24 mg of catalyst (for hybrid catalysts and Pt/C) on 1 cm 2 carbon fiber paper (purchased from Fuel Cell Store) from its 1 mg/ml ethanol dispersion with a 1:1 Nafion-to-catalyst ratio. It was cycled at least 2 times between and -.4 V vs SCE before data were recorded at a scan rate of 5mVs 1 for ORR measurement. To obtain both ORR and OER activities in.1 M KOH, the working electrode was scanned from -.3 V to.7 V vs SCE after ORR measurement. Multiple cycles were recorded for each sample. The initial anodic sweep was showed in Figure 4a. All the data from carbon fiber paper were ir-compensated. 5. RHE calibration We used saturated calomel electrode (SCE) as the reference electrode in all measurements. It was calibrated with respect to reversible hydrogen electrode (RHE). The calibration was performed in the high purity hydrogen saturated electrolyte with a Pt wire as the working electrode. CVs were run at a scan rate of 1 6 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S7 mv s 1, and the average of the two potentials at which the current crossed zero was taken to be the thermodynamic potential for the hydrogen electrode reactions. a).1 M KOH So in.1 M KOH, E (RHE) = E (SCE) +.99 V. b) 1 M KOH So the 1 M KOH, E (RHE) = E (SCE) + 1.51 V. c) 6 M KOH NATURE MATERIALS www.nature.com/naturematerials 7
DOI: 1.138/NMAT387 S8 So in 6 M KOH, E (RHE) = E (SCE) + 1.98 V. 8 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S9 Supplementary Figures (a) 2 nm (b) 1 nm Co 3 O 4 graphene (c) (d) 311 Intensity (a. u.) 22 4 511 44 2 3 4 5 6 2θ (degree) Fig. S1. Co 3 O 4 /rmgo hybrid prepared by the two step reaction. (a) SEM image of Co 3 O 4 /rmgo hybrid deposited on silicon substrate from a suspension in solution. (b) Low magnification and (c) high magnification TEM images of Co 3 O 4 /rmgo hybrid. (d) XRD spectrum of Co 3 O 4 /rmgo hybrid film. Transmission electron microscopy (TEM) revealed smaller particles in Co 3 O 4 /N-rmGO (~4-8 nm in size, Fig. 1b) than Co 3 O 4 /rmgo (~12-25 nm in size) shown here in (b) and (c). NATURE MATERIALS www.nature.com/naturematerials 9
DOI: 1.138/NMAT387 S1 a O Co Co Co Co 3 O 4 /rmgo b Co N-rmGO Co 3 O 4 O C Co O O Co Co Co Co C Co Co O N c 1 8 6 4 2 Binding Energy (ev) N 1s spectrum Co 3 O 4 /N-rmGO N-rmGO d 1 8 6 4 2 Co 2p spectrum Binding Energy (ev) 2P 3/2 2P 1/2 Co 3 O 4 /N-rmGO Co 3 O 4 46 44 42 4 398 396 394 Binding Energy (ev) 81 85 8 795 79 785 78 775 77 Binding Energy (ev) Fig. S2. XPS spectra of (a) Co 3 O 4 /rmgo hybrid and (b) Co 3 O 4 nanocrystal and N-rmGO. N-rmGO was made through the same steps as making Co 3 O 4 /N-rmGO without adding any Co salt in the first step. This produced N-doped reduced GO with N clearly resolved in the GO sample by XPS. (c) High resolution N 1s spectra of Co 3 O 4 /N-rmGO hybrid and N-rmGO. (d) High resolution Co 2p spectra of Co 3 O 4 /N-rmGO hybrid and Co 3 O 4 nanocrystal. The XPS spectra confirmed that N-dopants were on reduced GO sheets and not in Co 3 O 4 nanocrystals. High resolution XPS spectra of the N peak revealed pyridinic and pyrrolic nitrogen species in Co 3 O 4 /N-rmGO and in N-rmGO shown here in (c). 1 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S11 Co 3 O 4 N-rmGO rmgo 5 μa Co 3 O 4 /rmgo Co 3 O 4 /N-rmGO.3.6.9 Potential vs RHE (V) Fig. S3. CVs of Co 3 O 4 nanocrystal, rmgo, N-rmGO, Co 3 O 4 /rmgo and Co 3 O 4 /N-rmGO (all loaded on glassy carbon electrodes with the same mass loading) in oxygen (solid) or argon (dash) saturated.1 M KOH. Free Co 3 O 4 nanocrystals, rmgo or N-rmGO alone exhibited very poor ORR activity. In contrast, Co 3 O 4 /rmgo and Co 3 O 4 /N-rmGO hybrids showed much more positive ORR onset potentials and higher cathodic currents, suggesting synergistic ORR activity of the hybrid. NATURE MATERIALS www.nature.com/naturematerials 11
DOI: 1.138/NMAT387 S12 a Current Density (ma/cm 2 ) 1-1 -2-3 -4-5 -6 4 rpm 625 rpm 9 rpm 1225 rpm 16 rpm 225 rpm -7.2.3.4.5.6.7.8.9 1. Potential vs RHE (V) b J -1 (m 2 /A).4.3.2.75V.7V.65V.6V n = 4..1.6.8.1.12.14.16 ω -1/2 (s 1/2 /rad 1/2 ) Fig. S4.. (a) Rotating-disk voltammogram of Pt/C in O 2 -saturated.1 M KOH at a sweep rate of 5 mv/s and different rotation rates. The catalyst loading was.1 mg/cm 2. (b) Corresponding Koutecky Levich plot (J 1 versus ω.5 ) at different potentials. 12 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S13 a Current Density (ma/cm 2 ) -1-2 -3 4 rpm 625 rpm 9 rpm -4 1225 rpm 16 rpm 225 rpm -5.2.3.4.5.6.7.8.9 1. Potential (V vs RHE) b J -1 (m 2 /A).7.6.5.4.3.2.6.8.1.12.14.16 ω -1/2 (s 1/2 /rad 1/2 ).65V, n = 2.64.6V, n = 2.67.55V, n = 2.76.5V, n =2.86 Fig. S5. (a) Rotating-disk voltammogram of N-rmGO (loading.1 mg/cm 2 ) in O 2 -saturated.1 M KOH at a sweep rate of 5 mv/s and different rotation rates. (b) Corresponding Koutecky Levich lot (J 1 vs. ω.5 ) at different potentials. NATURE MATERIALS www.nature.com/naturematerials 13
DOI: 1.138/NMAT387 S14 Current density (ma/cm 2 ) -1-2 -3-4 -5.7.8.9 1. Potential vs RHE (V) Co 3 O 4 rmgo N-rmGO Fig. S6. Oxygen reduction currents of Co 3 O 4 nanocrystal, rmgo and N-rmGO on carbon fiber paper in O 2 -saturated.1 M KOH. The sample loadings were.24 mg/cm 2. At.7 V, Co 3 O 4 nanocrystal, rmgo and N-rmGO afforded an ORR current density of.12,.19 and 3.5 ma/cm 2 respectively, which were 1-3 orders of magnitude lower than hybrids (Co 3 O 4 /rmgo - 12.3 ma/cm 2 and Co 3 O 4 /N-rmGO - 52.6 ma/cm 2 as shown in Fig. 4a), confirming a synergetic coupling between two catalytically non-active components in our hybrids for ORR catalysis. 14 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S15 a Current density (ma/cm 2 ) -1-2 -3-4 Co 3 O 4 /N-rmGO Co 3 O 4 /rmgo.4.5.6.7.8.9 1. Potential (V vs RHE) b Potential (V vs RHE).95.9.85.8.75 Co 3 O 4 /N-rmGO Co 3 O 4 /rmgo b = 37 mv/dec b = 4 mv/dec.1 1 1 Current (ma/cm 2 ) Fig. S7. (a) Rotating-disk voltammograms of Co 3 O 4 /rmgo hybrid and Co 3 O 4 /N-rmGO hybrid (loading.1 mg/cm 2 ) in O 2 -saturated 1 M KOH with a sweep rate of 5 mv/s and rotation rate of 16 rpm. (b) Tafel plots of Co 3 O 4 /rmgo and Co 3 O 4 /N-rmGO hybrids derived by the mass-transport correction of corresponding RDE data. NATURE MATERIALS www.nature.com/naturematerials 15
DOI: 1.138/NMAT387 S16 a Current density (ma/cm 2 ) -2-4 -6 Co 3 O 4 /N-rmGO -8 Pt/C - 2% -1 Pt/C - 5% Fe-N/C -12 Pd/C - 1%.7.8.9 1. Potential (V vs RHE) b Normalized Current (%) 1 8 6 4 2 Co 3 O 4 /N-rmGO Pt/C - 2% Pt/C - 5% Fe-N/C Pd/C - 1% 5 1 Time (s) Fig. S8. ORR performance and stability of catalysts. (a) Oxygen reduction polarization curves of Co 3 O 4 /N-rmGO, Pt/C-2% (2 wt% Pt on Vulcan XC-72, Fuel Cell Store), Pt/C-5% (5 wt% Pt on Vulcan XC-72, Fuel Cell Store), Fe-N/C (Prepared followed Ref. [3] method) and Pd/C-1% (Palladium, 1% on activated carbon powder, Alfa Aesar) catalysts (catalyst loading ~.24 mg/cm 2 for all samples) dispersed on carbon fiber paper (CFP) in O 2 -saturated 1 M KOH electrolyte. (b) Chronoamperometric responses (percentage of current retained vs. operation time) of Co 3 O 4 /N-rmGO, Pt/C-2%, Pt/C-5%, Fe-N/C and Pd/C-1% on carbon fiber paper electrodes kept at.7 V vs. RHE in O 2 -saturated 1M KOH electrolytes respectively. Co 3 O 4 /N-rmGO hybrid showed comparable ORR catalytic activity to Pt/C and superior stability in alkaline solutions. 16 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S17 a Current density (ma/cm 2 ) -1-2 -3-4 -5-6 Fig. S9. (a) Rotating-disk voltammogram of Fe-N/C in O 2 -saturated.1 M KOH at a sweep rate of 5 mv/s and different rotation rates. The catalyst loading is 1 µg/cm 2. (b) Corresponding Koutecky Levich plot (J 1 versus ω.5 ) at different potentials. The used Fe-N/C catalyst is high quality, matching the performance reported in the literature 3..4.5.6.7.8.9 1. Potential (V vs RHE) Fe-N/C 4 rpm 625 rpm 9 rpm 1225 rpm 16 rpm 225 rpm b J -1 (m 2 /A).4.35.3.25.2.75 V.7 V.65 V.6 V.75 V: n = 4.8.7 V: n = 3.96.65 V: n = 3.95.6 V: n = 3.92.15.6.8.1.12.14.16 ω -1/2 (s 1/2 /rad 1/2 ) NATURE MATERIALS www.nature.com/naturematerials 17
DOI: 1.138/NMAT387 S18 Co L-edge Co 3 O 4 /N-rmGO Absorption (a.u.) Co 3 O 4 775 78 785 79 795 8 Photon Energy (ev) Fig. S1. Co L-edge XANES of Co 3 O 4 nanocrystal and Co 3 O 4 /N-rmGO hybrid. The increase in the normalized peak area in hybrid compared to Co 3 O 4 nanocrystal indicates increase of unoccupied Co 3d projected state in hybrid, suggesting lower electron density of Co site in hybrid. 18 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S19 Current density (ma/cm 2 ) -2-4 -6-8 -1 2 at% Co 1 at% Co 3 at% Co -12.6.7.8.9 1. Potential vs RHE (V) Fig. S11. Oxygen reduction currents of Co 3 O 4 /N-rmGO hybrids with various Co contents dispersed on carbon fiber paper in O 2 -saturated.1 M KOH. The sample loading was.24 mg/cm 2. Lowering Co loading from 2 at% to 3-1 at% led to systematic reduction in ORR activity, suggesting that the active reaction sites in hybrid materials could be Co oxide species interfaced with GO. NATURE MATERIALS www.nature.com/naturematerials 19
DOI: 1.138/NMAT387 S2 a Current Density (ma/cm 2 ) -4-8 Co -12 3 O 4 /rmgo Co 3 O 4 + rmgo mixture.7.8.9 1. Potential (V vs RHE) b Current Density (ma/cm 2 ) -2-4 Co 3 O 4 /N-rmGO Co 3 O 4 + N-rmGO mixture -6.7.8.9 1. Potential (V vs RHE) Fig. S12. Oxygen reduction currents of hybrids versus mixtures dispersed on carbon fiber paper in O 2 -saturated.1 M KOH. The sample loading was.24 mg/cm 2. Both rmgo and N-rmGO hybrids showed much high activity compared to corresponding physical mixtures, confirming the importance of intimate interaction in the hybrid materials for ORR performance. 2 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S21 Current Density (ma/cm 2 ) -2-4 Co 3 O 4 /N-rmGO Co 3 O 4 /carbon black Co 3 O 4 /GOH -6.7.8.9 1. Potential (V vs RHE) Fig. S13. ORR performance of Co 3 O 4 hybrid catalysts with different carbon materials dispersed on carbon fiber paper in O 2 -saturated.1 M KOH electrolyte (catalyst loading ~.24 mg/cm 2 for all samples). The nitrogen doped hybrid prepared from Hummer s GO (GOH) showed lower activity than Co 3 O 4 /N-rmGO hybrid, which could be due to the lower conductivity of GOH. Carbon black hybrid (composite) also showed much lower activity, likely due to the lack of functional groups in carbon black as anchored sites of the nanoparticles. These results indicated the high conductivity and surface area, as well as suitable anchoring sites of mgo are important for the high activity of the synthesized hybrid materials. NATURE MATERIALS www.nature.com/naturematerials 21
DOI: 1.138/NMAT387 S22 a Current density (ma/cm 2 ) 5 4 3 2 1 initial 5 cycle 1 cycle 15 cycle 1.3 1.4 1.5 1.6 1.7 Potential (V vs RHE) Co 3 O 4 /N-rmGO Fig. S14. OER stability test of Co 3 O 4 /N-rmGO, Co 3 O 4 /rmgo and Co 3 O 4 catalysts dispersed on glassy carbon electrode (with 15% of Nafion as binder) in 1M KOH electrolyte (catalyst loading.1 mg/cm 2 for all samples). Cycles were swept between 1.25 V and 1.65 V at.2 V/s. The anodic sweeps showed in the figures were measured from 1.25 V to 1.65 V at.5 V/s with IR compensation. All three catalysts suffered certain current decrease in the beginning cycles (about 2-3% at 1.65 V), which is mainly due to the blockage of some active sites by the gradual accumulation of evolved O 2 bubbles. The OER current did not decrease significantly after 1 cycles in all three catalysts, suggesting the Co 3 O 4 /N-rmGO and Co 3 O 4 /rmgo hybrid catalysts are inherently stable for OER. c Current density (ma/cm 2 ) 2 1 initial 5 cycle 1 cycle 15 cycle b Current density (ma/cm 2 ) 4 3 2 1 initial 5 cycle 1 cycle 15 cycle 1.3 1.4 1.5 1.6 1.7 Potential (V vs RHE) Co 3 O 4 /rmgo 1.3 1.4 1.5 1.6 1.7 Potential (V vs RHE) Co 3 O 4 22 NATURE MATERIALS www.nature.com/naturematerials
DOI: 1.138/NMAT387 S23 References 1. Bard, A. J. & Faulkner, L. R. Electrochemical Methods: Fundamentals and Aplications. (Wiley, 21). 2. Paulus, U. A., Schmidt, T. J., Gasteiger, H. A. & Behm, R. J. Oxygen reduction on a high-surface area Pt/Vulcan carbon catalyst: a thin-film rotating ring-disk electrode study. J. Electroanal. Chem. 495, 134-145 (21). 3. Meng, H., Jaouen, F., Proietti, E., Lefevre, M. & Dodelet, J. P. ph-effect on oxygen reduction activity of Fe-based electro-catalysts. Electrochem. Commun. 11, 1986-1989 (29). NATURE MATERIALS www.nature.com/naturematerials 23