Highly Active, CO-Tolerant, and Robust Hydrogen Anode Catalysts: Pt

Dec 6, 2016 - The electrocatalytic activity for the hydrogen oxidation reaction (HOR) in the presence of 1000 ppm of CO has been investigated on a ser...
0 downloads 0 Views 3MB Size
Research Article pubs.acs.org/acscatalysis

Highly Active, CO-Tolerant, and Robust Hydrogen Anode Catalysts: Pt−M (M = Fe, Co, Ni) Alloys with Stabilized Pt-Skin Layers Guoyu Shi,† Hiroshi Yano,‡ Donald A. Tryk,‡ Akihiro Iiyama,‡ and Hiroyuki Uchida*,‡,§ †

Interdisciplinary Graduate School of Medicine and Engineering, ‡Fuel Cell Nanomaterials Center, and §Clean Energy Research Center, University of Yamanashi, Takeda 4, Kofu, 400 8510, Japan S Supporting Information *

ABSTRACT: The electrocatalytic activity for the hydrogen oxidation reaction (HOR) in the presence of 1000 ppm of CO has been investigated on a series of binary Pt alloy catalysts Pt−M (M = Fe, Co, Ni), having two atomic layers of stabilized Pt skin (Pt2AL), supported on carbon black (Pt2AL−PtFe/C, Pt2AL− PtCo/C, and Pt2AL−PtNi/C) in 0.1 M HClO4 solution at 70 and 90 °C. It was found that Pt2AL−PtFe/C exhibited the highest CO-tolerant HOR activity (with respect to the area-specific activity js and the mass activity MA), followed by Pt2AL−PtCo/C and Pt2AL−PtNi/C. Such an order of the js values for the HOR with and without adsorbed CO can be correlated with density functional theory calculations, which have enabled us to propose a mechanism for the HOR on these surfaces. The apparent values of MA for the HOR on Pt2AL−PtFe/C at 20 mV vs RHE were 2−3 times larger than those for the conventional commercial catalyst c-Pt2Ru3/C over the whole CO coverage range from 0 to 0.7 at 70 and 90 °C. For an accelerated durability test simulating air exposure (2500 potential cycles between 0.02 and 0.95 V), the apparent js values for the CO-tolerant HOR on these Pt-skin catalysts were maintained completely, indicating that the dealloying of M components was virtually suppressed, whereas a significant reduction in js was observed for c-Pt2Ru3/C. A great mitigation of the particle agglomeration was also a highly attractive property of our catalysts in comparison with the commercial catalysts c-Pt/C and c-Pt2Ru3/C. KEYWORDS: Pt skin, alloy electrocatalysts, anode catalysts, CO tolerance, hydrogen oxidation reaction

1. INTRODUCTION Polymer electrolyte fuel cells (PEFCs) are being actively developed as primary power sources for fuel cell vehicles (FCVs) and residential cogeneration systems. For FCVs with highly purified H2 as the fuel, the hydrogen oxidation reaction (HOR) rate is sufficiently fast with an anode catalyst consisting of a small loading of Pt nanoparticles supported on carbon black (Pt/C, ca. 0.05 mgPt cm−2). For residential PEFCs operated with H2-rich fuel gas (reformate), however, the HOR at Pt catalysts is severely poisoned even by traces of CO (below 10 ppm) in the reformate due to strong adsorption, blocking the active sites.1 Thus far, Pt−Ru alloy anode catalysts have been used to mitigate the poisoning by low concentrations of CO. It has been demonstrated for some Pt alloys including Pt− Ru, by Fourier transform infrared (FTIR) spectroscopy and Xray photoelectron spectroscopy combined with an electrochemical cell (EC-XPS), that an electronic modification effect is essential for weakening CO adsorption to maintain the active sites for the HOR at a practical potential: i.e., E < 0.1 V vs RHE.2,3 However, Ru is not only costly but also unstable at E > 0.8 V, leaching into the acidic electrolyte membrane, followed by migration and deposition at the Pt cathode, a process that can attenuate the activity of both anode and cathode catalysts and lead to a loss of cell performance. Therefore, the Pt−Ru anode in a state of the art residential PEFC system such as © XXXX American Chemical Society

EneFarm has usually been protected from oxidation due to contact with air, which makes the system complicated and costly. Furthermore, the CO tolerance of the Pt−Ru catalyst is not sufficient when operated with high concentrations of CO (≥500 ppm) in a simplified, low-cost fuel processing system. To reduce the cost of the system and facilitate larger scale commercialization, a new anode catalyst with high CO tolerance and robustness in the presence of impurities such as oxygen must be developed. Recently, we have found that both a Pt−Co sputtered alloy film and a commercial c-Pt3Co/C catalyst exhibited CO tolerance comparable to that of commercial c-Pt2Ru3/C in 0.1 M HClO4 solution at elevated temperatures ( Pt−Co > Pt−Ni. All of the catalysts showed a decrease in MAk and jk at 90 °C in comparison to those at 70 °C, which is ascribed mainly to the decreased H2 concentration in the solution at high temperature, but the trend of the values was unchanged. Then, we calculated an apparent rate constant at 20 mV vs RHE per unit ECA, kapp, corrected for the change in hydrogen concentration dissolved in the bulk of the electrolyte solution, [H2], with temperature, in order to discuss the temperature dependence of the specific activity for the HOR. The kapp values were evaluated from the equation4

2. EXPERIMENTAL SECTION The Pt2AL−PtFe/C, Pt2AL−PtCo/C, and Pt2AL−PtNi/C catalysts were prepared by the use of a high-surface-area (specific surface area 780 m2 g−1) carbon black support (Denka Co., Ltd.) in the same manner as that described previously.9,10 Two commercial catalysts, c-Pt2Ru3/C and c-Pt/C, were used for comparison. From TEM images of the catalysts shown in Figure S1 in the Supporting Information, the average particle sizes dTEM and the standard deviations σd of the Pt2AL−PtFe/C, Pt2AL−PtCo/C, Pt2AL−PtNi/C, c-Pt2Ru3/C, and c-Pt/C catalysts were 2.9 ± 0.4, 3.3 ± 0.5, 3.2 ± 0.4, 3.5 ± 0.9, and 2.2 ± 0.5 nm, respectively. In comparison with c-Pt2Ru3/C and c-Pt/C (σd/dTEM > 20%), the Pt2AL−Pt−M nanoparticles were uniformly dispersed on the carbon black support, and their size distributions were fairly narrow, characteristics that can be ascribed to our use of the nanocapsule method.11,12 The experimental procedure by the use of a channel flow electrode cell (CFE) was the same as that described in ref 4. The working electrode was an Au substrate with geometric area of 0.04 cm2, upon which the respective Nafion-coated catalysts were uniformly dispersed. The amount of each catalyst was maintained at a constant loading of carbon support of 11 μg cm−2, which corresponds to a height of approximately two monolayers of the carbon black particles. The Nafion film was coated on the catalyst layer with an average thickness of 0.075 μm. The thickness of the Nafion film was calculated from its mass and the electrode surface area assuming a density of 1.98 g cm−3 in its dry state. All electrode potentials are referred to the reversible hydrogen electrode (RHE). The kinetically controlled current Ik at a given potential E was determined from the hydrodynamic voltammograms in the CFE by use of the equation4,13 1/I = 1/Ik + 1/IL

−1

jk /2F = kapp[H 2]

(2)

Figure 1 shows the Arrhenius plots of kapp for the HOR on various catalysts at 20 mV in H2-saturated (CO-free) 0.1 M HClO4 solution. Somewhat linear relationships between log kapp and 1/T were obtained for all of the catalysts but with a slight degree of curvature, which suggests a change in the ratedetermining step (rds) with increasing temperature or a change in the predominance of two steps in mixed kinetic control, as will be briefly discussed later. It can be seen that the kapp values increased in the order c-Pt/C ≤ c-Pt2Ru3/C < Pt2AL−PtNi/C < Pt2AL−PtCo/C < Pt2AL−PtFe/C in the whole temperature range. The value of kapp for Pt2AL−PtFe/C was larger than that for c-Pt/CB by a factor of ca. 2.3. The enhancement factors in kapp for Pt2AL−PtCo/C and Pt2AL−PtNi/C were 2.0 and 1.5, respectively. The apparent activation energies Ea at lower temperature for each electrode were found to be similar, in the range of ca. 9−13 kJ mol−1, indicating that the rds for the HOR might be similar among these electrodes. It is noteworthy that these Ea values are consistent with those obtained experimentally for the most HOR active surfaces, with 9.5 kJ mol−1

(1) 268

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274

Research Article

ACS Catalysis

for c-Pt2Ru3/C and all three Pt2AL−Pt−M/C catalysts. With elevation of the temperature to 90 °C, the CO tolerance for all of the catalysts was improved, but the order of MAapp was unchanged. To discuss the CO tolerance mechanism of these catalysts, we focus on the dependence of MAapp on CO coverage, θCO (Figure 2). The values of θCO at a given tad in Figure S2 in the Supporting Information were determined approximately from the number of Pt metal sites blocked by CO, irrespective of the type of bonding: e.g., linear (on-top), bridged, or others. As shown by the black line for the c-Pt/C electrode (Figure 2), the value of MAapp decreased rapidly with increasing θCO at 70 °C, probably due to a decrease in hydrogen adsorption sites caused by strongly adsorbed CO. In contrast, the Pt2AL−Pt−M/C and c-Pt2Ru3/C catalysts exhibited excellent CO tolerance at 70 °C, suggesting that the HOR active sites were not so rigidly blocked by COad, due to its enhanced mobility. Specifically, the CO tolerance increased in the order c-Pt/C < c-Pt2Ru3/C < Pt2AL− PtNi/C < Pt2AL−PtCo/C < Pt2AL−PtFe/C: e.g., at θCO = 0.7, the retention in MAapp was 47%, 74%, 75%, 86%, and 90%, respectively. The MAapp value on Pt2AL−PtFe/C at θCO = 0.7 was 2.9 times higher than that on c-Pt2Ru3/C and 3.3 times higher than that on c-Pt/C. It should be mentioned that we have seen one report of inferior CO tolerance at a Pt surface on enriched Pt−Ni in comparison to Pt−Co and Pt−Fe in a single cell test, which appears to be consistent with our experimental results.17 At 90 °C, high CO tolerance at the Pt2AL−PtFe/C catalyst was maintained: the MAapp value at θCO > 0.6 was as high as 95% of that at θCO = 0 (negligible loss). For c-Pt2Ru3/C, the CO tolerance was, as expected, improved by elevating the temperature to 90 °C, so that the dependence of MAapp on θCO became smaller than that at 70 °C. However, at the identical coverage θCO = 0.7, the enhancement factors of MAapp (compared with c-Pt2Ru3/C) for the Pt2AL−Pt−M/C catalysts described above were unchanged. These results convince us that the Pt2AL−Pt−M/C materials are superior CO-tolerant anode catalysts in the practical temperature range, which would make it possible to reduce the amount of platinum-group metals (Pt + Ru) in comparison with the case of the conventional c-Pt2Ru3/C catalyst. 3.3. DFT Studies for HOR Activity and CO Tolerance of Pt2AL-Pt-M/C. It has long been recognized that alloying can decrease the CO adsorption strength on Pt-based alloys, and this is thought to be the main reason for the increased CO tolerance in the practical potential range of the hydrogen anode (E < 0.1 V).18−20 Previous work has focused almost entirely on

Figure 1. Arrhenius plots for the apparent rate constant kapp for the HOR (CO-free) at the Nafion-coated supported catalysts. Dashed lines show the possible effect of lower activation energies at higher temperatures.

for Pt(110) and 12 kJ mol−1 for Pt(100),14 as well as for carbon-supported Pt and Pt alloys (9.6−10.9 kJ mol−1).15 The slight curvature means that, with an elevation of the temperature, the rds might change to one with a lower Ea value; these range from ca. 4 to 6 kJ mol−1. In any case, there is not a large thermal enhancement in the HOR kinetics at these catalysts, while for the oxygen reduction reaction (ORR), the kinetics are known to be more temperature dependent (with a typical Ea value of ca. 42 kJ mol−1).16 3.2. HOR Activities in the Presence of Adsorbed CO. Carbon monoxide was adsorbed on the catalysts by flowing 0.1 M HClO4 solution saturated with 1000 ppm of CO with H2 balance (1000 ppm of CO/H2) at 50 mV for a given period, tad, followed by the evaluation of the CO-tolerant HOR activity at 20 mV in pure H2-saturated 0.1 M HClO4 solution. The potential of 50 mV for CO adsorption has been chosen to maximize the CO coverage.4 As reported previously,7 the limiting currents for the HOR for all catalysts decreased due to the blocking of the Pt active sites by the adsorbed CO. Thus, it is not possible to obtain the true Ik values, and therefore, for convenience, we have calculated an apparent mass activity at 20 mV, MAapp, as a measure of the CO tolerance, by directly using the measured current I divided by the metal mass. We have measured changes in the MAapp with tad at 70 and 90 °C, as shown in Figure S2 in the Supporting Information. While MAapp for c-Pt/C operated at 70 °C decreased rapidly with increasing tad, the decreases in MAapp were much slower

Figure 2. Dependence of MAapp at 20 mV vs RHE on θCO at Nafion-coated electrodes measured in H2-saturated 0.1 M HClO4 solution at 70 and 90 °C. 269

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274

Research Article

ACS Catalysis

Figure 3. Adsorption energies for (A) H2 or 2H, (B) CO, and (C) H2O at step edges (triangles) and terraces (squares). In (A), the solid symbols denote dissociated 2H, while the open symbols denote undissociated H2. In all cases except for PtRu, undissociated H2 does not adsorb on the surface, at either step edges or terraces. On pure Pt(221), H2 dissociates spontaneously when close to the surface, while on Pt1AL−PtFe, Pt1AL−PtCo and Pt1AL−PtNi, it “floats” away from the surface. It should be noted that the surface unit cells for the Pt skin/Pt alloys are rhomboid to avoid a large unit cell, comprising two steps and two terraces, which would have been needed to obtain a rectangular geometry, due to the alternation of Pt and Co in the bulk structure. For Pt(221), a rectangular surface unit cell can be achieved with a single step and terrace. There is expected to be only a slight effect of the rhomboid geometry, and this would be the same for all of the alloy surfaces.

strength for the corresponding PtRu(221) step edge was calculated here and found to be stronger, only 0.22 eV weaker than that on the pure Pt step edge. We also examined the on-top CO adsorption on the (111) terraces, finding generally weaker adsorption than the values on the step edges, but with a larger range of values: i.e., 0.73 eV weaker than pure Pt for Pt1AL−PtFe, 0.70 eV for Pt1AL−PtCo, 0.54 eV for Pt1AL−PtNi, and 0.24 eV for PtRu. Due to the stronger adsorption on the step edges, CO should preferentially adsorb at those sites. The values obtained for Pt, PtRu, and Pt1AL−PtFe are similar to those obtained by Greeley et al. for near-surface alloys (NSA) with the (111) orientation, in which the top layer is pure Pt and the second layer is the pure alloying element27 (see Table S1 in the Supporting Information), with differences of at most 0.1 eV. Thus, we conclude that the calculated CO adsorption strengths are generally consistent with the experimental CO tolerances observed in the present work. As discussed later, the CO adsorption on both step edges and terraces is likely to be important. Regarding Pt-skin-covered Pt-alloy (111) surfaces of the Pt3M type, Stamenkovic et al. have reported DFT calculations for CO adsorption energies, which also became less negative (increasingly weaker adsorption) in the same order as that reported here, Pt(111) < Pt skin/Pt3Ni(111) (denoted as Pt/ Pt3Ni(111) in Table S1 in the Supporting Information) < Pt/ Pt3Co(111) < Pt/Pt3Fe(111),28 with comparable values. To our knowledge, our recent report is one of the first that has clearly shown the effect of alloying on the HOR activity itself in acid solution.7 In alkaline solution, Wang et al. have shown that alloying of Ru with Pt leads to enhanced HOR activity, and they attribute the enhancement to the weakening of the H adsorption on the Pt surface, specifically at (110) steps.29 In the present work, we extend this idea to additional alloys and seek to understand the underlying reasons for the enhancement in activity. First, we examine in more detail the hypothesis that H adsorption should be weakened for increased HOR activity. Similar to the case of CO adsorption, alloying has been recognized to weaken the hydrogen adsorption on Pt(111), according to DFT calculations for several Pt−M near-surface

the (111) surface, with the exception of our recent work, which focused on the (110) surface.7 In the present work, we have again carried out all-electron relativistic density functional theory (DFT) calculations, which in the past have been found to correctly predict the CO binding sites21,22 and HOR kinetics on various Pt(hkl) surfaces, including Pt(111),23,24 Pt(110),25 and Pt(100). 26 The Pt(110) surface has been found experimentally to exhibit the highest catalytic activity for the HOR, in comparison to those for Pt(111) and Pt(100).14 In the present work, we have selected the stepped surface (221), including narrow (111) terraces (three atoms wide) and (110) steps, which we believe is a realistic model for nanoparticles in a practical size range. The (110) steps resemble the ridges on the actual (110) surface, which we examined in a previous paper with DFT calculations.7 Those calculations showed that both CO and H adsorb more weakly on Pt2AL−PtCo(110) in comparison with Pt(110), consistent with the high CO tolerance and HOR activity of this catalyst.7 In extending that work to include Pt2AL−PtFe/C and Pt2AL−PtNi/C, however, we found that we were unable to fully explain the differences in CO tolerance and HOR activity and therefore have examined the stepped surface (221) as a more realistic model. As a first step in understanding these nanoparticle alloy surfaces, we examined the Pt1AL−PtM(221) surfaces with the second layer consisting of pure M (Figure 3; see also Figure S3 in the Supporting Information). On these surfaces, the adsorption strengths for on-top CO at the (110) step edges, −2.40 eV for pure Pt and −1.83 eV for Pt1AL−PtCo, were found to be similar to those we reported previously for the (110) surface of pure Pt, −2.21 eV, and Pt1AL−PtCo, −1.88 eV, due to their similar structures. Looking at the full array of alloys studied in the present work, also including Pt−Ru and pure Pt for comparison, we can see that the adsorption strengths for CO at step edges were significantly smaller, by 0.51, 0.57, and 0.39 eV, for M = Fe, Co Ni, than that on the pure Pt(221) step edges (see Table S1 in the Supporting Information). These values are somewhat consistent with the observed trend in increased CO tolerance for the present catalysts, except that the adsorption on the PtCo(221) step edges was weaker than that on Pt1AL−PtFe(221). For comparison, the CO adsorption 270

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274

Research Article

ACS Catalysis

Figure 4. (A) Changes in MAapp for the HOR with 30 min of CO adsorption in 1000 ppm of CO/H2 at Nafion-coated electrodes as a function of potential cycle number N in the durability test. The MAapp values were measured in the same manner as in Figure 2. (B) Plots of japp at 20 mV vs RHE for the CO-tolerant HOR (30 min of CO adsorption in 1000 ppm of CO/H2 saturated solution with the potential held at 50 mV) measured in H2-purged 0.1 M HClO4 solution at 70 °C as a function of potential cycle number N in the durability test.

alloys,30 by experiments for Pt3Ni(111),31 and for a series of Pt−Co(111) alloy single crystals.32 In the quest for COtolerant HOR catalysts, the focus has usually been on weakening CO adsorption with only secondary emphasis on H adsorption. For (110)-like steps and surfaces on Pt and its alloys, both CO and H adsorption are significantly stronger than that on (111) surfaces (see Table S1 in the Supporting Information). The (110) steps on (111) surfaces would be expected to behave in a manner similar to that for the (110) surface itself. These trends can be seen clearly for the step edges and terraces on the (221) surfaces (Figure 3). It is important to also emphasize the fact that, on nearly all of the surfaces studied, at both step edges and terraces, CO adsorption is stronger by approximately 1 eV in comparison with H adsorption. Thus, in order for H to compete effectively with CO for adsorption sites, the CO concentration must be maintained at very low levels. In addition, it is useful to take note of the adsorption strength of water at these sites (Figure 3). Then, it becomes clear that H adsorption can only compete with water adsorption at the step edges. Thus, we propose that H2 adsorption and dissociation occur at the step edges. With single-crystal alloy (110) surfaces, changes in the UPD H peaks have been observed in comparison with pure Pt(110), but they have not been as clear as expected,31,32 possibly due to the heterogeneity of H adsorption sites on Pt/Pt3Co(110): i.e., either Pt or Co can be underneath the surface Pt atom.7 However, in the present work, all adsorption sites are equivalent for the simulated Pt1AL−PtM(221) step edges, because the M atoms are present under all of the Pt atoms at the step. The adsorption strengths were computed for the “V” configuration, with two H atoms attached to a single Pt atom at the step (Figure 3). This configuration has been proposed to correspond to the low-potential voltammetric peak for H adsorption/desorption on Pt(110) and also to be involved in the HOR in a dissociative adsorption−oxidative desorption (Tafel-Volmer) mechanism, as discussed later.25 Therefore, if the simple hypothesis were correct that somewhat weakened H adsorption could lead to increased HOR activity,7,25,29,33−35 we would be able to see a clear trend in the DFT-calculated H adsorption energies. Indeed, we found that the calculated H adsorption was markedly weakened on all three alloy surfaces in comparison to that on Pt(221) step edges, consistent with the increased HOR activity observed experimentally, but to nearly the same extent: specifically, Pt1AL−PtFe(221) (by ca. 0.38 eV), Pt1AL−PtCo(221) (by 0.36 eV), and Pt1AL−PtNi(221) (by 0.35

eV). Thus, the differences are clearly too small to explain the much larger differences in HOR activity. However, the H adsorption energies on the (111) terraces varied over a much wider range, a difference of 0.99 eV going from pure Pt to Pt1AL−PtFe (Figure 3 and Table S1 in the Supporting Information). A similar trend was reported by Greeley et al. for a series of near-surface alloy (111) surfaces, but with a slightly smaller range, 0.74 eV.30 It should be noted that those values were based on a gas-phase H (hydrogen atom) reference, and we have converted the values to a gasphase H2 reference. After conversion, their value of −0.96 eV for Pt(111) compares well with the value calculated here for the (111) terrace of Pt(221), −0.91 eV. This wide variation in values going from Pt to Pt1AL−PtFe is highly suggestive of the (111) terrace being involved in a key role in the catalysis. To account for the involvement of the (111) terraces, we propose that, after the H2 dissociation at the step edges, the dissociated H atoms can “spill over” to the (111) terraces, which can accommodate larger numbers of atoms, prior to oxidative desorption: H 2,sol → H 2,ad(step)

H 2,ad(step) → 2Had(step) 2Had(step) → 2Had(terrace)

(3)

Tafel step

(4)

spillover step

2Had(terrace) → 2H+sol + 2e−

Volmer step

(5) (6)

Spillover could effectively remove the dissociated H atoms from the active sites at the step edges. Such a scheme could explain the significant variation in HOR activity of the series of Pt-skin-covered alloys. We note that, in the present DFT calculations, we have only considered a single Pt-skin layer (Pt1AL), whereas the experimental catalysts are thought to include two atomic layers (Pt2AL). Certainly, we expect that the range of variation will become smaller for Pt2AL but should still exhibit a variation wider than that of the step edges. We will examine the effect of skin layer thickness in the near future and plan to analyze in more detail the expected enhancement of the kinetics due to spillover. We note that the adsorption strength of the adsorbate on the (111) terrace depends to a quite small extent on the proximity to the step edge. For example, CO (atop) in the row nearer the ascending step has an adsorption energy of −1.74 eV, while that in the row nearer the descending step edge has an adsorption energy of −1.85 eV. However, there is one specific 271

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274

Research Article

ACS Catalysis

Figure 5. CO stripping voltammograms at Nafion-coated (A) c-Pt2Ru3/C, (B) c-Pt/C, and (C) Pt2AL−PtFe/C catalysts at a given potential cycle number of N measured in N2-saturated 0.1 M HClO4 solution at 70 °C and potential sweep rate of 20 mV s−1. The inset in (A) is an enlarged part of the potential range 0.2−0.8 V.

Figure 4B shows the variations in CO-tolerant HOR apparent specific activity, japp, upon 30 min of CO poisoning for these catalysts as a function of N. It was found that the japp value for Pt2AL−Pt−M/C was maintained at a high, constant value to 2500 cycles: e.g., the value for Pt2AL−PtFe/C was ca. 3.4-fold higher than that for c-Pt/C. Despite the fact that the japp value for c-Pt2Ru3/C showed a slight decline before N = 1000, it decreased gradually with increasing N (N > 1000), reaching the same value as that for c-Pt after 2500 cycles, suggesting severe dealloying of Ru during the potential cycling. The loss of Ru in c-Pt2Ru3/C can be also clearly manifested in the CO-stripping curves (Figure 5A). After 500 cycles of potential cycling, the CO electro-oxidation peak showed an apparent upshift in potential, even exhibiting a peak-splitting effect after 1000 cycles, which indicates the existence of a mixed phase for alloy and pure Pt. Finally, the CO oxidation peak became Pt-like (Figure 5B) after 2500 cycles, suggesting that Ru was dealloyed from the electrocatalyst during the potential scan in the 0.02−0.95 V region. In contrast, negative peak potential shifts were observed for c-Pt/C (Figure 5B) and Pt2AL−PtFe/C (Figure 5C), as well as for Pt2AL−PtCo/C and Pt2AL−PtNi/C (Figure S4 in the Supporting Information), which was probably due to the particle size increase during potential cycling,41 as will be described later. It should be noted that the aforementioned negative potential shift (generally corresponding to enhanced activity for CO oxidation) for c-Pt/ C and Pt2AL−Pt−M/C may not be directly related to the COtolerant HOR activity occurring at such a low potential E < 0.1 V vs RHE, as revealed from the constant japp vs N in Figure 4B and as seen also for those we reported earlier.4 In addition, it can be seen in Figure 5C that the peak shifting was nearly suppressed after 500 cycles for Pt2AL−PtFe/C, whereas continual shifting even up to N = 2500 was observed for cPt/C. This difference suggests more evidence for the robustness of Pt2AL−PtFe/C under high potentials in comparison with c-Pt/C. After 2500 cycles, the catalyst composition was analyzed by spot analysis with EDX (at 20 randomly selected particles). The result (Table S2 in the Supporting Information) shows that the average composition changed only slightly for the Pt2AL−Pt− M/C catalysts; in particular, the composition for Pt2AL−PtCo/ C was nearly unchanged, whereas the Ru level in c-Pt2Ru3/C decreased significantly. This clearly indicates that the uniform two atomic layers of Pt skin were able to protect the core to a large extent from the dissolution of nonprecious metals during potential cycling in acid solution. This robustness of Pt-skin layers versus high potential was probably due to the suppression of higher-order surface oxide formation, as suggested by Imai et al.42

case in which there is a strong effect of the step edge proximity: molecular H2 placed on the (111) terrace of Pt(221) always spontaneously dissociates due to the proximity of the step edge. However, on the alloy surfaces, H2 “floats” away for the surface on all but the Pt−Ru surface, on which H2 can actually adsorb molecularly. We also note that the presence of two distinct terrace-like rows on the (221) surface differs from the situation for (211) surfaces, on which there is only a single terracelike row available for adsorption in addition to the row corresponding to the (100) step edge.36 Thus, the (221) surface is more akin to the (533) surface, on which there are two terracelike rows in addition to the (100) step edge. Summarizing, the broad features of the experimental results can be rationalized with the DFT calculations: i.e., the CO tolerance trend is captured well, as well as the increased HOR activity for all three alloys. The main feature that distinguishes the HOR activities on these catalysts appears to involve the increasingly facile desorption of H atoms from the (111) terraces in a monotonic progression up to the most active catalyst, Pt skin-PtFe. The CO tolerance also follows essentially in the same order, with the highest tolerance for Pt skin-PtFe. Both step edges and terraces follow the same trend, with adsorption being stronger at the step edges, similar to the case of the H adsorption. Both are involved with the HOR; therefore, the CO tolerance should also involve both. 3.4. Catalyst Durability under Potential Cycling Degradation. Next, we have examined the robustness of these catalysts by applying a potential cycling between 0.02 and 0.95 V at a scan rate of 20 mV s−1 at 70 °C in N2-purged 0.1 M HClO4, which simulates the repeated exposure of the fuel cell anode to reformate gas and air during daily start/stop cycles. Figure 4A shows the cycle number (N) dependence of MAapp values for the HOR with respect to 30 min of CO poisoning. The degradation of CO-tolerant MAapp for Pt2AL−Pt−M/C was suppressed noticeably: the MAapp for Pt2AL−PtFe/C was maintained at 78% of the initial value after 2500 cycles and was maintained at a comparable level for Pt2AL−PtCo/C and Pt2AL−PtNi/C (80% retention). The significant decrease in MAapp at c-Pt/C (55% retention) can be ascribed to the socalled Ostwald ripening effect, where the Pt particle size increases via a dissolution−redeposition process as a result of the redox cycling, and the agglomeration of Pt particles, resulting in a marked loss of the electrochemical active area.37−40 The worst degradation was seen for c-Pt2Ru3/C (50% retention), probably due to severe Ru dissolution under oxidative conditions, accompanied by particle agglomeration, as mentioned above. 272

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274

Research Article

ACS Catalysis

Figure 6. TEM images and particle size distribution histograms for (A, a) Pt2AL−PtFe/C and (B, b) c-Pt2Ru3/C after the durability test at N = 2500.



Figure 6 shows TEM images and the particle size distribution histograms of Pt2AL−PtFe/C ((A) and (a)) and c-Pt2Ru3/C ((B) and (b)) after the durability test (N = 2500). It was found that a highly uniform distribution of the particles was still maintained for Pt2AL−PtFe/C after the durability test, although the average particle size increased to 4.7 ± 1.0 nm from 2.9 ± 0.4 nm for the pristine catalyst. The suppression of particle agglomeration was also seen for Pt2AL−PtCo/C and Pt2AL− PtNi/C (Figure S5 in the Supporting Information). In contrast, for c-Pt/C (Figure S5), and especially for c-Pt2Ru3/C, the particles became severely aggregated, with a large particle size distribution. To address the problem of Ru dissolution from the Pt−Ru alloy catalyst, several approaches have been proposed: e.g., utilization of functionalized supports,43 stabilization with ruthenium oxide in nanosheet form,44 and the incorporation of Au.45 However, for the sake of cost reduction, our proposed Pt2AL−Pt−M/C catalysts, without the use of Ru, appear to be more promising for practical use as robust hydrogen anode catalysts in reformate gas fuel cells.

Corresponding Author

*H.U.: e-mail, [email protected]; fax, +81-55-2208618; tel, +81-55-220-8619. ORCID

Hiroyuki Uchida: 0000-0001-6718-5431 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by funds for the ‘‘Superlative, Stable, and Scalable Performance Fuel Cells” (SPer-FC) project from the New Energy and Industrial Technology Development Organization (NEDO) of Japan.



REFERENCES

(1) Watanabe, M.; Igarashi, H.; Yoshioka, K. Electrochim. Acta 1995, 40, 329−334. (2) Igarashi, H.; Fujino, T.; Zhu, Y.; Uchida, H.; Watanabe, M. Phys. Chem. Chem. Phys. 2001, 3, 306−314. (3) Wakisaka, M.; Mitsui, S.; Hirose, Y.; Kawashima, K.; Uchida, H.; Watanabe, M. J. Phys. Chem. B 2006, 110, 23489−23496. (4) Uchida, H.; Izumi, K.; Aoki, K.; Watanabe, M. Phys. Chem. Chem. Phys. 2009, 11, 1771−1779. (5) Uchida, H.; Ozuka, H.; Watanabe, M. Electrochim. Acta 2002, 47, 3629−3636. (6) Wan, L. J.; Moriyama, T.; Ito, M.; Uchida, H.; Watanabe, M. Chem. Commun. 2002, 1, 58−59. (7) Shi, G. Y.; Yano, H.; Tryk, D. A.; Watanabe, M.; Iiyama, A.; Uchida, H. Nanoscale 2016, 8, 13893−13897. (8) Dai, Y.; Liu, Y.; Chen, S. Electrochim. Acta 2013, 89, 744−748. (9) Chiwata, M.; Yano, H.; Ogawa, S.; Watanabe, M.; Iiyama, A.; Uchida, H. Electrochemistry 2016, 84, 133−137. (10) Watanabe, M.; Yano, H.; Tryk, D. A.; Uchida, H. J. Electrochem. Soc. 2016, 163, F455−F463. (11) Yano, H.; Akiyama, T.; Bele, P.; Uchida, H.; Watanabe, M. Phys. Chem. Chem. Phys. 2010, 12, 3806−3814. (12) Okaya, K.; Yano, H.; Uchida, H.; Watanabe, M. ACS Appl. Mater. Interfaces 2010, 2, 888−895. (13) Levich, V. G. Physicochemical Hydrodynamics; Prentice Hall: Englewood Cliffs, NJ, 1962; p 112. (14) Markovic, N. M.; Grgur, B. N.; Ross, P. N. J. Phys. Chem. B 1997, 101, 5405−5413. (15) Mukerjee, S.; McBreen, J. J. Electrochem. Soc. 1996, 143, 2285− 2294. (16) Grgur, B. N.; Markovic, N. M.; Ross, P. N. Can. J. Chem. 1997, 75, 1465−1471. (17) Ehteshami, S. M.; Jia, Q.; Halder, A.; Chan, S. H.; Mukerjee, S. Electrochim. Acta 2013, 107, 155−163. (18) Ross, P. N. J. Vac. Sci. Technol., A 1992, 10, 2546−2550. (19) Liao, M. S.; Cabrera, C. R.; Ishikawa, Y. Surf. Sci. 2000, 445, 267−282. (20) Koper, M. T. M.; Shubina, T. E.; van Santen, R. A. J. Phys. Chem. B 2002, 106, 686−692.

4. CONCLUSIONS We have examined the effects of the non-precious-metal species M (M = Fe, Co, Ni) on the CO tolerance and robustness of novel Pt2AL−Pt−M/C HOR catalysts. We found that Pt2AL− PtFe/C exhibited the highest CO-tolerant HOR activity, followed by Pt2AL−PtCo/C and Pt2AL−PtNi/C. After 2500 potential cycles between 0.02 and 0.95 V, the CO-tolerant HOR area-specific activity was nearly maintained at Pt2AL−Pt− M/C, and the retention of mass activity was as high as 80%, which is far superior to those for c-Pt/C and c-Pt2Ru3/C, suggesting that the two uniform atomic layers of Pt skin can provide robustness during potential cycling. The area-specific HOR activities of these alloy catalysts were all superior to that for Pt/C, increasing in the order Pt < Pt2AL−PtNi < Pt2AL− PtCo < Pt2AL−PtFe, and were proposed to be due to a decreasing H adsorption strength on (111) terraces, on the basis of DFT calculations. The effects of particle size and/or core composition and an optimization of preparation methods are in progress in our laboratory.



AUTHOR INFORMATION

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.6b02794. TEM images and particle size distribution histograms of all catalysts, CO-poisoning time dependence of MAapp, and details of the DFT calculations (PDF) 273

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274

Research Article

ACS Catalysis (21) Orita, H.; Itoh, N.; Inada, Y. Chem. Phys. Lett. 2004, 384, 271− 276. (22) Yamagishi, S.; Fujimoto, T.; Inada, Y.; Orita, H. J. Phys. Chem. B 2005, 109, 8899−8908. (23) Ishikawa, Y.; Mateo, J. J.; Tryk, D. A.; Cabrera, C. R. J. Electroanal. Chem. 2007, 607, 37−46. (24) Mateo, J. J.; Tryk, D. A.; Cabrera, C. R.; Ishikawa, Y. Mol. Simul. 2008, 34, 1065−1072. (25) Santana, J. A.; Mateo, J. J.; Ishikawa, Y. J. Phys. Chem. C 2010, 114, 4995−5002. (26) Santana, J. A.; Saavedra-Arias, J. J.; Ishikawa, Y. Electrocatalysis 2015, 6, 534−543. (27) Greeley, J.; Mavrikakis, M. Catal. Today 2006, 111, 52−58. (28) van der Vliet, D.; Wang, C.; Li, D.; Paulikas, A. P.; Greeley, J.; Rankin, R. B.; Strmcnik, D.; Tripkovic, D.; Markovic, N. M.; Stamenkovic, V. R. Angew. Chem. 2012, 124, 3193−3196. (29) Wang, Y.; Wang, G.; Li, G.; Huang, B.; Pan, J.; Liu, Q.; Han, J.; Xiao, L.; Lu, J.; Zhuang, L. Energy Environ. Sci. 2015, 8, 177−181. (30) Greeley, J.; Mavrikakis, M. J. Phys. Chem. B 2005, 109, 3460− 3471. (31) Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G.; Ross, P. N.; Lucas, C. A.; Markovic, N. M. Science 2007, 315, 493−497. (32) Wakisaka, M.; Morishima, S.; Hyuga, Y.; Uchida, H.; Watanabe, M. Electrochem. Commun. 2012, 18, 55−57. (33) Conway, B. E.; Bockris, J. O. J. Chem. Phys. 1957, 26, 532−541. (34) Durst, J.; Siebel, A.; Simon, C.; Hasche, F.; Herranz, J.; Gasteiger, H. A. Energy Environ. Sci. 2014, 7, 2255−2260. (35) Sheng, W.; Zhuang, Z.; Gao, M.; Zhang, J.; Chen, J. G.; Yan, Y. Nat. Commun. 2015, 6, 5848. (36) Badan, C.; Koper, M. T. M.; Juurlink, L. B. F. J. Phys. Chem. C 2015, 119, 13551−13560. (37) Wang, Z.; Shi, G.; Xia, J.; Xia, Y.; Zhang, F.; Xia, L.; Song, D.; Liu, J.; Li, Y.; Xia, L.; Brito, M. E. Electrochim. Acta 2014, 121, 245− 252. (38) Rinaldo, S. G.; Urchaga, P.; Hu, J. W.; Lee, W.; Stumper, J.; Rice, C.; Eikerling, M. Phys. Chem. Chem. Phys. 2014, 16, 26876− 26886. (39) Shi, G.; Wang, Z.; Xia, J.; Bi, S.; Li, Y.; Zhang, F.; Xia, L.; Li, Y.; Xia, Y.; Xia, L. Electrochim. Acta 2014, 142, 167−172. (40) Wang, Z.; Shi, G.; Zhang, F.; Xia, J.; Gui, R.; Yang, M. Electrochim. Acta 2015, 160, 288−295. (41) Takasu, Y.; Iwazaki, T.; Sugimoto, W.; Murakami, Y. Electrochem. Commun. 2000, 2, 671−674. (42) Imai, H.; Matsumoto, M.; Miyazaki, T.; Kato, K.; Tanida, H.; Uruga, T. Chem. Commun. 2011, 47, 3538−3540. (43) Wang, S.; Wang, X.; Jiang, S. P. Langmuir 2008, 24, 10505− 10512. (44) Sugimoto, W.; Saida, T.; Takasu, Y. Electrochem. Commun. 2006, 8, 411−415. (45) Liang, Z. X.; Zhao, T. S.; Xu, J. B. J. Power Sources 2008, 185, 166−170.

274

DOI: 10.1021/acscatal.6b02794 ACS Catal. 2017, 7, 267−274