Pt/Co Alloy Nanoparticles Prepared by Nanocapsule Method Exhibit a High Oxygen Reduction Reaction Activity in the Alkaline Media.

Oxygen reduction reaction (ORR) catalysts are one of the main topics for fuel cells and metal/air batteries. We found that the platinum-cobalt alloy nanoparticles prepared by our original nanocapsule method exhibited a high ORR catalytic activity in alkaline solution, compared with that of the existing alloy nanoparticles prepared by different methods. The effect of alloy composition on the ORR activity was investigated to find the optimum composition (approximately 40 atom %). We also found that the enhancement of the catalytic activity in alkaline solution appeared in a very narrow range of Co content compared with that in acidic solution.

Introduction

Electrocatalysts for oxygen reduction reaction (ORR) have attracted considerable attention in particular for energy applications such as fuel cells and metal/air batteries. Current state-of-the-art ORR catalysts for polymer electrolyte membrane fuel cells are based on platinum and platinum alloys because they exhibit a high ORR activity and reasonable stability in the acidic media.[1−6] Because of the high cost and the limited availability of platinum resources, it is crucial to reduce the usage of platinum for wide-spread commercialization of fuel cells. The use of Pt alloys with transition metals has been recognized as an effective approach for this purpose. In addition, the alloying with some transition metals results in improved catalytic activity. For example, it was found that Pt3Co (atomic ratio) is approximately three times more active than Pt in terms of kinetically controlled mass activity (MA).[7,8]

We have developed a unique preparation method for platinum alloy (V, Ni, Cr, Co, and Fe) nanoparticles, in which metal precursors were encapsulated in reverse micelles (nanocapsules) and reduced in the presence of an electroconductive carbon black support.[9] The alloy nanoparticles thus obtained had small particle size distribution and controlled alloy composition and were well-distributed on the electroconductive support materials such as carbon black. PtCo/C catalysts exhibited an higher ORR activity than that of Pt/C in the acidic media; however, the activity decreased at an elevated temperature or potential cycling because of surface dealloying.[7,8]

Replacing acidic membranes with alkaline membranes is potentially advantageous for the use of transition metals as cathode materials because their oxides are stable in alkaline media.[10] It has also been reported that PtCo alloys exhibited a high ORR catalytic activity in the alkaline media.[11−14] In this study, we investigate the ORR activity in the alkaline media of the PtCo/C catalysts prepared by the nanocapsule method. A series of the alloy catalysts were prepared and characterized to find out the optimum alloy composition.

Results and Discussion

PtCo alloy nanoparticles were prepared via slightly modified nanocapsule method. Briefly, Pt(acac)2 and Co(acac)3 were used as metal precursors, oleic acid and oleylamine were used as surfactants, diphenyl ether was used as a solvent, and LiBEt3H was used as a reducing agent (see Supporting Information for details). Although Ketjenblack has often been used as a carbon support, acetylene black (AB800, 779 m2 g–1 of Brunauer–Emmett–Teller specific surface area) was adopted in the present study because the latter contains less interior nanopores and thus contributes to higher catalyst utilization (metal nanoparticles located in the interior nanopores of the carbon support are much less likely to function as a catalyst).[15] The loading amounts of the metals were determined by TG/DTA analyses to be from 16.3 to 20.6 wt %, which were within acceptable errors for the feed metal composition (projected value was 20.0 wt %). The average alloy compositions obtained by X-ray fluorescence (XRF) analyses were x = 0, 31, 39, 42, 51, and 73 atom % for Pt100–Co, which were similar to the feed values (0, 25, 40, 45, 50, and 75 atom %, respectively).

Figure shows transmission electron microscopy (TEM) image and particle size distribution histogram of Pt61Co39/AB800 as the typical example (see Figure S1 for the other samples). From the image, it was revealed that the nanoparticles with small diameter were well-distributed on the high-surface-area acetylene black carbon support. The average particle size (and the standard deviation) was 3.0 (±0.4) nm. Similar results were obtained for the other Pt100–Co/AB800 samples with different alloy compositions. The X-ray diffraction (XRD) patterns (Figure S2) of Pt100–Co showed diffraction peaks assignable to face-centered cubic (fcc) phase of Pt in addition to the characteristic broad peak of the carbon support (2θ = approximately 25°). No peaks assignable to Co metal and oxides were detected in any of these samples. The lattice constants for the alloy samples decreased when increasing the Co content (Figure S3). As expected, the change in the lattice constant was smaller than that calculated from the Vegard’s law because of the hexagonal close-packed (hcp) structure of Co metal. The nearly linear relationship between the lattice constant and the Co content indicates the formation of the solid solution with fcc structure. The average crystallite size was calculated by Scherrer’s equation using the peak at approximately 67° (200 plane) and is summarized in Table . These values were in fair agreement with the average particle sizes obtained from the TEM images. The results were very similar to those of our previously reported Pt100–Co alloy nanoparticles supported on a different carbon support (Ketjenblack).[9]

Figure 1

TEM image and particle size histogram of Pt61Co39/AB800.

Table 1

Properties of Pt100–Co/AB800 Catalysts

x (atom %)	dTEMa (nm)	dXRDb (nm)	MAc (A gmetal–1)	MAd (A gPt–1)	PH2O2e (%)
0	3.2	2.9	294	294	0.7
31	3.4	2.4	329	373	0.7
39	3.0	2.1	641	767	0.3
42	3.3	2.3	628	767	0.4
51	2.7	1.9	294	380	0.3
73	2.8	2.0	316	576	0.2

Average particle size obtained from the TEM images.

Average crystallite size calculated by Scherrer’s equation using the peak at approximately 67° (200 plane).

MA (per total metal) for the ORR at 0.85 V vs RHE.

MA (per platinum) for the ORR at 0.85 V vs RHE.

Percentage of H2O2 production with respect to the overall ORR at 0.80 V vs RHE.

The Pt100–Co/AB800 catalysts were loaded on a glassy carbon (GC) disk electrode and covered with a thin layer (approximately 50 nm thick) of a commercial anion exchange polymer (Tokuyama AS-4). The catalyst particles were seen to be well-dispersed and relatively uniform on the GC disk without notable aggregation. Figure shows cyclic voltammograms (CVs) of the electrodes in 0.1 M KOH aqueous solution purged with nitrogen at 25 °C. The CVs showed typical redox waves for surface Pt at approximately 0.8 V versus reversible hydrogen electrode (RHE) and hydrogen adsorption/desorption waves for polycrystalline Pt below approximately 0.3 V versus RHE. The presence of Co in the alloy samples contributed to the development of anodic and cathodic peaks at approximately 0.68 and 0.45 V versus RHE, respectively. This broad redox peak couple could be associated with the surface Co hydroxide species. The electric charge calculated from the reduction peak area increased when increasing the Co content from 13.5 C g–1 (x = 31), 25.1 C g–1 (x = 39), 29.3 C g–1 (x = 42), to 45.2 C g–1 (x = 51) because of the increased number of Co atoms on the surface. However, it decreased to 32.2 C g–1 for x = 73. The results suggest that dealloying was likely to take place rather easily and immediately in alkaline solution when Co content was high. Nevertheless, the potential cycling between 0.05 and 1.0 V versus RHE did not practically change the shapes of the CVs at least up to 80 cycles, indicating a stable surface structure. The electrochemically active surface areas (ECSAs) were calculated from the electric charge of the hydrogen desorption wave, assuming 0.21 mC cm–2 for smooth polycrystalline Pt. The ECSA was 23 m2 g–1 for x = 31 and decreased with increasing x (12 m2 g–1 for x = 73) (Figure S4). The values were significantly lower than those (approximately 80–100 m2 g–1) calculated from the diameter (dTEM) of the alloy nanoparticles and those (approximately 77 m2 g–1) of Pt67Co33/AB800 in acidic conditions.[16] The results suggest that, unlike in acid solution where Pt skin layer forms on the surface of PtCo nanoparticles by potential cycling, Co atoms were present on the surface of the nanoparticles (Pt100–Co/AB800) in alkaline solution. A similar result was obtained for the commercial Pt75Co25/AB800 catalyst (Tanaka Kikinzoku Kogyo TEC36F52) whose ECSA was also low, 50 m2 g–1 (based on dTEM) and 21 (ECSA) m2 g–1 (data not shown).

Figure 2

CVs of Pt100–Co/AB800 (x = 0, 31, 39, 42, 51, and 73) in 0.1 M KOH saturated with N2 at 25 °C. The scan rate was set at 50 mV s–1.

Figure shows hydrodynamic voltammograms for the ORR at the ionomer (AS-4)-covered Pt100–Co/AB800 working electrode in air-saturated 0.1 M KOH aqueous solution at 25 °C. (The rotation rate was 1750 rpm. See Figure S5 for the data of other rotation rates.) The ORR current started at 0.970 V and reached to diffusion limit at approximately 0.5 V for x = 0. With an increasing Co content, the ORR currents started at higher potential and reached at the maximum (0.995 V) with x = 39 and 42. Further increase in x resulted in lower potential for the commencement of ORR currents. The formation of hydrogen peroxide (PH) with respect to the overall ORR calculated from the current detected at the ring electrode at 0.8 V was negligibly small (<0.7%) and became even smaller with an increase in x (PH = 0.2% for x = 73), indicating that the four electron transfer process was dominant. The formation of H2O2 was similarly low in acidic conditions for PtCo nanoparticle catalysts prepared by the nanocapsule method.[7]

Figure 3

Hydrodynamic voltammograms for ORR in 0.1 M KOH saturated with air at 25 °C at Pt100–Co/AB800 catalysts deposited on a GC disk electrode (1750 rpm). The scan rate was set at 5 mV s–1. The ring potential was set at 1.1 V vs RHE to detect H2O2 produced at the catalysts.

To quantify the effect of the alloy composition in Pt100–Co/AB800 on the ORR catalytic activity, MA per platinum for ORR at 0.85 V was calculated and is plotted as a function of Co content in Figure . It was found that Pt100–Co/AB800 exhibited the maximum MA (767 A g–1) with x = 39 and 42. The MA values were comparable or somewhat higher than those reported for PtCo alloy nanoparticles deposited on the carbon support prepared by different methods.[12] Pt100–Co/AB800 also exhibited the maximum MAs (641 and 628 A g–1) per metal at the same compositions. As is discussed in the literature,[12] it is considered that the alloying effect on the enhancement of ORR catalytic activity is based on weakened chemical adsorption of OH species (OHads) onto surface Pt atoms, resulting in more available sites for ORR. However, further increasing Co contents (>approximately 40 atom % in the present case) reduced surface Pt atoms and thus lowered the ORR catalytic activity, resulting in the volcano-type dependence of the ORR activity on the Co content. Compared with the acidic conditions where the maximum MA was obtained at x = 25,[7] the optimum Co content was higher in the alkaline conditions. It is noticeable that the composition range of Co to show enhanced ORR activity is very narrow. The differences in the optimum Co content and its composition range between acidic and alkaline conditions might be associated with the surface Co hydroxide species. For reference, MA per platinum was also included in Figure for commercial Pt/C catalyst. Pt/AB800 prepared by the nanocapsule method exhibited higher catalytic activity (MA) than that of the commercial Pt/C catalysts presumably because the Pt nanoparticles were more uniform and better dispersed on the carbon support for the former catalyst. Similar results were obtained in acidic conditions as reported in our previous reports.[15,17]

Figure 4

MA (per platinum) of Pt100–Co/AB800 catalysts for the ORR reaction at 0.85 V vs RHE as a function of Co content. The open symbol at x = 0 was for the commercial catalyst, Pt/C (Tanaka Kikinzoku Kogyo TEC10E50E).

The durability of Pt61Co39/AB800 was evaluated with an accelerated durability test (ADT), which was comprised of potential step cycles between 0.6 and 1.0 V versus RHE in 0.1 M KOH aqueous solution saturated with nitrogen at 40 °C. During ADT, the redox peaks (ascribed to hydrogen adsorption/desorption and oxidation/reduction of surface Pt and Co atoms) in the CV curve became smaller (Figure S6). The ECSA was 40 m2 g–1 initially (note that the CVs and hydrodynamic voltammograms were measured at 40 °C, and thus, initial ECSA and MA values in Figure were higher than those measured at 25 °C in Figures and S4 and Table ) and decreased to approximately 12 m2 g–1 within 2000 cycles. Then, the ECSA gradually leveled off to 9 m2 g–1 at the end of testing (30 000 cycles) (Figure a). The commercial Pt/C exhibited similar behavior under the same conditions. Although the initial ECSAs were different, these two catalysts showed comparable ECSAs at the end of testing. The MA values for two catalysts decreased when increasing the step cycle numbers (Figure b), in a trend similar to ECSA values. The results would suggest that coarsening of the nanoparticles happened during the ADT. This idea was partly supported by the TEM image of the post-ADT sample (Figure S7), in which agglomeration of the nanoparticles was observed. However, 3.6 nm of the average particle size (dTEM) was slightly higher than the initial value (3.0 nm) and did not account for a large decrease in MA. Energy-dispersive X-ray (EDX) analyses suggested that the average Co content of the post-ADT sample was 26 and 13 atom % lower than that of the pristine sample. From the volcano dependence of ORR activity on Co content in Figure , large loss in MA seemed reasonable. Therefore, it is considered that both coarsening and dealloying in Pt61Co39/AB800 during ADT contributed to the loss in the ORR catalytic activity.

Figure 5

Changes in (a) ECSA and (b) MA (per platinum) for Pt61Co39/AB800 (closed symbols) and commercial Pt/C (open symbol) catalysts during the potential step cycles in the ADT.

Conclusions

A series of Pt100–Co alloy nanoparticles dispersed on a high-surface-area acetylene black carbon (AB800) support were prepared by our original nanocapsule method, and their ORR catalytic activity was evaluated in the alkaline media. Unlike the acidic media where PtCo alloys form Pt skin layer, cobalt atoms seemed rather stably present on the particle surface during the initial electrochemical potential cycling (at least 80 cycles) in alkaline solution that caused smaller ECSAs. The Pt100–Co alloy nanoparticles exhibited volcano-type dependence of the ORR activity on the alloy composition. At the optimum composition (x = approximately 40 atom %), the MA per platinum was 2.6 and 6.4 times higher than those of Pt/AB800 (prepared via the same method) and the commercial Pt/C, respectively. However, in a durability test comprised of potential step cycles between 0.6 and 1.0 V, the PtCo alloy catalysts lost ORR catalytic activity because of particle agglomeration and dealloying. Because the nanocapsule method is potentially applicable to other metals and alloys, we will continue to explore more active and stable ORR catalysts both in the acidic and alkaline media.

Experimental Section

Materials

Platinum(II) acetylacetonate (Pt(acac)2, Aldrich), cobalt(III) acetylacetonate (Co(acac)3, Aldrich), LiBEt3H (Kanto Chemical Co., Inc.), 1,2-hexadecanediol (TCI Co., Ltd.), oleic acid (Kanto Chemical Co., Inc.), oleylamine (ACROS), diphenyl ether (Kanto Chemical Co., Inc.), acetylene black (AB800, specific surface area = 800 m2 g–1, Denka Company Ltd.), and ultrapure potassium hydroxide aqueous solution (3 mol/L) (Kanto Chemical Co., Inc.) were commercially available reagents and used as-received. A 5 wt % aqueous solution of anion exchange polymer [AS-4, ion-exchange capacity (IEC) = 1.4 mequiv/g] was kindly supplied by Tokuyama Co.

Preparation of Carbon-Supported PtCo Nanoparticle (PtCo/AB800) Catalysts

PtCo/AB800 catalysts were prepared according to the following procedure slightly modified from our previous method. A typical procedure is as follows: Pt(acac)2 (0.13 mmol), Co(acac)3 (0.13 mmol), and 1,2-hexadecanediol (260 mg) were dissolved in 12.5 mL of diphenyl ether under a nitrogen flow in a 100 mL round-bottom flask equipped with a magnetic stirrer bar. The mixture was heated at 110 °C for 30 min followed by the addition of oleic acid (0.27 mmol) and oleylamine (0.24 mmol). The mixture was heated at 220 °C for 30 min. LiBEt3H (1.0 mL) was added dropwise into the mixture. The mixture was heated at 270 °C for 30 min and then cooled to 200 °C. The mixture was slowly added into a suspension of acetylene black AB800 (129.6 mg) in 12.5 mL of diphenyl ether. The mixture was heated again at 270 °C for 1 h and then, cooled to 50 °C. The product was recovered by filtration, washed with ethanol several times, and dried at 60 °C in a vacuum oven. The black powders thus obtained were heated at 400 °C for 4 h in 5% hydrogen (nitrogen balance).

Measurements

PtCo/AB800 catalysts were characterized by XRD (Rigaku RINT2000) with Cu Kα radiation (50 kV, 300 mA), TEM (Hitachi H-9500, acceleration voltage = 200 kV), and XRF (Shimadzu EDX-800). The loaded amounts of the metal on carbon were quantified by thermogravimetric analyses (TGA, Rigaku Thermo plus TG8120) in air from room temperature to 600 °C.

Preparation of the Electrodes

A typical rotating ring-disk electrode (RRDE) was used. The Pt ring electrode had a 3.5 mm inner radius and a 4.5 mm outer radius. The GC disk electrode had a 3 mm radius (0.283 cm2 geometric area Ageo). The electrode was polished with 1 μm alumina paste for 30 min, 0.3 μm alumina paste for 10 min, and 0.05 μm alumina paste for 10 min, to ensure a mirror finish, and was then washed with ethanol and water in an ultrasonic bath.

PtCo/AB800 (14.3 mg) was dispersed in 15 mL of 99% ethanol. The mixture was mixed in an ultrasonic bath to obtain a homogeneous suspension, of which 10 μL of aliquots was pipetted onto the GC disk electrode, which was dried slowly at room temperature under a nearly ethanol-saturated atmosphere to avoid aggregation of the PtCo/AB800 particles. The loading amount of PtCo was calculated to be 6.3 μg cm–2. A 2 μL of aliquot of Tokuyama AS-4 aqueous solution was pipetted onto the PtCo/AB800-loaded GC disk electrode. The concentration of Tokuyama AS-4 was 0.22 wt % to achieve the targeted thicknesses of Tokuyama AS-4 thin layer (approximately 0.05 μm), which was estimated based on the specific gravity of the dry membrane. The solutions were dried slowly at room temperature under a nearly ethanol-saturated atmosphere to obtain the Tokuyama AS-4-PtCo/AB800-loaded electrodes.

Electrochemical Techniques

RRDE equipment (Nikko Keisoku RRDE-1) with a gas-tight Pyrex glass cell was used to evaluate the ORR catalytic activity of the electrodes. A ring-shaped platinum wire and a RHE were used as the counter electrode and the reference electrode, respectively. All electrode potentials are stated relative to the RHE. The electrolyte solution, 0.1 M KOH, was prepared from 3 M KOH and Milli-Q water (Millipore). Before the ORR experiments, the working disk electrode was cleaned by cycling the potential between 0.05 and 1.0 V at a sweep rate of 0.5 V s–1 in deaerated 0.1 M KOH until steady voltammograms were obtained. The ECSA of Pt, APt, was evaluated from the electric charge of the hydrogen adsorption ΔQH in the negative-going potential scan from 0.05 to 0.40 V in cyclic voltammetry at a sweep rate of 0.05 V s–1 at 25 °C. After bubbling air in 0.1 M KOH for 30 min, hydrodynamic voltammograms for the ORR at the working disk electrodes were recorded by sweeping the potential from 0.2 to 1.0 V at a rate of 5 mV s–1 and at rotation rates of 1000, 1250, 1500, 1750, 2000, 2250, 2250, and 2750 rpm. The Pt ring collection electrode was potentiostatted at 1.1 V, where the anodic oxidation of hydrogen peroxide is diffusion-limited. The collection efficiency for this RRDE system was determined experimentally to be 0.365 using 1 mM K3[Fe(CN)6] in 0.5 M K2SO4 aqueous solution. All of these electrochemical experiments were performed at 25 °C.

Durability Test

The durability test of Pt61Co39/AB800 and commercial Pt/C catalysts were performed in 0.1 M KOH saturated with nitrogen at 40 °C. The potential was stepped between 0.6 and 1.0 V, with a holding period of 3 s at each potential (6 s for one cycle). After a given number of potential step cycles (N), RRDE measurement was taken as mentioned above to examine the changes in the CV, ECSA, and ORR catalytic activities at 40 °C.