Assessing the Potential of CoPt Bronze for Electrocatalysis in Acidic Media

An electron-conducting mixed oxide, Co-Pt bronze was synthesized and examined as a candidate for a highly durable electrocatalyst for both the polymer electrolyte fuel cells and electrolyzers. The motivation of this study comes from the fact that this material has not been studied as an electrocatalyst in acidic media, although past studies showed a high electronic conductivity and a high corrosion resistance. Co-Pt bronze without metallic Pt was obtained by solid-state synthesis and hot aqua regia rinsing. The OER activity was found to be among the highest as a material without Ir and Ru in acidic media, and it showed extremely high electrochemical stability in the OER potential range. Its oxygen reduction reaction (ORR) was obtained after potential cycles down to the hydrogen region, which formed a thin Pt metallic layer over the oxide. While its specific activity was not more than that of pure platinum nanoparticles, its durability against the potential cycles was much higher.


Introduction
Electrocatalysts facilitating electrochemical reactions are one of the most important components for realizing the hydrogen economy.In PEFCs, metallic platinum is mostly used for both the anode and cathode.While platinum is highly active in hydrogen oxidation reactions (HORs), its activity toward oxygen reduction reactions (ORR) is not sufficiently high for wider commercialization; therefore, various attempts have been made to enhance activity by alloying [1], forming core-shell structures [2] and shape-controlling [3,4].Generally speaking however, active samples show poor durability, while durable samples show lower activity [5].In practical usage, high activity without high durability is virtually useless; however, unlike the cases for activity improvement described above, no clear strategies have been proposed for durability improvement.In practical PEFC operation, other degradation modes were identified with regard to the electrocatalysts.One of those modes is called as the start-up/shut-down reverse current mode [6], in which an areal part is in H 2 /Air condition while others are in Air/Air condition.In this mode, the electromotive force in the H 2 /Air area causes a carbon oxidation reaction on the "cathode" (the air side electrode in the normal operation).To alleviate this mode of degradation, a high oxygen evolution reaction (OER) activity should be provided to the "cathode".Another degradation mode is called the reverse potential mode [7], in which if insufficient H 2 gas, compared to applied current, is supplied to some cells in a series connected stack, the potential is reversed to cause carbon oxidation in the anode.In this mode, again, a high OER activity should be provided to the anode.In both degradation modes, the key is the high OER activity.
In water electrolysis, platinum is sufficiently active for hydrogen evolution reactions (HERs), but not for oxygen evolution reaction (OER).While various nonprecious metals can be used as OER In water electrolysis, platinum is sufficiently active for hydrogen evolution reactions (HERs), but not for oxygen evolution reaction (OER).While various nonprecious metals can be used as OER catalysts for alkali electrolysis [8], only precious metals can be used for acidic polymer electrolyte membrane electrolysis, which allows much higher current density compared to alkali electrolysis, and therefore, arguably lower overall cost for hydrogen production.Among the OER catalysts in acidic media, iridium and/or ruthenium (mixed) oxides are regarded as the most active [9][10][11].For stabilities, ruthenium metal and oxides are regarded as unstable, since ruthenium forms volatile RuO4.Iridium, in contrast, does not form a volatile product, but can be dissolved as an ion, especially when the catalyst is electrochemically redoxable IrOx (x < 4), which is more OER active [12].Moreover, those two elements are among the scarcest (much scarcer than platinum) [13], and their usage will be an obstacle for wide distribution of water electrolyzers.Therefore, electrocatalysts without these two elements should be developed.
Another electrochemical device relating electrocatalysts and hydrogen is a reversible fuel cell, which operates as a fuel cell when electricity is necessary, and also as an electrolyzer when electricity is redundant.Reversible fuel cells are, if realized, as advantageous as a renewable energy system, since separate systems of fuel cells and electrolyzers are no longer necessary.In this application, however, an electrocatalyst that is active for both OER and ORR is required, since no such material is available [14].
Recently, we have started to study metal-platinum oxide bronzes (Figure 1, crystalline structure of Co-Pt bronze) as multi-purpose electrocatalysts for ORR, OER, HOR, and HER.From the late 1970s to early 1990s, platinum oxide bronzes of the type of MxPt3O4, where M can be Li, Na, Mg, Cs, Zn, Cd, Co, Ni, etc., were synthesized and explored as electrocatalysts for mainly reversible fuel cells in alkali solution [15][16][17][18] and limitedly in acidic media [19,20].Since then, however, further studies aiming at polymer electrolyte fuel cells and electrolysis have not been conducted.Although early studies did not exhibit high ORR activities [20], their significantly high anti-oxidation resistance (hot aqua regia resistant), electron conducting property, and non-use of the scarcest elements have attracted our attention, and we have decided to explore their potential as electrocatalysts.In this study, we synthesized Co-Pt bronze and conducted various characterization and electrochemical property examinations including ORR, OER, HER, and HOR activity, and durability tests to clarify whether this type of material has a potential as a practical electrocatalyst in acidic media and is worth further study.Note that this material, in which both Co and Pt are in their oxidized states, is totally different from Pt-Co alloy catalysts, in which both Co and Pt are in their metallic states, and which are known as being among the most active ORR catalyst systems [21][22][23].

Characterization
XRD patterns of Co-Pt bronze before and after the hot aqua regia treatment are shown in Figure 2. Before the treatment, peaks assigned to Co-Pt bronze and Pt metal are clearly seen, but the peak for Pt metal disappears after the treatment.This analysis indicates that metallic platinum was formed by the synthesis, but removed by the hot aqua regia, and that the final product is Co-Pt bronze only.

Characterization
XRD patterns of Co-Pt bronze before and after the hot aqua regia treatment are shown in Figure 2. Before the treatment, peaks assigned to Co-Pt bronze and Pt metal are clearly seen, but the peak for Pt metal disappears after the treatment.This analysis indicates that metallic platinum was formed by the synthesis, but removed by the hot aqua regia, and that the final product is Co-Pt bronze only.TEM images and EDS mappings of Co-Pt bronze before and after the hot aqua regia treatment are shown in Figure 3.The particle size is 20-50 nm, which roughly agrees with the diameter (24.6 nm) estimated by BET surface area analysis.EDS mappings show that intense Pt regions existing before the hot aqua regia treatment disappear afterwards.This is another indication of the removal of the Pt metal phase.In addition, spots having strong Co signals exist before the aqua regia treatment, and disappear afterwards.Therefore, unreacted Co content is considered to remain before the treatment and to be removed by it.
Figure 4 shows an electronic conductivity comparison among the Co-Pt bronze (after the hot aqua regia treatment), Vulcan XC-72, IrO2 (Furuya) and PtO2.At low pressure, conductivities are controlled by particle contact.Increasing pressure, however, decreases the contact resistance and reveals their bulk properties.The trend and absolute value for Vulcan agree well with published data [24].Unlike PtO2, which is an insulator, Co-Pt bronze shows a high electrical conductivity comparable with Vulcan XC-72 and IrO2.This property is crucial for a good electrocatalyst, especially aiming at devices operating at a high current density, including automotive PEFCs and water electrolyzers.TEM images and EDS mappings of Co-Pt bronze before and after the hot aqua regia treatment are shown in Figure 3.The particle size is 20-50 nm, which roughly agrees with the diameter (24.6 nm) estimated by BET surface area analysis.EDS mappings show that intense Pt regions existing before the hot aqua regia treatment disappear afterwards.This is another indication of the removal of the Pt metal phase.In addition, spots having strong Co signals exist before the aqua regia treatment, and disappear afterwards.Therefore, unreacted Co content is considered to remain before the treatment and to be removed by it.
Figure 4 shows an electronic conductivity comparison among the Co-Pt bronze (after the hot aqua regia treatment), Vulcan XC-72, IrO 2 (Furuya) and PtO 2 .At low pressure, conductivities are controlled by particle contact.Increasing pressure, however, decreases the contact resistance and reveals their bulk properties.The trend and absolute value for Vulcan agree well with published data [24].Unlike PtO 2 , which is an insulator, Co-Pt bronze shows a high electrical conductivity comparable with Vulcan XC-72 and IrO 2 .This property is crucial for a good electrocatalyst, especially aiming at devices operating at a high current density, including automotive PEFCs and water electrolyzers.

Electrochemical Properties
OER activities were measured by potential sweep and potential holding, as shown in Figure 5a,b, respectively, as current density per BET surface area.The specifications and OER activities of tested catalysts are listed in Tables 1 and 2, respectively.Although Pt black initially showed a comparable activity, that activity quickly faded because PtO 2 , an insulator, grew on the surface.Two iridium oxides showed significantly different behaviors.IrO 2 (Alfa Aesar) showed larger hysteresis in the voltammogram and maximum current of the potentiostat for initial ca. 10 s after the potential step from 0.4 to 1.7 V.Those observations are most likely due to a redox reaction of IrO 2 (Alfa Aesar), since the XRD pattern shown in Figure S1 shows that IrO 2 (Alfa Aesar) is actually a mixture of crystalline iridium metal and amorphous-like iridium oxide, both of which (or at least iridium metal) can be electrochemically active.In contrast, IrO 2 (Furuya) showed no hysteresis in the voltammogram, and reached the steady-state current instantly after the potential step, suggesting that crystalline IrO 2 , such as IrO 2 (Furuya) as shown in Figure S1, is electrochemically stable.The higher activity of IrO 2 (Alfa Aesar) compared to IrO 2 (Furuya) may stem from the presence of lower valent iridium, since Ir(III) is considered more active than Ir(IV) [12] for OER.Against potential cycles, however, catalysts containing Ir(III) are generally regarded as being less stable [12], although that trend is not seen in this short potential holding test.
Catalysts 2018, 8, x FOR PEER REVIEW 5 of 15 tested catalysts are listed in Tables 1 and 2, respectively.Although Pt black initially showed a comparable activity, that activity quickly faded because PtO2, an insulator, grew on the surface.Two iridium oxides showed significantly different behaviors.IrO2 (Alfa Aesar) showed larger hysteresis in the voltammogram and maximum current of the potentiostat for initial ca. 10 s after the potential step from 0.4 to 1.7 V.Those observations are most likely due to a redox reaction of IrO2 (Alfa Aesar), since the XRD pattern shown in Figure S1 shows that IrO2 (Alfa Aesar) is actually a mixture of crystalline iridium metal and amorphous-like iridium oxide, both of which (or at least iridium metal) can be electrochemically active.In contrast, IrO2 (Furuya) showed no hysteresis in the voltammogram, and reached the steady-state current instantly after the potential step, suggesting that crystalline IrO2, such as IrO2 (Furuya) as shown in Figure S1, is electrochemically stable.The higher activity of IrO2 (Alfa Aesar) compared to IrO2 (Furuya) may stem from the presence of lower valent iridium, since Ir(III) is considered more active than Ir(IV) [12] for OER.Against potential cycles, however, catalysts containing Ir(III) are generally regarded as being less stable [12], although that trend is not seen in this short potential holding test.

Specific Surface Area (BET) (m
Co-Pt bronze showed no hysteresis in the voltammograms, and exhibited stable OER currents, just like IrO2 (Furuya).These results also suggest that Co-Pt bronze can be an extremely stable OER catalyst.Its OER activity is lower than that of IrO2 below 1.62 V, but higher than that of IrO2 (Furuya) over 1.6 V. Activities at higher potential with higher current, however, are affected by the conductivity of powders, especially under RDE condition, since powders are not compressed and smaller powders tend to exhibit lower conductivity by contact resistance.
OER activities are also compared with published data [10,11] using IrO2(Alfa Aesar) in Table 2. IrO2 (Alfa Aesar) of this work showed lower OER activity than IrO2 (Alfa Aesar) of published data [10,11], especially at 1.60 V.This difference could be due to difference in experimental conditions  Co-Pt bronze showed no hysteresis in the voltammograms, and exhibited stable OER currents, just like IrO 2 (Furuya).These results also suggest that Co-Pt bronze can be an extremely stable OER catalyst.Its OER activity is lower than that of IrO 2 below 1.62 V, but higher than that of IrO 2 (Furuya) over 1.6 V. Activities at higher potential with higher current, however, are affected by the conductivity of powders, especially under RDE condition, since powders are not compressed and smaller powders tend to exhibit lower conductivity by contact resistance.
OER activities are also compared with published data [10,11] using IrO 2 (Alfa Aesar) in Table 2. IrO 2 (Alfa Aesar) of this work showed lower OER activity than IrO 2 (Alfa Aesar) of published data [10,11], especially at 1.60 V.This difference could be due to difference in experimental conditions (e.g., IR-corrected [10,11] vs. not corrected in this work), or the difference in thin film formation (e.g., with Nafion [10,11] vs. without it in this work).In this screening study, which aims to reveal the potential of Pt-Co bronze as an electrocatalyst, further analyses were not carried out.Longer durability tests, which are necessary for practical purposes, are desirable but were not carried out because powder catalysts deposited on glassy carbon used in this experiment (no ionomer added as a binder) were easily detached from glassy carbon by the oxygen evolution, carbon corrosion, and electrolyte turbulence.Longer durability test should be carried out using a membrane electrode assembly, which is only possible after a significant amount of catalysts is synthesized; this is beyond the scope of this paper.
Although the exact comparison of the results from these initial performance and short stability test among these samples is not possible because their surface areas and morphologies were not uniformly controlled (see Table 1), it is still safe to judge that Co-Pt bronze has a similar OER activity and durability with crystalline IrO 2 .These classes of catalysts have a great potential as OER catalysts without Ir and Ru, which should be further explored in future studies.
Voltammograms under argon and oxygen atmospheres are compared in Figure 6 to examine the ORR activity.Current below 0.8 V under oxygen atmosphere is clearly the ORR current, while current at 0.9 V, which is a standard potential to define the activity for Pt metal based catalyst, is negligible.This low activity agrees well with a previous study [15].Therefore, the intrinsic ORR activity is very limited, but as the following section shows, potential cycles to the hydrogen potential regions can provide a significant increase in ORR activity, which makes this material attractive for reversible fuel cells.(e.g., IR-corrected [10,11] vs. not corrected in this work), or the difference in thin film formation (e.g., with Nafion [10,11] vs. without it in this work).In this screening study, which aims to reveal the potential of Pt-Co bronze as an electrocatalyst, further analyses were not carried out.Longer durability tests, which are necessary for practical purposes, are desirable but were not carried out because powder catalysts deposited on glassy carbon used in this experiment (no ionomer added as a binder) were easily detached from glassy carbon by the oxygen evolution, carbon corrosion, and electrolyte turbulence.Longer durability test should be carried out using a membrane electrode assembly, which is only possible after a significant amount of catalysts is synthesized; this is beyond the scope of this paper.
Although the exact comparison of the results from these initial performance and short stability test among these samples is not possible because their surface areas and morphologies were not uniformly controlled (see Table 1), it is still safe to judge that Co-Pt bronze has a similar OER activity and durability with crystalline IrO2.These classes of catalysts have a great potential as OER catalysts without Ir and Ru, which should be further explored in future studies.
Voltammograms under argon and oxygen atmospheres are compared in Figure 6 to examine the ORR activity.Current below 0.8 V under oxygen atmosphere is clearly the ORR current, while current at 0.9 V, which is a standard potential to define the activity for Pt metal based catalyst, is negligible.This low activity agrees well with a previous study [15].Therefore, the intrinsic ORR activity is very limited, but as the following section shows, potential cycles to the hydrogen potential regions can provide a significant increase in ORR activity, which makes this material attractive for reversible fuel cells.

Surface Treated Co-Pt Bronze
The surface of the original Co-Pt bronze was electrochemically activated by potential cycles between 0.05 and 1.2 V vs. RHE under inert conditions.Voltammograms during these cycles are shown in Figure 7.As the number of cycles increases, typical voltammetric features of Pt metal emerge.After 100 cycles, the voltammogram becomes like that of Pt metal, and no further change was observed after 100 cycles.The platinum metal electrochemical surface area was calculated from the hydrogen desorption charge using a commonly used conversion factor of 210 microCoulomb/cm 2 , and the Pt metal coverage was obtained by dividing the electrochemical surface area by the BET surface area of loaded powder as 0.15.Therefore, roughly 15% of Pt-Co bronze was estimated to be reduced and covered with metallic platinum.

Surface Treated Co-Pt Bronze
The surface of the original Co-Pt bronze was electrochemically activated by potential cycles between 0.05 and 1.2 V vs. RHE under inert conditions.Voltammograms during these cycles are shown in Figure 7.As the number of cycles increases, typical voltammetric features of Pt metal After 100 cycles, the voltammogram becomes like that of Pt metal, and no further change was observed after 100 cycles.The platinum metal electrochemical surface area was calculated from the hydrogen desorption charge using a commonly used conversion factor of 210 microCoulomb/cm 2 , and the Pt metal coverage was obtained by dividing the electrochemical surface area by the BET surface area of loaded powder as 0.15.Therefore, roughly 15% of Pt-Co bronze was estimated to be reduced and covered with metallic platinum.

Characterization
Figure 8 shows Pt4f XPS spectra of the Co-Pt bronze before and after the potential cycles.Although two major peaks (Pt4f7/2: ca.72 eV and Pt4f5/2: ca.76 eV) corresponding to Pt(I) or (II) are unchanged, after the potential cycles, a peak (ca.78 eV) corresponding to Pt(IV) disappears and a shoulder (ca.71 eV) corresponding to Pt metal appears.Figure 9 shows XRD patterns of Co-Pt bronze loaded on glassy carbon after the potential cycles, which exhibits peaks of the bronze only but no Pt metal.These data, combined with CV patterns, suggest that the potential cycles induced formation of a Pt metal film that locates only on the surface of the Co-Pt bronze, and is thin enough not to be detected by XRD.This suggestion is confirmed by TEM observation and EDS analysis, shown in Figure 10.The TEM image shows, on a plate-like substrate, numerous bright flat areas (<10 nm), which were not observed before the potential cycles (Figure 3B), and which can be regarded as formed Pt metallic films.The lower portion of the EDS image exhibits a concentrated Pt area, which seems to correspond to part of the formed Pt metallic film islands.(TEM EDS analysis is not surface-sensitive as XPS.)Further element mapping under high magnitude and elemental point analysis were not carried out because the strong electron beam for those analyses was found to induce reduction reactions of the bronze, and to change the composition and shape during analysis.

Characterization
Figure 8 shows Pt4f XPS spectra of the Co-Pt bronze before and after the potential cycles.Although two major peaks (Pt4f7/2: ca.72 eV and Pt4f5/2: ca.76 eV) corresponding to Pt(I) or (II) are unchanged, after the potential cycles, a peak (ca.78 eV) corresponding to Pt(IV) disappears and a shoulder (ca.71 eV) corresponding to Pt metal appears.Figure 9 shows XRD patterns of Co-Pt bronze loaded on glassy carbon after the potential cycles, which exhibits peaks of the bronze only but no Pt metal.These data, combined with CV patterns, suggest that the potential cycles induced formation of a Pt metal film that locates only on the surface of the Co-Pt bronze, and is thin enough not to be detected by XRD.This suggestion is confirmed by TEM observation and EDS analysis, shown in Figure 10.The TEM image shows, on a plate-like substrate, numerous bright flat areas (<10 nm), which were not observed before the potential cycles (Figure 3B), and which can be regarded as formed Pt metallic films.The lower portion of the EDS image exhibits a concentrated Pt area, which seems to correspond to part of the formed Pt metallic film islands.(TEM EDS analysis is not surface-sensitive as XPS.)Further element mapping under high magnitude and elemental point analysis were not carried out because the strong electron beam for those analyses was found to induce reduction reactions of the bronze, and to change the composition and shape during analysis.11.The ORR activity is drastically improved after the potential cycles seemingly because of the Pt thin film formation.The active sites for ORR, therefore, should be the metallic platinum formed by the potential cycles.Figure 12 compared (1) specific surface areas determined by hydrogen desorption charge in Figure 7 and deposited Pt weight, (2) specific ORR activity determined by ORR current at 0.9 V and the surface area, and (3) ORR mass activity with those of Pt/C.Since the Co-Pt bronze has a much larger particle size (24.6 nm vs. 2.4 nm of Pt/C); its specific surface area is much smaller, resulting in the smaller mass activity.The specific activity, however, is similar with Pt/C, and no further improvement effects such as the alloying effect observed in Pt-Co alloy systems were observed.However, if a smaller size of Co-Pt bronze is synthesized, the mass activity is expected to significantly improve.11.The ORR activity is drastically improved after the potential cycles seemingly because of the Pt thin film formation.The active sites for ORR, therefore, should be the metallic platinum formed by the potential cycles.Figure 12 compared (1) specific surface areas determined by hydrogen desorption charge in Figure 7 and deposited Pt weight, (2) specific ORR activity determined by ORR current at 0.9 V and the surface area, and (3) ORR mass activity with those of Pt/C.Since the Co-Pt bronze has a much larger particle size (24.6 nm vs. 2.4 nm of Pt/C); its specific surface area is much smaller, resulting in the smaller mass activity.The specific activity, however, is similar with Pt/C, and no further improvement effects such as the alloying effect observed in Pt-Co alloy systems were observed.However, if a smaller size of Co-Pt bronze is synthesized, the mass activity is expected to significantly improve.Cyclic voltammograms during the ADT are shown in Figure 13 for Pt black, Pt/Vulcan and Co-Pt bronze.Pt black shows only a slight decrease in H-upd area, while Pt/Vulcan shows almost total disappearance in H-upd.Co-Pt bronze also showed a significant decrease, but even after 10,000 cycles, the voltammetric feature of H-upd is still recognizable.ECSAs, which are calculated by deducting double-layer discharge coulombs from H desorption coulombs, per unit weight of platinum were plotted in Figure 14, as well as their relative values from their initial values.In addition, dissolved Pt ion amounts in the ADT electrolyte were plotted in Figure 15.The ECSA decrease of Pt-black is limited, because it has a large particle size, and new platinum surface would   11.The ORR activity is drastically improved after the potential cycles seemingly because of the Pt thin film formation.The active sites for ORR, therefore, should be the metallic platinum formed by the potential cycles.Figure 12 compared (1) specific surface areas determined by hydrogen desorption charge in Figure 7 and deposited Pt weight, (2) specific ORR activity determined by ORR current at 0.9 V and the surface area, and (3) ORR mass activity with those of Pt/C.Since the Co-Pt bronze has a much larger particle size (24.6 nm vs. 2.4 nm of Pt/C); its specific surface area is much smaller, resulting in the smaller mass activity.The specific activity, however, is similar with Pt/C, and no further improvement effects such as the alloying effect observed in Pt-Co alloy systems were observed.However, if a smaller size of Co-Pt bronze is synthesized, the mass activity is expected to significantly improve.Cyclic voltammograms during the ADT are shown in Figure 13 for Pt black, Pt/Vulcan and Co-Pt bronze.Pt black shows only a slight decrease in H-upd area, while Pt/Vulcan shows almost total disappearance in H-upd.Co-Pt bronze also showed a significant decrease, but even after 10,000 cycles, the voltammetric feature of H-upd is still recognizable.ECSAs, which are calculated by deducting double-layer discharge coulombs from H desorption coulombs, per unit weight of platinum were plotted in Figure 14, as well as their relative values from their initial values.In addition, dissolved Pt ion amounts in the ADT electrolyte were plotted in Figure 15.The ECSA decrease of Pt-black is limited, because it has a large particle size, and new platinum surface would Cyclic voltammograms during the ADT are shown in Figure 13 for Pt black, Pt/Vulcan and Co-Pt bronze.Pt black shows only a slight decrease in H-upd area, while Pt/Vulcan shows almost total disappearance in H-upd.Co-Pt bronze also showed a significant decrease, but even after 10,000 cycles, the voltammetric feature of H-upd is still recognizable.ECSAs, which are calculated by deducting double-layer discharge coulombs from H desorption coulombs, per unit weight of platinum were plotted in Figure 14, as well as their relative values from their initial values.In addition, dissolved Pt ion amounts in the ADT electrolyte were plotted in Figure 15.The ECSA decrease of Pt-black is limited, because it has a large particle size, and new platinum surface would be available even after the surface is dissolved.The ECSA loss of 20% matched its dissolved portion in the electrolyte.Pt/Vulcan exhibited almost total dissolution and total loss in ECSA, and the remaining ECSAs are probably artefacts from the double layer charge estimation.Co-Pt bronze, having a large particle size and a small portion (15%) coverage with thin layered Pt metal, shows limited initial ECSA and ca.60% loss.The dissolution, however, is very limited (ca.5%).This discrepancy of Co-Pt bronze can be elucidated, assuming that the ADT dissolves only the metallic Pt formed by the activation cycles, leaving the bronze phase intact.This assumption was confirmed by TEM observation of the sample after the ADT, which showed no Pt islands anymore, as shown in Figure S3.Then, re-activation cycles were applied on the sample after ADT, as shown in the voltammogram of Figure 16, which clearly shows recovery of Pt metal surface.Although the recovered area is smaller than that shown in Figure 7, this is a remarkable and unique feature as an electrocatalyst for PEFCs.
Catalysts 2018, 8, x FOR PEER REVIEW 10 of 15 be available even after the surface is dissolved.The ECSA loss of 20% matched its dissolved portion in the electrolyte.Pt/Vulcan exhibited almost total dissolution and total loss in ECSA, and the remaining ECSAs are probably artefacts from the double layer charge estimation.Co-Pt bronze, having a large particle size and a small portion (15%) coverage with thin layered Pt metal, shows limited initial ECSA and ca.60% loss.The dissolution, however, is very limited (ca.5%).This discrepancy of Co-Pt bronze can be elucidated, assuming that the ADT dissolves only the metallic Pt formed by the activation cycles, leaving the bronze phase intact.This assumption was confirmed by TEM observation of the sample after the ADT, which showed no Pt islands anymore, as shown in Figure S3.Then, re-activation cycles were applied on the sample after ADT, as shown in the voltammogram of Figure 16, which clearly shows recovery of Pt metal surface.Although the recovered area is smaller than that shown in Figure 7, this is a remarkable and unique feature as an electrocatalyst for PEFCs.Catalysts 2018, 8, x FOR PEER REVIEW 10 of 15 be available even after the surface is dissolved.The ECSA loss of 20% matched its dissolved portion in the electrolyte.Pt/Vulcan exhibited almost total dissolution and total loss in ECSA, and the remaining ECSAs are probably artefacts from the double layer charge estimation.Co-Pt bronze, having a large particle size and a small portion (15%) coverage with thin layered Pt metal, shows limited initial ECSA and ca.60% loss.The dissolution, however, is very limited (ca.5%).This discrepancy of Co-Pt bronze can be elucidated, assuming that the ADT dissolves only the metallic Pt formed by the activation cycles, leaving the bronze phase intact.This assumption was confirmed by TEM observation of the sample after the ADT, which showed no Pt islands anymore, as shown in Figure S3.Then, re-activation cycles were applied on the sample after ADT, as shown in the voltammogram of Figure 16, which clearly shows recovery of Pt metal surface.Although the recovered area is smaller than that shown in Figure 7, this is a remarkable and unique feature as an electrocatalyst for PEFCs.HOR and OER activities of the surface treated Co-Pt bronze were roughly examined by cyclic voltammetry under a hydrogen atmosphere, and the result is shown in Figure 17.Diffusion limiting current was easily achieved for HOR between 0.05 V and 0.9 V in the anodic sweep and 0.1 V and 0.5 V in the cathodic sweep.A sharp decrease in HOR current toward 0 V suggests its activity toward HER.These HOR and HER activities are reasonable, since metallic Pt is formed this potential range.Depression of HOR current above these potential range is most likely due to platinum oxide formation observed on Pt/C [25].Above 1.5 V, similar OER current with untreated Co-Pt bronze was observed.This OER activity is due to the native activity of Co-Pt bronze shown in Figure 5.   HOR and OER activities of the surface treated Co-Pt bronze were roughly examined by cyclic voltammetry under a hydrogen atmosphere, and the result is shown in Figure 17.Diffusion limiting current was easily achieved for HOR between 0.05 V and 0.9 V in the anodic sweep and 0.1 V and 0.5 V in the cathodic sweep.A sharp decrease in HOR current toward 0 V suggests its activity toward HER.These HOR and HER activities are reasonable, since metallic Pt is formed in this potential range.Depression of HOR current above these potential range is most likely due to platinum oxide formation observed on Pt/C [25].Above 1.5 V, similar OER current with untreated Co-Pt bronze was observed.This OER activity is due to the native activity of Co-Pt bronze shown in Figure 5.  HOR and OER activities of the surface treated Co-Pt bronze were roughly examined by cyclic voltammetry under a hydrogen atmosphere, and the result is shown in Figure 17.Diffusion limiting current was easily achieved for HOR between 0.05 V and 0.9 V in the anodic sweep and 0.1 V and 0.5 V in the cathodic sweep.A sharp decrease in HOR current toward 0 V suggests its activity toward HER.These HOR and HER activities are reasonable, since metallic Pt is formed in this potential range.Depression of HOR current above these potential range is most likely due to platinum oxide formation observed on Pt/C [25].Above 1.5 V, similar OER current with untreated Co-Pt bronze was observed.This OER activity is due to the native activity of Co-Pt bronze shown in Figure 5.  HOR and OER activities of the surface treated Co-Pt bronze were roughly examined by cyclic voltammetry under a hydrogen atmosphere, and the result is shown in Figure 17.Diffusion limiting current was easily achieved for HOR between 0.05 V and 0.9 V in the anodic sweep and 0.1 V and 0.5 V in the cathodic sweep.A sharp decrease in HOR current toward 0 V suggests its activity toward HER.These HOR and HER activities are reasonable, since metallic Pt is formed in this potential range.Depression of HOR current above these potential range is most likely due to platinum oxide formation observed on Pt/C [25].Above 1.5 V, similar OER current with untreated Co-Pt bronze was observed.This OER activity is due to the native activity of Co-Pt bronze shown in Figure 5.

Synthesis
Platinum oxide (3.952 g) and cobalt (II) nitrate hexahydrate (1.640 g) (Pt:Co = 3:1 mol) was mixed using a pestle and mortar, and the mixture was heated in an alumina boat placed in an atmosphere-controlled tube at 650 • C under air stream for 5 h, according to a past study [15].The obtained powder (3.536 g) was immersed in hot (80 • C) aqua regia for 30 min and filtrated to remove soluble components.This cleaning process was carried out three times before it was thoroughly rinsed with ultrapure water (Milli-Q).Cobalt-platinum bronze powder (1.20 g) was recovered.The recovery was poor because a significant part of the powder attached to the filter could not be retrieved.

Characterization
TEM (transmission electron microscope) observation and EDS (energy dispersive spectroscopy) characterization were carried out with JEOL JEM-2100F (Tokyo, Japan) and FEI Titan G2 60-300 (Columbus, OH, USA).Intense element mapping under high magnitude and elemental point analysis were not carried out because the strong electron beam used for those analyses was found to induce reduction reactions of the bronze, and to change the composition and shape during the analysis.SEM observation was carried out with Hitachi S5500 (Hitachi High-Technologies Corporation, Tokyo, Japan).XRD (X-ray diffractometry) was carried out with Rigaku Ultima IV (Rigaku Corporation, Tokyo, Japan) and the data were analyzed based on the ICDD database.XPS (X-ray photoelectron spectroscopy) was conducted with ULVAC PHI Quantera SXM (Kanagawa, Japan).Electric conductivities of powder samples were measured with a compression cell shown in Figure 18 that allows to apply a controlled pressure to the powder samples [24].Brunauer-Emmett-Teller (BET) surface area was determined by Quantachrome Autosorb-1 (Boynton Beach, FL, USA).

Electrochemical Tests
Obtained Co-Pt bronze powder was deposited on a glassy carbon (GC) electrode (0.246 cm 2 ) from a dispersion in acetone (100%) to form a thin film of the bronze containing 13 μgcm −2 of platinum (assumed Co0.5Pt3O4).For ORR and HOR activity comparison, a commercial platinum/carbon (Vulcan XC-72) catalyst (TEC10V30E) was deposited on GC similarly.For OER activity comparison, Pt black and two iridium oxides were also deposited on GC.The electrochemical tests were conducted in the rotating disk electrode set-up in 0.1 M HClO4 electrolyte.The detailed electrochemical conditions are described in Table S1.
OER test was conducted by cyclic voltammetry and chronoamperometry whose detailed conditions are listed in Table S2.
ORR test procedures shown in Table S3 are similar with the one described elsewhere [26].To avoid the reduction of the surface of the bronze from oxide to metal, the electrode potential was held over 0.6 V at the first stage.Then, the potential was cycled between 0.05 and 1.2 V before re-examining the ORR activity.An accelerated durability test (ADT) was carried out in a chloride-containing electrolyte by a procedure described in Table S4.Dissolved Pt ion concentration was determined by stripping voltammetry on a hanging mercury drop electrode [27][28][29], as shown in Table S5.HOR examination was also carried out by cyclic voltammetry (CV) under a hydrogen atmosphere, and an OER test was conducted by CV and potential holding conditions under an argon atmosphere.

Conclusions
An electron-conducting mixed oxide, Co-Pt bronze, was synthesized by a solid state reaction and hot aqua regia treatment.The obtained 20-50 nm size particles showed high conductivity, high OER, and low ORR activities.Although the OER activity was lower than those of two commercial iridium oxides, its electrochemical stability seemed extremely high, since Co-Pt bronze was electrochemically inert in the potential of interest for OER.Potential cycles to the hydrogen region promoted the formation of thin Pt metallic films on the bronze surface, which is highly active to ORR, HOR, and HER.ADT did not totally dissolve the activated Co-Pt bronze, unlike Pt metal catalysts, and reactivation was possible by additional potential cycles.These unique features can be advantageously utilized for:

•
Water electrolyzer anode with high OER activity without Ir or Ru, • PEFC cathode catalyst with high load-cycle durability and high start-up shutdown durability due to the high OER activity [7],

Electrochemical Tests
Obtained Co-Pt bronze powder was deposited on a glassy carbon (GC) electrode (0.246 cm 2 ) from a dispersion in acetone (100%) to form a thin film of the bronze containing 13 µgcm −2 of platinum (assumed Co 0.5 Pt 3 O 4 ).For ORR and HOR activity comparison, a commercial platinum/carbon (Vulcan XC-72) catalyst (TEC10V30E) was deposited on GC similarly.For OER activity comparison, Pt black and two iridium oxides were also deposited on GC.The electrochemical tests were conducted in the rotating disk electrode set-up in 0.1 M HClO 4 electrolyte.The detailed electrochemical conditions are described in Table S1.
OER test was conducted by cyclic voltammetry and chronoamperometry whose detailed conditions are listed in Table S2.
ORR test procedures shown in Table S3 are similar with the one described elsewhere [26].To avoid the reduction of the surface of the bronze from oxide to metal, the electrode potential was held over 0.6 V at the first stage.Then, the potential was cycled between 0.05 and 1.2 V before re-examining the ORR activity.An accelerated durability test (ADT) was carried out in a chloride-containing electrolyte by a procedure described in Table S4.Dissolved Pt ion concentration was determined by stripping voltammetry on a hanging mercury drop electrode [27][28][29], as shown in Table S5.HOR examination was also carried out by cyclic voltammetry (CV) under a hydrogen atmosphere, and an OER test was conducted by CV and potential holding conditions under an argon atmosphere.

Conclusions
An electron-conducting mixed oxide, Co-Pt bronze, was synthesized by a solid state reaction and hot aqua regia treatment.The obtained 20-50 nm size particles showed high conductivity, high OER, and low ORR activities.Although the OER activity was lower than those of two commercial iridium oxides, its electrochemical stability seemed extremely high, since Co-Pt bronze was electrochemically inert in the potential of interest for OER.Potential cycles to the hydrogen region promoted the formation of thin Pt metallic films on the bronze surface, which is highly active to ORR, HOR, and HER.ADT did not totally dissolve the activated Co-Pt bronze, unlike Pt metal catalysts, and reactivation was possible by additional potential cycles.These unique features can be advantageously utilized for:

•
Water electrolyzer anode with high OER activity without Ir or Ru, • PEFC cathode catalyst with high load-cycle durability and high start-up shutdown durability due to the high OER activity [7],

Figure 1 .
Figure 1.Crystalline structure of Co Pt bronze.Figure 1. Crystalline structure of Co Pt bronze.

Figure 1 .
Figure 1.Crystalline structure of Co Pt bronze.Figure 1. Crystalline structure of Co Pt bronze.

Figure 9 .
Figure 9. XRD pattern of Co-Pt bronze on glassy carbon after the activation treatment compared with those of Co-Pt bronze powder, glassy carbon and Pt metal.

Figure 9 .
Figure 9. XRD pattern of Co-Pt bronze on glassy carbon after the activation treatment compared with those of Co-Pt bronze powder, glassy carbon and Pt metal.

Figure 9 .
Figure 9. XRD pattern of Co-Pt bronze on glassy carbon after the activation treatment compared with those of Co-Pt bronze powder, glassy carbon and Pt metal.

Figure 9 .
Figure 9. XRD pattern of Co-Pt bronze on glassy carbon after the activation treatment compared with those of Co-Pt bronze powder, glassy carbon and Pt metal.

Figure 11 .
Figure 11.ORR activity measurement voltammogram of activated Co-Pt bronze compared with original Co-Pt bronze and Pt/Vulcan.

Figure 12 .
Figure 12.(a) ECSA; (b) ORR mass activity and (c) ORR specific activity comparisons between Co-Pt bronze and Pt/Vulcan.

Figure 11 .
Figure 11.ORR activity measurement voltammogram of activated Co-Pt bronze compared with original Co-Pt bronze and Pt/Vulcan.

Figure 11 .
Figure 11.ORR activity measurement voltammogram of activated Co-Pt bronze compared with original Co-Pt bronze and Pt/Vulcan.

Figure 12 .
Figure 12.(a) ECSA; (b) ORR mass activity and (c) ORR specific activity comparisons between Co-Pt bronze and Pt/Vulcan.

Figure 12 .
Figure 12.(a) ECSA; (b) ORR mass activity and (c) ORR specific activity comparisons between Co-Pt bronze and Pt/Vulcan.

Figure 14 .
Figure 14.ECSA retention during potential cycle accelerated durability test.Figure 14.ECSA retention during potential cycle accelerated durability test.

Figure 14 .
Figure 14.ECSA retention during potential cycle accelerated durability test.Figure 14.ECSA retention during potential cycle accelerated durability test.

Figure 15 .
Figure 15.Pt dissolution ratio (=dissolved Pt/load Pt on glassy carbon) during potential cycle accelerated durability test.Inserted is an expansion.

Figure 17 .
Figure 17.Cyclic Voltammogram of activated Co-Pt bronze showing its HOR and OER activities.

Figure 15 . 15 Figure 15 .
Figure 15.Pt dissolution ratio (=dissolved Pt/load Pt on glassy carbon) during potential cycle accelerated durability test.Inserted is an expansion.

Figure 17 .
Figure 17.Cyclic Voltammogram of activated Co-Pt bronze showing its HOR and OER activities.

Figure 17 .
Figure 17.Cyclic Voltammogram of activated Co-Pt bronze showing its HOR and OER activities.

Figure 17 .
Figure 17.Cyclic Voltammogram of activated Co-Pt bronze showing its HOR and OER activities.

Figure 18 .
Figure 18.Schematic diagram of a compression cell for measuring conductivity of powder samples.

Figure 18 .
Figure 18.Schematic diagram of a compression cell for measuring conductivity of powder samples. 2

•g −1 ) Mean Diameter * (nm) Loading/GC (µg) Loaded Surface Area/GC (cm 2 )
* Mean diameter was calculated from BET surface area assuming monodispersed spherical particles with bulk density of respective material as:

Table 1 .
Specifications of tested OER catalysts.
1: BET surface area per unit weight; * 2 : electrochemical surface area per unit weight of iridium.
* 1 : BET surface area per unit weight; * 2 : electrochemical surface area per unit weight of iridium.