Carbon Monoxide Oxidation on Pt-Ru Electrocatalysts Supported on High Surface Area Carbon

Este trabalho descreve a preparação e caracterização de ligas dispersas de Pt-Ru sobre carbono de alta área superficial, as quais foram avaliadas para a oxidação de CO em eletrodos de disco rotatório/camada fina porosa e para a oxidação de hidrogênio em células a combustível de eletrólito polimérico alimentadas com hidrogênio contendo 100 ppm de CO. Tratamentos térmicos (H 2 , 300 °C) aplicados aos catalisadores melhoram a tolerância a pequenas quantidades de CO e, em alguns casos, reduzem o potencial necessário para promover a oxidação de CO durante a varredura do potencial. Sob condições operacionais em uma célula a combustível na presença de CO, foi observado que os melhores resultados foram obtidos quando a liga Pt-Ru/C foi preparada por redução simultânea dos íons Pt (IV) e Ru (III), diferentemente da redução seqüencial.


Introduction
In recent years there has been a steadily growing concern with the degradation of the environment and the effect of pollutants on human health. 1 High levels of pollutants are produced by internal combustion engines, particularly those running on diesel, in large urban centers.Today, the use of clean energy sources is considered an urgent necessity and, among the alternatives, fuel cells are attracting much interest. 2In particular, polymer electrolyte membrane fuel cells (PEMFC), are considered good candidates for transportation and portable applications because they are capable of delivering high power densities and can start operating at room temperature.
In spite of the efforts to develop the direct methanol fuel cell (DMFC), which has the advantage of using a liquid fuel, the most efficient low temperature fuel cells still use hydrogen as fuel.The cheapest way of producing hydrogen is by reforming fossil fuels or low molecular weight alcohols. 3This process produces 6-7% CO, which can be reduced to 1-2% by a shift reaction and to levels smaller than 100 ppm by partial oxidation. 4This CO adsorbs strongly on the Pt catalyst of the fuel cell electrode inhibiting the anodic reaction.The DMFC is not free of this problem, because the oxidation of methanol on Pt produces CO as an intermediate. 5The possibility of using PEMFC in transportation applications is a strong reason to search for solutions of the CO poisoning problem.For the DMFC a solution of this problem would allow to obtain power densities over 0.3 W cm -2 , necessary to make the fuel cell powered car competitive.
When impure hydrogen is used as a fuel, several approaches have been tested to make PEMFC's run with up to 100 ppm CO. [6][7][8][9][10][11][12][13][14][15][16][17][18][19] These involve: i) adding small amounts of oxygen (up to 5%) to the hydrogen entering the anode, [6][7] ii) performing a catalytic oxidation using Co oxide, 8 iii) using in the anode Pt alloys with other metals like Ru, [9][10][11][12][13] Sn, 14 Rh, 15 Mo, 16 Re, 17 or some noble metals like Au 18 and iv) using Pt-Ru alloys and adding small amounts of hydrogen peroxide, as an oxygen carrier, to the water hydrating the membrane. 19,20lthough some reports point to the Pt-Mo alloy as being the more effective, 21 Pt-Ru alloys have been more extensively investigated.The use of Pt alloys is based on the fact that the less noble metal forms the hydrated oxides necessary to oxidize CO at lower potentials.Thus, the CO adsorbed on Pt is oxidized by the second metal through the so-called bifunctional mechanism: 22 Pt-CO + 2Ru-OH ® CO 2 + Pt + 2Ru + H 2 O (1) In order to understand better the oxidation of CO to CO 2 several studies have been carried out on the adsorption of CO on Pt, using techniques as FTIR 23 and DEMS. 24It is observed that CO adsorbs on Pt in two forms: bridge and linear bonding. 25,26Morimoto et al. 27 used in situ FTIR and cyclic voltammetry (CV) to study the characteristics of the adsorption of CO on surfaces with different morphologies.They found that on smooth surfaces the bridge form amounts to 12%, while for rough surfaces this figure increases to 29%.
Ianiello et al. 28 studied the oxidation of CO on smooth Pt and Pt-Ru alloys by in situ FTIR identifying the OH species that participate in the process.The lowest oxidation potential was found for the 50:50 Pt-Ru material.A similar result was presented by Arico et al. 29 for Pt-Ru alloys supported on carbon.On smooth rotating disk electrodes (RDE) Gasteiger et al. 9 proposed that the oxidation of CO on Pt-Ru follows a Langmuir-Hinshelwood mechanism.Later, Schmidt et al. 30 proposed a method for the evaluation of catalysts for the oxidation of H 2 and CO by working with Pt and Pt-Ru particles supported on carbon and covered with a Nafion Ò film.A kinetic study of the oxidation of CO was presented later by the same authors. 31Recently, it was demonstrated in this laboratory that carbon supported Pt and Pt-Ru electrodes show a better performance for the oxidation of H 2 containing 100 ppm CO when submitted to a thermal treatment. 32ne of the electrochemical methods that has allowed a close study of certain electrode reactions on supported catalysts is the thin porous coating rotating disk electrode (TPC/RDE). 33This technique was used for the study of oxygen reduction, [34][35][36] and methanol oxidation 37 on carbon supported catalysts, allowing kinetic and mechanistic characterizations.In this work, the TPC/RDE technique was used to study the activity of several Pt-Ru/C electrocatalysts for the oxidation of CO prepared by different methods which were eventually submitted to a thermal treatment in a hydrogen atmosphere.The catalysts that showed the better performances on the TPC/RDE system were also tested in single PEMFC operating with hydrogen containing 100 ppm CO.

Experimental
Carbon supported Pt-Ru catalysts were prepared by the following methods, using carbon powder (Vulcan XC-72) pre-treated as described elsewhere: 38 Formic acid method (FAM) 39 This method was developed in this laboratory and uses formic acid for the reduction of Pt and Ru.Developed initially as a simple method not requiring thermal treatments, it was found later that an adequate thermal treatment (here designed as T. T.) improves the performance of the catalyst for H 2 /CO mixtures. 32For the preparation of Pt-Ru supported alloys freshly prepared aqueous solution of RuCl 3 as precursor were used, to avoid the formation of complexes RuO[(H 2 O) 4 ] +2 present in aged solutions of RuCl 3 . 40 is similar to a method proposed earlier by Watanabe et al. (WM) 41 but includes a thermal treatment in a hydrogen atmosphere at 300 o C for one hour.

Radmilovic et al. method (RM)
In all cases the metal content with respect to carbon was 20% by weight.For the purpose of making comparisons, commercial carbon supported Pt and 50:50 Pt-Ru (E-TEK) catalysts were also used.
The composition of the prepared catalysts was examined by energy dispersive X-ray (EDX) analysis using a scanning electron microscope (DSM 960 Zeiss) with a 20 keV electron beam and provided with a microanaliser Link Analytical QX 2000 and a detector of SiLi.
The particle sizes of the catalysts were approximately evaluated by X-ray diffraction (XRD) using a URD-6 Carl Zeiss-Jena diffractometer.The X-ray diffractograms were obtained with a low scan rate of 0.05 degree s -1 for 2q values between 30 and 100°.The particle size determinations were made using the peak associated to the (220) face of the fcc platinum lattice at 2q values between 60 and 80° with a scan rate of 0.02 degree s -1 .In all cases the incident wavelength (KaCu) was 1.5406 Å.
5][36] The TPC/RDE was constructed using a PTFE cylinder with a cavity 0.15 mm deep and 0.19 cm 2 area.The carbon supported catalyst was agglomerated with a 6% PTFE suspension (DuPont) and placed in the cavity.The cylinder was joined to a Pine Instrument AFMSRE rotating system with a control unit ASR2E.A reversible hydrogen electrode (RHE) was used as reference and a 2 cm 2 Pt foil as secondary electrode.The base electrolyte was 0.5 mol L -1 H 2 SO 4 (Mallinckrodt) prepared with purified water in a Milli-Q (Millipore) system.Electrochemical experiments were done with a EG&G PAR 273 potentiostat/galvanostat coupled to a personal computer, using the software M270 (EG&G).
Gas diffusion electrodes for single cells were prepared using a carbon cloth (Stackpole) purified by a thermal treatment at 450 o C followed by treatment with 0.5 mol L -1 HNO 3 at 80 o C. On the carbon cloth, a diffusion layer consisting of carbon powder (Vulcan XC-72R, Cabot) and 15% PTFE was applied first.A catalytic layer containing Nafion ® and Pt supported alloys for the anode and Pt/C for the cathode was brushed on the diffusion layer.In all cases the metal content was 0.4 mg cm -2 .The Nafion ® content was 1.1 mg cm -2 .Membrane and electrode assemblies were prepared with a Nafion ® 115 membrane (DuPont) by hot pressing the electrodes at 50 atm and 125 o C. Single PEMFC experiments were carried out galvanostatically at 85 o C with pure oxygen at 1.7 atm and humidified at 90 o C and with either pure hydrogen or hydrogen containing 100 ppm CO, at 2 atm and humidified at 100 o C.

EDX results
Table 1 shows the atomic composition of the different Pt-Ru/C catalysts employed in this work determined by EDX.The emitted radiation energies are La = 9.441 kV and M = 2.048 kV for Pt and La = 2.558 kV and M = 0.461 kV for Ru.Thus, the value of the L energy for Ru is near the value of the M radiation for Pt, and this may introduce errors.Table 1 shows the composition values determined.Some Pt-Ru/C catalysts, with an approximate composition 80:20 were prepared by either simultaneous or successive deposition of the metals on carbon.The amount of Ru was similar in both cases.
Previous results 32 show that when the FAM method is used to prepare Pt-Ru/C catalysts it is very difficult to obtain alloy compositions with more than 30% Ru.On the other hand, catalysts prepared with the RM method allow the preparation of alloys with an atomic Ru content up to 50%.

XRD results
Figure 1 shows the diffractograms for the Pt-Ru/C catalysts prepared with the FAM.The peaks at 2q = 40, 47, 67 and 82 are associated to the (111), ( 200), ( 220) and (311) planes, respectively, of the fcc structure of platinum, 42 or a Pt-Ru phase rich in platinum that retains the fcc structure.It was not observed any peak due to metallic ruthenium or to materials rich in Ru with hexagonal structure.These structures would produce a peak at 2q = 44, due to a reflection of the (101) plane, the most intense peak of hcp Ru, that could indicate their presence.However, it is known that for equal quantities of Pt and Ru the intensities of Ru peaks are lower than those of Pt peaks  and for this reason metallic Ru may not be detected in the carbon support.] When the supported alloys were treated with H 2 for one hour at 300 o C it was observed that the peaks in the diffractograms become narrower, which could be due to: i) an increase in the cristallinity, ii) a lowering of the internal stress in the net of the alloy caused by a possible migration of Ru (as a substitution defect) or iii) an increase of the particle size.The same features are observed in the diffractograms of Figure 2, with the materials prepared by anchoring either Pt or Ru first.The profiles of the spectra were unchanged even when the catalyst powders were submitted to thermal treatment.In these cases it was expected to see the Ru peaks because the metals were anchored separately onto the carbon.Hovewer, no Ru peaks were observed probably due to the reasons discussed above.Figure 3 shows the diffractograms of the material prepared by the RM method and the commercial E-TEK electrocatalysts.For the RM material peaks associated to hexagonal Ru or species rich in Ru are not present, which suggests that in these materials the association between Pt and Ru is predominant.The diffractograms for the E-TEK alloys (Figure 3) are rather featureless, indicating either low crystalinity or very small particle sizes, preventing calculations of particle size or cell parameter.

Mean particle size
The mean particle size of the catalysts was determined from the X-ray diffractograms using Scherrer's equation 45 and assuming that the particles are spherical.In this case, the (220) peak of fcc platinum was chosen because it appears in a region where the carbon substrate contributions can be removed, allowing a gaussian adjustment of the peaks to eliminate the background contributions.In order to obtain more precise measurements, the patterns for particle size determinations were obtained for 2q between 60 and 80 ° with a scan rate of 0.02 degree s -1 .The mean particle sizes for all the materials used in this work are collected in Table 1.For the materials prepared with the FAM, particle sizes vary between 2.5 and 5 nm, being comparable to those resulting from other preparation methods.The particle sizes for catalysts prepared with the RM are close to those published by Watanabe et al. (3-4 nm) 41 and Radmilovic et al. (1.9-2.3 nm). 40Table 1 also shows that in all cases the particle size increases with the thermal treatment, probably due to sintering and agglomeration of the particles.

Cell parameter "a"
From the X-ray diffractograms, the cell parameter a was also determined for the catalysts used in this work, and the values are shown in Table 1.In all cases the values of a are smaller than those found for pure Pt (0.3923 nm), which indicates the presence of Ru substituting Pt in the lattice and leading to a smaller value of a.It must be noted that the value of a is slightly larger for the materials submitted to thermal treatment which may indicate some segregation in the alloy.Considering the Pt-Ru/C 80:20 alloy, the lattice parameter is smaller when both metals are anchored on the carbon particles simultaneously than when Pt is anchored first.This is a consequence of the presence of Ru in the fcc lattice of Pt, as further demonstrated by the still smaller value of a when Ru is anchored first.When the materials are thermally treated, the values of a for the catalysts obtained by simultaneous deposition and with  the previous deposition of Pt increase, approaching the value for pure Pt.On the other hand, the a parameter for the catalysts obtained with the initial deposition of Ru remains lower.In this case, a predominance of Ru in the lattice prevails, even after the thermal treatment.
7][48] Arc melted Pt-Ru alloys show a = 0.3904 nm for Pt-Ru 80:20 and 0.3864 nm for Pt-Ru 50:50, while this last composition has a value of 0.38965 nm when prepared by chemical reduction with sodium borohydride on a carbon support. 47Radmilovic et al. 40 found 0.3898 nm for the 75:25 and 0.3884 nm for the 50:50 Pt-Ru/C alloys.Thus, this method seems to favor the formation of solid solutions more than the FAM.Recently, it was reported by Takasu et al. 48that the Ru precursor used to produce the catalysts has an effect on the structure of the material.Thus, by using RuCl 3 to produce a Pt-Ru/C 50:50 alloy the value of a was 0.391 nm, similar to that found in this work.With Ru 3 (CO) 12 the values of a were 0.3882 and 0.390 nm for Pt-Ru/C 75:25 and 50:50 respectively, and with RuNO (NO 3 ) x the values were even smaller: 0.3862 and 0.3865 nm, respectively.The possible explanation for the low incorporation of Ru when RuCl 3 is used, is the formation of [Ru(CO)(H 2 O)Cl 4 ] -2 49 from RuCl 3 and HCOOH in solution when the reaction time is increased.The materials containing Ru do not present the peaks in the hydrogen region observed for pure Pt, which is due to the formation of Ru oxides at those potentials. 50Also, currents in the double layer are larger for these materials because of oxide formation.Some differences are apparent between the Pt-Ru/C E-TEK and the Pt-Ru/C RM.The former has a more characteristic hydrogen region and lower currents in the double layer region which is probably due to the different preparation methods.The Pt/C E-TEK material shows the typical behavior for this material in acid medium and it does not show the peak at 0.18 V due to the oxidation of hydrogen in the submonolayer. 51igure 5 shows the CV's for the materials in which the metals were anchored simultaneously and separately by the FAM.Without any thermal treatment and when the metals are incorporated simultaneously (Figure 5A) the profile of the CV is that expected for an alloy in the sense that there is no definition of the hydrogen peaks.The profiles are slightly better defined when Ru was anchored first but the response is different from that of Pt/C E-TEK catalyst (Figure 5B).After the thermal treatment there are clear changes in the profiles that can be the result of: i) a reduced amount of Ru oxides, ii) segregation of Ru, iii) elimination of impurities, which may increase the area of the clean active surface and iv) an increase of particle size.Effects (i-iii) are expected to increase the current, while effect (iv) should result in a decrease.

Electrochemical characterization of the catalysts Cyclic voltammetry.
The thermal treatment results in a marked increase in the current levels for the material in which the metals were anchored simultaneously and for the material in which Ru was anchored first (Figure 5A).It is probable that, in spite of the increase in particle size, effects i-iii are predominant for these materials.The opposite must be the case for the material in which Pt was anchored first (Figure 5B) because a decrease of current levels is observed as a consequence of the thermal treatment.
Linear sweep voltammetry.The dependence of the amount of CO adsorbed with the electrode potential was studied by linear sweep voltammetry.In order to guarantee equilibrium conditions for the CO adsorption process with the TPC/RDE, the electrode was rotated at 1600 rpm for an adequate time.This was necessary because of the low solubility (0.96 x 10 -3 mol L -1 ) and diffusion coefficient (1.8 x 10 -5 cm 2 s -1 ) of CO in 0.5 mol L -1 H 2 SO 4 . 9Also, before applying the linear voltage scan it was necessary to eliminate the CO in the solution and in the pores of the TPC/RDE.This was done by bubbling N 2 with the electrode rotating at 1600 rpm until the response to the linear voltage scan was the same with and without rotation.Figure 6 shows a linear sweep voltammogram on Pt/C E-TEK for the oxidation of CO previously adsorbed at different potentials.Between 50 and 400 mV there is no dependence of the profiles on the potential at which CO was adsorbed.This is because at these potentials the CO is not oxidized on Pt, which needs the participation of adsorbed OH species formed at potentials above 0.7 V vs. RHE.Figure 7 shows the same experiment with Pt-Ru/C (50:50 E-TEK).In this case, the charge due to CO oxidation decreases slightly for increasing values of the CO adsorption potential.This is due to the oxidation of CO to CO 2 at potentials over 200 mV, induced by oxygenated species formed on Ru, and the extent of oxidation increases for increasing potentials.Because of this effect, all experiments were done adsorbing CO at 50 mV.
The stripping of CO on Pt presents an oxidation peak at a potential that depends on several factors.On the Pt/C electrodes used in this work, the peak potential is 0.815 V at 10 mV s -1 , which is much more positive than the value of 0.73 V observed on smooth Pt 9 and platinized Pt. 52 Schmidt et al. 30 found a peak potential of 0.81 V at 20 mV s -1 on a thin supported catalyst electrode covered with Nafion Ò , with no influence of diffusional effects, which agrees with the value of 0.80 V found on Pt (111) by the same authors.For scan rates of 50 mV s -1 other authors found a peak potential of 0.87 V. 53 As discussed by Schmidt et al. 30 the oxidation peak potential of CO on Pt nanoparticles appears around 0.81 V due to the fact that in small particles there is a predominance of (111) facets on the surface. 40,54igure 8A shows linear sweep voltammetry curves for the oxidation of CO on the Pt-Ru/C 80:20 catalysts prepared by the FAM and 50:50 prepared by the Radmilovic method.Also, the curves for the commercial catalysts Pt/C and Pt-Ru/C 50:50 E-TEK are shown.Neglecting double layer charging contributions, no currents are observed up to 0.2 V. On Pt/C a pre-wave can be observed between 0.20 and 0.75 V.This pre-wave was attributed to the oxidation of weakly adsorbed CO, 55 and it can also be seen up to 0.5 V on Pt-Ru/C materials (Figure 8A).Between 0.4 and 0.63 V the most active electrocatalyst is the commercial material Pt-Ru/C 50:50 E-TEK.Between 0.5 and 0.63 V the Pt-Ru/C 80:20 prepared by the FAM  and submitted to thermal treatment and the material prepared by the Radmilovic method present similar activity for the oxidation of CO.In spite of the improvement of the FAM material due to the thermal treatment the performance is not as good as that of the commercial E-TEK catalysts.Although a strict comparison cannot be made due to the differences in Ru content, it is probable that the better performance of the commercial material is due to a smaller particle size and a better Ru distribution on the surface of the electrocatalyst.
Figure 8B shows linear sweep voltammograms for the oxidation of CO on Pt-Ru/C 80:20 materials in which one of the metals was anchored first (Pt-Ru/C or Ru-Pt/C).It also shows the effect of the thermal treatment on these materials.The materials show some evidences of inhomogeneity.For example, in the material where Ru was anchored first, the peak at 0.5 V may be characteristic of CO oxidation on Ru particles. 9Without a thermal treatment, the alloy in which Pt was anchored first shows the best performance.That of the Ru-Pt/C material is poor, as is that of the material with simultaneous anchoring (Figure 8A).Evidently, the thermal treatment introduces qualitative changes in the performance of the electrocatalysts for the oxidation of CO.The performance of the Ru-Pt/C material and that of the material with simultaneous anchoring show a marked improvement, while that of the Pt-Ru/C materials decreases somewhat.This is mainly the result of the combined effects of segregation of the materials and the increase in particle size due to sintering and agglomeration.The conclusion is that the Ru-Pt/C material with thermal treatment presents the lower potential for the oxidation of CO and currents comparable with those of the other materials.
The charge associated to the oxidation of CO, corrected for the double layer charging, was used to calculate the active areas of the catalysts and the values are presented in Table 2.The catalysts prepared by the FAM present an active area similar to that of Pt/C E-TEK, which is smaller than that of Pt-Ru/C E-TEK, probably due to the smaller particle size for this last material.For the catalysts prepared by the FAM, the thermal treatment promotes an increase in the active area, particularly for the material where the two metals were anchored simultaneously.This material without the thermal treatment presents active areas similar to those reported by Takasu et al., 48 who found active areas of 34 m 2 g -1 for Pt-Ru/C materials prepared by impregnation using RuCl 3 as precursor.

Experiments in single PEMFC
In order to extrapolate the results obtained with the TPC/RDE to practical systems, experiments were carried out in single PEMFC using Pt-Ru/C 80:20 catalysts prepared by the FAM. Figure 9 shows the performance of single cells operating with pure H 2 and with H 2 containing 100 ppm CO.The catalysts used in the gas diffusion electrodes were all thermally treated for 1 h in a H 2 atmosphere at 300 o C. For comparison, Figure 9 incorporates results obtained with Pt/C and Pt-Ru/C E-TEK.

Figure 3 .
Figure 3. X-ray diffractograms for Pt-Ru/C alloys prepared by the Radmilovic method and commercial E-TEK electrocatalysts.

Figure 5 .
Figure 5. Cyclic voltammetries of Pt-Ru/C 80:20 alloys prepared by the FAM, with and without thermal treatment (T.T.) (A) For the two metals anchored simultaneously (s) and Ru anchored first (Ru-Pt/ C). (B) For the Pt anchored first (Pt-Ru/C) and, for comparison, for Pt/C E-TEK.

Figure 6 .
Figure 6.Stripping curves for CO oxidation on Pt/C E-TEK with adsorption of CO at several potential values indicated in the Figure in mV.v =10 mV s -1 .A cyclic voltammogram without CO is also included.

Figure 7 .
Figure 7. Stripping curves for CO oxidation on Pt-Ru/C 50:50 E-TEK with adsorption of CO at several potential values.v = 10 mV s -1 .A cyclic voltammogram without CO is also included.

Table 2 .
Electroactive areas of the different materials prepared by the FAM determined through the charge to oxidise a monolayer of CO.T.T.: thermally treated materials.The areas were normalized per g of Pt for all the : simultaneously anchoring of Pt and Ru; T. T.: Thermal treatment; a Ru anchored first; b Pt anchored first.

Figure 8 .
Figure 8. (A) Stripping curves of CO on Pt-Ru/C electrocatalysts, prepared by different methods, and E-TEK materials.Adsorption of CO at 50 mV vs RHE.v = 10 mV s -1 .T. T.: thermal treatment.(B) Stripping curves of CO on Pt-Ru/C FAM electrocatalysts anchored in succession (with the order indicated in the material).Adsorption of CO at 50 mV vs RHE.v = 10 mV s -1 .T.T.: thermal treatment.

Table 1 .
Composition (EDX), particle size (XRD) and cell parameter a (XRD) for the several Pt-Ru/C alloys c From reference 40; d Unpublished TEM results.