(Manganese, Iron, Cobalt, and Nickel) Phosphates for

DOI: 10.1002/cctc.201500952

Full Papers

Activity of Transition-Metal (Manganese, Iron, Cobalt, and Nickel) Phosphates for Oxygen Electrocatalysis in Alkaline Solution Yi Zhan,[a] Meihua Lu,[a] Shiliu Yang,[a] Chaohe Xu,[a] Zhaolin Liu,[b] and Jim Yang Lee*[a] Although transition-metal oxides are common non-platinum group metal catalysts for the industrially important oxygen reduction reaction (ORR) and oxygen evolution reaction (OER), the performance gap between transition-metal oxides and platinum group metal catalysts is still substantial and there is a continuing need to search for alternatives. In this study, transition-metal (Mn, Fe, Co, and Ni) phosphates prepared by a solution chemistry method under ambient conditions are found to display interesting electrocatalytic properties for the ORR and OER in alkaline solution. Among them, manganese phos-

phate is more active than most state-of-the-art manganese oxides for the ORR, and nickel phosphate is as active as the best Ni-based catalysts for the OER. Hence these phosphates can be used as tandem catalysts for rechargeable metal–air batteries in which both the ORR and OER take place. The good performance may be attributed to the stabilization of the catalytic centers by the phosphate framework. This study establishes phosphates as yet another class of highly active low-cost non-platinum group metal alternatives for oxygen electrocatalysis in alkaline solution.

Introduction of 21 mA mg¢1 at 0.75 V (vs. the reversible hydrogen electrode; RHE) in alkaline solution, which is the best-performing simple MnOx catalyst reported to date, even though its performance is approximately 1/10 of that of Pt/C ( 200 mA mg¢1 at 0.90 V).[6] Encouragingly, metal oxides could also be optimized through structural and composition modifications to approach the performance of Ir- or Ru-based catalysts in the OER.[4c,d, 7] For example, the mass activity of the perovskite OER catalyst Ba0.5Sr0.5Co0.8Fe0.2O3¢d is comparable to that of an IrO2 reference catalyst.[7] These promising results indicate that effective nonPGM catalysts for oxygen electrochemical reactions do exist. Recent studies of the OER in neutral solutions (for artificial photosynthesis) suggested transition-metal phosphates as yet another possibility.[1a, 8] In particular Mn3(PO4)2·3 H2O was used as the oxygen-evolving catalyst in a neutral solution. The phosphate framework was found to stabilize the MnIII active sites much better than in oxides, in which the MnIII centers are prone to disproportionation to MnII and MnIV at neutral pH.[8a] The OER activity of Mn3(PO4)2·3 H2O in neutral solution is too low to be useful for energy conversion. However, the more facile oxygen electrochemical reactions in alkaline solution could provide greater opportunities for the development of non-PGM catalysts (the carbonation problem of alkaline electrolytes can be addressed by using CO2-impermeable membranes or by periodic electrolyte replacement).[9] A change of the electrolyte system may also alter the oxygen reaction pathways beneficially, as has been shown in the case of a Ag catalyst that had a rather poor ORR activity in an acid electrolyte but became a fairly good ORR catalyst in an alkaline electrolyte.[10] We present here a fairly extensive study of the activity of phosphates of the first-row transition metals (Mn, Fe, Co, and

The electrochemical oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) are central to energy conversion and storage processes such as water electrolysis, artificial photosynthesis, and rechargeable metal–air batteries.[1] The kinetics of these oxygen electrochemical reactions under typical operating conditions is inherently slow, which necessitates the use of catalysts.[2] Thus far the platinum group metals (PGM) have shown the highest activity in both acid and alkaline solutions.[1e, 3] Specifically Pt-based catalysts are the most active for the ORR, and Ru- and Ir-based catalysts are the most active for the OER. However, the scarcity of PGMs and their high cost are prohibitive to any large-scale application. Therefore, there is a definite need for earth-abundant and low-cost alternatives to PGM catalysts. There are already non-PGM catalysts that can catalyze the ORR and OER in alkaline solution with reasonably good activity.[4] Among them, transition-metal oxides, and Mn-based oxides in particular, have shown a fairly good ORR performance.[5] For example, the free-standing, ultrathin MnO2 nanoflakes prepared by Wei et al. demonstrated a mass activity [a] Y. Zhan, Dr. M. Lu, Dr. S. Yang, Dr. C. Xu, Prof. J. Y. Lee Department of Chemical and Biomolecular Engineering National University of Singapore 10 Kent Ridge Crescent Singapore 119260 (Singapore) E-mail: [email protected] [b] Dr. Z. Liu Institute of Materials Research and Engineering (IMRE) Agency of Science, Technology, and Research (A*STAR) 3 Research Link Singapore 117602 (Singapore) Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/cctc.201500952.

ChemCatChem 2016, 8, 372 – 379

372

Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers (PO4)3¢.[11] The XPS results confirmed that the phosphates were formed by a simple anion exchange without any redox reaction. The transition-metal phosphates were also analyzed by X-ray diffraction (XRD). As-prepared MPO and FPO were well crystallized and their diffraction peaks could all be indexed to Mn3(PO4)2·3 H2O (JCPDS file #03-0426) and Fe3(PO4)2·8 H2O (JCPDS file #30-0662), respectively (Figure 2). No spurious diffraction peaks were detected, and hence the phosphates were phase pure. However, no diffraction peak was found in the XRD patterns of CPO and NPO, and hence this deposition method could only form these phosphates in the amorphous form. This is consistent with previous findings that cobalt phosphate and nickel phosphate could not crystallize in the temperature range of 25–400 8C.[11, 12] Inductively coupled plasma optical emission spectroscopy (ICP-OES) was used to determine the metal/P ratios in CPO and NPO. The ratio was 1.5 for CPO, the same as the metal/P ratio used in the preparation, and as such the molecular formula Co3(PO4)2 may be assigned to CPO. For NPO, the Ni/P ratio was 1.167, which best fits the molecular formula 4 NiHPO4·Ni3(PO4)2. The deviation from the M3(PO4)2 stoichiometry appears to be a characteristic of Ni; Pratt et al. found that the precipitate obtained from the titration of a mixture of NiCl2 and KH2PO4 with NaOH at pH 7 was an amorphous compound with the composition of 4 NiHPO4·Ni3(PO4)2.[12] Thermogravimetric analysis (TGA) and differential thermogravimetric analysis (DTA) were performed in the temperature range of 25–750 8C to determine the hydration numbers of CPO and NPO. Only endothermic events (dehydration of adsorbed water and water of crystallization) were observed for CPO (Figure S2), whereas NPO exhibited a broad exothermic peak at 400 8C likely to be caused by the conden-

Ni and their corresponding phosphates denoted as MPO, FPO, CPO, and NPO, respectively) for oxygen electrocatalysis in an alkaline solution. The metal phosphates were prepared by a facile deposition reaction under ambient conditions and all exhibited very high surface areas. Their morphology and structures were characterized as comprehensively as possible for future reference. Electrochemical measurements identified MPO as a good ORR catalyst with a performance that surpassed the best of simple manganese oxide catalysts. Furthermore, NPO was the most active OER catalyst among the metal phosphates evaluated in this study and had a performance close to that of the best OER catalysts.

Results and Discussion The first-row transition-metal phosphates in this study were prepared by a simple deposition method under ambient conditions in near-neutral solutions. These phosphates are unstable in acid but could be precipitated as an insoluble solid by adding 0.1 m phosphate buffer solution to a metal sulfate solution in 2:3 molar ratio based on the normal tribasic phosphate stoichiometry. X-ray photoelectron spectroscopy (XPS) was used to determine the metal oxidation states in the phosphates (Figure 1, Figure S1). The deconvoluted metal 2p XPS spectra showed characteristic satellite peaks only for Co 2p and Ni 2p and not for Mn 2p and Fe 2p as expected. The binding energies (BEs) of 642.4, 711.9, 781.5, and 856.8 eV correspond well with Mn 2p3/2, Fe 2p3/2, Co 2p3/2, and Ni 2p3/2, respectively, to indicate that the transition metals were all in the +2 oxidation state.[8a, 11] The P 2p spectra of all metal phosphates also showed peaks centered at BE 133.5 eV, which confirmed the presence of

Figure 1. a) Mn 2p XPS spectrum of MPO, b) Fe 2p XPS spectrum of FPO, c) Co 2p XPS spectrum of CPO, and d) Ni 2p XPS spectrum of NPO.

ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

373


Full Papers

Figure 2. XRD patterns of a) MPO, b) FPO, c) CPO, and d) NPO.

sation of NiHPO4 to Ni2P2O7 with some dehydration.[13] XRD measurements indicated that both CPO and NPO remained amorphous after the TGA measurements. The dehydration of CPO led to a weight loss of 16.83 % (which corresponds to 4 moles of water per mole of Co3(PO4)2), and the dehydration and condensation of NPO led to a weight loss of 31.0 % (which corresponds to 24 moles of water per mole of 4 NiHPO4·Ni3(PO4)2 and includes 2 moles of water from the condensation of NiHPO4). Therefore, Co3(PO4)2·4 H2O and 4 NiHPO4·Ni3(PO4)2·22 H2O could be assigned as the molecular formulas of CPO and NPO, respectively. Transmission electron microscopy (TEM) images of the metal phosphates are shown in Figure 3. Both MPO and CPO exhibited an ultrathin sheetlike morphology that extends up to sever-

al micrometers. The ultrathin structure was confirmed independently by atomic force microscopy (AFM), which were used to estimate the thickness of the MPO and CPO nanosheets to be 5 nm (Figure S3). This value is small by comparison with many 2 D materials such as SnS2.[14] The ultrathin architecture suggests that MPO and CPO should have very high surface areas for catalysis. The Brunauer–Emmett–Teller (BET) surface areas measured from N2 adsorption–desorption isotherms confirm this prediction (Figure S4). The BET surface areas were in the region of 300 m2 g¢1 for MPO, CPO, and NPO and 154 m2 g¢1 for FPO. As NPO did not have a filmlike morphology and its particle size was larger than that of FPO, its high surface area implied the presence of mesopores, the outline of which is faintly visible in the TEM images (Figure S5 b). The low surface area of FPO, however, was the result of extensive particle aggregation (Figure S5 a). These values, which are considerably higher than that of the corresponding transition-metal oxides, were accomplished easily without complex surface enlargement technique. A high surface area is desirable as it increases the possibility of a higher proportion of surface defects, which are considered to be more active for oxygen electrocatalysis.[15] The ORR performance of the metal phosphates was first evaluated by cyclic voltammetry (CV) and benchmarked against a reference PGM catalyst (Pt/C) in 0.1 m KOH. Although the ORR onset and peak potentials on MPO were more negative than that of the 20 wt % Premetek Pt/C catalyst (Figure 4 a), they were the most positive of the transition-metal phosphates in this study. The linear sweep voltammetry (LSV) plots shown in Figure 4 b measured at 1600 rpm confirmed the performance in the following order: Pt/C > MPO > other phosphates. Rotating disk electrode (RDE) measurements in the range of 400–2500 rpm (Figure S6) were used to extract the

Figure 3. TEM images of a) MPO, b) FPO, c) CPO, and d) NPO.

ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

374


Full Papers

Figure 4. a) CVs of catalysts in N2- (dashed line) and O2-saturated (solid line) 0.1 m KOH solution; b) ORR polarization curves of different catalysts measured in O2-saturated 0.1 m KOH at 1600 rpm; c) chronoamperogram of MPO and Pt/C in O2-saturated 0.1 m KOH solution at 0.73 V and 1600 rpm; d) Tafel plots of specific activities of the catalysts for ORR.

ORR kinetic parameters under the experimental conditions. The number of electrons involved in the ORR as calculated from the Koutechy–Levich plots were 3.9 for MPO, 2.2–3.5 for FPO, 3.5–3.6 for CPO, and 2.1–2.3 for NPO. Hence the fourelectron reduction pathway was prevalent only on MPO. As a result of its good ORR performance among the phosphates tested, MPO was tested further for operational stability. The test consisted of an extended operation in the ORR region at 0.73 V (the approximate cathode operating potential during the discharge of a metal–air cell) for 40 000 s. A 29 % loss of initial current density was observed for MPO, which compares favorably with the 41 % loss of initial activity for the benchmark Pt/C catalyst (Figure 4 c). Generally, the activity loss is caused by the loss of contact between the catalyst and the support and/or carbon black because of carbon oxidation under the ORR operating conditions.[16] The oxidation of carbon is catalyst dependent, for which PGMs are likely to be the most active.[16b] The fact that the MPO catalyst was used as an unsupported system (and only blended with carbon black) also reduced the risk of catalyst–support detachment. The Tafel plots of intrinsic specific activities are given in Figure 4 d. Intrinsic specific activity is the kinetic current calculated from the Koutechy–Levich equation divided by the catalyst BET surface area. As a result, it is an intrinsic catalytic property free from mass transfer effects and surface area contributions. The Tafel slopes were all similar and close to 59 mV dec¢1 (i.e., 2.3 RT/F), which is a typical value for Pt/C at a low current density. This is often taken to suggest that the first electrochemical step in the ORR is the rate-determining step. In this step, the catalyst surface has an intermediate coverage of adsorbed oxygen species described by Temkin adsorption.[17] Under this ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

condition, the ORR activation barrier is linearly dependent on the surface coverage because of adsorbate–adsorbate interactions. The intrinsic specific activity of MPO was the highest among the transition-metal phosphates. Despite this, the Tafel plot of MPO was still shifted vertically downwards by 120 mV relative to that of Pt/C, which indicates that the intrinsic specific activity of MPO is two orders of magnitude lower than that of Pt/C in the same potential region. The performance gap between Pt and non-PGM alternatives reflects that the higher intrinsic activity of the Pt site (higher onset potential for the ORR) and the active site density of Pt are yet to be matched by non-PGM catalysts.[18] The good ORR performance of MPO relative to other metal phosphates resembles the performance of manganese oxides as the most ORR-active transition-metal oxide catalysts.[19] The ORR performance of MPO was compared with that of state-ofthe-art Mn oxide catalysts (Table S1). The comparison was based on catalyst mass activity in which a high catalyst mass activity is desirable to reduce the mass loading in the air cathode to prevent loading-induced mass transfer limitations.[5a, 6, 20] For a fair comparison, in Table S1 we only include pure manganese oxides and exclude manganese oxides that are doped or modified with a conducting substrate (e.g., reduced graphene oxide). The specific activity of MPO is of the same order of magnitude as that of a-MnO2, the best of the manganese oxides (Table S1). As MnIII is commonly accepted as the active centers of manganese oxides in ORR catalysis,[21] it may also be the catalytic center in MPO. Although MPO was created in the MnII oxidation state, the oxidation of surface metal sites to higher oxidation states could have occurred under the ORR condi375


Full Papers

Figure 5. a) OER polarization curves of different catalysts at 1600 rpm in O2-saturated 0.1 m KOH electrolyte; b) comparison of Tafel slopes from polarization measurements.

tions. This has been found previously for MnO and CoO in which Mn2O3 and Co3O4, respectively, were detected by X-ray absorption near-edge structure (XANES) in ORR catalysis.[22] Interestingly, the mass activity, the most practical performance indicator in real operations, is higher for MPO than the other manganese oxides, although the former has only a moderately large surface area (in the hundreds). The high mass activity of MPO, together with the smaller number of metal sites per gram of MPO (relative to manganese oxides), suggests that the turnover frequency (TOF) was higher in MPO, that is, the MnIII centers in MPO are more active than the MnIII centers in manganese oxides. It is known that MnIII is highly susceptible to Jahn–Teller distortion.[8a, 23] The Jahn–Teller effect can affect the ORR by altering the electronic structure of the catalyst to bind with various oxygen species. For example, Chen and coworkers proved that a tetragonal spinel has a weaker binding with O2 than a cubic spinel.[15a] Hence if the MnIII in MnO6 octahedra is subjected to a strong Jahn–Teller effect, for example, in the case of spinel Co3¢xMnxO4 (1 x 2), the transformation from a cubic spinel to a tetragonal spinel would result in a substantial reduction of the spinel ORR activity.[23] Although PO43¢ anions may appear to be less atom efficient than O2¢ anions on a metal atom basis, PO43¢ anions are more structurally robust than O2¢ anions because of the bulkiness of the phosphate polyhedra and the presence of strong P¢O bonds. Specifically, the induction effect from the covalent P¢O bonds increases the ionic character of the Mn¢O bonds to stabilize the Jahn–Teller effect.[8a, 24] Hence the PO43¢ framework is more able to maintain the d4 MnIII ion in an octahedral geometry with a better tolerance against the Jahn–Teller distortion, a feature that has been utilized in the design of transition-metal-based Li storage compounds for Li-ion batteries.[25] The general stabilization effect of PO43¢ anions against Jahn–Teller distortion has been demonstrated by Nam and coworkers for the OER.[8a] Therefore, the good ORR activity of MPO could be attributed to the stabilization of MnIII centers by the phosphate framework, which thereby facilitates oxygen binding with the MnIII centers.[21] The good performance of MPO together with its ease of preparation (in near-neutral solution under ambient conditions) makes MPO a good alternative to manganese oxides for ORR catalysis. ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

As the OER is as important as the ORR in oxygen electrocatalysis for rechargeable metal–air batteries, the suitability of the transition-metal phosphates for the OER was evaluated. The polarization curves of the transition-metal phosphates in 0.1 m KOH solution are shown in Figure 5 a. A prominent pair of oxidation and reduction peaks emerged in the voltammogram of NPO, which suggests the involvement of Ni redox reactions in the OER that is associated with NiII/NiIII conversion in other Ni-based OER catalysis.[26] Further study is current underway to investigate this observation further. Conversely, no distinct redox features other than the monotonic increase in current density with potential was observed for the other metal phosphates. However, this does not rule out the possibility of concealed redox features under the OER polarization curves.[26a] Among the tested catalysts NPO and Ir/C had the most negative onset potential for the OER, followed by CPO, FPO, and MPO in that order.[4d] Despite the closely matched OER onset potential at 1.50 V, NPO showed a more rapid increase in current density with potential than Ir/C. This rapid increase resulted in a small Tafel slope, which is desired for the OER.[4c] The Tafel slopes measured in this study are shown in Figure 5 b. The Tafel slope of NPO at 48 mV dec¢1 was the smallest among the catalysts, which includes the Ir/C reference catalyst. The OER activity of the transition-metal phosphates could, therefore, be ranked in the order of NPO > CPO > FPO > MPO. The onset potentials and Tafel slopes are comparable to several effective non-PGM OER catalysts. For example, delithiated LiCo0.33Ni0.33Fe0.33O2, one of the best non-PGM OER catalysts, has an onset potential of 1.48 V and a Tafel slope of 35 mV dec¢1. These values also compare favorably with those of Ir/C (1.48 V and 46 mV dec¢1).[4d] Furthermore, the OER stability as measured by continuous CV (an accelerated test for stability)[27] showed that NPO was superior to Ir/C in extended cycling (800 cycles; Figure 6). The performance decay might be caused by carbon black oxidation in extended operation under oxidizing conditions[1b, 28] that deteriorated the integration between the catalyst and carbon black. Unlike the ORR, the OER activities of the phosphate were dependent on the OH¢ concentration with the following “apparent reaction orders” in the pH range of 11–13: 1.0 for NPO and CPO, and 0.8 for both FPO and MPO. Hence the OH¢ concentration is important to the OER performance of transi376


Full Papers Conclusions Transition-metal phosphates (Mn3(PO4)2·3 H2O, Fe3(PO4)2·8 H2O, Co3(PO4)2·4 H2O, and 4 NiHPO4·Ni3(PO4)2·22 H2O; abbreviated as MPO, FPO, CPO, and NPO, respectively) with high surface areas were prepared by a simple solution chemistry method and evaluated as catalysts for the oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) in a pH 13 alkaline solution. Electrochemical measurements revealed that the ORR and OER activities were the highest for MPO and NPO, respectively. Interestingly, MPO was even more active for the ORR than state-of-the-art manganese oxide catalysts and was more stable in the ORR than Pt/C in prolonged operations. Furthermore, the OER activity of NPO surpassed most OER catalysts based on non-platinum group metals and it was also more stable than Ir/C in extended operations. Although the performance of MPO and NPO may be attributed in part to the use of Mn for the ORR and Ni for the OER, their better performance than the corresponding manganese and nickel oxides is most likely because of the phosphate stabilization of the catalytic centers. This study revealed MPO and NPO as strong alternatives to catalysts based on platinum group metals for the ORR and OER, respectively, based on their good activity, high stability, ease of synthesis, and low cost.

Figure 6. OER polarization curves of NPO and Ir/C before and after 800 CV cycles in the potential range of 1.3–1.8 V at a rotation speed of 1600 rpm in O2-saturated 0.1 m KOH solution. The 800 CV cycles were performed at a scan rate of 100 mV s¢1.

tion-metal phosphates. In a neutral phosphate electrolyte, the OER activity of MPO decreased, but the OER specific activity of 0.1 mA cm¢2 catalyst at an overpotential of 680 mV was reasonably close to the value measured by Jin et al. for flower-like Mn3(PO4)2·3 H2O (OER specific activity of 0.316 mA cm¢2 catalyst at the same overpotential).[8a] The potential to reach 10 mA cm¢2 is often used to evaluate the performance of OER catalysts.[3] Our literature survey of state-of-the-art OER catalysts is summarized in Table S2. The OER potential of NPO at 10 mA cm¢2 places this catalyst in the category of “good OER catalysts”. The good OER activity of NPO coincides with the trend that most Ni-based compounds are effective for OER (and some of them even surpass the performance of PGM-based OER catalysts).[4c,d, 26b] We then performed a comparison between NPO and nickel oxides (Table S3). Similar to the comparison between MPO and manganese oxides, only simple nickel oxides were included in the comparison. Based on this comparison, the TOF of NPO is approximately one order of magnitude higher than that of nickel oxides, which again may be attributed to stabilization effects of the phosphate framework. Specifically, b-NiOOH has often been considered as the most active OER component of nickel oxides under the OER conditions.[29] A computational study by Li and Selloni revealed that although the openness of the layer-structured b-NiOOH provides more accessible NiIII active centers during catalysis, Jahn–Teller distortion from the lowspin NiIII (t2g6eg1) could render the less-open tunnel structure of b-NiOOH more energetically favorable to undermine the OER performance.[30] As it is highly likely that the NiII in NPO was oxidized to NiIII before the OER (as shown by the redox features in the OER polarization curve), the NiIII centers in NPO may likewise benefit from the stabilization effect of the phosphate framework relative to the less stable environment of MO6 octahedra in nickel oxides.[8a] A better accommodation of the Jahn–Teller distortion could be the reason for the stronger performance of NPO over the simple NiOx.

ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

Experimental Section Chemicals Cobalt(II) sulfate heptahydrate (CoSO4·7 H2O, > 98.0 %), manganese(II) sulfate monohydrate (MnSO4·H2O, > 99.0 %), iron(II) sulfate heptahydrate (FeSO4·7 H2O, > 99.0 %), nickel(II) sulfate hexahydrate (NiSO4·6 H2O, > 99.0 %), potassium phosphate monobasic (KH2PO4, 99 %), potassium phosphate dibasic (K2HPO4, > 98 %), potassium hydroxide (KOH, > 85 %), and potassium nitrate (KNO3, 99.0 %) were from Sigma–Aldrich, carbon black (Ketjen black EC 600JD) was from Shanghai Tengmin Corp, and platinum on Vulcan XC-72 (Pt/C, 20 wt %) and Ir on Vulcan XC-72 (Ir/C, 20 wt %) were from Premetek and used as received. Deionized water was the universal solvent.

Preparation of metal phosphates Typically, MnSO4·H2O (3 mmol) was dissolved in deionized water (20 mL), followed by the addition of 0.1 m potassium phosphate buffer solution (that contained 61.5 mmol K2HPO4 and 38.5 mmol KH2PO4 in 1 L of deionized water, pH 7.0; 20 mL) under vigorous stirring. A solid product formed instantly, which was removed, washed with water several times, and then freeze-dried. Other phosphates were prepared similarly by replacing MnSO4·H2O with the corresponding metal sulfate.

Morphology and structure characterization TEM was performed by using a JEOL JEM-2010 microscope that operated at an accelerating voltage of 200 kV. XRD patterns were measured by using a Bruker GADDS XRD powder diffractometer with a CuKa source (l = 1.5418 æ) at 40 kV and 30 mA. XPS was performed by using a Kratos Axis Ultra DLD spectrometer, and all binding energies were corrected by referencing the C 1s peak of

377


Full Papers Acknowledgements

adventitious carbon to BE = 284.5 eV. Elemental analysis was based on ICP-OES by using a Thermo Scientific iCAP 6200 Duo Inductively Coupled Plasma Optical Emission Spectrometer. A Shimadzu automatic simultaneous thermal analyzer DTG-60AH was used for TGA and DTA in air in the temperature range of 25–750 8C, which was scanned at 10 8C min¢1. A Quadrasorb SI Surface Area and Pore Size Analyzer was used to obtain N2 adsorption–desorption isotherms at 77 K. The specific surface areas of the samples were calculated by the multipoint BET method. AFM images were obtained by using a Bruker Dimension ICON microscope that was operated in the tapping mode. The AFM examinations were performed in air at RT by using Nanoscope Analysis 1.50r2 software for thickness calculations whenever relevant.

This work was supported by a grant from the Advanced Energy Storage Program of the Science and Engineering Research Council (SERC), Singapore (R-265-000-436-305). Keywords: cobalt · electrochemistry · iron · manganese · nickel [1] a) M. W. Kanan, D. G. Nocera, Science 2008, 321, 1072; b) L. Jçrissen, J. Power Sources 2006, 155, 23; c) M. E. G. Lyons, A. Cakara, P. O’Brien, I. Godwin, R. L. Doyle, Int. J. Electrochem. Sci. 2012, 7, 11768; d) Y. Zhan, C. H. Xu, M. H. Lu, Z. L. Liu, J. Y. Lee, J. Mater. Chem. A 2014, 2, 16217; e) G. Chen, D. A. Delafuente, S. Sarangapani, T. E. Mallouk, Catal. Today 2001, 67, 341. [2] I. Katsounaros, S. Cherevko, A. R. Zeradjanin, K. J. Mayrhofer, Angew. Chem. Int. Ed. 2014, 53, 102; Angew. Chem. 2014, 126, 104. [3] Y. Gorlin, T. F. Jaramillo, J. Am. Chem. Soc. 2010, 132, 13612. [4] a) Y. Liang, Y. Li, H. Wang, J. Zhou, J. Wang, T. Regier, H. Dai, Nat. Mater. 2011, 10, 780; b) Y. Liang, H. Wang, J. Zhou, Y. Li, J. Wang, T. Regier, H. Dai, J. Am. Chem. Soc. 2012, 134, 3517; c) F. Song, X. Hu, Nat. Commun. 2014, 5, 4477; d) Z. Lu, H. Wang, D. Kong, K. Yan, P. C. Hsu, G. Zheng, H. Yao, Z. Liang, X. Sun, Y. Cui, Nat. Commun. 2014, 5, 4345. [5] a) Y. Meng, W. Song, H. Huang, Z. Ren, S. Y. Chen, S. L. Suib, J. Am. Chem. Soc. 2014, 136, 11452; b) F. Cheng, J. Shen, W. Ji, Z. Tao, J. Chen, ACS Appl. Mater. Interfaces 2009, 1, 460; c) F. Cheng, Y. Su, J. Liang, Z. Tao, J. Chen, Chem. Mater. 2010, 22, 898. [6] C. Wei, L. Yu, C. Cui, J. Lin, C. Wei, N. Mathews, F. Huo, T. Sritharan, Z. Xu, Chem. Commun. 2014, 50, 7885. [7] J. Suntivich, K. J. May, H. A. Gasteiger, J. B. Goodenough, Y. Shao-Horn, Science 2011, 334, 1383. [8] a) K. Jin, J. Park, J. Lee, K. D. Yang, G. K. Pradhan, U. Sim, D. Jeong, H. L. Jang, S. Park, D. Kim, N. E. Sung, S. H. Kim, S. Han, K. T. Nam, J. Am. Chem. Soc. 2014, 136, 7435; b) S. W. Lee, C. Carlton, M. Risch, Y. Surendranath, S. Chen, S. Furutsuki, A. Yamada, D. G. Nocera, Y. Shao-Horn, J. Am. Chem. Soc. 2012, 134, 16959. [9] Z. W. Chen, D. Higgins, A. P. Yu, L. Zhang, J. J. Zhang, Energy Environ. Sci. 2011, 4, 3167. [10] B. B. Blizanac, P. N. Ross, N. M. Markovic, Electrochim. Acta 2007, 52, 2264. [11] D. Yang, Z. Lu, X. Rui, X. Huang, H. Li, J. Zhu, W. Zhang, Y. M. Lam, H. H. Hng, H. Zhang, Q. Yan, Angew. Chem. Int. Ed. 2014, 53, 9352; Angew. Chem. 2014, 126, 9506. [12] P. Pratt, F. Bair, G. McLean, Soil Sci. Soc. Am. J. 1964, 28, 363. [13] H. Onoda, T. Sakumura, Mater. Sci. Appl. 2011, 2, 1578. [14] M. Xu, T. Liang, M. Shi, H. Chen, Chem. Rev. 2013, 113, 3766. [15] a) F. Cheng, J. Shen, B. Peng, Y. Pan, Z. Tao, J. Chen, Nat. Chem. 2011, 3, 79; b) S. H. Oh, R. Black, E. Pomerantseva, J.-H. Lee, L. F. Nazar, Nat. Chem. 2012, 4, 1004. [16] a) H. Yano, T. Akiyama, P. Bele, H. Uchida, M. Watanabe, Phys. Chem. Chem. Phys. 2010, 12, 3806; b) Y. Y. Shao, G. P. Yin, Y. Z. Gao, J. Power Sources 2007, 171, 558. [17] a) D. Sepa, M. Vojnovic, A. Damjanovic, Electrochim. Acta 1980, 25, 1491; b) G. Wu, G. F. Cui, D. Y. Li, P. K. Shen, N. Li, J. Mater. Chem. 2009, 19, 6581. [18] H. A. Gasteiger, N. M. Markovic, Science 2009, 324, 48. [19] a) A. D¦bart, A. J. Paterson, J. Bao, P. G. Bruce, Angew. Chem. Int. Ed. 2008, 47, 4521; Angew. Chem. 2008, 120, 4597; b) P. Z˙ûłtowski, D. Drazˇic´, L. Vorkapic´, J. Appl. Electrochem. 1973, 3, 271. [20] F. Cheng, J. Chen, Chem. Soc. Rev. 2012, 41, 2172. [21] I. Roche, E. Chanet, M. Chatenet, J. Vondrk, J. Phys. Chem. C 2007, 111, 1434. [22] a) F. H. B. Lima, M. L. Calegaro, E. A. Ticianelli, J. Electroanal. Chem. 2006, 590, 152; b) Y. Liang, H. Wang, P. Diao, W. Chang, G. Hong, Y. Li, M. Gong, L. Xie, J. Zhou, J. Wang, T. Z. Regier, F. Wei, H. Dai, J. Am. Chem. Soc. 2012, 134, 15849. [23] C. Li, X. Han, F. Cheng, Y. Hu, C. Chen, J. Chen, Nat. Commun. 2015, 6, 7345.

Electrochemical measurements Extrinsic conductivity is important to electrocatalysts[1b] as electrons can only travel through conductive pathways between the catalytically active sites and the current collector. In this study, the transition-metal phosphates are compounded with conducting carbon black. Specifically, the catalyst (1.5 mg) and carbon black (3.5 mg) were added to a Nafion mixture (which contained 1:1:0.08 by volume water, ethanol, and commercial 5 wt % Nafion solution; 1.25 mL) and sonicated for 30 min to form a 4.0 mg mL¢1 catalyst ink. The catalyst ink (5 mL) was drop-cast onto a clean glassy carbon (GC; 5 mm in diameter) electrode to achieve a total solid loading of 0.1 mg cm¢2. For performance comparison, 20 wt % Pt/C and 20 wt % Ir/C were also used to prepare electrodes with a total solid loading of 0.14 mg cm¢2 (to maintain the same metal loading at approximately 0.028 mg cm¢2 in all cases). All electrochemical measurements were performed in 0.1 m KOH solution (except in the study of the pH effects on activity for which the KOH concentration was varied to adjust the pH, a calculated amount of KNO3 was added to keep the ionic strength constant at 0.1 m) using a Ag/AgCl (in aq. 3 m KCl) reference electrode and a Pt foil counter electrode in a standard three-electrode setup. O2 was flowed continuously through the electrolyte to maintain an O2-saturated environment during the measurements. CV, LSV, RDE measurements, and chronoamperometry (CA) were performed by using a computer-controlled Autolab type III potentiostat/galvanostat. Before measurements, the working electrode was first cycled 10 times until a stable response was obtained. The scan rate was 5 mV s¢1 for CV, LSV, and RDE. All potentials in this study were corrected for internal resistance and quoted with respect to the RHE at pH 13 (E = +0.9783 V vs. Ag/AgCl/3 m KCl). The current density was based on the electrode geometric area unless stated otherwise. Equation (1) was used to calculate the catalyst TOF:[26b] TOF ¼

j A n F m

ð1Þ

in which j is the current density [A cm¢2] at a given overpotential, A is the area of the GC electrode, n is the number of electrons exchanged in the overall reaction, F is the Faraday constant (96 485 C mol¢1), and m is the number of moles of catalytic metal deposited on the GC electrode. The TOF calculated in this way represents the lower boundary of the true value as the equation assumes complete accessibility to all of the catalytic metal in catalysis. ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

378


Full Papers [24] A. K. Padhi, K. S. Nanjundaswamy, J. B. Goodenough, J. Electrochem. Soc. 1997, 144, 1188. [25] D. Y. Wang, H. Buqa, M. Crouzet, G. Deghenghi, T. Drezen, I. Exnar, N. H. Kwon, J. H. Miners, L. Poletto, M. Graetzel, J. Power Sources 2009, 189, 624. [26] a) L. Wang, C. Lin, D. Huang, F. Zhang, M. Wang, J. Jin, ACS Appl. Mater. Interfaces 2014, 6, 10172; b) M. Gong, Y. Li, H. Wang, Y. Liang, J. Z. Wu, J. Zhou, J. Wang, T. Regier, F. Wei, H. Dai, J. Am. Chem. Soc. 2013, 135, 8452. [27] M. R. Gao, Y. F. Xu, J. Jiang, Y. R. Zheng, S. H. Yu, J. Am. Chem. Soc. 2012, 134, 2930.

ChemCatChem 2016, 8, 372 – 379

www.chemcatchem.org

[28] Y. Lee, J. Suntivich, K. J. May, E. E. Perry, Y. Shao-Horn, J. Phys. Chem. Lett. 2012, 3, 399. [29] K. Fominykh, P. Chernev, I. Zaharieva, J. Sicklinger, G. Stefanic, M. Doblinger, A. Muller, A. Pokharel, S. Bocklein, C. Scheu, T. Bein, D. Fattakhova-Rohlfing, ACS Nano 2015, 9, 5180. [30] a) Y. F. Li, A. Selloni, J. Phys. Chem. Lett. 2014, 5, 3981; b) L. Trotochaud, S. W. Boettcher, Scr. Mater. 2014, 74, 25.

Received: August 27, 2015 Published online on November 16, 2015

379
