- Email: [email protected]
Surfactant stabilized Pt and Pt alloy electrocatalyst for polymer electrolyte fuel cells
Electrochimica Acta 47 (2002) 2981 /2987 www.elsevier.com/locate/electacta
Surfactant stabilized Pt and Pt alloy electrocatalyst for polymer electrolyte fuel cells Xin Wang, I-Ming Hsing * Department of Chemical Engineering, Hong Kong University of Science and Technology, Room 4551 Clear Water Bay, Kowloon, Hong Kong Received 17 December 2001; received in revised form 20 March 2002
Abstract A simple process for the synthesis of carbon supported Pt and Pt/Ru electrocatalysts was investigated. Borrowing from the homogeneous catalyst preparation, this process uses a surfactant as a stabilizer which prevents the metal colloids from aggregation during the reduction process without influencing the deposition of the colloids onto the carbon support. Chemical, morphological and crystallographic properties of the newly prepared electrocatalysts were characterized using various surface techniques including X-ray diffraction (XRD), Transmission electron microscopy (TEM) and X-ray photoelectron spectroscopy (XPS). These techniques show that mono-size, well-dispersed metal colloids can be formed and successfully supported on the carbon black. Moreover, the size of metallic colloids prepared by this method can be manipulated by controlling the synthesis temperature and is independent of the catalyst loading. Electrochemical characterizations show that in comparison with commercial E-TEK electrocatalysts, surfactant-based Pt/C electrocatalysts possess similar catalytic activity in terms of oxygen reduction and higher CO tolerance performance can be obtained by the surfactant stabilized Pt,Ru/C. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Surfactant; Stabilizer; Pt; Electrocatalyst; Fuel cell
1. Introduction One of the obstacles preventing polymer electrolyte fuel cells from commercialization is the high cost of noble metals to be used as catalyst, such as platinum and ruthenium. To effectively use these metals, they have to be well dispersed to small particles on conductive carbon supports. Conventional methods, such as impregnation, co-precipitation and ion exchange, are not adequate from the viewpoint of offering high metal surface area or good dispersion of metal particles [1]. To avoid the above disadvantages, an alternative method has been investigated, involving the use of prefabricated metal colloids and subsequent deposition on the support material. Different fabrication routes are proposed based on this methodology, among which, the most common one is the sulfite-complex route [1 /3]. Recently, Bonnemann and coworkers introduced a new
* Corresponding author. Tel.: /852-2358-7131; fax: /852-23580054 E-mail address: [email protected] (I.-M. Hsing).
route using an organometallic compound to synthesize colloidal precursors for the carbon-supported catalyst, where the organic molecules are used to prevent the agglomeration and coalescence of the particles [4]. However, these methods are complex and require a delicate control of operating conditions. In this study, we report a simple procedure for the preparation of Pt and Pt,Ru/C catalysts based on an alcohol reduction method. This alcohol reduction method [5 /7] has been widely used in the preparation of metal colloids for homogeneous catalysis, where mono-size, well dispersed metal colloids can be formed and stabilized in aqueous solution with the existence of polymer via steric mechanism [8]. Here, its application is extended to the preparation of heterogeneous catalyst. Preliminary experiment shows that the formed metal colloids cannot be immobilized onto the carbon support since the steric effect of polymer is so strong that the carbon black is also stabilized by polymer. Considering the size difference between the metal colloid (2 /3 nm) and the carbon black (30 nm), a surfactant Dodecyldimethyl (3-sulfo-propyl) ammonium hydroxide (SB12) is carefully chosen as the stabilizer where the stabilization
0013-4686/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 3 - 4 6 8 6 ( 0 2 ) 0 0 1 9 9 - 8
2982
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
effect is only sufficient to prevent metal colloids from aggregation and will not influence the immobilization of colloids onto the carbon black. Different from the polymer-based stabilizer, the stabilization mechanism of SB12 surfactant is possibly related to the electrostatic effect [9] with the hydrophobic end of the surfactant oriented towards the particle surface and the charged hydrophilic part pointing towards the solution. Physicochemical and electrochemical characterizations of the prepared catalysts are carried out to determine the chemical, morphological, crystallographic, as well as electrochemical properties of the catalysts.
2. Experimental 2.1. Catalyst preparation The surfactant-protected Pt/C and Pt,Ru/C catalysts (denoted as SFT) were prepared by a modified alcoholreduction method. Surfactant SB12, chloroplatinic acid H2PtCl6, Vulcan XC-72 carbon black and RuCl3 (used only for Pt,Ru/C case) were mixed in a methanol /water (volumetric ratio 1:3) solvent mixture to form a suspension. The resulting suspension was stirred and refluxed under air at elevated temperature for 1 h and then filtered and washed. The filtrate was analyzed with inductively coupled plasma atomic emission spectroscopy (ICP-AES), which indicates that most Pt and Ru have been adsorbed on the carbon black. For Pt/C catalyst, two different loadings (20 and 40 wt.%) were prepared by varying the amount of carbon black. Experiments were also carried out at different temperatures to determine their effects on the Pt colloids size. As to Pt,Ru/C catalyst, two batches with different atomic ratios (1:1 and 7:3) were prepared. 2.2. Physicochemical characterization Powder X-ray diffraction (XRD) patterns for the catalysts were obtained on a Philips Powder Diffraction System (Model PW1830) using Cu Ka-source operated at 40 keV and at a scan rate of 0.0258 s 1. Transmission electron microscope (TEM) analysis was carried out using JEOL 2010 high-resolution TEM system with LaB6 filament at 200 keV. To obtain the electron micrographs, the catalyst samples were ultrasonically dispersed in methanol. A drop of the dispersion was deposited onto a carbon film with 400 mesh Cu grid and subsequently dried. X-ray photoelectron spectroscopic (XPS) analysis of the catalysts was carried out with PHI 5600 (Physical Electronics) multi-technique system using Al monochromatic X-ray at a power of 350 W. Survey and regional spectra were obtained with pass energy of 187.85 and 23.50 eV, respectively.
2.3. Membrane electrode assembly fabrication and electrochemical characterization The catalytic activity of the SFT Pt/C for O2 reduction and Pt,Ru/C for H2/CO oxidation were evaluated and compared with commercial E-TEK catalysts in a 5 cm2 single fuel cell testing station using an Autolab potentiostat. Membrane electrode assemblies (MEAs) were prepared following the procedure reported previously [10]. To evaluate the activity of Pt/C for O2 reduction, the SFT 20 wt.% Pt/C (prepared at 90 8C) was used for cathode fabrication at a loading of 0.2 mg cm 2. A Nafion 117 membrane and an E-TEK 20 wt.% Pt/C electrode (0.35 mg Pt cm 2 with 1.2 mg cm2 Nafion impregnation) were used as the electrolyte and anode, respectively. For comparison, another MEA was made with a cathode containing the E-TEK 20 wt.% Pt/C instead of the SFT catalyst. Electrochemical measurements including the polarization behavior and impedance spectroscopy were then carried out at cell temperature of 60 8C and humidifier temperature of 75 8C. Flowrates of H2 and O2 were fixed at 90 ml min 1, well above the stoichiometry. To evaluate the activity of SFT Pt,Ru/C for CO tolerance, two more MEAs were prepared. Identical ETEK 20 wt.% Pt/C electrodes (0.35 mg Pt cm 2 with 1.2 mg cm 2 Nafion impregnation) were employed as the cathodes of these two MEAs. 20 wt.% SFT Pt,Ru/C (1:1) and E-TEK Pt,Ru/C (1:1) were used for anodes of these two MEAs, respectively. The cell polarization testing, impedance measurement and CO stripping voltammetry were carried out in the same 5 cm2 single fuel cell configuration. For the polarization measurement, 100 ppm CO/H2 mixture and O2 were fed to the anode and cathode, respectively. For the impedance measurement and CO stripping voltammetry, 2% CO/ H2 mixture was fed to the anode, and H2 was fed to the cathode, serving as a counter and reference electrode. During the CO stripping voltammetry, CO was first adsorbed onto the Pt,Ru/C catalyst by flowing 2% CO/ H2 mixture at 90 ml min1 through the anode for 30 min, while holding the working electrode potential at 0.05 V. The gas was then switched to N2 to remove residual H2 and nonadsorbed CO from the gas phase with the potential fixed at 0.05 V for 30 min. The CO stripping voltammogram was recorded by scanning potential from 0.05 to 0.98 V at 20 mV s 1.
3. Result and discussion 3.1. Physicochemical characterization The chemical, morphological and crystallographic properties of the prepared catalysts are examined using
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
physicochemical techniques including XRD, TEM and XPS. The detailed discussions are given below. 3.1.1. XRD Curve (a) in Fig. 1 shows the XRD patterns of the SFT 20 wt.% Pt/C catalyst prepared at 90 8C. The characteristic diffraction peaks of the fcc Pt demonstrate that a successful reduction of Pt precursor to metallic form has been achieved. Importantly, the lack of sharp peak rules out the presence of a significant mass */ fraction of much larger crystallites. The high intensity appearing at the 2u of 258 is associated with the use of Vulcan XC-72 carbon black as support. From the extent of the line broadening of (111) at 2u of /398, the average crystallite size is estimated to be 2.5 nm with the use of Scherrer equation after background subtraction [11]. XRD results of the SFT Pt,Ru/C samples (1:1 and 7:3) were shown in curve (b) and (c) of Fig. 1. The lack of peak associated with Ru and its oxide could indicate the alloying of Pt and Ru. However, it is possible that some Ru could also be present in amorphous form [12]. It can be observed that the diffraction peaks slightly shift to higher Bragg angles in the order of Pt/C, Pt,Ru/ C (7:3) and Pt,Ru/C (1:1), which indicates the decrease of the lattice constant with the increase of Ru concentration. Such evidence accounts for the presence of a Pt /Ru alloy in the catalyst, where the platinum atoms on the lattice points of the face centered cubic lattice are replaced by the smaller ruthenium atoms [1]. Furthermore, the higher alloying content for Pt,Ru/C (1:1) compared with Pt,Ru/C (7:3) can be identified. Theoretically, from the contraction of lattice constant and Vegard’s law, the Pt/Ru alloy composition can be determined. However, the estimated experimental error makes the quantitative assessment unreliable [11]. From the line broadening of peak (111), the average particle
Fig. 1. XRD patterns of the SFT electrocatalysts. (a) 20 wt.%; (b) Pt,Ru/C (7:3); (c) Pt,Ru/C (1:1).
2983
sizes for Pt,Ru/C (7:3) and Pt,Ru/C (1:1) are estimated to be around 2.5 and 2.6 nm, respectively, which shows negligible increase compared with Pt/C. 3.1.2. TEM Results obtained from the TEM of these samples (Fig. 2) show the morphology and dispersion of Pt (Fig. 2a) and Pt /Ru alloy (Fig. 2b) on the carbon support. As it can be seen, the particle size distribution is uniform with an average particle size of 2 /3 nm and no agglomerates of metal particles were observed. 3.1.3. XPS The nature of the surface species of the prepared Pt/C and Pt,Ru/C sample were investigated by XPS analysis. Both survey spectra from 0 to 1400 eV binding energy and regional spectra were recorded. For 20% SFT Pt/C, the survey spectra confirmed the near surface presence of Pt, C, O and S. The presence of S impurities indicates that some surfactants remain on the surface of catalyst after filtration and washing. From the deconvolution of the regional spectra of Pt4f (Fig. 3), two pairs of Pt peaks can be found. The most intensive peaks (71.3 eV and 74.6 V) are attributed to metallic Pt at a ratio of 68%. The second pair of peaks observed at binding energy 1 eV higher with a ratio of 32% can be assigned to the Pt 2/chemical state in PtOads. Such existence of PtOads was also reported by other researchers [1,13,14], indicating that the surface sites of fine metallic platinum are often covered with chemisorbed oxygen. The surface composition determined by XPS shows a Pt loading of 28.5%, which is higher than the nominal one. This can be attributed to the nature of XPS technique, which is a surface analyzing technique with sampling depth of 5/8 nm. The XPS analysis for two SFT Pt,Ru/C (1:1 and 7:3) and E-TEK Pt,Ru/C shows that the real atomic ratio between Pt and Ru is in agreement with the nominal composition (Table 1). Since the XPS spectra are similar for all the samples, only those of Pt,Ru/C (1:1) are presented in Fig. 4. It is shown from Fig. 4a that the Ru3d peak is overlapped by the large C 1s signal, which prevents the accurate determination of the Ru species in this spectrum. To determine the chemical state of Ru, Ru3p spectrum was investigated and it shows that the 3p3/2 signal is derived from the contributions of two components with binding energies of 462.7 and 466.2 eV at a concentration of 82 and 18%, respectively. Compared with the results obtained for E-TEK 20% Pt,Ru/C catalyst (463.2 and 466.7 eV), the binding energies are 0.5 eV lower. Arico et al. have done a similar analysis on the commercial Pt,Ru/C (Electrochem. Inc) and found that the 3p3/2 peak is derived from the contributions of two components with binding energies of 463.37 and 466.9 eV, respectively [13]. Based on the observation by Goodenough et al. [15] that the close association of Ru
2984
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
Fig. 3. X-ray photoelectron spectra (Pt4f) of the SFT 20 wt.% Pt/C.
Table 1 XPS analysis summary of atomic concentration in Pt,Ru/C electrocatalysts Catalysts
C1s
O1s
S2p
Ru3p
Pt4f
E-TEK (1:1) SFT (1:1) SFT (7:3)
83.3 80.3 80.6
13.2 13.7 13.8
/ 1.8 1.5
1.7 2.2 1.2
1.8 2.0 2.9
components to metallic Ru in the alloy and Ru(IV) species. Since the binding energies obtained in this work for E-TEK catalyst is similar, it is reasonable to conclude that Ru in E-TEK catalyst also exists in the same chemical state. Furthermore, for the SFT catalyst, the lower binding energies indicate the presence of a lower-valence state for the surface Ru species. Therefore, more Ru exists in the form of alloy with Pt, which allows an easier chemisorption of labile oxygen species, and thus enhances CO oxidation kinetics [14].
Fig. 2. Transmission electron micrograph of the SFT electrocatalysts. (a) 20 wt.% Pt/C; (b) 20 wt.% Pt,Ru/C (1:1).
with Pt in carbon supported Pt/Ru catalyst produced a significant positive shift of the Ru3p signal compared with a Ru/C catalyst (ca. 1 eV), he attributed the two
3.1.4. Temperature effect on particle size The temperature effect on the catalyst particle size is shown in Fig. 5. XRD technique was used for the determination of average particle size. It can be observed that the size of particle can be controlled by manipulating the synthesis temperature and with the increase of temperature, a smaller size can be achieved. This effect can be explained by the theory of crystallization. When Pt precursor is reduced to Pt metal, the solution becomes supersaturated with the Pt metal. The formation of the colloids is then a process of crystallization. The rate of nucleation determines the number and size of the primary particles generated in a supersaturated solution. [16,17]. The higher the reaction rate (i.e. higher temperature), the higher the supersaturation, consequently the number of nucleus is larger and the size can be smaller. If temperature is too low (e.g. 50 8C), the reaction rate can be so slow that even after several
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
2985
In our process, the loading shows little effect on the particle size as the colloid is pre-formed in the solution and the loading is determined by the amount of carbon added. To prove this, an experiment was carried out to prepare Pt/C at a loading of 40 wt.%. The XRD result shows that the particle size is not affected by the loading and is at the average of 2.5 nm. 3.2. Electrochemical characterization
Fig. 4. X-ray photoelectron spectra of the SFT 20 wt.% Pt,Ru/C (1:1). (a) C1s; (b) Ru3p.
The electrocatalytic activity of the SFT 20 wt.% Pt/C electrocatalyst and E-TEK catalyst were measured and their results are presented in Fig. 6. Both catalysts show similar catalytic activities. Considering the same catalyst loading and the similar Pt particle size, the SFT Pt/C exhibits comparable catalytic activity with respect to the commercial E-TEK catalyst. Moreover, no noticeable activity difference was observed between the newly prepared catalyst and the catalyst after the removal of surfactant. Therefore, the existence of surfactant will have negligible effect on the activity of Pt. A similar conclusion was also obtained by Bonnemann et al. [9] who measured the Pt surface accessibility by CO chemisorption and Pt catalytic activity by hydrogenation experiment. They found that although the surface accessibility by CO is altered by the existence of surfactant, the catalytic activity is still high enough as small molecules like H2 and O2 can penetrate through surfactant layer and react at the catalytic sites. The impedance plots in Fig. 7 show that similar high frequency resistance was obtained for both MEAs prepared with the E-TEK catalyst and SFT catalyst, respectively. This indicates that the surfactant will not introduce additional resistance to the MEA. The H2/CO oxidation polarization curve test (Fig. 8) shows the overall activity of two catalysts, SFT 20% Pt,Ru/C (1:1) and E-TEK 20% Pt,Ru/C (1:1). At low current density, a better performance is obtained for
Fig. 5. Pt crystalline sizes at different synthesis temperatures.
hours no visible change of solution color, i.e. reaction, can be observed. 3.1.5. Effect of catalyst loading Conventionally the growth of particle size is inevitable when preparing high loading (e.g. 40%) electrocatalyst.
Fig. 6. Cell polarization curves for SFT 20 wt.% Pt/C and E-TEK 20 wt.% Pt/C electrode. Operating conditions: Tcell 60 8C, atmospheric pressure, O2 and H2 flow rate at 90 ml min 1.
2986
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
Fig. 7. Experimental impedance plots for MEAs prepared with SFT 20 wt.% Pt/C and E-TEK 20 wt.% Pt/C. Operating conditions: working potential 0.7 V, Tcell 60 8C, atmospheric pressure, O2 and H2 flow rate at 90 ml min 1. Z?, real part of the impedance Z; Zƒ, imaginary part of the impedance Z.
Fig. 8. Cell polarization curves for MEAs with SFT 20 wt.% Pt,Ru/C and E-TEK 20 wt.% Pt,Ru/C electrodes. Operating conditions: Tcell 60 8C, atmospheric pressure, 100 ppm CO/H2 mixture and O2 flow rate at 90 ml min 1.
SFT catalyst, while a similar performance is observed at high current density region. The enhanced CO oxidation can also be observed from the impedance measurement (Fig. 9), where the onset potential with inductive behavior shifts negatively 0.05 V for the SFT Pt,Ru/C. Previously we have investigated the potential dependence of the impedance pattern for 2% CO/H2 and reported the appearance of this inductive pattern can be used as a criterion for the onset of CO oxidation by the coincidence of the potential with inductive behavior with the ignition potential for continuous CO oxidation [18]. Therefore, this negative shift of potential indicates that the SFT catalyst has a higher catalytic activity than the E-TEK catalyst.
Fig. 9. Impedance plots for MEAs with SFT 20 wt.% Pt,Ru/C and ETEK 20 wt.% Pt,Ru/C electrodes. Operating conditions: Tcell 60 8C, atmospheric pressure, 2% CO/H2 mixture and H2 flow rate at 90 ml min 1.
The CO stripping voltammgrams for the E-TEK Pt/C, SFT and E-TEK Pt,Ru/C are presented in Fig. 10. It is clear that both bimetallic catalysts enhance the oxidation of the pre-adsorbed CO compared with the Pt/C, evidenced by the cathodic shift in the onset of oxidation. Furthermore, the onset potential of CO oxidation for the SFT Pt,Ru/C is more negative than the E-TEK Pt,Ru/C, in agreement with the result from the impedance measurement. However, there exists difference between the potential value with inductive behavior and the onset potential for CO stripping. This discrepancy arises from the condition of CO supply. For the impedance measurement, CO will continuously adsorb on reaction sites and compete with water for the same sites after the originally adsorbed CO is oxidized. Similar phenomenon has also been observed on smooth Pt /Ru alloy surface by other researchers [19]. Therefore, the potential difference between inductive behavior
Fig. 10. CO stripping voltammetry for three different MEAs. Operating conditions: Tcell 60 8C, atmospheric pressure, all gas flow rates at 90 ml min1.
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
and CO stripping is a consequence of the competition between CO and water for the same sites. It can be further observed from Fig. 10 that the total desorption charge of the SFT Pt,Ru/C is smaller than that of the E-TEK Pt/C. Considering the surface accessibility of CO as described by Bonnemann et al. [9], this smaller total charge of SFT Pt,Ru/C might be caused by the blocking effect of the surfactant. Due to the higher alloy component in the SFT catalyst, the more negative onset potential and a slightly higher polarization behavior can be explained by a lower potential needed for electro-oxidation of weakly adsorbed CO on Pt, Ru alloy surface, where the energy barrier for surface CO desorption process is reduced as a result of lowering the electronic density state of Pt, as suggested by Lin and coworkers [20]. The presence of weakly adsorbed CO on Pt alloy surface has been discussed in detail by Wang et al. [19] and Markovic et al. [21].
4. Conclusion A surfactant-based alcohol reduction method is delineated for the synthesis of polymer electrolyte fuel cell catalysts. Particle sizes of Pt/C and Pt,Ru/C electrocatalysts prepared by this method can be controlled by synthesis temperature and they are in the range of 2 /3.5 nm. Electrochemical characterizations illustrate that the surfactant based Pt/C and Pt,Ru/C exhibit high catalytic activities in a fuel cell environment.
Acknowledgements The authors gratefully acknowledge the Innovation and Technology Commission of the Hong Kong SAR Government (UIM/15) for the funding support.
2987
References [1] M. Watanabe, M. Uchida, S. Motoo, J. Electroanal. Chem. 229 (1987) 395. [2] H.G. Petrow, R.J. Allen, in US Patent. 1977. p. No. 4044193. [3] A.M.C. Luna, G.A. Camara, V.A. Paganin, E.A. Ticianelli, E.R. Gonzalez, Electrochem. Commun. 2 (2000) 222. [4] U.A. Paulus, U. Endruschat, G.J. Feldmeyer, T.J. Schmidt, H. Bonnemann, R.J. Behm, J. Catalysis 195 (2000) 383. [5] N. Toshima, M. Kuriyama, Y. Yamada, H. Hirai, Chem. Lett. (1981) 793. [6] N. Toshima, K. Hirakawa, Polym. J. 31 (1999) 1127. [7] D. Duff, T. Mallat, M. Schneider, A. Baiker, Appl. Catalysis A: General 133 (1995) 133. [8] M. Kerker, J. Coll. Interf. Sci. 112 (1986) 302. [9] H. Bonnemann, G. Braun, W. Brijoux, R. Brinkmann, A. Tilling, K. Seevogel, K. Siepen, J. Organomet. Chem. 520 (1996) 143. [10] X. Wang, I.M. Hsing, P.L. Yue, J. Power Sources 96 (2001) 282. [11] V. Radmilovic, H.A. Gasteiger, J.P.N. Ross, J. Catalysis 154 (1995) 98. [12] C.Z. He, H.E. Kunz, J.M. Fenton, in: The Electrochemical Society Proceedings, vol. 13, 1997, p. 52. [13] A.S. Arico, P. Creti, H. Kim, R. Mantegna, N. Giordano, V. Antonucci, J. Electrochem. Soc. 143 (1996) 3950. [14] A.S. Arico, P. Creti, E. Modica, G. Monforte, V. Baglio, V. Antonucci, Electrochim. Acta 45 (2000) 4319. [15] J.B. Goodenough, R. Manoharan, A.K. Shukla, K.V. Ramesh, Chem. Mater. 1 (1989) 391. [16] A.D. Randolph, M.A. Larson, Theory of Particulate Processes, Academic Press, 1988, p. 35. [17] A. Mersmann, Crystallization Technology Handbook, Marcel Dekker, 1995, p. 150. [18] X. Wang, Y.J. Leng, I.M. Hsing, P.L. Yue, Electrochim. Acta 46 (2001) 4397. [19] K. Wang, H.A. Gasteiger, N.M. Markovic, P.N. Ross, Electrochim. Acta 41 (1996) 2587. [20] S.D. Lin, T.C. Hsiao, J.R. Chang, A.S. Lin, J. Phys. Chem. B 103 (1999) 97. [21] N.M. Markovic, H.A. Gasteiger, C.A. Lucas, P.N. Ross, J. Phys. Chem. B 103 (1999) 487.
Surfactant stabilized Pt and Pt alloy electrocatalyst for polymer electrolyte fuel cells Xin Wang, I-Ming Hsing * Department of Chemical Engineering, Hong Kong University of Science and Technology, Room 4551 Clear Water Bay, Kowloon, Hong Kong Received 17 December 2001; received in revised form 20 March 2002
Abstract A simple process for the synthesis of carbon supported Pt and Pt/Ru electrocatalysts was investigated. Borrowing from the homogeneous catalyst preparation, this process uses a surfactant as a stabilizer which prevents the metal colloids from aggregation during the reduction process without influencing the deposition of the colloids onto the carbon support. Chemical, morphological and crystallographic properties of the newly prepared electrocatalysts were characterized using various surface techniques including X-ray diffraction (XRD), Transmission electron microscopy (TEM) and X-ray photoelectron spectroscopy (XPS). These techniques show that mono-size, well-dispersed metal colloids can be formed and successfully supported on the carbon black. Moreover, the size of metallic colloids prepared by this method can be manipulated by controlling the synthesis temperature and is independent of the catalyst loading. Electrochemical characterizations show that in comparison with commercial E-TEK electrocatalysts, surfactant-based Pt/C electrocatalysts possess similar catalytic activity in terms of oxygen reduction and higher CO tolerance performance can be obtained by the surfactant stabilized Pt,Ru/C. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Surfactant; Stabilizer; Pt; Electrocatalyst; Fuel cell
1. Introduction One of the obstacles preventing polymer electrolyte fuel cells from commercialization is the high cost of noble metals to be used as catalyst, such as platinum and ruthenium. To effectively use these metals, they have to be well dispersed to small particles on conductive carbon supports. Conventional methods, such as impregnation, co-precipitation and ion exchange, are not adequate from the viewpoint of offering high metal surface area or good dispersion of metal particles [1]. To avoid the above disadvantages, an alternative method has been investigated, involving the use of prefabricated metal colloids and subsequent deposition on the support material. Different fabrication routes are proposed based on this methodology, among which, the most common one is the sulfite-complex route [1 /3]. Recently, Bonnemann and coworkers introduced a new
* Corresponding author. Tel.: /852-2358-7131; fax: /852-23580054 E-mail address: [email protected] (I.-M. Hsing).
route using an organometallic compound to synthesize colloidal precursors for the carbon-supported catalyst, where the organic molecules are used to prevent the agglomeration and coalescence of the particles [4]. However, these methods are complex and require a delicate control of operating conditions. In this study, we report a simple procedure for the preparation of Pt and Pt,Ru/C catalysts based on an alcohol reduction method. This alcohol reduction method [5 /7] has been widely used in the preparation of metal colloids for homogeneous catalysis, where mono-size, well dispersed metal colloids can be formed and stabilized in aqueous solution with the existence of polymer via steric mechanism [8]. Here, its application is extended to the preparation of heterogeneous catalyst. Preliminary experiment shows that the formed metal colloids cannot be immobilized onto the carbon support since the steric effect of polymer is so strong that the carbon black is also stabilized by polymer. Considering the size difference between the metal colloid (2 /3 nm) and the carbon black (30 nm), a surfactant Dodecyldimethyl (3-sulfo-propyl) ammonium hydroxide (SB12) is carefully chosen as the stabilizer where the stabilization
0013-4686/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 3 - 4 6 8 6 ( 0 2 ) 0 0 1 9 9 - 8
2982
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
effect is only sufficient to prevent metal colloids from aggregation and will not influence the immobilization of colloids onto the carbon black. Different from the polymer-based stabilizer, the stabilization mechanism of SB12 surfactant is possibly related to the electrostatic effect [9] with the hydrophobic end of the surfactant oriented towards the particle surface and the charged hydrophilic part pointing towards the solution. Physicochemical and electrochemical characterizations of the prepared catalysts are carried out to determine the chemical, morphological, crystallographic, as well as electrochemical properties of the catalysts.
2. Experimental 2.1. Catalyst preparation The surfactant-protected Pt/C and Pt,Ru/C catalysts (denoted as SFT) were prepared by a modified alcoholreduction method. Surfactant SB12, chloroplatinic acid H2PtCl6, Vulcan XC-72 carbon black and RuCl3 (used only for Pt,Ru/C case) were mixed in a methanol /water (volumetric ratio 1:3) solvent mixture to form a suspension. The resulting suspension was stirred and refluxed under air at elevated temperature for 1 h and then filtered and washed. The filtrate was analyzed with inductively coupled plasma atomic emission spectroscopy (ICP-AES), which indicates that most Pt and Ru have been adsorbed on the carbon black. For Pt/C catalyst, two different loadings (20 and 40 wt.%) were prepared by varying the amount of carbon black. Experiments were also carried out at different temperatures to determine their effects on the Pt colloids size. As to Pt,Ru/C catalyst, two batches with different atomic ratios (1:1 and 7:3) were prepared. 2.2. Physicochemical characterization Powder X-ray diffraction (XRD) patterns for the catalysts were obtained on a Philips Powder Diffraction System (Model PW1830) using Cu Ka-source operated at 40 keV and at a scan rate of 0.0258 s 1. Transmission electron microscope (TEM) analysis was carried out using JEOL 2010 high-resolution TEM system with LaB6 filament at 200 keV. To obtain the electron micrographs, the catalyst samples were ultrasonically dispersed in methanol. A drop of the dispersion was deposited onto a carbon film with 400 mesh Cu grid and subsequently dried. X-ray photoelectron spectroscopic (XPS) analysis of the catalysts was carried out with PHI 5600 (Physical Electronics) multi-technique system using Al monochromatic X-ray at a power of 350 W. Survey and regional spectra were obtained with pass energy of 187.85 and 23.50 eV, respectively.
2.3. Membrane electrode assembly fabrication and electrochemical characterization The catalytic activity of the SFT Pt/C for O2 reduction and Pt,Ru/C for H2/CO oxidation were evaluated and compared with commercial E-TEK catalysts in a 5 cm2 single fuel cell testing station using an Autolab potentiostat. Membrane electrode assemblies (MEAs) were prepared following the procedure reported previously [10]. To evaluate the activity of Pt/C for O2 reduction, the SFT 20 wt.% Pt/C (prepared at 90 8C) was used for cathode fabrication at a loading of 0.2 mg cm 2. A Nafion 117 membrane and an E-TEK 20 wt.% Pt/C electrode (0.35 mg Pt cm 2 with 1.2 mg cm2 Nafion impregnation) were used as the electrolyte and anode, respectively. For comparison, another MEA was made with a cathode containing the E-TEK 20 wt.% Pt/C instead of the SFT catalyst. Electrochemical measurements including the polarization behavior and impedance spectroscopy were then carried out at cell temperature of 60 8C and humidifier temperature of 75 8C. Flowrates of H2 and O2 were fixed at 90 ml min 1, well above the stoichiometry. To evaluate the activity of SFT Pt,Ru/C for CO tolerance, two more MEAs were prepared. Identical ETEK 20 wt.% Pt/C electrodes (0.35 mg Pt cm 2 with 1.2 mg cm 2 Nafion impregnation) were employed as the cathodes of these two MEAs. 20 wt.% SFT Pt,Ru/C (1:1) and E-TEK Pt,Ru/C (1:1) were used for anodes of these two MEAs, respectively. The cell polarization testing, impedance measurement and CO stripping voltammetry were carried out in the same 5 cm2 single fuel cell configuration. For the polarization measurement, 100 ppm CO/H2 mixture and O2 were fed to the anode and cathode, respectively. For the impedance measurement and CO stripping voltammetry, 2% CO/ H2 mixture was fed to the anode, and H2 was fed to the cathode, serving as a counter and reference electrode. During the CO stripping voltammetry, CO was first adsorbed onto the Pt,Ru/C catalyst by flowing 2% CO/ H2 mixture at 90 ml min1 through the anode for 30 min, while holding the working electrode potential at 0.05 V. The gas was then switched to N2 to remove residual H2 and nonadsorbed CO from the gas phase with the potential fixed at 0.05 V for 30 min. The CO stripping voltammogram was recorded by scanning potential from 0.05 to 0.98 V at 20 mV s 1.
3. Result and discussion 3.1. Physicochemical characterization The chemical, morphological and crystallographic properties of the prepared catalysts are examined using
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
physicochemical techniques including XRD, TEM and XPS. The detailed discussions are given below. 3.1.1. XRD Curve (a) in Fig. 1 shows the XRD patterns of the SFT 20 wt.% Pt/C catalyst prepared at 90 8C. The characteristic diffraction peaks of the fcc Pt demonstrate that a successful reduction of Pt precursor to metallic form has been achieved. Importantly, the lack of sharp peak rules out the presence of a significant mass */ fraction of much larger crystallites. The high intensity appearing at the 2u of 258 is associated with the use of Vulcan XC-72 carbon black as support. From the extent of the line broadening of (111) at 2u of /398, the average crystallite size is estimated to be 2.5 nm with the use of Scherrer equation after background subtraction [11]. XRD results of the SFT Pt,Ru/C samples (1:1 and 7:3) were shown in curve (b) and (c) of Fig. 1. The lack of peak associated with Ru and its oxide could indicate the alloying of Pt and Ru. However, it is possible that some Ru could also be present in amorphous form [12]. It can be observed that the diffraction peaks slightly shift to higher Bragg angles in the order of Pt/C, Pt,Ru/ C (7:3) and Pt,Ru/C (1:1), which indicates the decrease of the lattice constant with the increase of Ru concentration. Such evidence accounts for the presence of a Pt /Ru alloy in the catalyst, where the platinum atoms on the lattice points of the face centered cubic lattice are replaced by the smaller ruthenium atoms [1]. Furthermore, the higher alloying content for Pt,Ru/C (1:1) compared with Pt,Ru/C (7:3) can be identified. Theoretically, from the contraction of lattice constant and Vegard’s law, the Pt/Ru alloy composition can be determined. However, the estimated experimental error makes the quantitative assessment unreliable [11]. From the line broadening of peak (111), the average particle
Fig. 1. XRD patterns of the SFT electrocatalysts. (a) 20 wt.%; (b) Pt,Ru/C (7:3); (c) Pt,Ru/C (1:1).
2983
sizes for Pt,Ru/C (7:3) and Pt,Ru/C (1:1) are estimated to be around 2.5 and 2.6 nm, respectively, which shows negligible increase compared with Pt/C. 3.1.2. TEM Results obtained from the TEM of these samples (Fig. 2) show the morphology and dispersion of Pt (Fig. 2a) and Pt /Ru alloy (Fig. 2b) on the carbon support. As it can be seen, the particle size distribution is uniform with an average particle size of 2 /3 nm and no agglomerates of metal particles were observed. 3.1.3. XPS The nature of the surface species of the prepared Pt/C and Pt,Ru/C sample were investigated by XPS analysis. Both survey spectra from 0 to 1400 eV binding energy and regional spectra were recorded. For 20% SFT Pt/C, the survey spectra confirmed the near surface presence of Pt, C, O and S. The presence of S impurities indicates that some surfactants remain on the surface of catalyst after filtration and washing. From the deconvolution of the regional spectra of Pt4f (Fig. 3), two pairs of Pt peaks can be found. The most intensive peaks (71.3 eV and 74.6 V) are attributed to metallic Pt at a ratio of 68%. The second pair of peaks observed at binding energy 1 eV higher with a ratio of 32% can be assigned to the Pt 2/chemical state in PtOads. Such existence of PtOads was also reported by other researchers [1,13,14], indicating that the surface sites of fine metallic platinum are often covered with chemisorbed oxygen. The surface composition determined by XPS shows a Pt loading of 28.5%, which is higher than the nominal one. This can be attributed to the nature of XPS technique, which is a surface analyzing technique with sampling depth of 5/8 nm. The XPS analysis for two SFT Pt,Ru/C (1:1 and 7:3) and E-TEK Pt,Ru/C shows that the real atomic ratio between Pt and Ru is in agreement with the nominal composition (Table 1). Since the XPS spectra are similar for all the samples, only those of Pt,Ru/C (1:1) are presented in Fig. 4. It is shown from Fig. 4a that the Ru3d peak is overlapped by the large C 1s signal, which prevents the accurate determination of the Ru species in this spectrum. To determine the chemical state of Ru, Ru3p spectrum was investigated and it shows that the 3p3/2 signal is derived from the contributions of two components with binding energies of 462.7 and 466.2 eV at a concentration of 82 and 18%, respectively. Compared with the results obtained for E-TEK 20% Pt,Ru/C catalyst (463.2 and 466.7 eV), the binding energies are 0.5 eV lower. Arico et al. have done a similar analysis on the commercial Pt,Ru/C (Electrochem. Inc) and found that the 3p3/2 peak is derived from the contributions of two components with binding energies of 463.37 and 466.9 eV, respectively [13]. Based on the observation by Goodenough et al. [15] that the close association of Ru
2984
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
Fig. 3. X-ray photoelectron spectra (Pt4f) of the SFT 20 wt.% Pt/C.
Table 1 XPS analysis summary of atomic concentration in Pt,Ru/C electrocatalysts Catalysts
C1s
O1s
S2p
Ru3p
Pt4f
E-TEK (1:1) SFT (1:1) SFT (7:3)
83.3 80.3 80.6
13.2 13.7 13.8
/ 1.8 1.5
1.7 2.2 1.2
1.8 2.0 2.9
components to metallic Ru in the alloy and Ru(IV) species. Since the binding energies obtained in this work for E-TEK catalyst is similar, it is reasonable to conclude that Ru in E-TEK catalyst also exists in the same chemical state. Furthermore, for the SFT catalyst, the lower binding energies indicate the presence of a lower-valence state for the surface Ru species. Therefore, more Ru exists in the form of alloy with Pt, which allows an easier chemisorption of labile oxygen species, and thus enhances CO oxidation kinetics [14].
Fig. 2. Transmission electron micrograph of the SFT electrocatalysts. (a) 20 wt.% Pt/C; (b) 20 wt.% Pt,Ru/C (1:1).
with Pt in carbon supported Pt/Ru catalyst produced a significant positive shift of the Ru3p signal compared with a Ru/C catalyst (ca. 1 eV), he attributed the two
3.1.4. Temperature effect on particle size The temperature effect on the catalyst particle size is shown in Fig. 5. XRD technique was used for the determination of average particle size. It can be observed that the size of particle can be controlled by manipulating the synthesis temperature and with the increase of temperature, a smaller size can be achieved. This effect can be explained by the theory of crystallization. When Pt precursor is reduced to Pt metal, the solution becomes supersaturated with the Pt metal. The formation of the colloids is then a process of crystallization. The rate of nucleation determines the number and size of the primary particles generated in a supersaturated solution. [16,17]. The higher the reaction rate (i.e. higher temperature), the higher the supersaturation, consequently the number of nucleus is larger and the size can be smaller. If temperature is too low (e.g. 50 8C), the reaction rate can be so slow that even after several
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
2985
In our process, the loading shows little effect on the particle size as the colloid is pre-formed in the solution and the loading is determined by the amount of carbon added. To prove this, an experiment was carried out to prepare Pt/C at a loading of 40 wt.%. The XRD result shows that the particle size is not affected by the loading and is at the average of 2.5 nm. 3.2. Electrochemical characterization
Fig. 4. X-ray photoelectron spectra of the SFT 20 wt.% Pt,Ru/C (1:1). (a) C1s; (b) Ru3p.
The electrocatalytic activity of the SFT 20 wt.% Pt/C electrocatalyst and E-TEK catalyst were measured and their results are presented in Fig. 6. Both catalysts show similar catalytic activities. Considering the same catalyst loading and the similar Pt particle size, the SFT Pt/C exhibits comparable catalytic activity with respect to the commercial E-TEK catalyst. Moreover, no noticeable activity difference was observed between the newly prepared catalyst and the catalyst after the removal of surfactant. Therefore, the existence of surfactant will have negligible effect on the activity of Pt. A similar conclusion was also obtained by Bonnemann et al. [9] who measured the Pt surface accessibility by CO chemisorption and Pt catalytic activity by hydrogenation experiment. They found that although the surface accessibility by CO is altered by the existence of surfactant, the catalytic activity is still high enough as small molecules like H2 and O2 can penetrate through surfactant layer and react at the catalytic sites. The impedance plots in Fig. 7 show that similar high frequency resistance was obtained for both MEAs prepared with the E-TEK catalyst and SFT catalyst, respectively. This indicates that the surfactant will not introduce additional resistance to the MEA. The H2/CO oxidation polarization curve test (Fig. 8) shows the overall activity of two catalysts, SFT 20% Pt,Ru/C (1:1) and E-TEK 20% Pt,Ru/C (1:1). At low current density, a better performance is obtained for
Fig. 5. Pt crystalline sizes at different synthesis temperatures.
hours no visible change of solution color, i.e. reaction, can be observed. 3.1.5. Effect of catalyst loading Conventionally the growth of particle size is inevitable when preparing high loading (e.g. 40%) electrocatalyst.
Fig. 6. Cell polarization curves for SFT 20 wt.% Pt/C and E-TEK 20 wt.% Pt/C electrode. Operating conditions: Tcell 60 8C, atmospheric pressure, O2 and H2 flow rate at 90 ml min 1.
2986
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
Fig. 7. Experimental impedance plots for MEAs prepared with SFT 20 wt.% Pt/C and E-TEK 20 wt.% Pt/C. Operating conditions: working potential 0.7 V, Tcell 60 8C, atmospheric pressure, O2 and H2 flow rate at 90 ml min 1. Z?, real part of the impedance Z; Zƒ, imaginary part of the impedance Z.
Fig. 8. Cell polarization curves for MEAs with SFT 20 wt.% Pt,Ru/C and E-TEK 20 wt.% Pt,Ru/C electrodes. Operating conditions: Tcell 60 8C, atmospheric pressure, 100 ppm CO/H2 mixture and O2 flow rate at 90 ml min 1.
SFT catalyst, while a similar performance is observed at high current density region. The enhanced CO oxidation can also be observed from the impedance measurement (Fig. 9), where the onset potential with inductive behavior shifts negatively 0.05 V for the SFT Pt,Ru/C. Previously we have investigated the potential dependence of the impedance pattern for 2% CO/H2 and reported the appearance of this inductive pattern can be used as a criterion for the onset of CO oxidation by the coincidence of the potential with inductive behavior with the ignition potential for continuous CO oxidation [18]. Therefore, this negative shift of potential indicates that the SFT catalyst has a higher catalytic activity than the E-TEK catalyst.
Fig. 9. Impedance plots for MEAs with SFT 20 wt.% Pt,Ru/C and ETEK 20 wt.% Pt,Ru/C electrodes. Operating conditions: Tcell 60 8C, atmospheric pressure, 2% CO/H2 mixture and H2 flow rate at 90 ml min 1.
The CO stripping voltammgrams for the E-TEK Pt/C, SFT and E-TEK Pt,Ru/C are presented in Fig. 10. It is clear that both bimetallic catalysts enhance the oxidation of the pre-adsorbed CO compared with the Pt/C, evidenced by the cathodic shift in the onset of oxidation. Furthermore, the onset potential of CO oxidation for the SFT Pt,Ru/C is more negative than the E-TEK Pt,Ru/C, in agreement with the result from the impedance measurement. However, there exists difference between the potential value with inductive behavior and the onset potential for CO stripping. This discrepancy arises from the condition of CO supply. For the impedance measurement, CO will continuously adsorb on reaction sites and compete with water for the same sites after the originally adsorbed CO is oxidized. Similar phenomenon has also been observed on smooth Pt /Ru alloy surface by other researchers [19]. Therefore, the potential difference between inductive behavior
Fig. 10. CO stripping voltammetry for three different MEAs. Operating conditions: Tcell 60 8C, atmospheric pressure, all gas flow rates at 90 ml min1.
X. Wang, I.-M. Hsing / Electrochimica Acta 47 (2002) 2981 /2987
and CO stripping is a consequence of the competition between CO and water for the same sites. It can be further observed from Fig. 10 that the total desorption charge of the SFT Pt,Ru/C is smaller than that of the E-TEK Pt/C. Considering the surface accessibility of CO as described by Bonnemann et al. [9], this smaller total charge of SFT Pt,Ru/C might be caused by the blocking effect of the surfactant. Due to the higher alloy component in the SFT catalyst, the more negative onset potential and a slightly higher polarization behavior can be explained by a lower potential needed for electro-oxidation of weakly adsorbed CO on Pt, Ru alloy surface, where the energy barrier for surface CO desorption process is reduced as a result of lowering the electronic density state of Pt, as suggested by Lin and coworkers [20]. The presence of weakly adsorbed CO on Pt alloy surface has been discussed in detail by Wang et al. [19] and Markovic et al. [21].
4. Conclusion A surfactant-based alcohol reduction method is delineated for the synthesis of polymer electrolyte fuel cell catalysts. Particle sizes of Pt/C and Pt,Ru/C electrocatalysts prepared by this method can be controlled by synthesis temperature and they are in the range of 2 /3.5 nm. Electrochemical characterizations illustrate that the surfactant based Pt/C and Pt,Ru/C exhibit high catalytic activities in a fuel cell environment.
Acknowledgements The authors gratefully acknowledge the Innovation and Technology Commission of the Hong Kong SAR Government (UIM/15) for the funding support.
2987
References [1] M. Watanabe, M. Uchida, S. Motoo, J. Electroanal. Chem. 229 (1987) 395. [2] H.G. Petrow, R.J. Allen, in US Patent. 1977. p. No. 4044193. [3] A.M.C. Luna, G.A. Camara, V.A. Paganin, E.A. Ticianelli, E.R. Gonzalez, Electrochem. Commun. 2 (2000) 222. [4] U.A. Paulus, U. Endruschat, G.J. Feldmeyer, T.J. Schmidt, H. Bonnemann, R.J. Behm, J. Catalysis 195 (2000) 383. [5] N. Toshima, M. Kuriyama, Y. Yamada, H. Hirai, Chem. Lett. (1981) 793. [6] N. Toshima, K. Hirakawa, Polym. J. 31 (1999) 1127. [7] D. Duff, T. Mallat, M. Schneider, A. Baiker, Appl. Catalysis A: General 133 (1995) 133. [8] M. Kerker, J. Coll. Interf. Sci. 112 (1986) 302. [9] H. Bonnemann, G. Braun, W. Brijoux, R. Brinkmann, A. Tilling, K. Seevogel, K. Siepen, J. Organomet. Chem. 520 (1996) 143. [10] X. Wang, I.M. Hsing, P.L. Yue, J. Power Sources 96 (2001) 282. [11] V. Radmilovic, H.A. Gasteiger, J.P.N. Ross, J. Catalysis 154 (1995) 98. [12] C.Z. He, H.E. Kunz, J.M. Fenton, in: The Electrochemical Society Proceedings, vol. 13, 1997, p. 52. [13] A.S. Arico, P. Creti, H. Kim, R. Mantegna, N. Giordano, V. Antonucci, J. Electrochem. Soc. 143 (1996) 3950. [14] A.S. Arico, P. Creti, E. Modica, G. Monforte, V. Baglio, V. Antonucci, Electrochim. Acta 45 (2000) 4319. [15] J.B. Goodenough, R. Manoharan, A.K. Shukla, K.V. Ramesh, Chem. Mater. 1 (1989) 391. [16] A.D. Randolph, M.A. Larson, Theory of Particulate Processes, Academic Press, 1988, p. 35. [17] A. Mersmann, Crystallization Technology Handbook, Marcel Dekker, 1995, p. 150. [18] X. Wang, Y.J. Leng, I.M. Hsing, P.L. Yue, Electrochim. Acta 46 (2001) 4397. [19] K. Wang, H.A. Gasteiger, N.M. Markovic, P.N. Ross, Electrochim. Acta 41 (1996) 2587. [20] S.D. Lin, T.C. Hsiao, J.R. Chang, A.S. Lin, J. Phys. Chem. B 103 (1999) 97. [21] N.M. Markovic, H.A. Gasteiger, C.A. Lucas, P.N. Ross, J. Phys. Chem. B 103 (1999) 487.