- Email: [email protected]
C electrocatalysts for polymer electrolyte fuel cells
Electrochimica Acta 48 (2003) 1627 /1634 www.elsevier.com/locate/electacta
Erratum
A novel route to prepare stable Pt Ru/C electrocatalysts for polymer electrolyte fuel cells /
A. Pozio *, R.F. Silva, M. De Francesco, F. Cardellini, L. Giorgi ENEA, C.R. Casaccia, IDROCOMB, Via Anguillarese 301, S. Maria di Galeria, Rome 00060, Italy Received 24 July 2002; received in revised form 30 September 2002
Abstract A new method for preparing high surface Pt /Ru/C catalysts at low temperature is described. Pt /Ru on carbon was prepared from carbonaceous material, Pt(NH3)4Cl2 and RuNO(NO3)x (OH)y with borohydride as a reducing agent. Simultaneous reduction of both metals was provided by means of heat treatment. Small and homogeneously dispersed catalyst particles were obtained. XRD and electrochemical measurements show that the performance of the catalyst prepared was similar to that of commercial E-Tek samples, but with increased stability. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Electrocatalysts; PEFCs; DAFCs; Hydrogen oxidation; Platinum alloy
1. Introduction Thanks to their high power-density performance at low temperature (70 /90 8C), Polymer Electrolyte Fuel Cells (PEFCs) and Direct Alcohol Fuel Cells (DAFCs) are very promising as energy sources for electric vehicles and portable devices. In PEFCs the anode gas stream is hydrogen-rich, containing also CO2 and CO produced by reforming or partial oxidation of hydrocarbons. In DAFCs a direct oxidation of alcohol (ethanol or methanol) to CO2 occurs with the production of derived carbonyl species, mainly CO. In the presence of CO, the Pt/C anode catalyst is subjected to poisoning even at very low concentrations (10 ppm). A possible solution for this problem consists in the use of CO tolerant electrocatalysts formed by alloying Pt with a second transition metal (e.g. Ru, Mo) [1 /4]. So far, the best CO tolerant electrocatalyst appears to be the Pt/Ru alloy which is more active than pure Pt in a H2/100 ppm CO anode gas stream. Carbon supported Pt /Ru catalysts can be prepared using several processes [5 /8]. The
PII of original article: S 0 0 1 3 - 4 6 8 6 ( 0 2 ) 0 0 6 2 2 - 9 * Corresponding author. Tel.: /39-0630-484071; fax: /39-0630486357. E-mail address: [email protected] (A. Pozio).
electrocatalytic activity of Pt/Ru/C is intimately related to the preparative process. Castro Luna et al. [5] obtained Pt /Ru/C catalysts with best performance by using Watanabe’s modified method, which consists of providing the sample a thermal treatment at 300 8C in H2 atmosphere after the original procedure proposed by Watanabe [8]. The performance of their catalysts is similar to that of the commercial E-Tek Pt/Ru/C samples in pure hydrogen. Conversely, different preparative methods using other reduction agents such as formic acid or borohydride furnished worse results with respect to the E-Tek samples [5]. However, the commercially available E-Tek 50:50 Pt/Ru catalyst (20% in weight of catalyst) has been shown to be poorly stable in ethanol and methanol at room temperature and its solubility is enhanced by increasing temperature or by using ultrasonic treatment, as predicted by many authors [5 /7,9 /11], thus forming an homogeneous ink. An improvement of catalyst synthesis is needed to eliminate this problem in order to control the exact atomic ratio and ruthenium loading in the electrode, especially in DAFCs where ruthenium can be easily dissolved in ethanol or methanol and is thus lost during the cell life-time. Additionally, ionic ruthenium can diffuse into the polymeric membrane, linking to the sulphonic acid groups and consequently increasing the ohmic resistance [12,13].
0013-4686/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0013-4686(02)00779-X
1628
Erratum
A new proprietary method for preparing stable high surface Pt/Ru/C catalysts at low temperatures and in short times has been employed. Unlike Watanabe’s modified method [5], the present route does not make use of thermal treatment in hydrogen atmosphere and gives rise to catalysts with well defined atomic ratios capable of performing similarly to commercial ones. The route (Fig. 1) involves a proprietary heat treatment that allows very fast reduction kinetics of metal ions, with reduced thermal and diffusive gradients, and permits the production of more homogeneous catalysts. The method has a lower cost than traditional ones and appears to be easily scaled up to manufacturing levels. In this paper, the description of the method is presented as well as the characterisation of the catalyst produced and its comparison with commercial E-Tek.
2. Experimental 2.1. Catalyst and electrode preparation Carbon black powder (Vulcan XC-72, Cabot International) with a specific surface area (BET) of 250 m2 g1 was used as a support for the catalyst. All the samples contained 20% in weight of catalyst dispersed on carbon. Three grams of 50:50 at.% Pt/Ru catalyst were obtained using a proprietary method briefly explained here. A 300 ml slurry composed of carbon black powder, Pt(NH3)4Cl2, RuNO(NO3)x (OH)y and NaOH 2 M solution was prepared. The ink was stirred at 90 8C for 30 min and then cooled in water flow at room temperature. A solution of the reducing agent sodium borohydride (0.5 M) was added to the ink and the bath was quickly heated to the boiling temperature. The mixtures were cooled, dried and washed repeatedly with distilled water. The catalyst powders were heated overnight at 110 8C in an air oven. The platinum and ruthenium percentages (wt.%) were controlled by means of two spectrophotometric methods described elsewhere [14,15]. Commercially available 50:50 Pt /Ru catalyst
powder (20% w/w) on carbon black Vulcan XC-72 was obtained from E-Tek Inc. and used as a reference. Three-layer (substrate/diffusive layer/catalyst layer) gas diffusion electrodes (GDE) were prepared. The substrate was Toray TGPH090 (CP) carbon paper. As already described in our previous work [11], the diffusive layer (DL) was prepared with a weight composition of 80% w/w carbon, 20% w/w PTFE and a loading of about 2 and 0.5 mg cm 2, respectively. The active layer on the GDE was prepared as follows: a mixture containing about 0.1 g of Pt /Ru/C catalyst powder and 3 /4 ml of isopropanol was ultrasonically blended in a glass vial at 40 8C for 14 min to obtain the catalyst paint. A volume of 1 /2 ml of this ink was spread on the surface of the weighted ‘CP/DL’ disk using a micropipette and dried in air flow at 70 8C for about 30 min to eliminate solvent and obtain a thin active layer. The GDE was weighted again and the Pt /Ru loading was calculated for every sample. A platinum loading of 2.1 mg cm 2 was achieved for all the electrodes. An appropriate volume of 5 wt.% NafionTM solution (Du Pont) was spread on the active layers and dried in air flow at room temperature for about 30 min in order to obtain a loading of about 10 mg cm 2. Nafion acts as a protective layer and avoids the loss of catalyst powder into the electrolyte [16]. 2.2. Electrochemical characterisation Galvanostatic polarisation and electrochemical impedance spectroscopy (EIS) measurements were carried out in a conventional three-electrode cell (anode/counter-electrode/reference-electrode) described elsewhere [11,17] containing a 1 M H2SO4 electrolyte solution at 25 8C. The GDE was placed inside a teflon holder provided with a platinum-ring current collector and gas back-feeding. A stream of 16 ml min 1 of H2 or H2/ 100 ppm CO was used in all the experiments of hydrogen oxidation. Methanol oxidation measurements were conducted in the same cell without H2 gas backfeeding and containing a 5 M CH3OH/1.5 M H2SO4 electrolyte at 65 8C [18]. The electrode geometric area
Fig. 1. Flow diagram of the Pt /Ru/C catalyst preparative process.
Erratum
exposed to the electrolyte was 1 cm2. A large area platinum flat electrode was used as a counter electrode. A Hg/HgSO4 reference electrode was connected to the cell through a Luggin capillary whose tip was placed as close as possible to the working electrode surface. The potential values quoted are reported with respect to the normal hydrogen electrode (NHE). The electrochemical cell was connected to a Solartron mod. 1287 potentiostat/galvanostat and a Solartron mod. 1260 frequency response analyser, both interfaced with a GPIB card to a personal computer. EIS measurements were carried out in the frequency range 20 kHz /1 Hz at open circuit potential (OCP). The amplitude of the ac signal was always 10 mVpp.
2.3. Physical chemical characterisation X-ray diffractograms were registered with a Philips diffractometer operating in the Bragg/Brentano para˚ ) radiation focusing geometry. A Cu Ka (l /1.5418 A source was used together with a graphite monochromator on the diffracted beam. A Philips mod. PW1729 high tension generator was operated at 40 kV and 30 mA in the step scan mode with a 0.05 step, acquisition time 36 s per step and in the 2u range 15 /908. Thermogravimetric (TG) analysis of the powder samples was carried out in a DuPont 2000 Thermal Analysis System with 1600 DTA and 951 TGA modules. TG measurements were run in dynamic air or nitrogen streams of 100 ml min 1 and at a heating rate of 5 8C min 1 in the range 25 /300 8C. About 15 mg samples were used and a-alumina powder was kept as the standard material. To control the ruthenium loss after the ultrasonic treatment in the ENEA and E-Tek Pt /Ru/C catalysts the following procedure was used: 30 mg of catalyst powder was dispersed in 10 ml of ethanol and the ink formed was ultrasonically blended at 25 8C for 14 min and then left overnight. The sample was filtered and divided into two parts. On each part a different spectrophotometric method, described elsewhere in detail [14,15], was used to selectively detect the presence of platinum and ruthenium. Particularly, the presence of ruthenium in the second part was determined [15] by measuring sample absorption at l/620 nm with a UV / VIS spectrophotometer Beckman mod. DU65. A method for controlling the ruthenium loss in ethanol and methanol without ultrasonic treatment was also used by leaving 30 mg of the catalyst powder in 10 ml of the two alcohols at 25 8C for 11 days and at 65 8C for 1 h and then analysing the solution with the above method.
1629
3. Results and discussion Spectrophotometric analysis of the ethanol solution after 14 min of ultrasonic treatment at 25 8C shows that the E-Tek catalyst loses 2.38 wt.% of ruthenium while the ENEA sample remains stable (Ru loss B/0.05 wt.%). The ruthenium loss of the E-Tek catalyst in ethanol or methanol without ultrasonic treatment is also verified. By leaving the catalyst powder in ethanol at room temperature for some hours, the solution changed from uncoloured to yellowish and analysis after 11 days showed a 9.02 wt.% loss of Ru in ethanol and 10.47 wt.% in methanol. At 65 8C in ethanol, the effect becomes faster and in only 1 h the E-Tek catalyst lost 8.189/0.05 wt.% of ruthenium. At the same temperature, the Ru loss was 5.929/0.05 wt.% in methanol. On the other hand, the ENEA catalyst shows a very low ruthenium loss: 5/0.4 wt.% after 11 days in ethanol or methanol and 0.09 wt.% after 1 h at 65 8C in ethanol, confirming a very good stability. Another difference between the two bimetallic catalysts is the presence of 2 /3 wt.% of sulphur in the E-Tek catalyst as indicated by our XPS analysis not shown here and from literature data [19,20]. We attribute the presence of sulfur to the E-Tek preparative method [21], similar to that of Watanabe [8], which makes use of sodium bisulphite to convert the PtCl2 precursor to a 6 sulphitic complex prior to the reduction to metallic platinum and ruthenium. In contrast, platinum and ruthenium in the ENEA catalyst are reduced simultaneously with only one reducing agent (NaBH4) and a crystalline alloy is formed without sulphur impurities, as indicated by the XRD data shown below. The anodic galvanostatic polarisation data in pure H2 at 25 8C in 1 M H2SO4 solution are shown in Fig. 2. To compare the two catalysts, the electrolyte resistance (Re) for each electrode was determined by means of EIS and the iRe corrected data (obtained by subtracting the term Re /i from the experimental E vs. i plot) were reported. The ENEA and E-Tek Pt /Ru electrodes show the same performance, characterised by a very low activation polarisation and the absence of diffusion limitation in the analysed current range. Fig. 3 shows the anode galvanostatic polarisation data in methanol at 65 8C. Once again the electrolyte resistance (Re) for each electrode was obtained by means of EIS and the iRe corrected data were reported. The ENEA and E-Tek Pt/Ru electrodes show the same performance and at 150 mA cm 2 the electrode potential corrected for iRe was in the range 430/435 mV versus NHE. Fig. 4 illustrates the Bode plot for the E-Tek and ENEA catalysts at 65 8C and OCP. The difference between Z ? (the real part of impedance) at 20 kHz and at 1 Hz represents the polarisation resistance, R 8p, which is the sum of all possible phenomena: charge transfer, diffusion and adsorption. The R 8p is in the
1630
Erratum
Fig. 2. Galvanostatic steady-state polarisation curves for the H2 oxidation of Pt /Ru/C ENEA and E-Tek catalysts in 1 M H2SO4 at 25 8C. iRe corrected polarisation data are also presented.
Fig. 3. Galvanostatic steady-state (250 s at each current) polarisation curves for the methanol oxidation of Pt /Ru/C ENEA and E-Tek catalysts in 5 M CH3OH/1.5 M H2SO4 at 65 8C. iRe corrected polarisation data are also presented.
range 0.27 /0.30 V cm 2, indicating the same methanol oxidation rate for the two catalysts. EIS was also used to monitoring and comparing the degradation effects of carbon monoxide on the electrode performance at OCP [11,18]. Figs. 5 and 6 show the Bode plot registered continuously while the gas stream was changed from pure to poisoned hydrogen with 100 ppm of carbon monoxide. The ohmic resistance (Re) can be obtained at high frequency and appears to be constant for both catalysts during contamination. The difference between Z ? at low and high frequencies (R 8p), which can be ascribed to the hydrogen oxidation reaction in the catalytic sites, is strictly dependent on the active area. As carbon monoxide contaminates the platinum sites, the electrochemical active area decreases and the polarisation resistance increases rapidly. Fig. 7 reports the trend of R 8p versus time of contamination for both catalysts and shows an increase from 30 mV cm 2 to 1.10 V cm 2 in only 80 min due to a reduction of the electrochemical active area. At zero time both catalysts have very low R 8p values, as already stated from galvanostatic polarisation data (44 mV cm 2). Electrochemical characterisation confirms the same behaviour in pure H2, in H2 contaminated with CO and in methanol for the ENEA and E-Tek catalysts.
Fig. 8 shows the X-ray diffraction patterns of the commercial E-Tek and ENEA catalysts. In both cases, the five characteristic peaks of the face-centred cubic crystalline structure of Pt [22,23] are seen, namely the (111), (200), (220), (311) and the poorly resolved (222). The first broad peak at 2u/258 is assigned to the
Fig. 4. Bode plots of the ENEA and E-Tek 50:50 Pt /Ru GDEs at OCP vs. NHE in 5 M CH3OH/1.5 M H2SO4 at 65 8C.
Erratum
Fig. 5. Bode plots of the ENEA 50:50 Pt /Ru GDEs at OCP vs. NHE in 1 M H2SO4 at 25 8C, under H2/CO 100 ppm stream at different times.
Fig. 6. Bode plots of the E-Tek 50:50 Pt /Ru GDEs at OCP vs. NHE in 1 M H2SO4 at 25 8C, under H2/CO 100 ppm stream at different times.
Vulcan XC-72 support material, as demonstrated from diffractograms recorded for carbon only (data not presented here). Peaks deconvolution of XRD patterns have been performed both with the Rietveld and X-Fit [24] programs to compute the lattice parameter, peak integrated intensity and peak width. Both crystal size and strain produce the broadening of the diffraction peaks, but the crystal size gives a contribution proportional to cos(u ) while the strain contribution increases with tan(u ). The Williamson et Hall [25] plot (not shown) of peak width allows the evaluation of the two
1631
contributions and for both samples the strain can be neglected. Under these conditions, the grain size (D ) was calculated using the Scherrer formula: b cos(u )/kl/D , where b is the full width at half maximum of the diffraction peak in the 2u scale and k is a constant close to unit (1 in this work). The algebraic mean of the grain size (D ) computed from the (111) and (220) diffraction peaks (the best resolved in the diffractograms) was 2.69/ 0.2 nm for both the catalysts. Platinum and ruthenium form a fcc solid solution in the Pt-rich side of the composition range, i.e. [Ru] B/60 at.% [26]. Since the atomic volume of ruthenium is ˚ 3 and that of platinum 15.094 A ˚ 3, the introduc13.577 A tion of Ru in the Pt fcc structure reduces the lattice parameter afcc. According to the variation of afcc with composition for Pt/Ru bulk alloys [27 /29], it is possible to obtain the atomic fraction of Ru present in the alloy. For the ENEA sample, we measured a value of 3.8569/ ˚ for the lattice parameter of the cubic phase, as 0.005 A determined by the peak profile fitting of the (220) reflection, corresponding to 549/5 at.% of Ru in the alloy according to literature data [26,28]. For the E-Tek ˚ was found, corresample, a value of 3.8839/0.005 A sponding to 329/5 at.% of Ru in the solid solution ˚ ) for the E[27,29]. A similar value of afcc (3.883 /3.884 A Tek catalyst was reported by Radmilovic et al. [21] and Martelli et al. [30]. Mukerjee et al. [31] measured a ˚ ) corresponding to 26.1 at.% of Ru. higher afcc (3.890 A Arico` et al. [32] have found a different value for the ˚ ) corresponding to 42.7 at.% lattice parameter (3.870 A of Ru. Finally, Rolison et al. [19] observed that the Ru metal content in E-Tek catalyst is significantly lower than the nominal Ru bulk content, implying that the majority of Ru is not included in the alloy, but present in an oxide phase. Comparing our results with the ones from literature, we can suppose that part of the ruthenium in the E-Tek catalyst is outside the crystalline lattice of the alloy and not present in the Ru hcp crystalline structure (absence of Ru peaks in XRD patterns). This Ru may instead form amorphous ruthenium oxide. In agreement with this result, XPS measurements [17] revealed the presence of typical oxide and hydroxide bonds on E-Tek catalyst. To evaluate the degree of crystalline phase in both samples, a fitting procedure was performed by the deconvolutions of carbon and alloy diffraction peaks and by measuring their integrated intensity, which is proportional to the crystalline fraction of the alloy. Comparing the intensity ratio of carbon and alloy diffraction peaks for both catalysts, it is possible to estimate the amount of alloy that crystallises. Since the (111) and (200) fcc reflections are not well resolved, we used the sum of their intensities for the computation of the quantity of crystalline phase. The integrated intensities of the carbon and (111)/(200) peaks of the ENEA and E-Tek catalysts are listed in Table 1. The intensity
1632
Erratum
Fig. 7. Polarisation resistance at OCP (from EIS data of Figs. 5 and 6) vs. time of the ENEA and E-Tek GDEs in 1 M H2SO4 at 25 8C, under H2/ CO 100 ppm stream.
Fig. 8. XRD patterns of the ENEA and E-Tek Pt /Ru/C catalysts.
Table 1 Integrated intensities of carbon and (111)/(200) XRD peaks of the ETek and ENEA Pt /Ru/C catalysts Catalyst
Carbon
111/200
[Pt/Ru]crystalline (%)
E-Tek ENEA
44 243 44 243
67 927 111 825
60 100
ratio shows an amount of crystalline alloy 409/3% lower in the E-Tek sample with respect to the ENEA catalyst. This result confirms that part of metal in the E-Tek
catalyst is not present as a crystalline phase and agrees with the low Ru content (at.%) present in the alloy with respect to the reported value [19,21,32]. Fig. 9 presents the TG analysis of both catalysts in the range 25/300 8C in air flow. A weight decrease was observed for both catalysts mainly due to the combustion of carbon. We can exclude a predominant water desorption phenomena because the weight loss was absent in nitrogen flow. A significant difference is the presence of an exothermic peak at approximately 270 8C for the ENEA catalyst. This peak is not present in the measurements carried out in nitrogen flow, most likely indicating an oxidative phenomena. Additionally, TG analyses not shown here of the pure Vulcan XC carbon or the ENEA Pt/C catalyst produced with the same process do not show this peak. Therefore, we can conclude that the oxidation of ruthenium occurs at this temperature. XRD data of the ENEA Pt /Ru/C calcined at 270 8C for 1 h in air (Fig. 10) supports this hypothesis showing the presence of ruthenium oxide characteristic peaks at 2u values of 28, 35.2 and 54.48 [22]. In contrast, it was noticed that the behaviour of the E-Tek catalyst treated in the same manner (Fig. 11) is quite similar to that of the untreated sample. These
Fig. 9. TG scans of the ENEA and E-Tek Pt /Ru/C catalysts at 5 8C min1 in flowing air.
Erratum
1633
Fig. 10. XRD patterns of the as-produced ENEA catalyst and after heat treatment at 270 8C for 1 h in air flow.
. a thermal treatment is provided up to 100 8C; thermal and concentration gradients are decreased in the suspension and small catalyst particles are obtained.
Fig. 11. XRD patterns of the as-received E-Tek catalyst and after heat treatment at 270 8C for 1 h in air flow.
results confirm that ruthenium in the ENEA catalyst is completely alloyed with platinum and part of it can be transformed to crystalline oxide by heating in air flow. This leads us to conclude that the ruthenium in the ETek sample cannot be oxidised by heat treatment because it is already present as amorphous oxide [19] or perhaps even as RuS (unidentifiable by XRD). XRD and TG characterisations have thus indicated a marked difference between the two catalysts that can explain their different stabilities.
The produced catalyst is a real 50 at.% Pt /Ru alloy with much more crystalline phase with respect to the commercial E-Tek sample. The electrochemical performances in methanol or hydrogen and the carbon monoxide tolerance are similar to those of E-Tek catalyst. Additionally, the new catalyst is much more stable in ethanol and methanol even if heated or ultrasonicated.
Acknowledgements This work was supported by the Italian University and Scientific Research Ministry (MIUR). The authors would like to thank Dr. M. Carewska (ENEA) for the TG analyses.
References 4. Conclusions A new route for preparing high surface Pt /Ru/C catalysts at low temperature was defined. A satisfactory reduction process of the metals was achieved by means of a proprietary low temperature heat treatment. The performance of the catalysts prepared was similar to that of the commercial E-Tek sample, but with increased stability. The process has a lower cost than the traditional ones and appears to be easily scaled up to manufacturing levels. The main characteristics of the process include: . metal precursors, reducing agents and carbon support are homogeneously and intimately mixed at room temperature in a highly stable single bath;
[1] H.F. Oetjen, V.M. Schmidt, U. Stimming, F. Trila, J. Electrochem. Soc. 143 (1996) 3838. [2] A. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Phys. Chem. 98 (1994) 617. [3] A. Gasteiger, N. Markovic, P.N. Ross, J. Phys. Chem. 99 (1995) 16757. [4] C. Lamy, A. Lima, V. Le Rhun, F. Delime, C. Coutanceau, J.M. Leger, J. Power Source 105 (2002) 283. [5] A.M.C. Luna, G.A. Camara, V.A. Paganin, E.A. Ticianelli, E.R. Gonzalez, Electrochem. Commun. 2 (2000) 222. [6] H. Bo¨nnemann, R. Brinkmann, P. Britz, U. Endruschat, R. Mo¨rtel, U.A. Paulus, G.J. Feldmeyer, T.J. Schmidt, H.A. Gasteiger, R.J. Behm, J. New Mater. Electrochem. Syst. 3 (2000) 199 /206. [7] T.J. Schmidt, H.A. Gasteiger, R.J. Behm, Electrochem. Commun. 1 (1999) 1. [8] M. Watanabe, M. Uchida, S. Motoo, J. Electroanal. Chem. 229 (1987) 395.
1634
Erratum
[9] S. Mukerjee, S.J. Lee, E.A. Ticianelli, J. McBreen, B.N. Grgur, N.M. Markovic, P.N. Ross, J.R. Giallombardo, E.S. De Castro, Electrochem. Solid State Lett. 2/1 (1999) 12. [10] T. Ioroi, N. Fujiwara, Z. Siroma, K. Yasuda, Y. Miyazaki, Electrochem. Commun. 4 (2002) 442. [11] L. Giorgi, E. Antolini, A. Pozio, E. Passalacqua, Electrochim. Acta 43 (1998) 3675. [12] T. Okada, Y. Ayato, M. Yuasa, I. Sekine, J. Phys. Chem. B 103 (1999) 3315. [13] T. Okada, Y. Ayato, H. Satou, M. Yuasa, I. Sekine, J. Phys. Chem. B 105 (2001) 6980. [14] G. Ayres, F. Young, Anal. Chem. 23 (1951) 299. [15] G. Ayres, F. Young, Anal. Chem. 22/10 (1950) 1277. [16] A. Pozio, M. De Francesco, A. Cemmi, F. Cardellini, L. Giorgi, J. Power Sources 105 (2002) 13. [17] L. Giorgi, A. Pozio, C. Bracchini, R. Giorgi, S. Turtu`, J. Appl. Electrochem. 31 (2001) 325. [18] G.T. Burstein, C.J. Barnett, A.R. Kucernak, K.R. Williams, Catalysis Today 38 (1997) 425. [19] D.R. Rolison, P.L. Hagans, K.E. Swider, J.W. Long, Langmuir 15 (1999) 774. [20] A.S. Arico`, P. Cretı`, E. Modica, G. Manforte, V. Baglio, V. Antonucci, Electrochim. Acta 45 (2000) 4319.
[21] V. Radmilovic, H.A. Gasteiger, P.N. Ross, Jr., J. Catal. 154 (1995) 98. [22] K. Lasch, G. Hayn, L. Jorissen, J. Garche, O. Besenhardt, J. Power Source 105 (2002) 305. [23] H. Chunzi, H.R. Kunz, J.M. Fenton, J. Electrochem. Soc. 144 (1997) 970. [24] A.A. Coelho, R.W. Cherary, in: X-ray Line Profile Fitting Program, School of Phisical Sciences University of Technology, Sydney, Australia, 1996. [25] G.K. Williamson, W.H. Hall, Acta Metall. 1 (1953) 22. [26] T.B. Massalski, Binary Alloy Phase Diagrams, vol. 2, American Society for Metals, 1986. [27] G. Lalande, M.C. Denis, J.P. Dodelet, R. Schulz, J. Alloys Compounds 292 (1999) 301. [28] W.B. Pearson, Handbook of Lattice Spacing and Structures of Metals and Alloys, Pergamon Press, London, 1956. [29] H.A. Gasteiger, P.N. Ross, Jr., E.J. Cairns, Surf. Sci. 293 (1993) 67. [30] S. Martelli, P.E. Di Nunzio, Proceedings from the Sixth Italian Meeting on Nanophase Materials, Rome, Italy, May 2001. [31] S. Mukerjee, R.C. Urian, Electrochim. Acta 47 (2002) 3219. [32] A.S. Arico`, P. Cretı´, P.L. Antonucci, J. Cho, H. Kim, V. Antonucci, Electrochim. Acta 43/24 (1998) 3719.
Erratum
A novel route to prepare stable Pt Ru/C electrocatalysts for polymer electrolyte fuel cells /
A. Pozio *, R.F. Silva, M. De Francesco, F. Cardellini, L. Giorgi ENEA, C.R. Casaccia, IDROCOMB, Via Anguillarese 301, S. Maria di Galeria, Rome 00060, Italy Received 24 July 2002; received in revised form 30 September 2002
Abstract A new method for preparing high surface Pt /Ru/C catalysts at low temperature is described. Pt /Ru on carbon was prepared from carbonaceous material, Pt(NH3)4Cl2 and RuNO(NO3)x (OH)y with borohydride as a reducing agent. Simultaneous reduction of both metals was provided by means of heat treatment. Small and homogeneously dispersed catalyst particles were obtained. XRD and electrochemical measurements show that the performance of the catalyst prepared was similar to that of commercial E-Tek samples, but with increased stability. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Electrocatalysts; PEFCs; DAFCs; Hydrogen oxidation; Platinum alloy
1. Introduction Thanks to their high power-density performance at low temperature (70 /90 8C), Polymer Electrolyte Fuel Cells (PEFCs) and Direct Alcohol Fuel Cells (DAFCs) are very promising as energy sources for electric vehicles and portable devices. In PEFCs the anode gas stream is hydrogen-rich, containing also CO2 and CO produced by reforming or partial oxidation of hydrocarbons. In DAFCs a direct oxidation of alcohol (ethanol or methanol) to CO2 occurs with the production of derived carbonyl species, mainly CO. In the presence of CO, the Pt/C anode catalyst is subjected to poisoning even at very low concentrations (10 ppm). A possible solution for this problem consists in the use of CO tolerant electrocatalysts formed by alloying Pt with a second transition metal (e.g. Ru, Mo) [1 /4]. So far, the best CO tolerant electrocatalyst appears to be the Pt/Ru alloy which is more active than pure Pt in a H2/100 ppm CO anode gas stream. Carbon supported Pt /Ru catalysts can be prepared using several processes [5 /8]. The
PII of original article: S 0 0 1 3 - 4 6 8 6 ( 0 2 ) 0 0 6 2 2 - 9 * Corresponding author. Tel.: /39-0630-484071; fax: /39-0630486357. E-mail address: [email protected] (A. Pozio).
electrocatalytic activity of Pt/Ru/C is intimately related to the preparative process. Castro Luna et al. [5] obtained Pt /Ru/C catalysts with best performance by using Watanabe’s modified method, which consists of providing the sample a thermal treatment at 300 8C in H2 atmosphere after the original procedure proposed by Watanabe [8]. The performance of their catalysts is similar to that of the commercial E-Tek Pt/Ru/C samples in pure hydrogen. Conversely, different preparative methods using other reduction agents such as formic acid or borohydride furnished worse results with respect to the E-Tek samples [5]. However, the commercially available E-Tek 50:50 Pt/Ru catalyst (20% in weight of catalyst) has been shown to be poorly stable in ethanol and methanol at room temperature and its solubility is enhanced by increasing temperature or by using ultrasonic treatment, as predicted by many authors [5 /7,9 /11], thus forming an homogeneous ink. An improvement of catalyst synthesis is needed to eliminate this problem in order to control the exact atomic ratio and ruthenium loading in the electrode, especially in DAFCs where ruthenium can be easily dissolved in ethanol or methanol and is thus lost during the cell life-time. Additionally, ionic ruthenium can diffuse into the polymeric membrane, linking to the sulphonic acid groups and consequently increasing the ohmic resistance [12,13].
0013-4686/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0013-4686(02)00779-X
1628
Erratum
A new proprietary method for preparing stable high surface Pt/Ru/C catalysts at low temperatures and in short times has been employed. Unlike Watanabe’s modified method [5], the present route does not make use of thermal treatment in hydrogen atmosphere and gives rise to catalysts with well defined atomic ratios capable of performing similarly to commercial ones. The route (Fig. 1) involves a proprietary heat treatment that allows very fast reduction kinetics of metal ions, with reduced thermal and diffusive gradients, and permits the production of more homogeneous catalysts. The method has a lower cost than traditional ones and appears to be easily scaled up to manufacturing levels. In this paper, the description of the method is presented as well as the characterisation of the catalyst produced and its comparison with commercial E-Tek.
2. Experimental 2.1. Catalyst and electrode preparation Carbon black powder (Vulcan XC-72, Cabot International) with a specific surface area (BET) of 250 m2 g1 was used as a support for the catalyst. All the samples contained 20% in weight of catalyst dispersed on carbon. Three grams of 50:50 at.% Pt/Ru catalyst were obtained using a proprietary method briefly explained here. A 300 ml slurry composed of carbon black powder, Pt(NH3)4Cl2, RuNO(NO3)x (OH)y and NaOH 2 M solution was prepared. The ink was stirred at 90 8C for 30 min and then cooled in water flow at room temperature. A solution of the reducing agent sodium borohydride (0.5 M) was added to the ink and the bath was quickly heated to the boiling temperature. The mixtures were cooled, dried and washed repeatedly with distilled water. The catalyst powders were heated overnight at 110 8C in an air oven. The platinum and ruthenium percentages (wt.%) were controlled by means of two spectrophotometric methods described elsewhere [14,15]. Commercially available 50:50 Pt /Ru catalyst
powder (20% w/w) on carbon black Vulcan XC-72 was obtained from E-Tek Inc. and used as a reference. Three-layer (substrate/diffusive layer/catalyst layer) gas diffusion electrodes (GDE) were prepared. The substrate was Toray TGPH090 (CP) carbon paper. As already described in our previous work [11], the diffusive layer (DL) was prepared with a weight composition of 80% w/w carbon, 20% w/w PTFE and a loading of about 2 and 0.5 mg cm 2, respectively. The active layer on the GDE was prepared as follows: a mixture containing about 0.1 g of Pt /Ru/C catalyst powder and 3 /4 ml of isopropanol was ultrasonically blended in a glass vial at 40 8C for 14 min to obtain the catalyst paint. A volume of 1 /2 ml of this ink was spread on the surface of the weighted ‘CP/DL’ disk using a micropipette and dried in air flow at 70 8C for about 30 min to eliminate solvent and obtain a thin active layer. The GDE was weighted again and the Pt /Ru loading was calculated for every sample. A platinum loading of 2.1 mg cm 2 was achieved for all the electrodes. An appropriate volume of 5 wt.% NafionTM solution (Du Pont) was spread on the active layers and dried in air flow at room temperature for about 30 min in order to obtain a loading of about 10 mg cm 2. Nafion acts as a protective layer and avoids the loss of catalyst powder into the electrolyte [16]. 2.2. Electrochemical characterisation Galvanostatic polarisation and electrochemical impedance spectroscopy (EIS) measurements were carried out in a conventional three-electrode cell (anode/counter-electrode/reference-electrode) described elsewhere [11,17] containing a 1 M H2SO4 electrolyte solution at 25 8C. The GDE was placed inside a teflon holder provided with a platinum-ring current collector and gas back-feeding. A stream of 16 ml min 1 of H2 or H2/ 100 ppm CO was used in all the experiments of hydrogen oxidation. Methanol oxidation measurements were conducted in the same cell without H2 gas backfeeding and containing a 5 M CH3OH/1.5 M H2SO4 electrolyte at 65 8C [18]. The electrode geometric area
Fig. 1. Flow diagram of the Pt /Ru/C catalyst preparative process.
Erratum
exposed to the electrolyte was 1 cm2. A large area platinum flat electrode was used as a counter electrode. A Hg/HgSO4 reference electrode was connected to the cell through a Luggin capillary whose tip was placed as close as possible to the working electrode surface. The potential values quoted are reported with respect to the normal hydrogen electrode (NHE). The electrochemical cell was connected to a Solartron mod. 1287 potentiostat/galvanostat and a Solartron mod. 1260 frequency response analyser, both interfaced with a GPIB card to a personal computer. EIS measurements were carried out in the frequency range 20 kHz /1 Hz at open circuit potential (OCP). The amplitude of the ac signal was always 10 mVpp.
2.3. Physical chemical characterisation X-ray diffractograms were registered with a Philips diffractometer operating in the Bragg/Brentano para˚ ) radiation focusing geometry. A Cu Ka (l /1.5418 A source was used together with a graphite monochromator on the diffracted beam. A Philips mod. PW1729 high tension generator was operated at 40 kV and 30 mA in the step scan mode with a 0.05 step, acquisition time 36 s per step and in the 2u range 15 /908. Thermogravimetric (TG) analysis of the powder samples was carried out in a DuPont 2000 Thermal Analysis System with 1600 DTA and 951 TGA modules. TG measurements were run in dynamic air or nitrogen streams of 100 ml min 1 and at a heating rate of 5 8C min 1 in the range 25 /300 8C. About 15 mg samples were used and a-alumina powder was kept as the standard material. To control the ruthenium loss after the ultrasonic treatment in the ENEA and E-Tek Pt /Ru/C catalysts the following procedure was used: 30 mg of catalyst powder was dispersed in 10 ml of ethanol and the ink formed was ultrasonically blended at 25 8C for 14 min and then left overnight. The sample was filtered and divided into two parts. On each part a different spectrophotometric method, described elsewhere in detail [14,15], was used to selectively detect the presence of platinum and ruthenium. Particularly, the presence of ruthenium in the second part was determined [15] by measuring sample absorption at l/620 nm with a UV / VIS spectrophotometer Beckman mod. DU65. A method for controlling the ruthenium loss in ethanol and methanol without ultrasonic treatment was also used by leaving 30 mg of the catalyst powder in 10 ml of the two alcohols at 25 8C for 11 days and at 65 8C for 1 h and then analysing the solution with the above method.
1629
3. Results and discussion Spectrophotometric analysis of the ethanol solution after 14 min of ultrasonic treatment at 25 8C shows that the E-Tek catalyst loses 2.38 wt.% of ruthenium while the ENEA sample remains stable (Ru loss B/0.05 wt.%). The ruthenium loss of the E-Tek catalyst in ethanol or methanol without ultrasonic treatment is also verified. By leaving the catalyst powder in ethanol at room temperature for some hours, the solution changed from uncoloured to yellowish and analysis after 11 days showed a 9.02 wt.% loss of Ru in ethanol and 10.47 wt.% in methanol. At 65 8C in ethanol, the effect becomes faster and in only 1 h the E-Tek catalyst lost 8.189/0.05 wt.% of ruthenium. At the same temperature, the Ru loss was 5.929/0.05 wt.% in methanol. On the other hand, the ENEA catalyst shows a very low ruthenium loss: 5/0.4 wt.% after 11 days in ethanol or methanol and 0.09 wt.% after 1 h at 65 8C in ethanol, confirming a very good stability. Another difference between the two bimetallic catalysts is the presence of 2 /3 wt.% of sulphur in the E-Tek catalyst as indicated by our XPS analysis not shown here and from literature data [19,20]. We attribute the presence of sulfur to the E-Tek preparative method [21], similar to that of Watanabe [8], which makes use of sodium bisulphite to convert the PtCl2 precursor to a 6 sulphitic complex prior to the reduction to metallic platinum and ruthenium. In contrast, platinum and ruthenium in the ENEA catalyst are reduced simultaneously with only one reducing agent (NaBH4) and a crystalline alloy is formed without sulphur impurities, as indicated by the XRD data shown below. The anodic galvanostatic polarisation data in pure H2 at 25 8C in 1 M H2SO4 solution are shown in Fig. 2. To compare the two catalysts, the electrolyte resistance (Re) for each electrode was determined by means of EIS and the iRe corrected data (obtained by subtracting the term Re /i from the experimental E vs. i plot) were reported. The ENEA and E-Tek Pt /Ru electrodes show the same performance, characterised by a very low activation polarisation and the absence of diffusion limitation in the analysed current range. Fig. 3 shows the anode galvanostatic polarisation data in methanol at 65 8C. Once again the electrolyte resistance (Re) for each electrode was obtained by means of EIS and the iRe corrected data were reported. The ENEA and E-Tek Pt/Ru electrodes show the same performance and at 150 mA cm 2 the electrode potential corrected for iRe was in the range 430/435 mV versus NHE. Fig. 4 illustrates the Bode plot for the E-Tek and ENEA catalysts at 65 8C and OCP. The difference between Z ? (the real part of impedance) at 20 kHz and at 1 Hz represents the polarisation resistance, R 8p, which is the sum of all possible phenomena: charge transfer, diffusion and adsorption. The R 8p is in the
1630
Erratum
Fig. 2. Galvanostatic steady-state polarisation curves for the H2 oxidation of Pt /Ru/C ENEA and E-Tek catalysts in 1 M H2SO4 at 25 8C. iRe corrected polarisation data are also presented.
Fig. 3. Galvanostatic steady-state (250 s at each current) polarisation curves for the methanol oxidation of Pt /Ru/C ENEA and E-Tek catalysts in 5 M CH3OH/1.5 M H2SO4 at 65 8C. iRe corrected polarisation data are also presented.
range 0.27 /0.30 V cm 2, indicating the same methanol oxidation rate for the two catalysts. EIS was also used to monitoring and comparing the degradation effects of carbon monoxide on the electrode performance at OCP [11,18]. Figs. 5 and 6 show the Bode plot registered continuously while the gas stream was changed from pure to poisoned hydrogen with 100 ppm of carbon monoxide. The ohmic resistance (Re) can be obtained at high frequency and appears to be constant for both catalysts during contamination. The difference between Z ? at low and high frequencies (R 8p), which can be ascribed to the hydrogen oxidation reaction in the catalytic sites, is strictly dependent on the active area. As carbon monoxide contaminates the platinum sites, the electrochemical active area decreases and the polarisation resistance increases rapidly. Fig. 7 reports the trend of R 8p versus time of contamination for both catalysts and shows an increase from 30 mV cm 2 to 1.10 V cm 2 in only 80 min due to a reduction of the electrochemical active area. At zero time both catalysts have very low R 8p values, as already stated from galvanostatic polarisation data (44 mV cm 2). Electrochemical characterisation confirms the same behaviour in pure H2, in H2 contaminated with CO and in methanol for the ENEA and E-Tek catalysts.
Fig. 8 shows the X-ray diffraction patterns of the commercial E-Tek and ENEA catalysts. In both cases, the five characteristic peaks of the face-centred cubic crystalline structure of Pt [22,23] are seen, namely the (111), (200), (220), (311) and the poorly resolved (222). The first broad peak at 2u/258 is assigned to the
Fig. 4. Bode plots of the ENEA and E-Tek 50:50 Pt /Ru GDEs at OCP vs. NHE in 5 M CH3OH/1.5 M H2SO4 at 65 8C.
Erratum
Fig. 5. Bode plots of the ENEA 50:50 Pt /Ru GDEs at OCP vs. NHE in 1 M H2SO4 at 25 8C, under H2/CO 100 ppm stream at different times.
Fig. 6. Bode plots of the E-Tek 50:50 Pt /Ru GDEs at OCP vs. NHE in 1 M H2SO4 at 25 8C, under H2/CO 100 ppm stream at different times.
Vulcan XC-72 support material, as demonstrated from diffractograms recorded for carbon only (data not presented here). Peaks deconvolution of XRD patterns have been performed both with the Rietveld and X-Fit [24] programs to compute the lattice parameter, peak integrated intensity and peak width. Both crystal size and strain produce the broadening of the diffraction peaks, but the crystal size gives a contribution proportional to cos(u ) while the strain contribution increases with tan(u ). The Williamson et Hall [25] plot (not shown) of peak width allows the evaluation of the two
1631
contributions and for both samples the strain can be neglected. Under these conditions, the grain size (D ) was calculated using the Scherrer formula: b cos(u )/kl/D , where b is the full width at half maximum of the diffraction peak in the 2u scale and k is a constant close to unit (1 in this work). The algebraic mean of the grain size (D ) computed from the (111) and (220) diffraction peaks (the best resolved in the diffractograms) was 2.69/ 0.2 nm for both the catalysts. Platinum and ruthenium form a fcc solid solution in the Pt-rich side of the composition range, i.e. [Ru] B/60 at.% [26]. Since the atomic volume of ruthenium is ˚ 3 and that of platinum 15.094 A ˚ 3, the introduc13.577 A tion of Ru in the Pt fcc structure reduces the lattice parameter afcc. According to the variation of afcc with composition for Pt/Ru bulk alloys [27 /29], it is possible to obtain the atomic fraction of Ru present in the alloy. For the ENEA sample, we measured a value of 3.8569/ ˚ for the lattice parameter of the cubic phase, as 0.005 A determined by the peak profile fitting of the (220) reflection, corresponding to 549/5 at.% of Ru in the alloy according to literature data [26,28]. For the E-Tek ˚ was found, corresample, a value of 3.8839/0.005 A sponding to 329/5 at.% of Ru in the solid solution ˚ ) for the E[27,29]. A similar value of afcc (3.883 /3.884 A Tek catalyst was reported by Radmilovic et al. [21] and Martelli et al. [30]. Mukerjee et al. [31] measured a ˚ ) corresponding to 26.1 at.% of Ru. higher afcc (3.890 A Arico` et al. [32] have found a different value for the ˚ ) corresponding to 42.7 at.% lattice parameter (3.870 A of Ru. Finally, Rolison et al. [19] observed that the Ru metal content in E-Tek catalyst is significantly lower than the nominal Ru bulk content, implying that the majority of Ru is not included in the alloy, but present in an oxide phase. Comparing our results with the ones from literature, we can suppose that part of the ruthenium in the E-Tek catalyst is outside the crystalline lattice of the alloy and not present in the Ru hcp crystalline structure (absence of Ru peaks in XRD patterns). This Ru may instead form amorphous ruthenium oxide. In agreement with this result, XPS measurements [17] revealed the presence of typical oxide and hydroxide bonds on E-Tek catalyst. To evaluate the degree of crystalline phase in both samples, a fitting procedure was performed by the deconvolutions of carbon and alloy diffraction peaks and by measuring their integrated intensity, which is proportional to the crystalline fraction of the alloy. Comparing the intensity ratio of carbon and alloy diffraction peaks for both catalysts, it is possible to estimate the amount of alloy that crystallises. Since the (111) and (200) fcc reflections are not well resolved, we used the sum of their intensities for the computation of the quantity of crystalline phase. The integrated intensities of the carbon and (111)/(200) peaks of the ENEA and E-Tek catalysts are listed in Table 1. The intensity
1632
Erratum
Fig. 7. Polarisation resistance at OCP (from EIS data of Figs. 5 and 6) vs. time of the ENEA and E-Tek GDEs in 1 M H2SO4 at 25 8C, under H2/ CO 100 ppm stream.
Fig. 8. XRD patterns of the ENEA and E-Tek Pt /Ru/C catalysts.
Table 1 Integrated intensities of carbon and (111)/(200) XRD peaks of the ETek and ENEA Pt /Ru/C catalysts Catalyst
Carbon
111/200
[Pt/Ru]crystalline (%)
E-Tek ENEA
44 243 44 243
67 927 111 825
60 100
ratio shows an amount of crystalline alloy 409/3% lower in the E-Tek sample with respect to the ENEA catalyst. This result confirms that part of metal in the E-Tek
catalyst is not present as a crystalline phase and agrees with the low Ru content (at.%) present in the alloy with respect to the reported value [19,21,32]. Fig. 9 presents the TG analysis of both catalysts in the range 25/300 8C in air flow. A weight decrease was observed for both catalysts mainly due to the combustion of carbon. We can exclude a predominant water desorption phenomena because the weight loss was absent in nitrogen flow. A significant difference is the presence of an exothermic peak at approximately 270 8C for the ENEA catalyst. This peak is not present in the measurements carried out in nitrogen flow, most likely indicating an oxidative phenomena. Additionally, TG analyses not shown here of the pure Vulcan XC carbon or the ENEA Pt/C catalyst produced with the same process do not show this peak. Therefore, we can conclude that the oxidation of ruthenium occurs at this temperature. XRD data of the ENEA Pt /Ru/C calcined at 270 8C for 1 h in air (Fig. 10) supports this hypothesis showing the presence of ruthenium oxide characteristic peaks at 2u values of 28, 35.2 and 54.48 [22]. In contrast, it was noticed that the behaviour of the E-Tek catalyst treated in the same manner (Fig. 11) is quite similar to that of the untreated sample. These
Fig. 9. TG scans of the ENEA and E-Tek Pt /Ru/C catalysts at 5 8C min1 in flowing air.
Erratum
1633
Fig. 10. XRD patterns of the as-produced ENEA catalyst and after heat treatment at 270 8C for 1 h in air flow.
. a thermal treatment is provided up to 100 8C; thermal and concentration gradients are decreased in the suspension and small catalyst particles are obtained.
Fig. 11. XRD patterns of the as-received E-Tek catalyst and after heat treatment at 270 8C for 1 h in air flow.
results confirm that ruthenium in the ENEA catalyst is completely alloyed with platinum and part of it can be transformed to crystalline oxide by heating in air flow. This leads us to conclude that the ruthenium in the ETek sample cannot be oxidised by heat treatment because it is already present as amorphous oxide [19] or perhaps even as RuS (unidentifiable by XRD). XRD and TG characterisations have thus indicated a marked difference between the two catalysts that can explain their different stabilities.
The produced catalyst is a real 50 at.% Pt /Ru alloy with much more crystalline phase with respect to the commercial E-Tek sample. The electrochemical performances in methanol or hydrogen and the carbon monoxide tolerance are similar to those of E-Tek catalyst. Additionally, the new catalyst is much more stable in ethanol and methanol even if heated or ultrasonicated.
Acknowledgements This work was supported by the Italian University and Scientific Research Ministry (MIUR). The authors would like to thank Dr. M. Carewska (ENEA) for the TG analyses.
References 4. Conclusions A new route for preparing high surface Pt /Ru/C catalysts at low temperature was defined. A satisfactory reduction process of the metals was achieved by means of a proprietary low temperature heat treatment. The performance of the catalysts prepared was similar to that of the commercial E-Tek sample, but with increased stability. The process has a lower cost than the traditional ones and appears to be easily scaled up to manufacturing levels. The main characteristics of the process include: . metal precursors, reducing agents and carbon support are homogeneously and intimately mixed at room temperature in a highly stable single bath;
[1] H.F. Oetjen, V.M. Schmidt, U. Stimming, F. Trila, J. Electrochem. Soc. 143 (1996) 3838. [2] A. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Phys. Chem. 98 (1994) 617. [3] A. Gasteiger, N. Markovic, P.N. Ross, J. Phys. Chem. 99 (1995) 16757. [4] C. Lamy, A. Lima, V. Le Rhun, F. Delime, C. Coutanceau, J.M. Leger, J. Power Source 105 (2002) 283. [5] A.M.C. Luna, G.A. Camara, V.A. Paganin, E.A. Ticianelli, E.R. Gonzalez, Electrochem. Commun. 2 (2000) 222. [6] H. Bo¨nnemann, R. Brinkmann, P. Britz, U. Endruschat, R. Mo¨rtel, U.A. Paulus, G.J. Feldmeyer, T.J. Schmidt, H.A. Gasteiger, R.J. Behm, J. New Mater. Electrochem. Syst. 3 (2000) 199 /206. [7] T.J. Schmidt, H.A. Gasteiger, R.J. Behm, Electrochem. Commun. 1 (1999) 1. [8] M. Watanabe, M. Uchida, S. Motoo, J. Electroanal. Chem. 229 (1987) 395.
1634
Erratum
[9] S. Mukerjee, S.J. Lee, E.A. Ticianelli, J. McBreen, B.N. Grgur, N.M. Markovic, P.N. Ross, J.R. Giallombardo, E.S. De Castro, Electrochem. Solid State Lett. 2/1 (1999) 12. [10] T. Ioroi, N. Fujiwara, Z. Siroma, K. Yasuda, Y. Miyazaki, Electrochem. Commun. 4 (2002) 442. [11] L. Giorgi, E. Antolini, A. Pozio, E. Passalacqua, Electrochim. Acta 43 (1998) 3675. [12] T. Okada, Y. Ayato, M. Yuasa, I. Sekine, J. Phys. Chem. B 103 (1999) 3315. [13] T. Okada, Y. Ayato, H. Satou, M. Yuasa, I. Sekine, J. Phys. Chem. B 105 (2001) 6980. [14] G. Ayres, F. Young, Anal. Chem. 23 (1951) 299. [15] G. Ayres, F. Young, Anal. Chem. 22/10 (1950) 1277. [16] A. Pozio, M. De Francesco, A. Cemmi, F. Cardellini, L. Giorgi, J. Power Sources 105 (2002) 13. [17] L. Giorgi, A. Pozio, C. Bracchini, R. Giorgi, S. Turtu`, J. Appl. Electrochem. 31 (2001) 325. [18] G.T. Burstein, C.J. Barnett, A.R. Kucernak, K.R. Williams, Catalysis Today 38 (1997) 425. [19] D.R. Rolison, P.L. Hagans, K.E. Swider, J.W. Long, Langmuir 15 (1999) 774. [20] A.S. Arico`, P. Cretı`, E. Modica, G. Manforte, V. Baglio, V. Antonucci, Electrochim. Acta 45 (2000) 4319.
[21] V. Radmilovic, H.A. Gasteiger, P.N. Ross, Jr., J. Catal. 154 (1995) 98. [22] K. Lasch, G. Hayn, L. Jorissen, J. Garche, O. Besenhardt, J. Power Source 105 (2002) 305. [23] H. Chunzi, H.R. Kunz, J.M. Fenton, J. Electrochem. Soc. 144 (1997) 970. [24] A.A. Coelho, R.W. Cherary, in: X-ray Line Profile Fitting Program, School of Phisical Sciences University of Technology, Sydney, Australia, 1996. [25] G.K. Williamson, W.H. Hall, Acta Metall. 1 (1953) 22. [26] T.B. Massalski, Binary Alloy Phase Diagrams, vol. 2, American Society for Metals, 1986. [27] G. Lalande, M.C. Denis, J.P. Dodelet, R. Schulz, J. Alloys Compounds 292 (1999) 301. [28] W.B. Pearson, Handbook of Lattice Spacing and Structures of Metals and Alloys, Pergamon Press, London, 1956. [29] H.A. Gasteiger, P.N. Ross, Jr., E.J. Cairns, Surf. Sci. 293 (1993) 67. [30] S. Martelli, P.E. Di Nunzio, Proceedings from the Sixth Italian Meeting on Nanophase Materials, Rome, Italy, May 2001. [31] S. Mukerjee, R.C. Urian, Electrochim. Acta 47 (2002) 3219. [32] A.S. Arico`, P. Cretı´, P.L. Antonucci, J. Cho, H. Kim, V. Antonucci, Electrochim. Acta 43/24 (1998) 3719.