Boron-Doped Diamond Powder as a Durable Support for Platinum-Based Cathode Catalysts in Polymer Electrolyte Fuel Cells

Platinum nanoparticle-supported boron-doped diamond powder (Pt/BDDP) was prepared and investigated as a durable polymer electrolyte fuel cell (PEFC) cathode catalyst. The use of the nanocapsule method enabled dense deposition of Pt nanoparticles (2– 5 nm in size) on a boron-doped diamond (BDD) powder < 500 nm in size. The Pt/BDDP cathode catalyst showed oxygen reduction reaction activity comparableto Pt-supportedcarbonblack(Pt/C),indicating sufﬁcient conductivityoftheBDDP asa catalyst support. Potential cycling in a highly positive potential region ( + 1.0– + 1.5 V vs. NHE) that simulates the start–stop operations of the PEFC was performed to investigate the durability of the Pt/BDDP catalyst. Decreases in the electrochemically active surface area of the Pt/BDDP were suppressed compared to that of Pt/C. Corrosion resistance of BDD against potential cycling was demonstrated by testing a BDD thin-ﬁlm electrode. The corrosion resistance should be responsible for the improved durability of the BDDP support, possibly by lowering the Pt nanoparticle association process (e.g., agglomeration). © The Author(s) 2018.

The polymer electrolyte fuel cell (PEFC) is a promising power generating system for portable devices, electric vehicles, and electric transportation applications because of its low operating temperature, high power density, high theoretical energy efficiency, and fast start-up advantages. 1,2 Improvement of PEFC durability has been desired for large-scale commercialization; a major obstacle is the deterioration of the carbon support for PEFC cathode catalysts. In frequent start-stop operations, particularly for automobiles, the cathode is exposed to a highly positive potential and the carbon support can be corroded via oxidation. 1,[3][4][5][6] As solutions to this durability issue, SnO 2 , Nb-SnO 2 , 7 TiO x and TiN, 8 and polymer-coated carbon nanotubes 9 have all been reported as candidates for alternative support materials with superior tolerance to corrosion from electrochemical oxidation. Boron-doped diamond (BDD) has also attracted attention as an electrode material because it possesses excellent chemical/electrochemical stability. The BDD electrode is corrosion resistant; even in electrolytic water treatment at highly positive potentials, BDD electrode exhibits long service lifetimes with stable high performance. [10][11][12][13] To apply BDD as a PEFC catalyst support, BDD powder (BDDP) should be used. Several reports exist on the application of BDDP as a catalyst support for the direct methanol fuel cell anode catalyst. The Pt or Pt-Ru particle catalysts were deposited onto BDDP, and the catalysts exhibited high activity and stability to anodic methanol oxidation. [14][15][16][17][18] Recently, Kim reported that Pt/BDDP can be useful as a durable cathode catalyst for PEFC. 19 They fabricated a unit cell using the Pt/BDDP cathode catalyst and found that the decrease in Pt loading during an accelerated long-term test (0.6 V at 90 • C) was less on the BDD support than on the Vulcan XC-72 or multi-walled carbon nanotube supports. Thus, BDDP should be a very promising material for construction of durable PEFCs. To further improve the durability of the Pt/BDDP system, understanding the deterioration mechanism of the catalyst is essential, particularly under severe PEFC operating conditions. This study investigates the deterioration behavior of the Pt/BDDP cathode catalyst against highly positive potentials. A potential cycle test simulating the start-stop operation of PEFC was performed using a * Electrochemical Society Member. z E-mail: t-kondo@rs.noda.tus.ac.jp half-cell containing the Pt/BDDP catalyst, and the deterioration behavior was compared with commercial Pt-supported Vulcan XC-72 (Pt/C).

Experimental
Preparation of Pt/BDDP.-A BDD layer was grown on the surface of insulating diamond powder using microwave plasma chemical vapor deposition (CVD) to obtain BDDP. The CVD conditions were identical to that reported previously. 20 To minimize the sp 2 carbon component in the as-deposited powder, the sample was heat treated in air using a muffle furnace at 450 • C for 3 h. This also resulted in the oxidation of the BDDP surface. For the diamond powder substrate, Micron+ MDA M0.5 (Element Six, average particle size = 300 nm) and MD-200 (Tomei Diamond, average particle size = 200 nm) were used. The particle size distribution of BDDP was evaluated using dynamic light scattering (DLS, Nicomp 380, Particle Sizing Systems). The sample prepared using Micron+ MDA M0.5 was labeled as BDDP-300, and the sample prepared using MD-200 was labeled as BDDP-200.
Pt-supported BDDP (Pt/BDDP) was prepared using the colloid adsorption and nanocapsule methods. For the colloid adsorption method, 50 mg of BDDP was added to 125 mL of ethanol and dispersed using ultrasonication. Then, 2.85 mL of a Pt nanoparticle dispersion (Renaissance Energy Research, nominal particle size = 1-6 nm, concentration = 10 mM) was added to the BDDP dispersion. The mixed dispersion was subjected to ultrasonication and stirring for 30 min. After the solvent evaporated, the sample was vacuum-dried at 70 • C and pulverized. Then, to remove the colloidal protecting agent (polyvinylpyrrolidone) contained in the Pt nanoparticle dispersion, the sample was heat treated in air at 200 • C for 8 h to obtain Pt/BDDP.
The reaction temperature and the reducing agent addition process for the nanocapsule method was optimized from that reported by Yano 21 as follows: A mixture of 12.5 mL diphenyl ether, 260 mg 1,2hexadecanediol, and 99.0 mg platinum acetylacetonate (Pt(acac) 2 ) in a three-necked flask was stirred under argon at 110 • C for 20 min. Then, 80 μL of oleylamine, 85 μL of oleic acid, and 130 mL of ethanol dispersion containing 50 mg of BDDP was added to the mixture followed F3073 by stirring at 170 • C for 30 min. Then, a tetrahydrofuran (3 mL)/water (0.5 mL) mixture solution containing 322 mg of NaBH 4 (as a reducing agent) was added to the mixture for depositing Pt nanoparticles onto the BDDP surface. After stirring the mixture for 10 min, the dispersion was refluxed at 180 • C for 30 min. After cooling to room temperature and suction filtration (washing with ethanol), the sample was vacuum-dried at 80 • C for 5 h. Furthermore, a heat-treatment was performed in air at 230 • C for 4 h to remove the organic contents, such as surfactants, and to obtain Pt/BDDP. Pt/BDDP samples from both synthetic methods were characterized using transmission electron microscopy (TEM, H-7650, Hitachi) and X-ray diffraction (XRD, X'Pert Pro, Philips).
Electrochemical characterization.-500 μL of 35 wt% ethanol aqueous solution and 5 wt% Nafion solution (Aldrich) were added to 2 mg of Pt/BDDP with a Nafion-to-carbon (N/C) ratio of one, and the mixture was ultrasonicated for 3 h to obtain a dispersion. 20 μL of the dispersion was then cast uniformly on the surface of an edge-oriented pyrolytic graphite (EPG) disk electrode (6 mm in diameter) and dried to prepare the Pt/BDDP electrode. For comparison, a Pt/C electrode was prepared using 10 wt% Pt-supported carbon black (Vulcan XC-72) Electrochemical measurements of the Pt/BDDP electrodes were performed in 1.0 M HClO 4 using a platinum ring counter electrode and a Ag/AgCl reference electrode. The electrochemically active surface area (ECA) of the platinum on the Pt/BDDP electrode was determined using cyclic voltammetry (CV) in the electrolyte purged via argon bubbling for 30 min. CV was performed with an initial potential of +80 mV vs. NHE, an initial potential retention time of 0.8 s, and a switching potential of +1200 mV vs. NHE at a scan rate of 30 mV/s. The ECA was then calculated using the following equation from the charge amount of the cathodic adsorption peaks of H 2 (typically at +0.1-+0.3 V vs. NHE), Q H [μC], and the hydrogen adsorption charge amount on a smooth polycrystalline platinum surface (210 μC cm −2 ). 22,23 ECA m 2 g-Pt where w Pt [g] is the weight of the platinum contained in the Pt/BDDP electrodes.
The oxygen reduction reaction (ORR) activity of the Pt/BDDP electrode was evaluated via linear sweep voltammetry (LSV) using the rotating disk electrode (RDE) method at a 1000-rpm rotational speed. After holding the potential at +1.0 V vs. NHE in O 2 -saturated 1.0 M HClO 4 for 60 s, the electrode potential was scanned to the negative direction at a potential sweep rate of 10 mV/s. The potential cycle test simulating the start-stop operation of a fuel cell was adopted as proposed by the Fuel Cell Commercialization Conference of Japan (FCCJ). 24,25 In argon-purged 1.0 M HClO 4 , the CV was first recorded for the evaluation of ECA. Then, the potential was stepped from the open circuit potential to +1.0 V vs. NHE, and held a the potential for 30 s. This was followed by potential cycling from +1.0 V-+1.5 V vs. NHE at a scan rate of 0.5 V/s. After a certain number of potential cycles, the CV was recorded repetitively for estimating ECA until 10,000 cycles were accumulated. The durability of the catalyst against highly positive potentials was then estimated by the ECA change with respect to the number of potential cycles.

Results and Discussion
Characterization of Pt/BDDP.-The size distribution of the newly prepared BDDP was evaluated using DLS (Fig. 1). The particle sizes of BDDP-300 and BDDP-200 were distributed in the 200-500-and 150-400-nm ranges, respectively. Brunauer-Emmett-Teller (BET) specific surface areas were determined using the nitrogen gas adsorption method to be 62.1 and 92.5 m 2 g −1 , respectively. The electric conductivity of the BDDP was calculated to be typically 0.2-0.4 S cm −1 from the electric resistance of BDDP packed in a glass capillary. As previously reported, the doping level of the BDD layer on the BDDP should be in the range of 10 20 -10 21 boron atoms/cm 3 from the estimation using Hall effect measurement of a thin film sample prepared under the same CVD condition. 26 The electric conductivity of hydrogenated BDDP was 0.47 S cm −1 , and thus, there was almost no influence of the surface termination (hydrogen or oxygen) on the electric conductivity. In addition, electric conductivity of Pt/BDDP did not change substantially from the original BDDP. These fact indicates that the electric conductivity of the Pt/BDDP was derived from that of the BDD layer on the BDDP surface.
At first, Pt/BDDP was prepared via the colloid adsorption method using BDDP-300. TEM observations revealed that the Pt nanoparticles (particle diameter = 5-10 nm) were found to be densely supported on the BDDP surface (Fig. 2a). Although the nominal particle size of the Pt nanoparticles in the initial suspension was 1-6 nm, the Pt nanoparticles deposited on the BDDP surface were found to be larger than this size range. The increase in size was attributed to Pt nanoparticle agglomeration during the Pt/BDDP preparation procedure. In contrast, TEM observations revealed that the Pt/BDDP prepared using the nanocapsule method comprised 2-5 nm diameter Pt nanoparticle that were densely supported on the BDDP surface (Fig. 2b). Pt nanoparticles produced using the nanocapsule method have been reported to be relatively small in particle size, and the size distribution can be relatively narrow. 21,27 Even using a modified method in this study, small Pt nanoparticles were successfully deposited on the BDDP surface. Pt/BDDP was also prepared using the nanocapsule method under the same conditions with BDDP-200. Therefore, this method was indicated to be a reproducible preparation method for Pt/BDDP. Figure  3 shows the XRD patterns of Pt/BDDP-300. The deposition of platinum was confirmed by the appearance of Pt diffraction patterns. The Pt nanoparticle size was estimated to be 4.5 nm from the XRD peaks using the Scherrer equation; this value was consistent with the TEM  observations. Thus, the nanocapsule method is a useful method for the preparation of Pt/BDDP and allows Pt nanoparticles of relatively small particle sizes to be densely supported.

Electrochemical properties of Pt/BDDP.-The electrochemical
properties of the Pt/BDDP catalyst were evaluated using a modified EPG disk electrode coated with Pt/BDDP and Nafion. Figure 4 shows the CV recorded in 1.0 M HClO 4 at the Pt/BDDP-modified electrode. Redox peaks that were characteristics of Pt were observed at both the Pt/BDDP-300 and Pt/BDDP-200 electrodes as well as the Pt/C electrode. Therefore, Pt nanoparticles supported on the BDDP was shown to be electrochemically active. The ECAs calculated from the cathodic peak for reductive adsorption of hydrogen at +0.1-+0.3 V vs. NHE were 62.3, 11.7, and 31.4 m 2 g −1 for the Pt/C, Pt/BDDP-300, and Pt/BDDP-200 electrodes, respectively. The smaller ECA for Pt/BDDP than that for the Pt/C is because of the smaller specific surface area of Pt/BDDP resulting in a lower Pt nanoparticle loading. The larger Pt/BDDP-200 ECA than that at the Pt/BDDP-500 results from the slightly larger specific surface area from the smaller BDDP particle size. Next, the ORR activity of the Pt/BDDP catalyst was evaluated using the RDE method. Figures 5 shows the LSV in O 2 -saturated 1.0 M HClO 4 . BDDP-300 (before Pt deposition) showed no reduction current for ORR. In contrast, an ORR current was observed at Pt/BDDP-300 with a near-equivalent onset potential to that at the Pt/C electrode. A similar behavior was observed for BDDP-200 and Pt/BDDP-200 electrodes. Thus, the Pt nanoparticles supported on the BDDP demonstrated equivalent catalytic activity to that supported on carbon black, indicating that the use of BDDP as a catalyst support does not deteriorate the Pt catalytic activity.
Durability test.-The durabilities of the catalysts against potential cycling tests in a highly positive potential region were investigated. At first, the durability of BDD itself against the potential cycle was tested. Figures 6a and 6b show CVs recorded at a glassy carbon (GC) electrode and a BDD thin-film electrode after a specified number of potential cycles in 1.0 M HClO 4 . In the CVs at the GC electrode, the current for electric double-layer charging increased with cycle number. In addition, a redox peak pair at +0.6 V vs. NHE appeared, and the peak   test (2,000, 4,000, 6,000, 8,000, and 10,000 cycles). Arrows indicate this order of the curves. Potential sweep rate was 50 mV/s. (c and d) SEM images of the BDD thin-film and GC electrode surfaces before the potential cycle test, respectively. (e and f) SEM images of the BDD thin-film and GC electrode surfaces after the potential cycle test (10,000 cycles), respectively.  current increased with cycle number. Comparing the SEM images of the GC electrode surface before and after the cycle test, the surface morphology greatly changed after the cycle testing, showing increased surface roughness (Figs. 6d and 6f). Therefore, the most vulnerable parts on the GC surface were etched during the potential cycling, thereby increasing the specific surface area and the electric doublelayer capacitance. Furthermore, the redox peak pair at +0.6 V vs. NHE originated from a redox reaction of the quinone/hydroquinone groups on the GC surface. Such oxygen-containing surface functional groups should be generated in tandem with the oxidative etching. In contrast, in the BDD thin-film case, no changes were observed in the CV shape and current before and after the potential cycle testing. Furthermore, no substantial differences were found in the SEM images before and after the testing (Figs. 6c and 6e). This result indicates that BDD has sufficient durability against the potential cycle test.
Next, potential cycle tests were conducted at the Pt/BDDP electrode. Figures 7a-7c show CVs at Pt/C, Pt/BDDP-300, and Pt/BDDP-200 electrodes before the cycle test and after 1,000, 5,000, and 10,000 cycles. For every electrode, the peak current derived from Pt was found to decrease with increasing cycle number. Therefore, removal and/or agglomeration of Pt was implied to occur during the potential cycle testing. In the case wherein the Pt/C electrode was tested, an increase of the redox peak pair at around +0.6 V vs. NHE, thought to be associated with the quinone/hydroquinone group, was observed with increasing cycle number. This indicated that oxidative etching of the carbon support occurred, as was seen for the GC electrode (Figs. 6b).
In contrast, such a change was not observed for the Pt/BDDP electrode. Figures 7d and 7e illustrate the ECA and normalized ECA at the Pt/C, Pt/BDDP-300, and Pt/BDDP-200 electrodes. The normalized ECA was calculated against the ECA values before cycle testing began. For the Pt/C electrode, a remarkable decrease in ECA was observed in the initial stages (up to 1000 cycles). This behavior was similar to that reported by Monzó 28 and could be because of Pt nanoparticle association processes (agglomeration/Ostwald ripening) caused, in part, by corrosion of the support. In contrast, such an early deterioration was not seen for the Pt/BDDP electrodes. From BDD thin-film electrode investigations shown above, the BDDP was observed to be resistant to potential cycling-induced corrosion. Thus, the association process of the Pt nanoparticle could be suppressed, and the dissociation process (dissolution) 28 could be responsible for deterioration of the ECA at the Pt/BDDP. The ECA retention ratio after 10,000 cycles at Pt/C, Pt/BDDP-300, and Pt/BDDP-200 were 53%, 61%, and 78%, respectively. Presently, although the reason for a difference in the retention ratio observed between Pt/BDDP-300 and Pt/BDDP-200 is unclear, the degradation behavior was similar and indicates a common feature of BDDP as a catalyst support. Although further improvement in the durability may be desired (increased stability based on understanding the degradation mechanisms and increased mass activity by increasing the specific surface area), the newly prepared Pt/BDDP, which was synthesized using the nanocapsule method, exhibited sufficient ORR activity and good durability against highly positive potentials.

Conclusions
Pt nanoparticles with a size of 2-5 nm were densely deposited on a BDDP surface for the preparation of Pt/BDDP using the nanocapsules method. Pt nanoparticles deposited on the Pt/BDDP showed nearly equivalent ORR activity to a commercial Pt/C catalyst, indicating that BDDP can be a useful cathode catalyst support. The examination of the ECA changes after a potential cycle test were performed;