In Situ Ftir Analysis of Co-tolerance of a Pt-fe Alloy with Stabilized Pt Skin Layers as a Hydrogen Anode Catalyst for Polymer Electrolyte Fuel Cells

The CO-tolerance mechanism of a carbon-supported Pt-Fe alloy catalyst with two atomic layers of stabilized Pt-skin (Pt 2AL –PtFe/C) was investigated, in comparison with commercial Pt 2 Ru 3 /C (c-Pt 2 Ru 3 /C), by in situ attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy in 0.1 M HClO 4 solution at 60 • C. When 1% CO (H 2-balance) was bubbled continuously in the solution, the hydrogen oxidation reaction (HOR) activities of both catalysts decreased severely because the active sites were blocked by CO ad , reaching the coverage θ CO ≈ 0.99. The bands in the IR spectra observed on both catalysts were successfully assigned to linearly adsorbed CO (CO L) and bridged CO (CO B), both of which consisted of multiple components (CO L or CO B at terraces and step/edge sites). The Pt 2AL –PtFe/C catalyst lost 99% of its initial mass activity (MA) for the HOR after 30 min, whereas about 10% of the initial MA was maintained on c-Pt 2 Ru 3 /C after 2 h, which can be ascribed to a suppression of linearly adsorbed CO at terrace sites (CO L, terrace). In contrast, the HOR activities of both catalysts with pre-adsorbed CO recovered appreciably after bubbling with CO-free pure H 2. We clarify, for the first time, that such a recovery of activity can be ascribed to an increased number of active sites by a transfer of CO L, terrace to CO L, step/edge , without removal of CO ad from the surface. The Pt 2AL –PtFe/C catalyst showed a larger decrease in the band intensity of CO L, terrace. A possible mechanism for the CO-tolerant HOR is also discussed.


Introduction
For the applications of fuel cell vehicles (FCVs) and stationary cogeneration systems (FC-CG), polymer electrolyte fuel cells (PEFCs) have been actively developed.In 2014, a strategic roadmap for hydrogen and fuel cells was formulated by the Agency for Natural Resources and Energy, the Ministry of Economy, Trade, and Industry (METI), Japan [1].Its Phase 1 is an expansion of the scope of applications for FCVs and FC-CG to achieve dramatic energy conservation.While the number of residential PEFC systems installed has been increasing continuously in Japan and has also commenced to increase in Europe, the reduction of the system cost is essential for larger scale commercialization, while maintaining the performance and durability.
Therefore, far, Pt-Ru alloy anode catalysts have been employed for the hydrogen oxidation reaction (HOR) to lessen the poisoning by low concentrations of CO contained in the reformate (hydrogen-rich gas produced by reforming hydrocarbons, followed by a purification).Indeed, the state-of-the-art commercial anode catalyst used in the commercial FC-CG system Ene•Farm ® is nano-sized Pt 2 Ru 3 dispersed on high-surface-area carbon black (c-Pt 2 Ru 3 /C).However, because the CO-tolerant HOR mass activity (based on the mass of both noble metals, Pt and Ru) and durability of c-Pt 2 Ru 3 /C are not sufficient, it is very important to develop novel anode catalysts, which would simplify the system, leading to cost reduction.As the support or co-catalyst for Pt-Ru alloys, the use of metal oxide materials has been effective in increasing the CO-tolerance [2][3][4][5][6][7].It was found in our previous work that the CO-tolerance of Pt 2 Ru 3 nanoparticles was improved by the use of an Sb-SnO 2 support, in place of the conventional carbon black support [7].By use of in situ attenuated total reflection Fourier transform infrared reflection-adsorption spectroscopy (ATR-FTIRAS), we clarified that the adsorption states of CO were changed due to electronic modification by the Sb-SnO 2 support.Regarding the improvement of durability of c-Pt 2 Ru 3 /C, the instability of Ru at high potentials E > 0.8 V (vs.reversible hydrogen electrode, RHE) is difficult to overcome: the Ru component leaches into the electrolyte membrane, migrating and depositing at the cathode catalyst layer made from Pt or Pt-alloy.Thus, the loss of Ru reduces the cell performance by decreases in not only the CO-tolerance of the anode but also the activity for the oxygen reduction reaction (ORR) at the cathode.Hsieh et al. reported CO-tolerance of Pt shell/Ru core catalysts designed to suppress the dissolution of Ru [8], but it is expected to be difficult to completely protect Ru from dissolution.
Recently, a new Ru-free hydrogen anode catalyst has been developed in our laboratory.Carbon-supported PtCo alloy particles with two atomic layers of stabilized Pt skin (Pt 2AL -PtCo/C) exhibited high mass activity MA for the CO-tolerant HOR, together with high robustness versus air exposure [9].Very recently, we have reported our research on the effect of the non-noble metal species M (M = Fe, Co, Ni) in Pt 2AL -Pt-M/C on CO-tolerance and the robustness by the use of channel flow electrode (CFE) method in 0.1 M HClO 4 .It was found that Pt 2AL -PtFe/C exhibited the highest CO-tolerant HOR activity (with respect to the area-specific activity j s and the MA).We considered the possibility that such a CO-tolerance could be ascribed to a modification of the electronic structure of the Pt-skin layer due to the presence of the alloy beneath the surface [10], but this needs to be analyzed experimentally.
In the present research, we have investigated the CO-tolerance mechanism on Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C by the use of in situ ATR-FTIRAS in 0.1 M HClO 4 at 60 • C. We, for the first time, demonstrate that the recovery of the HOR activity that occurs on these catalysts when changing the gas from CO/H 2 to pure H 2 can be ascribed to the mobility of adsorbed CO, resulting in increased numbers of active sites.

FTIR Analysis of CO Adsorption on Catalysts
TEM images and particle size distribution histograms of Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C (Figure 1) show that the average particle sizes and the standard deviations (σ d ) of the Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C, which were determined from 500 particles in several TEM images, were 2.9 ± 0.4 nm and 3.5 ± 0.9 nm, respectively.As reported previously [11], the Pt 2AL -PtFe nanoparticles were more uniform in size (smaller σ d ) and highly dispersed on the carbon black support, compared with c-Pt 2 Ru 3 /C.By the use of an in situ FTIR cell with the attenuated total reflection configuration (ATR-FTIR) [7,12], the IR spectra on both catalysts at 0.02 V (practical operating potential in PEFCs) versus an RHE and 60 °C by bubbling 1% CO (H2-balance) continuously in 0.1 M HClO4 solution were measured together with the HOR current.The electrolyte solution was first saturated with pure H2 to measure the initial HOR current and the reference IR spectrum at 0.02 V, followed by changing the gas to 1% CO/H2.Figure 2 shows change in the HOR mass activity (MA, HOR current per unit metal mass) at 0.02 V vs. RHE during the in situ ATR-FTIR measurement of CO adsorption.The initial MA value (measured in pure H2-saturated solution) on the Pt2AL-PtFe/C catalyst (35 A gmetal −1 ) was ca.1.3 times larger than that on the c-Pt2Ru3/C.Such an enhancement factor in the MA for the HOR of pure H2 on the Pt2AL-PtFe/C was smaller than that (ca.2.5 times) measured in the CFE at 70 and 90 °C [10].This is ascribed with certainty to the fact that it is difficult for all of the catalyst particles to work effectively for the HOR, because the thickness of the catalyst layer loaded on the ATR prism corresponded to about 6.5 monolayers of the carbon black particles (to increase the signal/noise ratio of the IR spectrum), whereas the amount of the catalyst loaded in the CFE cell was ca.two monolayers of carbon black particles to obtain the HOR activity in the ideal electrochemical condition.Focusing on the changes in the MA values during CO adsorption on both catalysts in Figure 2, the CO-poisoning rate at the Pt2AL-PtFe/C was rapid, losing 99% of MA after 30 min, whereas that at the c-Pt2Ru3/C was slower, so that about 10% of the initial MA was still maintained after 2 h.Then, the CO coverage θCO on the surface just after the in situ ATR-FTIR measurement was evaluated from the CO-stripping voltammogram in N2-purged solution, as shown in Figure 3.The loss of MA on the Pt2AL-PtFe/C catalyst is reasonably explained by the θCO value of 0.99.The value of θCO on the c-Pt2Ru3/C catalyst was also 0.99, but, as stated above, the HOR activity was still maintained.It is clear that the adsorbed CO (COad) cannot be oxidized during the HOR at 0.02 V, even on c-Pt2Ru3/C, because the values of onset potential for the COad oxidation on Pt2AL-PtFe/C and c-P2tRu3/C were ca.0.50 V and 0.33 V, respectively.The difference in the CO-tolerance will be discussed in more detail below.By the use of an in situ FTIR cell with the attenuated total reflection configuration (ATR-FTIR) [7,12], the IR spectra on both catalysts at 0.02 V (practical operating potential in PEFCs) versus an RHE and 60 • C by bubbling 1% CO (H 2 -balance) continuously in 0.1 M HClO 4 solution were measured together with the HOR current.The electrolyte solution was first saturated with pure H 2 to measure the initial HOR current and the reference IR spectrum at 0.02 V, followed by changing the gas to 1% CO/H 2 .Figure 2 shows change in the HOR mass activity (MA, HOR current per unit metal mass) at 0.02 V vs. RHE during the in situ ATR-FTIR measurement of CO adsorption.The initial MA value (measured in pure H 2 -saturated solution) on the Pt 2AL -PtFe/C catalyst (35 A g metal −1 ) was ca.1.3 times larger than that on the c-Pt 2 Ru 3 /C.Such an enhancement factor in the MA for the HOR of pure H 2 on the Pt 2AL -PtFe/C was smaller than that (ca.2.5 times) measured in the CFE at 70 and 90 • C [10].This is ascribed with certainty to the fact that it is difficult for all of the catalyst particles to work effectively for the HOR, because the thickness of the catalyst layer loaded on the ATR prism corresponded to about 6.5 monolayers of the carbon black particles (to increase the signal/noise ratio of the IR spectrum), whereas the amount of the catalyst loaded in the CFE cell was ca.two monolayers of carbon black particles to obtain the HOR activity in the ideal electrochemical condition.Focusing on the changes in the MA values during CO adsorption on both catalysts in Figure 2, the CO-poisoning rate at the Pt 2AL -PtFe/C was rapid, losing 99% of MA after 30 min, whereas that at the c-Pt 2 Ru 3 /C was slower, so that about 10% of the initial MA was still maintained after 2 h.Then, the CO coverage θ CO on the surface just after the in situ ATR-FTIR measurement was evaluated from the CO-stripping voltammogram in N 2 -purged solution, as shown in Figure 3.The loss of MA on the Pt 2AL -PtFe/C catalyst is reasonably explained by the θ CO value of 0.99.The value of θ CO on the c-Pt 2 Ru 3 /C catalyst was also 0.99, but, as stated above, the HOR activity was still maintained.It is clear that the adsorbed CO (CO ad ) cannot be oxidized during the HOR at 0.02 V, even on c-Pt 2 Ru 3 /C, because the values of onset potential for the CO ad oxidation on Pt 2AL -PtFe/C and c-P 2 tRu 3 /C were ca.0.50 V and 0.33 V, respectively.The difference in the CO-tolerance will be discussed in more detail below.Figure 4 shows changes in the IR spectra observed on both catalysts during CO adsorption measured simultaneously with the HOR current shown in Figure 2. The bands observed for both catalysts around 2050-1950 cm −1 and 1900-1750 cm −1 were assigned to linearly adsorbed (on-top) CO on Pt [CO L ] and bridged CO on Pt-Pt pair sites [CO B (Pt-Pt)], respectively [7,[12][13][14].A small band around 1950 cm −1 on c-Pt 2 Ru 3 /C (see Figure 5) was assigned to CO ad in a bridged configuration on Pt-Ru and/or Ru-Ru sites [CO-Ru, CO B (Pt-Ru) or CO B (Ru-Ru)] [13].
A nearly identical CO-Ru band around 1950 to 2020 cm −1 at high CO coverage was reported, but it was previously assigned to on-top CO on Ru sites for Ru-decorated Pt [15].However, based on FTIR experiments carried out by the use of size-controlled Pt, Ru, and Pt-Ru particles, Baranova et al. clearly assigned such a band to CO ad bridged on Pt-Ru and/or Ru-Ru sites [16].As has been observed for the CO adsorption process on various catalysts, the changes in the IR spectra in Figure 4 indicate that the bands consists of multiple components, which can be ascribed to CO ad in slightly different configurations or environments, specifically, terraces or step/edge sites [12,13].
A nearly identical CO-Ru band around 1950 to 2020 cm −1 at high CO coverage was reported, but it was previously assigned to on-top CO on Ru sites for Ru-decorated Pt [15].However, based on FTIR experiments carried out by the use of size-controlled Pt, Ru, and Pt-Ru particles, Baranova et al. clearly assigned such a band to COad bridged on Pt-Ru and/or Ru-Ru sites [16].As has been observed for the CO adsorption process on various catalysts, the changes in the IR spectra in Figure 4 indicate that the bands consists of multiple components, which can be ascribed to COad in slightly different configurations or environments, specifically, terraces or step/edge sites [12,13].region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.
Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt2AL-PtFe/C and c-Pt2Ru3/C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The COL band on c-Pt2Ru3/C was deconvoluted into three components, i.e., COL on Pt terrace sites (COL, terrace, 2031 cm −1 ) and two types of COL, on Pt step/edge sites (COL, step/edge-1, 2011 cm −1 and COL, step/edge-2, 1993 cm −1 ).Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt2AL-PtFe/C and c-Pt2Ru3/C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The COL band on c-Pt2Ru3/C was deconvoluted into three components, i.e., COL on Pt terrace sites (COL, terrace, 2031 cm −1 ) and two types of COL, on Pt step/edge sites (COL, step/edge-1, 2011 cm −1 and COL, step/edge-2, 1993 cm −1 ). ) experimental spectrum, ( Figure 5. Deconvolution of FTIR spectra observed on (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C after 2 h of 1% CO/H2 gas bubbling in 0.1 M HClO4.The COL bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.
Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt2AL-PtFe/C and c-Pt2Ru3/C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The COL band on c-Pt2Ru3/C was deconvoluted into three components, i.e., COL on Pt terrace sites (COL, terrace, 2031 cm −1 ) and two types of COL, on Pt step/edge sites (COL, step/edge-1, 2011 cm −1 and COL, step/edge-2, 1993 cm −1 ).Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt2AL-PtFe/C and c-Pt2Ru3/C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The COL band on c-Pt2Ru3/C was deconvoluted into three components, i.e., COL on Pt terrace sites (COL, terrace, 2031 cm −1 ) and two types of COL, on Pt step/edge sites (COL, step/edge-1, 2011 cm −1 and COL, step/edge-2, 1993 cm −1 ).region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.
Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt2AL-PtFe/C and c-Pt2Ru3/C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The COL band on c-Pt2Ru3/C was deconvoluted into three components, i.e., COL on Pt terrace sites (COL, terrace, 2031 cm −1 ) and two types of COL, on Pt step/edge sites (COL, step/edge-1, 2011 cm −1 and COL, step/edge-2, 1993 cm −1 ).Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt2AL-PtFe/C and c-Pt2Ru3/C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The COL band on c-Pt2Ru3/C was deconvoluted into three components, i.e., COL on Pt terrace sites (COL, terrace, 2031 cm −1 ) and two types of COL, on Pt step/edge sites (COL, step/edge-1, 2011 cm −1 and COL, step/edge-2, 1993 cm −1 ). ) CO B peaks.
Then, these bands were deconvoluted into several symmetric Gaussian peaks, and curve fitting was carried out for all spectra assuming the full width at half maximum (FWHM) to be constant and allowing the peak maxima (wavenumber) and areas to vary, as described our previous work [7,12,13].As shown in Figure 5, the FTIR spectra on Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C were successfully deconvoluted into five and six components, respectively, with close correspondence between the fitted and measured spectra (see also Figure S2 in the Supplementary Materials, as another example of deconvolution of FTIR spectra measured after 1 h).The values of peak wavenumber and FWHM of all peaks are summarized in Table S1 in the Supplementary Materials.The CO L band on c-Pt 2 Ru 3 /C was deconvoluted into three components, i.e., CO L on Pt terrace sites (CO L, terrace , 2031 cm −1 ) and two types of CO L , on Pt step/edge sites (CO L, step/edge-1, 2011 cm −1 and CO L, step/edge-2 , 1993 cm −1 ).These peak wavenumbers were very close to those previously reported for c-Pt 2 Ru 3 /C (same composition, but different lot number) measured at 25 • C [13].The values of peak wavenumbers for the three types of CO L on Pt 2AL -PtFe/C were similar to those of c-Pt 2 Ru 3 /C.However, the ratio of integrated intensities of CO L components was quite different so that the intensity of CO L, terrace , I[CO L, terrace ], on c-Pt 2 Ru 3 /C was smaller than that for Pt 2AL -PtFe/C.With respect to the ratio I[CO L, step/edge ] to I[CO L, terrace ], the number of atoms on the terraces and step/edges was calculated, based on a cuboctahedral model, for simplicity, for the fcc nanoparticles Pt 2AL -PtFe with d = 2.9 nm and Pt 2 Ru 3 with d = 3.5 nm.The calculation method [17,18] is shown in Section S1 in the Supplementary Materials.The number ratio of atoms at the step/edge to those at the terrace for Pt 2AL -PtFe/C was calculated to be 57% (see in Table S2), which is in accord with the percentage of I[CO L, step/edge ]/I[CO L, terrace ] of 57% after CO adsorption for 2 h.In contrast, the value of I[CO L, step/edge ]/I[CO L, terrace ] at c-Pt 2 Ru 3 /C was as large as 73%, although the number ratio of atoms at the step/edge to those at the terrace was estimated to be 45%.Hence, even at θ CO ≈ 0.99, the adsorption of CO L at the terrace sites was suppressed on c-Pt 2 Ru 3 /C.This effect of the suppression of CO L, terrace will be discussed later.
The bridged CO ad on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with a peak at 1967 cm −1 , which will be denoted as CO-Ru.The CO B (Pt-Pt) band on both catalysts was deconvoluted into two components: CO B on Pt-Pt pairs on terraces and step/edges.The integrated intensity of CO B (Pt-Pt), specifically on terraces, on the Pt 2AL -PtFe/C was larger than that on c-Pt 2 Ru 3 /C, which is ascribed to the fact that two atomic layers of Pt-skin layers were formed on the PtFe alloy.However, it should be noted that the intensity ratio of I[CO B (Pt-Pt)s] to I[CO L s] observed on c-Pt 2 Ru 3 /C in Figures 3 and 4 was larger than that reported previously [13], suggesting that a Pt-rich surface layer was formed on the present c-Pt 2 Ru 3 /C catalyst, while Ru sites were still present on the surface, because CO-Ru was observed.Thus, we found definite differences in the adsorption state of CO on Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C in the CO adsorption experiment using in situ ATR-FTIR.

FTIR Analysis of Recovery of HOR Activity on CO-Adsorbed Catalysts
It is important to note that the apparently inferior CO-tolerance of the Pt 2AL -PtFe/C catalyst is entirely inconsistent with that evaluated by the CFE [10].Specifically, there were three differences in experimental conditions, i.e., temperatures (60 • C in the present work vs. 70 and 90 • C for the CFE experiment), CO concentrations (1% vs. 0.1%), and protocols.In the CFE experiment, CO was adsorbed on the catalysts by flowing 0.1 M HClO 4 solution saturated with 0.1% CO/H 2 for various time intervals, and the HOR activity was evaluated by a hydrodynamic voltammogram in pure H 2 -saturated solution, followed by the CO-stripping voltammogram in N 2 -purged solution to evaluate θ CO .This protocol was employed to determine the dependence of HOR activity on θ CO with a minimum change in θ CO during the HOR measurement with a slow potential scan rate (1 mV•s −1 ).Then, we adopted a similar dynamic change of the atmosphere to the present in situ ATR-FTIR measurements at 60 • C.After CO adsorption by bubbling 1% CO/H 2 for 30 min, pure H 2 was introduced to saturate the electrolyte solution, to remove dissolved CO.Lastly, the θ CO values on the catalyst were evaluated by measuring the CO-stripping voltammogram.
Changes in the IR spectra during CO adsorption on both catalysts for 30 min (shown in Figure S1 in the Supplementary Materials) were, of course, nearly identical with the corresponding time intervals shown in Figure 4.As shown in Figure 6, small changes in the shape of the CO L band were observed by the introduction of pure H 2 in the solution.The MA for the HOR on both catalysts decreased by CO adsorption similarly to the case of Figure 2, but the MA recovered appreciably just after bubbling pure H 2 , as shown in Figure 7A.For example, the MA on the Pt 2AL -PtFe/C increased from only 1% of the initial value to as high as 22%.To examine the changes in the IR spectra, time courses of integrated intensities of all peaks were plotted in Figure 7B,C.When the CO-adsorbed Pt 2AL -PtFe/C catalyst was contacted with CO-free pure H 2 , the I[CO L, terrace ] decreased by ca.20% after 60 min of H 2 bubbling, accompanied by an increase in I[CO L, step/edge-1 ] by ca.13%.Only a slight decrease in I[CO B, step/edge ] was seen, while I[CO B, terrace ] was nearly unchanged.It is striking that the θ CO evaluated (solid black line in Figure 3A) was still 0.99 even after H 2 bubbling for 60 min, suggesting that the recovery of the HOR activity can be ascribed to a transfer of CO ad within the surface, not by a removal of CO ad from the surface.For the case of c-Pt 2 Ru 3 /C, the MA increased from 14% of the initial value to 52% after 60 min of H 2 bubbling, where I[CO L, terrace ] decreased by ca.12% with an increase in I[CO L, step/edge-1 ] by 8%.The I[CO-Ru] was nearly unchanged during the recovery of the MA.A decrease in I[CO B, step/edge ] (about 10%) was larger than that for Pt 2AL -PtFe/C.Because the value of θ CO evaluated was 0.74, in accord with the remaining MA (24%) at 30 min of CO adsorption, the θ CO on c-Pt 2 Ru 3 /C was, with certainty, unchanged even after H 2 bubbling for 60 min, similar to the case for Pt 2AL -PtFe/C.Here, we discuss a mechanism for the recovery of the HOR activity on CO-adsorbed catalysts by contacting with CO-free pure H2.It has been reported for Pt2Ru3/C catalysts that the presence of a Pt-rich surface overlaid upon Pt-Ru alloy was essential in providing the HOR active sites by weakening the CO adsorption on Pt [13].The Pt-rich surface of the present c-Pt2Ru3/C satisfies this criterion.In our previous work on the CO-tolerance of Pt2Ru3 nanoparticles dispersed on carbon black and Sb-SnO2, we proposed a possible mechanism that the electronic modification (ligand effect) of Pt2Ru3 nanoparticles by the Sb-SnO2 support gave rise to a weakening of the COL, terrace adsorption and suppression of COL, step/edge [7].In ref. [19], the Pt(110) surface exhibited the highest catalytic activity for the HOR, in comparison to those for Pt(111) and Pt(100).Truncated octahedral or truncated cuboctohedral Pt-based (fcc) nanoparticles include (111) and (100) terraces, with (110)-like sites at the edges between two (111) facets.In addition, (110) steps exist on the (111) terraces.Both of these types of (110) sites can be HOR-active.Thus, we considered that, in order to improve CO-tolerant HOR activity on the Pt2Ru3/Sb-SnO2 catalyst, lowering the coverage of COL, step/edge could be beneficial.However, very recently, we proposed a modified HOR mechanism [10] in which, after H2 dissociates 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: H2,sol  H2,ad(step) H2,ad(step)  2Had(step) (Tafel step) 2Had(step)  2Had(terrace) (spillover step) 2Had(terrace)  2H + sol + 2e − (Volmer step) In this mechanism, H2 adsorbs and spontaneously dissociates at step sites due to stronger adsorption of H2 at these sites, since H2 cannot compete with water adsorption on the terraces.After dissociation, however, the H atoms can compete with water more successfully on the terraces, and, Here, we discuss a mechanism for the recovery of the HOR activity on CO-adsorbed catalysts by contacting with CO-free pure H 2 .It has been reported for Pt 2 Ru 3 /C catalysts that the presence of a Pt-rich surface overlaid upon Pt-Ru alloy was essential in providing the HOR active sites by weakening the CO adsorption on Pt [13].The Pt-rich surface of the present c-Pt 2 Ru 3 /C satisfies this criterion.In our previous work on the CO-tolerance of Pt 2 Ru 3 nanoparticles dispersed on carbon black and Sb-SnO 2 , we proposed a possible mechanism that the electronic modification (ligand effect) of Pt 2 Ru 3 nanoparticles by the Sb-SnO 2 support gave rise to a weakening of the CO L, terrace adsorption and suppression of CO L, step/edge [7].In ref. [19], the Pt(110) surface exhibited the highest catalytic activity for the HOR, in comparison to those for Pt(111) and Pt(100).Truncated octahedral or truncated cuboctohedral Pt-based (fcc) nanoparticles include (111) and (100) terraces, with (110)-like sites at the edges between two (111) facets.In addition, (110) steps exist on the (111) terraces.Both of these types of (110) sites can be HOR-active.Thus, we considered that, in order to improve CO-tolerant HOR activity on the Pt 2 Ru 3 /Sb-SnO 2 catalyst, lowering the coverage of CO L, step/edge could be beneficial.However, very recently, we proposed a modified HOR mechanism [10] in which, after H 2 dissociates 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,ad(step) → 2H ad(step) (Tafel step) 2H ad(step) → 2H ad(terrace) (spillover step) 2H ad(terrace) → 2H + sol + 2e − (Volmer step) In this mechanism, H 2 adsorbs and spontaneously dissociates at step sites due to stronger adsorption of H 2 at these sites, since H 2 cannot compete with water adsorption on the terraces.After dissociation, however, the H atoms can compete with water more successfully on the terraces, and, even though the adsorption strength would still be larger at the steps at the same coverage, at high coverage, decreased adsorption strength at the steps might allow H to move to the terraces.
This mechanism was derived to explain the orders of CO-tolerant HOR activities of various catalysts (Pt 2AL -PtFe > Pt 2AL -PtCo > Pt 2AL -PtNi > PtRu > Pt) in the CFE experiments [10].Increased HOR activity was considered to correlate with decreased H adsorption strength on (111) terraces.However, the results shown in Figure 7 can also be well explained by this HOR mechanism.The decrease in the coverage of CO L, terrace would result in an increased number of active sites for reaction (4) on the terrace, assuming that the dissociation rate of H 2 were maintained at the step/edge, for example, step/edge-2.A decrease in the coverage of CO B, step/edge would also be beneficial for reactions (1) and ( 2), because two Pt active sites (Pt-Pt pair site) are formed by the desorption of one CO B .Furthermore, the mobility of CO ad would be enhanced with increasing temperature, consistent with our observation of excellent CO-tolerance for Pt 2AL -PtFe/C at 70 and 90 • C. According to our DFT calculations on nearly all of the surfaces studied, we found that at both step edges and terraces, CO adsorption is stronger by approximately 1 eV compared with the adsorption of atomic hydrogen.Hence, as is well known, the CO concentration must be maintained at very low levels in order for H to compete effectively with CO for adsorption sites.
Even though the use of 1% CO was a very challenging condition for both Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C, we have, for the first time, observed the recovery of the HOR activity of these CO-adsorbed catalysts (without removal of CO ad ) and clarified the mechanism correlated with the mobility of CO ad to create HOR active sites.Such an enhanced mobility of CO ad can be ascribed to a modification of the electronic structure of the Pt 2AL skin on PtFe alloy and the Pt-rich layer on PtRu alloy.
In situ attenuated total reflection Fourier transform infrared reflection spectroscopy (ATR-FTIR) was employed to analyze the CO-tolerant mechanism at the Pt 2AL -PtFe/C and c-Pt 2 Ru 3 /C catalysts at 60 • C. The spectro-electrochemical cell was first reported by Ataka et al. [21], and we modified it to measure the IR spectrum on practical electrocatalysts of Pt or Pt-alloy nanoparticles supported on carbon black.Details of the experimental setup and the procedure of the ATR-FTIR can be found in our previous paper [13]: the cell used is schematically shown in Figure S3 of the Supplementary Materials.The Nafion-coated Pt 2AL -PtFe/C (15 µg Pt •cm −2 ) or c-Pt 2 Ru 3 /C (10 µg Pt •cm −2 ) layer was prepared on an Au film electrode deposited on an Si ATR prism.This amount of catalyst was chosen to ensure a high signal/noise ratio of the IR spectrum.The average thickness of Nafion was 0.013 µm and the geometric surface area of the working electrode was 1.72 cm 2 .The Nafion-coated working electrode was finally heated at 130 • C for 30 min in air.The electrolyte solution used for all experiments was 0.1 M HClO 4 prepared from suprapur-grade HClO 4 (Merck, Frankfurt, Germany) and Milli-Q water.
The ATR-FTIR measurements were carried out in a class 1000 clean room maintaining a constant temperature of 25 • C and humidity of 40% RH.A spectrometer (FTS7000, DIGILAB, Inc., Holliston, MA, USA) with an MCT detector was used.The spectral resolution was set at 4 cm −1 with an interferometer scan of 40 kHz.All IR spectra are displayed in absorbance units, log (I 0 /I), where I 0 and I are the spectral intensities of the reference state and the sample, respectively.The reference electrode employed was an RHE.Prior to all measurements, the working electrode surface was cleaned by

Figure 2 .
Figure 2. Change in the mass activity (MA) for the HOR at 0.02 V on Nafion-coated Pt2AL-PtFe/C and c-Pt2Ru3/C electrodes at 0.02 V and 60° C in 0.1 M HClO4 during CO adsorption by bubbling 1% CO/H2.

Figure 3 .Figure 2 .Figure 2 .
Figure 3. CO stripping voltammograms at Nafion-coated (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C electrodes measured in N2-purged 0.1 M HClO4 at 60 °C and potential sweep rate of 0.02 V•s −1 .Dotted lines indicate the blank CVs on the CO-free electrodes.CVs were measured at 120 min in Figure 2 (red solid line) and 90 min in Figure 7 see below (black solid line).

Figure 3 .Figure 3 .
Figure 3. CO stripping voltammograms at Nafion-coated (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C electrodes measured in N2-purged 0.1 M HClO4 at 60 °C and potential sweep rate of 0.02 V•s −1 .Dotted lines indicate the blank CVs on the CO-free electrodes.CVs were measured at 120 min in Figure 2 (red solid line) and 90 min in Figure 7 see below (black solid line).

Figure 5 .
Figure5.Deconvolution of FTIR spectra observed on (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C after 2 h of 1% CO/H2 gas bubbling in 0.1 M HClO4.The COL bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.

Figure 5 .Figure 5 .
Figure5.Deconvolution of FTIR spectra observed on (A) Pt 2AL -PtFe/C and (B) c-Pt 2 Ru 3 /C at 0.02 V and 60 • C after 2 h of 1% CO/H 2 gas bubbling in 0.1 M HClO 4 .The CO L bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to CO L on terrace and step/edge sites, respectively.The CO ad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt 2 Ru 3 /C.The CO B bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to CO B on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to CO L , I[CO L ]; ( Figure 5. Deconvolution of FTIR spectra observed on (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C after 2 h of 1% CO/H2 gas bubbling in 0.1 M HClO4.The COL bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.

Figure 5 .
Figure5.Deconvolution of FTIR spectra observed on (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C after 2 h of 1% CO/H2 gas bubbling in 0.1 M HClO4.The COL bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.

Figure 5 .
Figure5.Deconvolution of FTIR spectra observed on (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C after 2 h of 1% CO/H2 gas bubbling in 0.1 M HClO4.The COL bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.

Figure 5 .
Figure5.Deconvolution of FTIR spectra observed on (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C after 2 h of 1% CO/H2 gas bubbling in 0.1 M HClO4.The COL bands in the 2050-1950 cm −1 region were deconvoluted into three components, which were assigned to COL on terrace and step/edge sites, respectively.The COad bridged on Pt-Ru and/or Ru-Ru sites was approximated by a broad single band with peak at 1967 cm −1 , which will be denoted as CO-Ru on c-Pt2Ru3/C.The COB bands in the 1850-1790 cm −1 region were also deconvoluted into two components, which were assigned to COB on terrace and step/edge sites.The experimental spectra in (A) and (B) were normalized to the total intensities of peaks assigned to COL, I[COL]; (-------) experimental spectrum, () sum of all peaks, () COL peaks, () CO-Ru peaks, and () COB peaks.

Figure 6 .Figure 6 .
Figure 6.Changes in FTIR spectra observed on CO-adsorbed electrodes of (A) Pt2AL-PtFe/C and (B) c-Pt2Ru3/C at 0.02 V and 60 °C in 0.1 M HClO4 during a recovery process of HOR by bubbling CO-free pure H2 for 60 min (see Figure7 below).CO was pre-adsorbed by bubbling 1% CO/H2 for 30 min, and time intervals shown in (A) and (B) correspond to those from 30 min, at which time the gas was changed from 1% CO/H2 to pure H2.

Figure 7 .
Figure 7. Time courses of the mass activity (MA) for the HOR (A) and integrated intensities of I[COL] (B), and I[COB] (C) observed at 0.02 V and 60 °C in 0.1 M HClO4.First, CO was adsorbed on the working electrode at 0.02 V by bubbling 1% CO (H2-balance) at a flow rate of 10 mL•min −1 for 30 min, and CO-free pure H2 gas was then bubbled in the solution at a flow rate 10 mL•min −1 for an additional 60 min.

Figure 7 .
Figure 7. Time courses of the mass activity (MA) for the HOR (A) and integrated intensities of I[CO L ] (B), and I[CO B ] (C) observed at 0.02 V and 60 • C in 0.1 M HClO 4 .First, CO was adsorbed on the working electrode at 0.02 V by bubbling 1% CO (H 2 -balance) at a flow rate of 10 mL•min −1 for 30 min, and CO-free pure H 2 gas was then bubbled in the solution at a flow rate 10 mL•min −1 for an additional 60 min.
surface, not by a removal of COad from the surface.For the case of c-Pt2Ru3/C, the MA increased from 14% of the initial value to 52% after 60 min of H2 bubbling, where I[COL, terrace] decreased by ca.12% with an increase in I[COL, step/edge-1] by 8%.The I[CO-Ru] was nearly unchanged during the recovery of the MA.A decrease in I[COB, step/edge] (about 10%) was larger than that for Pt2AL-PtFe/C.Because the value of θCO evaluated was 0.74, in accord with the remaining MA (24%) at 30 min of CO adsorption, the θCO on c-Pt2Ru3/C was, with certainty, unchanged even after H2 bubbling for 60 min, similar to the case for Pt2AL-PtFe/C. the