Suppression of CO Adsorption on PtRu/C Catalysts Modiﬁed with Metallic Ruthenium Nanosheets

PtRu/C catalyst were modiﬁed with one-atom-thick metallic Ru nanosheets, and the electrocatalytic activity as well as long term durability were studied for use as an anode catalyst in polymer electrolyte fuel cells. Metallic Ru nanosheet-PtRu/C composite catalysts were obtained by hydrogen reduction of RuO 2.1 nanosheet-modiﬁed PtRu/C. The hydrogen oxidation reaction activity in 300 ppm CO/H 2 saturated 0.1 M HClO 4 at 20 mV vs. RHE increased from 89 A (g-PtRu) − 1 for PtRu/C to 124 A (g-PtRu) − 1 for metallic Ru nanosheet modiﬁed PtRu/C catalyst. After the accelerated durability test, the hydrogen oxidation reaction activity for the modiﬁed catalyst was 1.8 times higher than that of PtRu/C. The results suggest that the CO tolerance and durability of PtRu/C can be easily improved by adding metallic Ru nanosheets.

PtRu binary alloy nanoparticles supported on carbon (PtRu/C) is presently used as the anode catalyst for polymer electrolyte fuel cells in residential fuel cell systems. [1][2][3][4][5][6][7][8][9] Carbon monoxide, which is present in the reformate as a trace impurity, rapidly and strongly adsorbs on the Pt surface, blocking the hydrogen oxidation reaction (HOR) sites. 10 The onset potential for the oxidation of adsorbed CO on PtRu is about 200 mV lower than Pt. 11 However, the onset potential is higher than the typical anode potential when operated at low current density, 12 which means that adsorption of CO on the PtRu surface will lead to gradual loss of cell performance. Therefore, catalysts with enhanced CO tolerance at a potential near the H + /H 2 potential are necessary to improve the performance of present residential fuel cells.
CO tolerance can be enhanced either by developing catalysts with higher CO oxidation capability at low potential or by suppressing the adsorption of CO on the catalyst surface. Decreasing the overpotential for the oxidation of adsorbed CO has been conducted by fine control of the nanostructure and composition, as well as extension to ternary and more complicated alloys. 12,13 For example, decoration of Pt nanoparticles with Ru atoms was shown to give rise to high CO tolerance. [13][14][15][16][17][18] PtRu nanoplatelets synthesized by galvanostatic replacement of Pb 0 supported on carbon by Pt and Ru was reported to show high methanol oxidation activity. 19 The high CO tolerance has been attributed to the synergistic effect between Pt and Ru atoms, i.e., the CO mobility on the catalyst surface from Pt to Ru was increased by the intimate contact of the atoms. Despite such improved performance, the Ru decoration simultaneously decreases the amount of Pt on the surface and leads to reduced HOR sites. Metal oxides, including TaO x , 20 MoO x , 21 and SnO 2 , 22 have also been reported to act as effective co-catalysts to enhance CO tolerance. However, in many cases, the metal oxides partially cover the catalyst surface, leading to decrease in the electrochemically active Pt surface area and consequently HOR active sites. Therefore, the design of CO tolerant catalysts should be considered by an additive that does not block HOR sites, while reducing CO adsorption, and/or enhancing the CO oxidation activity.
We recently reported a concept that is different to conventional alloys or composites, which is based on the addition of RuO 2.1 nanosheets to commercial PtRu/C. The composite catalyst effectively improves the HOR activity and CO tolerance in 300 ppm CO containing H 2 gas. 23 The enhanced catalytic behavior was attributed to the suppression of CO adsorption to active sites for HOR.
In this study, we have pursued the use of metallic Ru nanosheets 24 instead of RuO 2.1 nanosheets as an additive to enhance the CO tolerance of PtRu/C. By the metallization of RuO 2.1 nanosheets to metallic Ru nanosheets, metallic Ru should be in intimate contact with the surface of PtRu nanoparticles, while still maintaining the high surface area of Pt, which we anticipated would lead to an enhancement of CO tolerance.  Electrochemical characterization.-Electrochemical measurements were conducted with a rotating disk electrode (Nikko Keisoku, SC-5) connected to an automatic polarization system (Hokuto Denko, HSV-100). Catalyst suspensions were prepared by dispersing 18.5 mg of the catalyst in 25 mL of 2-propanol/water solution (75/25 volume ratio). A 5 wt% Nafion solution (100 μL) was added to the catalyst suspension as a proton conducting binder to ensure adhesion. The working electrode was prepared by depositing 5.5 μg-carbon cm −2 on a mirror-polished glassy carbon rod (6 mm diameter), which corresponds to approximately a monolayer in height of the carbon black particles, [28][29][30] and vacuum dried at 60 • C for 30 min. A carbon fiber (Toho Tenax Co., HTA40 E13 3K 200tex) was used as a counter electrode, and a reversible hydrogen electrode (RHE) was used as a reference electrode. All electrochemical measurements were performed in 0.1 M HClO 4 at 25 • C.
Potential cycling between 5 and 800 mV vs RHE at a scan rate of 100 mV s −1 was conducted for 20 cycles to clean the catalyst surface. The electrochemically active surface area (ECSA) of the catalyst was characterized by CO stripping voltammetry. Gaseous CO was purged into the electrolyte for 40 min while maintaining a constant potential of 100 mV vs. RHE. Excess CO in the electrolyte was purged out by bubbling N 2 gas for 40 min. Pre-adsorbed CO (CO ad ) was electro-oxidized by voltammetry at a scan rate of 10 mV s −1 . The amount of CO ad was estimated by integration of the CO ad stripping peak, corrected for the electrical double-layer capacitance, assuming a monolayer of linearly adsorbed CO on the metal surface and the Columbic charge necessary for oxidation as 420 μC cm −2 . The HOR activity in pure H 2 was measured by chronoamperometry at 20 mV vs. RHE at a rotation rate of 400 rpm. After characterization in pure H 2 , H 2 gas was changed to 300 ppm CO containing H 2 gas to evaluate the CO tolerance of the catalyst. After admission of CO/H 2 into the electrolyte for 40 min at 0 mV vs RHE, the potential was swept between 5 to 800 mV vs. RHE at 20 mV s −1 for 3 cycles. Chronoamperometry was successively measured at 20 mV vs. RHE with a rotation rate of 400 rpm. The quasi-steady state current after 5 hours was taken as the hydrogen oxidation reaction (HOR) activity of the catalyst. Accelerated durability test (ADT) in 300 ppm CO/H 2 was conducted by square-wave potential stepping between 5 and 400 mV vs. RHE for up to 3000 cycles with a holding time of 3 seconds at each potential.
Copper-stripping voltammetry was conducted in 2 mM CuSO 4 + 0.5 M H 2 SO 4 solution to estimate the surface Ru composition of the catalysts. 31,32 The potential was kept at 300 mV vs. RHE for 120 sec, and then the potential was swept up to 800 mV vs. RHE at a scan rate of 100 mV s −1 . Surface Ru composition was determined from the ratio of the charge from the first and second peak, which are associated with the Cu oxidation charge on Ru and Pt, respectively. Figure 1 shows typical HR-SEM and TEM images of metallic Ru nanosheet-modified PtRu/C and the corresponding histogram of the PtRu nanoparticle size distribution obtained from TEM. The average diameter of PtRu nanoparticles for Ru(ns)-PtRu/C was 4.3 ± 0.4 nm, comparable to that of as-received PtRu/C (4.2 ± 0.3 nm) (Fig. S1). This shows that the PtRu nanoparticles do not grow by the metallization process. The XRD patterns for all catalyst give the typical diffraction peaks of the face-centered cubic phase of PtRu (Fig. S2). The fcc(220) diffraction peak at 2θ = 68.5 • (d = 0.1369 nm) in Ru(ns)-PtRu/C was identical in peak position and width to PtRu/C and RuO 2.1 ns-PtRu/C (Fig. 2), indicating the alloying state of PtRu nanoparticles is the same for all catalysts. Peaks associated with metallic Ru nanosheets (2D-hcp) are not observed due to the low content. Figure 3 shows the Ru 3p 3/2 XPS data for PtRu/C, RuO 2.1 ns-PtRu/C and Ru(ns)-PtRu/C. The Ru 3p 3/2 spectra could be deconvoluted into two components for metallic Ru and ruthenium in a higher oxidation state (hydrous ruthenium oxide) (Table I). XPS analysis of Ru(ns)/C (Fig. S3) reveals that all of Ru n+ is reduced to Ru 0 by the H 2 treatment. Thus, Ru n+ is assigned to partially oxidized ruthenium in the pristine PtRu/C that cannot be reduced with H 2 at 200 • C. XPS data shows that 45 at.% of Ru in PtRu/C is oxidized due to atmospheric exposure, in agreement with other studies. [33][34][35] The ratio of Ru n+ is naturally higher for RuO 2.1 ns-PtRu/C (53 at.%) than that of PtRu/C (45 at.%), owing to the addition of Ru 4+ O 2.1 nanosheet. After H 2 reduction, the Ru 0 content for Ru(ns)-PtRu/C increased to 61 at.% from 47 at.% for RuO 2.1 ns-PtRu/C. No change in the Pt 4f spectra is evident (Fig. S4  Based on these data and our previous work, 24 it is concluded that H 2 treatment of RuO 2.1 ns-PtRu/C at 200 • C successfully converted the nanosheet to its metal form leading to Ru(ns)-PtRu/C. Figure 4 shows the CO ad stripping voltammograms of the catalysts. The ECSA of RuO 2.1 ns-PtRu/C is 54 m 2 (g-PtRu) −1 , comparable to PtRu/C (52 m 2 (g-PtRu) −1 ). The ECSA of Ru(ns)-PtRu/C increased to 83 m 2 (g-PtRu) −1 due to the reduction of RuO 2.1 nanosheets to metallic Ru nanosheets. Metallic Ru nanosheets has a large ECSA, which is evidenced from the control sample Ru(ns)/C (109 m 2 (g-Ru) −1 ). From the peak separation of CO ad oxidation charge associated with oxidation of CO ad on Ru(ns) and PtRu, the ratio for Ru(ns) and PtRu in Ru(ns)-PtRu/C was estimated as 15.3% and 84.7%, respectively (Fig. S5). This is in excellent agreement with the nominal atomic ratio of 15.6% for Ru(ns) and 84.4% for PtRu. The ECSA of Ru(ns) and PtRu/C can be calculated from the deconvoluted CO ad oxidation charge related to CO ad on Ru(ns) and PtRu/C, which were 115 m 2 (g-Ru) −1 and 55 m 2 (g-PtRu) −1 , respectively. This is again in good agreement with the ECSA of Ru(ns)/C (109 m 2 (g-Ru) −1 ) and PtRu/C (52 m 2 (g-PtRu) −1 ). Thus, the higher ECSA of Ru(ns)-PtRu/C is attributed to the combination of metallic Ru nanosheets and PtRu nanoparticles. The onset and peak potential of the CO ad oxidation for Ru(ns)-PtRu/C were observed at 300 and 520 mV vs. RHE, respectively. These potentials are shifted 30 mV negative compared to PtRu/C (onset potential at 330 mV vs. RHE and peak potential at 550 mV vs. RHE). The results suggest that the modification of metallic Ru nanosheet effectively enhanced the oxidation of CO ad for PtRu/C. It is well known that Ru promotes the oxidation of CO ad by supplying an oxygen source to adjacent Pt sites via the so-called bi-functional mechanism. 36 The lower onset potential may be a signature of a bi-functional effect between metallic Ru nanosheet and PtRu alloy nanoparticles. The lower peak potential and the sharper peak suggests improved kinetics of the CO ad oxidation for Ru(ns)-PtRu/C. As shown in Fig. 4d, the peak potential for Ru(ns)/C and Ru(np)/C was 520 and 530 mV vs. RHE; Ru(ns)/C exhibits slightly faster reaction kinetics for oxidation of CO ad than Ru(np)/C. From the result, the negative shift of onset and peak potential for Ru(ns)-PtRu/C in comparison with PtRu/C is attributed to the modification with metallic Ru nanosheet, which acts as a promoter for CO ad oxidation.

Electrochemically active surface area (ECSA) and CO ad oxidation activity of the catalysts.-
HOR activity and CO tolerance.-Hydrodynamic voltammograms of HOR in pure H 2 for PtRu/C, RuO 2.1 ns-PtRu/C and Ru(ns)-PtRu/C are compared in Fig. 5a. Diffusion limited current (j lim ) is obtained above 100 mV vs. RHE. The voltammograms of RuO 2.1 ns-PtRu/C show comparable behavior to PtRu/C, indicating that the addition of RuO 2 nanosheets does not obstruct the HOR activity. The j lim for Ru(ns)-PtRu/C was also similar to that of PtRu/C. Thus, the modification with metallic Ru nanosheet does not interfere with HOR sites on PtRu surface.
Hydrodynamic voltammograms of HOR in the presence of trace CO (300 ppm CO/H 2 ) are compared in Fig. 5b. The HOR current for all of the catalysts were smaller than that in pure H 2 , which is attributed  to CO poisoning of the catalyst surface. In the case of PtRu/C, the HOR current below the onset of CO oxidation was less than half of that in pure H 2 . Upon polarization above 300 mV vs. RHE, the current increases due to the oxidation of weakly adsorbed CO on the catalyst surface. 37 CO ad oxidation in 300 ppm CO/H 2 is observed at a lower potential than the CO ad oxidation onset potential from CO ad stripping voltammetry (Fig. 4), due to the low CO coverage. 37 In sharp contrast, high HOR current is observed in the case of RuO 2.1 ns-PtRu/C and Ru(ns)-PtRu/C even at low potential. The HOR current at 200 mV vs. RHE for Ru(ns)-PtRu/C was higher than that of PtRu/C and RuO 2.1 ns-PtRu/C. The results indicate that the modification with metallic Ru nanosheets effectively improves the CO tolerance. Figure 6a shows chronoamperograms in pure H 2 at 20 mV vs. RHE (ω = 400 rpm) for PtRu/C, RuO 2.1 ns-PtRu/C, and Ru(ns)-PtRu/C. The HOR activity is normalized to the total mass of metal. The HOR activity in pure H 2 (steady-state current after 20 min) for RuO 2.1 ns-PtRu/C and Ru(ns)-PtRu/C was similar to or slightly higher than PtRu/C. Chronoamperograms in 300 ppm CO/H 2 are shown in Fig.  6b. The HOR activity of all catalysts decreased sharply in the first 60 min due to accumulation of CO on the catalyst surface, and then leveled off to a quasi-steady current (notice that the potential applied of 20 mV vs. RHE is much lower than the onset potential of CO ad oxidation). Taking the quasi-steady state HOR current at 5 hours as a measure of HOR activity, Ru(ns)-PtRu/C has 1.3 times higher activity (124 A (g-PtRu) −1 ) than that of PtRu/C (93 A (g-PtRu) −1 ). Excluding the mass of metallic Ru nanosheets (which contributes marginally to HOR), the HOR activity is 132 A (g-PtRu) −1 . The decreasing rate of HOR activity, estimated from the change in the HOR current divided by the time from 3 to 5 hours was 1.7 A (g-PtRu) −1 hour −1 for PtRu/C. In the case of Ru(ns)-PtRu/C, a steady-state is obtained after 3 hours (decreasing rate is negligible, 0 A (g-PtRu) −1 hour −1 ). Fig. 6c shows the ratio of current decay normalized by the current at 0 sec. It can be seen that the CO poisoning in the initial 60 min is impeded for Ru(ns)-PtRu/C. More importantly, steady-state current was obtained at 3 hours for Ru(ns)-PtRu/C, while for PtRu/C, a continuous decrease in current was observed even after 5 hours of polarization. Since the onset of CO ad oxidation occurs above 200 mV vs. RHE (Figs. 4 and 5), electronic and bi-functional effects cannot explain the enhanced HOR activity at 20 mV vs. RHE in 300 ppm CO/H 2 . Conversely, the enhancement in CO tolerance by the modification with metallic Ru nanosheets is most likely due to the suppression of CO adsorption on the Pt surface.
Durability of the catalysts.-The HOR activity (measured at E = 20 mV vs. RHE with ω = 400 pm after 60 min) after accelerated durability test (ADT) at 1000 and 3000 steps are compared with the initial activity in Table II. The retention in HOR activity is more pronounced for the Ru(ns)-PtRu/C. After the 3000 potential step ADT test, the HOR current of Ru(ns)-PtRu/C was 97 A (g-PtRu) −1 , 1.8 and 1.3 times higher than PtRu/C (55 A (g-PtRu) −1 ) and RuO 2.1 ns-PtRu/C (75 A (g-PtRu) −1 ), respectively. Based on Cu-stripping voltammetry, the surface Ru composition of PtRu/C decreased from 16 to  13 atomic% after 3000 potential step ADT testing ( Fig. S7 and Table  SI). On the other hand, the surface Ru composition for Ru(ns)-PtRu/C is unchanged at 28 atomic%. The results indicate that Ru dissolution is suppressed by the modification with metallic Ru nanosheets. Degradation of PtRu/C is accelerated in the presence of CO (Fig. S8), which may be linked to Ru dissolution. Accordingly, the lesser extent of CO adsorption on the PtRu surface by the addition of metallic Ru nanosheets would lead to enhanced durability of Ru(ns)-PtRu/C.

Conclusions
The CO tolerance of commercial PtRu/C was improved by the modification with metallic Ru nanosheets, which was obtained by the metallization of RuO 2.1 nanosheets derived from layered K 0.2 RuO 2.1 .
In the presence of trace CO (300 ppm CO/H 2 ), the HOR current at 20 mV vs. RHE of PtRu/C increased to 124 A (g-PtRu) −1 from 89 A (g-PtRu) −1 by the modification with metallic Ru nanosheets. The durability of the catalyst was also improved by the modification with metallic Ru nanosheets. The HOR current of Ru(ns)-PtRu/C after accelerated durability test was 97 A (g-PtRu) −1 , which is 1.8 times higher than PtRu/C (55 A (g-PtRu) −1 ). The enhancement in HOR activity and durability in the presence of CO by the modification with metallic Ru nanosheets most likely suggests suppression of CO adsorption on the catalyst surface.