Electrooxidation of Carbon Monoxide and Methanol on Platinum

May 24, 2007 - Atomically Dispersed Platinum on Gold Nano-Octahedra with High Catalytic .... Platinum-Decorated Nanoporous Gold Leaf for Methanol ...
0 downloads 0 Views 361KB Size
Download PDF
Langmuir 2007, 23, 7365-7371

7365

Electrooxidation of Carbon Monoxide and Methanol on Platinum-Overlayer-Coated Gold Nanoparticles: Effects of Film Thickness Sachin Kumar and Shouzhong Zou* Department of Chemistry and Biochemistry, Miami UniVersity, Oxford, Ohio 45056 ReceiVed December 22, 2006. In Final Form: April 12, 2007 The electrooxidation of carbon monoxide and methanol on Pt-coated Au nanoparticles attached to 3-aminopropyl trimethoxysilane-modified indium tin oxide electrodes was examined as a function of Pt film thickness and Au particle coverage. For the electrodes with medium and high Au particle coverages, the CO stripping peak position shifts to more negative values with increasing Pt film thickness, from ca. 0.8 V (vs Ag/AgCl) at 1 ML to 0.45 V at 10 ML. Accompanying this peak potential shift is the sharpening of the peak width from more than 150 to 65 mV. For the electrode with low Au particle coverage, similar peak width narrowing was also observed, but the peak potential shift is much smaller, from 0.85 V at 1 ML of Pt to 0.65 V at 10 ML. These observations are compared with the CO oxidation on bulk Pt electrodes and on Pt films deposited on bulk Au electrodes. The film-thickness-dependent CO oxidation is explained by d band theory in terms of strain and ligand effects, the particle size effect, and the particle aggregation induced by Pt film growth. Corresponding to the increasing CO oxidation activity, the current density of methanol oxidation grows with the Pt film thickness. The peak potential and current density reach the same values as those obtained on a polycrystalline bulk Pt electrode when more than 4 ML of Pt is deposited on the Au particle electrodes with a particle coverage higher than 0.25. These results suggest that it is feasible to reduce Pt loading in methanol fuel cells by using Pt thin films as the anode catalyst.

Introduction One of the bottlenecks in methanol fuel cell development is the high loading of Pt catalysts due to the slow kinetics of oxygen reduction and methanol oxidation.1 To overcome this hurdle, different approaches have been explored, including the use of Pt-based alloys,2,3 Ru-decorated Pt,4 and non-Pt catalysts.5,6 A recently developed method uses Pt thin films deposited on a second metal for O2 reduction.7,8 Theoretically, depending on the chemical-physical properties of the substrate, the catalytic activity of the thin film may be drastically modified from that of the corresponding bulk material. This modification is explained by Norskov et al. as the shift of d band center energy (average d band energy), which determines the metal-adsorbate bonding strength.9,10 Experimentally, whereas the reactivity and adsorption properties of metal monolayers have been extensively explored in ultrahigh vacuum (UHV),11-14 there has been little work done * To whom correspondence should be addressed. E-mail: zous@ muohio.edu. Tel: 513-529-8084. Fax: 513-529-5715. (1) Dillon, R.; Srinivasan, S.; Arico, A. S.; Antonucci, V. J. Power Sources 2004, 127, 112-126. (2) Markovic, N. M.; Ross, P. N. Surf. Sci. Rep. 2002, 45, 121-229. (3) Adzic, R. In Electrocatalysis; Lipkowski, J., Ross, P. N., Eds.; WileyVCH: New York, 1998. (4) Maillard, F.; Lu, G. Q.; Wieckowski, A.; Stimming, U. J. Phys. Chem. B 2005, 109, 16230-16243. (5) Bashyam, R.; Zelenay, P. Nature 2006, 443, 63-66. (6) Fernandez, J. L.; Raghuveer, V.; Manthiram, A.; Bard, A. J. J. Am. Chem. Soc. 2005, 127, 13100-13101. (7) Zhang, J. L.; Vukmirovic, M. B.; Sasaki, K.; Nilekar, A. U.; Mavrikakis, M.; Adzic, R. R. J. Am. Chem. Soc. 2005, 127, 12480-12481. (8) Zhang, J. L.; Vukmirovic, M. B.; Xu, Y.; Mavrikakis, M.; Adzic, R. R. Angew. Chem., Int. Ed. 2005, 44, 2132-2135. (9) Hammer, B.; Norskov, J. K. AdVances in Catalysis; Academic Press: San Diego, CA, 2000; Vol. 45, pp 71-129. (10) Hammer, B.; Morikawa, Y.; Norskov, J. K. Phys. ReV. Lett. 1996, 76, 2141-2144. (11) Pedersen, M. O.; Helveg, S.; Ruban, A.; Stensgaard, I.; Laegsgaard, E.; Norskov, J. K.; Besenbacher, F. Surf. Sci. 1999, 426, 395-409. (12) Larsen, J. H.; Chorkendorff, I. Surf. Sci. 1998, 405, 62-73. (13) Rodriguez, J. A.; Goodman, D. W. Science 1992, 257, 897-903. (14) Hager, T.; Rauscher, H.; Behm, R. J. Surf. Sci. 2004, 558, 181-194.

on electrochemical systems until very recently, presumably because of the difficulty of preparing uniform metal monolayers in the latter environment. By utilizing an intuitive galvanic replacement method, Adzic and co-workers were able to prepare a single monolayer of Pt on various metal substrates and studied the oxygen reduction on the Pt monolayers.7,8,15-17 Their results showed that depending on the substrate the Pt monolayer can have higher reactivity than bulk Pt.8,15-17 For example, a monolayer of Pt deposited on a Pd(111) surface exhibits higher O2 reduction activity than on Pt(111) surface.8 The higher activity was explained in terms of a delicate balance of the activities for O-O bond breaking and O-H bond formation, which are determined by the d band center energy and can be fine tuned by depositing a Pt monolayer on different metal substrates.8 Using a different approach, Du and Tong deposited submonolayer Pt on polycrystalline Au and found that carbon monoxide (CO) oxidation occurs at a much higher potential than on a bulk Pt electrode.18 They attributed this observation to stronger CO and OH adsorption due to the tensile strain of the Pt layer that raises the d band center energy.18 These studies demonstrate that in the electrochemical environment the adsorption and reactivity of a metal monolayer can also be significantly different from those of the corresponding bulk material and the changes can be explained by d band theory as well. In this article, the electrooxidation of CO and methanol (MeOH) on Pt thin films deposited onto 3 nm gold particles was studied. We try to address two important questions regarding thin film reactivity. First, how does the reactivity of the Pt thin films toward the above two reactions change with film thickness, and at what thickness will the thin film behave like the corresponding (15) Zhang, J.; Mo, Y.; Vukmirovic, M. B.; Klie, R.; Sasaki, K.; Adzic, R. R. J. Phys. Chem. B 2004, 108, 10955-10964. (16) Zhang, J.; Vukmirovic, M. B.; Sasaki, K.; Uribe, F.; Adzic, R. R. J. Serb. Chem. Soc. 2005, 70, 513-525. (17) Zhang, J.; Lima, F. H. B.; Shao, M. H.; Sasaki, K.; Wang, J. X.; Hanson, J.; Adzic, R. R. J. Phys. Chem. B 2005, 109, 22701-22704. (18) Du, B. C.; Tong, Y. Y. J. Phys. Chem. B 2005, 109, 17775-17780.

10.1021/la0637216 CCC: $37.00 © 2007 American Chemical Society Published on Web 05/24/2007

7366 Langmuir, Vol. 23, No. 13, 2007

bulk material? Second, do the monolayers deposited on the nanoparticle surface have the same catalytic activity as those coated on the corresponding bulk electrode? There are only a few electrochemical studies of the adsorption and reaction on Pt thin films deposited on nanoparticles. Weaver et al. studied CO adsorption on Pt-coated 70 nm Au particles by surface-enhanced Raman spectroscopy.19 Dong and co-workers examined oxygen reduction on Pt monolayers deposited by galvanic replacement on Au nanoparticles.20 Brankovic et al. studied hydrogen oxidation on submonolayers of Pt spontaneously deposited on Ru nanoparticles.21 By using a solution-phase seed-growth method, Tian’s group explored surface-enhanced Raman activity and CO oxidation on Aucore-Ptshell nanoparticles.22,23 The nominal Pt shell thickness can be as thin as 1 ML.22 This provides an alternative way to prepare Pt-coated Au particles with varying overlayer thickness. The oxidation of CO and methanol on Pt thin films deposited on Au particles as a function of film thickness has not been explored. Given that nanoparticles are used as the catalysts in fuel cells,1 it is of both fundamental and practical interest to address the above issues. Experimental Section Chemicals. Hydrogen tetrachloroaurate trihydrate (HAuCl4‚ 3H2O), sodium borohydride (NaBH4), trisodium citrate dihydrate, (3-amino-propyl)trimethoxysilane, potassium tetrachloroplatinate (K2PtCl4), and copper sulfate (CuSO4) were obtained from SigmaAldrich (St. Louis, MO). Ethanolamine was obtained from Alfa Aesar (Ward Hill, MA). HPLC-grade methanol was from Pharmco (Brookfield, CT), and semiconductor-grade (99.997%) carbon monoxide was from Spectra Gas (Branchburg, NJ). The aqueous solutions were prepared using Milli-Q water from a Millipore water purification system (Milli-Q A10, Millipore, MA). Instruments. UV-vis absorption spectra of Au colloids were obtained with an Agilent 8453 spectrometer. Atomic force microscopy (AFM) images of Au nanoparticles assembled on indium tin oxide (ITO)-coated glass slides (resistance 4-6 Ω·cm, Delta Technologies, Stillwater, MN) were obtained using a PicoPlus II scanning probe microscope (Molecular Imaging, Tempe, AZ). Cyclic voltammograms were recorded in a conventional two-compartment threeelectrode cell using an electrochemical analyzer (CHI 630, CH Instruments, Austin, TX). A Pt wire served as the counter electrode, and a KCl-saturated Ag/AgCl electrode was used as the reference electrode. The bulk Pt and Au electrodes were procured from CH Instruments. The cell resistance was compensated with the iR compensation function in the analyzer software. All measurements were made at room temperature (22 ( 1 °C). Preparation of 3 nm Diameter Colloidal Au. All glassware used for colloid preparation was thoroughly cleaned in aqua regia (3:1 HCl/HNO3), rinsed with water, and soaked in water for 1 h prior to use. The colloidal Au nanoparticles were prepared according to the reported method and stored in the dark at 4 °C.24 A UV-vis spectrum of the Au colloid was taken, and the appearance of an absorption peak between 508 and 511 nm confirms that the nanoparticle size is approximately 3 nm.24 Surface Derivatization of Substrate. ITO glass slides were cleaned by sonication for 30 min sequentially in a 20% v/v ethanolamine aqueous solution and methanol. The slides were then thoroughly rinsed with methanol. The subsequent surface deriva(19) Park, S.; Yang, P. X.; Corredor, P.; Weaver, M. J. J. Am. Chem. Soc. 2002, 124, 2428-2429. (20) Jin, Y. D.; Shen, Y.; Dong, S. J. J. Phys. Chem. B 2004, 108, 8142-8147. (21) Brankovic, S. R.; Wang, J. X.; Adzic, R. R. Electrochem. Solid-State Lett. 2001, 4, A217-A220. (22) Li, J. F.; Yang, Z. L.; Ren, B.; Liu, G. K.; Fang, P. P.; Jiang, Y. X.; Wu, D. Y.; Tian, Z. Q. Langmuir 2006, 22, 10372-10379. (23) Zhang, B.; Li, J. F.; Zhong, Q. L.; Ren, B.; Tian, Z. Q.; Zou, S. Z. Langmuir 2005, 21, 7449-7455. (24) Brown, K. R.; Walter, D. G.; Natan, M. J. Chem. Mater. 2000, 12, 306313.

Kumar and Zou tization with 3-aminopropyl trimethoxysilane (3-APTMS) was done by soaking the cleaned ITO slides in a 40 mM methanol solution of 3-APTMS overnight.25 The slides were then sonicated for 1 min in CH3OH followed by thorough rinsing with CH3OH to remove any loosely attached APTMS molecules. Attachment of Au Nanoparticles. After derivatization with 3-APTMS, the slides were dried in an oven at 110 °C for 10 min, followed by immersing the slides in a 3 nm Au colloid solution with varying concentrations and different soaking times to obtain different particle coverages.26 The slides were then rinsed thoroughly with water and dried in an oven at 110 °C for 10 min. The drying step is critical for obtaining good-quality AFM images. Deposition of Platinum on Au Nanoparticles. Platinum films were deposited by means of the galvanic replacement technique.19,27 Briefly, an atomic layer of Cu was deposited on Au nanoparticles with underpotential deposition (UPD) at -0.02 V in 1 mM CuSO4 + 0.1 M H2SO4 for 2 min. The UPD potential was chosen at the foot of the Cu bulk deposition peak. As an example, a cyclic voltammogram obtained on a Au nanoparticle ensemble electrode in the Cu deposition solution is shown in Supporting Information (Figure S1). The particle coverage has little effect on the Cu UPD potential. The Cu-coated Au particles were then immersed in a vigorously deaerated 5 mM K2PtCl4 + 0.1 M HClO4 solution for 10 min. In this process, the Cu layer was replaced by Pt through a galvanic replacement process, and an atomic layer of Pt was then formed on the Au particles. Repeating these procedures with Cu UPD at a potential 40-60 mV more positive than that on Au leads to the formation of multiple atomic layers of Pt on Au, with the addition of one atomic layer in each deposition.28

Results and Discussion Characterization of Pt-Coated Au Nanoparticles. Previously, we reported that gold nanoparticle ensemble electrodes with varying particle coverage can be obtained by immersing APTMS-covered ITO slides in a 3 nm gold particle solution with different duration times.26 The particle coverage (θp) in terms of the ratio between the electrochemically accessible Au area and the ITO geometric area can be varied from 0.05 to 0.5, as estimated from the gold oxide reduction charge in the cyclic voltammograms. The charge for 1 ML of gold oxide is assumed to be 723 µC cm-2, which is obtained theoretically from the gold density (19.3 g cm-3) and atomic weight.29 These Au nanoparticle ensemble electrodes are used in the present study. Shown in Figure 1 are a representative set of cyclic voltammograms obtained in deaerated 0.1 M HClO4 for Pt-coated Au nanoparticles attached on APTMS-modified ITO slides, with a particle coverage of 0.28. For comparison, the CV of the Au particle electrode acquired in 0.1 M HClO4 immediately prior to Pt deposition is also included. Before Pt deposition, the Au particle electrode was subjected to potential cycling from -0.25 to +1.35 V in 0.1 M HClO4 until a stable voltammogram was available. This usually takes five to six potential cycles. During this process, the loosely attached Au particles are removed, and the residual citrate on the particle surface is oxidized to CO2.30,31 After this treatment, the Au oxide reduction charge typically decreases about 15 to 20%. Starting with the particle electrode coated with 1 ML of Pt (green trace), the cyclic voltammogram shows the characteristic (25) Grabar, K. C.; Freeman, R. G.; Hommer, M. B.; Natan, M. J. Anal. Chem. 1995, 67, 735-743. (26) Kumar, S.; Zou, S. Z. J. Phys. Chem. B 2005, 109, 15707-15713. (27) Brankovic, S. R.; Wang, J. X.; Adzic, R. R. Surf. Sci. 2001, 474, L173L179. (28) Mrozek, M. F.; Xie, Y.; Weaver, M. J. Anal. Chem. 2001, 73, 59535960. (29) Oesch, U.; Janata, J. Electrochim. Acta 1983, 28, 1237-1246. (30) Nichols, R. J.; Burgess, I.; Young, K. L.; Zamlynny, V.; Lipkowski, J. J. Electroanal. Chem. 2004, 563, 33-39. (31) Floate, S.; Hosseini, M.; Arshadi, M. R.; Ritson, D.; Young, K. L.; Nichols, R. J. J. Electroanal. Chem. 2003, 542, 67-74.

Electrooxidation of Carbon Monoxide and Methanol

Langmuir, Vol. 23, No. 13, 2007 7367

Figure 1. Cyclic voltammograms of Pt-coated Au nanoparticles recorded in deaerated 0.1 M HClO4 solution. The dotted trace was obtained before Pt deposition. Au particle coverage: 0.28. ITO geometric area: 0.7 cm2. Scan rate: 0.1 V s-1.

symmetric hydrogen adsorption/desorption peaks between -0.25 and 0 V and the surface oxidation current above 0.55 V in the anodic potential scan together with the corresponding surface oxide/hydroxide reduction between 0.3 and 0.4 V on the cathodic scan. The Au oxide reduction peak at 0.85 V is commonly used as an indicator of the presence of pinholes in the Pt film.19 This is a rather stringent criterion because the Au oxidation potential is much higher than that of Pt. At the potential where Au oxidation occurs, the Pt film uniformity is inevitably disrupted. Nevertheless, the cathodic peak from the Au oxide reduction is very small (Figure 1), signifying that the Au particle surface is nearly completely covered with Pt. The charge under this peak is only about 10% of that before the Pt coating is applied, indicating that the Pt coverage on the Au particles is at least 90%. This Pt coverage agrees very well with that estimated from the ratio of the Pt film area assessed from either the CO oxidation or hydrogen adsorption/desorption charges to the Au particle surface area obtained from Au oxide reduction charge. Compared with the monolayer Pt film on a bulk Au electrode (Supporting Information 2) or on a particle ensemble electrode of 70 nm Au particles,19 the Pt film deposited on these smaller particles oxidizes at a more positive potential. In addition, the corresponding oxide reduction peak current density is smaller and appears at a more negative potential. These interesting oxidation-reduction differences likely arise from the distinctly different size of the Au particles. A systematic study of the oxidation and reduction of the Pt monolayer film on Au particles of varying size is required to comprehend these observations fully and is underway in our laboratory. With increasing Pt coating thickness by repeating the galvanic replacement cycle, the hydrogen adsorption/desorption peak grows, indicating that the surface area increases. In addition, the surface oxidation starts at more negative potentials, and the corresponding surface oxide reduction peak shifts to more positive potentials. With 5 ML of Pt, the oxide reduction peak is at 0.4 V, which is close to that on a bulk Pt electrode. The Au oxide reduction peak decreases with increasing Pt film thickness and finally disappears at 4 ML of Pt. The positive shift of the Pt surface oxide reduction peak potential has been observed on Pt nanoparticles with increasing particle size.32-35 We have also (32) Maillard, F.; Eikerling, M.; Cherstiouk, O. V.; Schreier, S.; Savinova, E.; Stimming, U. Faraday Discuss. 2004, 125, 357-377. (33) Mayrhofer, K. J. J.; Arenz, M.; Blizanac, B. B.; Stamenkovic, V.; Ross, P. N.; Markovic, N. M. Electrochim. Acta 2005, 50, 5144-5154.

Figure 2. AFM images (500 nm × 500 nm) of 3 nm Au particles supported on ITO slides before (A) and after (B) being coated with 5 ML of Pt. Au particle coverage: 0.30.

observed similar cyclic voltammograms on Pt-coated Au nanoparticle ensemble electrodes with higher particle coverage but not on the electrodes with particle coverage below 0.1 because the double-layer charging current obscures the observation of the much smaller faradaic current. Figure 2A,B displays the AFM images of 3 nm Au particles on ITO slides before and after the Pt deposition, respectively. The Au particle coverage, again estimated from the Au oxide reduction charge, is about 0.30. The Au particles appear to be larger than 3 nm because of the limited AFM tip radius. However, the height of the particles, whose measurement is not affected by the tip radius, is around 3 nm. It can be seen from Figure 2A that Au nanoparticles are distributed over the entire ITO slide and individual particles are distinguishable. Similar images were obtained after the deposition of one or two monolayers of Pt because the AFM tip cannot resolve the small changes. However, after depositing 5 ML of Pt, the particles cannot be distinguished because the gap between the individual Au nanoparticles is filled with Pt. Because of the limiting resolution of the AFM tip, these connected particles appear to be a single large particle. Carbon Monoxide Oxidation. Figure 3 shows the oxidative stripping voltammograms of the irreversibly adsorbed CO on Pt-coated Au nanoparticle ensemble electrodes (Figure 3A-C) and a bulk Au electrode (Figure 3D) in 0.1 M HClO4. Three (34) Mayrhofer, K. J. J.; Blizanac, B. B.; Arenz, M.; Stamenkovic, V. R.; Ross, P. N.; Markovic, N. M. J. Phys. Chem. B 2005, 109, 14433-14440. (35) Arenz, M.; Mayrhofer, K. J. J.; Stamenkovic, V.; Blizanac, B. B.; Tomoyuki, T.; Ross, P. N.; Markovic, N. M. J. Am. Chem. Soc. 2005, 127, 6819-6829.

7368 Langmuir, Vol. 23, No. 13, 2007

Kumar and Zou

representative electrodes with particle coverage (θp) of 0.45, 0.28, and 0.08 were used. Hereafter, these particle ensemble electrodes are referred to as high, medium, and low θp electrodes, respectively. CO stripping voltammograms were recorded on each of these electrodes with sequentially increasing Pt film thickness from 1 to 10 ML. Before the cyclic voltammogram was acquired, the Pt film electrodes were cleaned by CO annealing33-35 (i.e., cycling the electrode potential at 0.1 V s-1 between -0.25 and +1 V in CO-saturated 0.1 M HClO4 until a stable voltammogram is obtained). This pretreatment typically takes 5 min and is essential for obtaining reproducible results for CO and methanol oxidation. In addition to cleaning the surface, CO annealing was found to remove defects on the Pt(111) surface.2 After pretreatment, the CO adlayer was formed by purging N2 for 10 min while the electrode potential was held at -0.2 V. For clarity, only the segment showing CO oxidation is presented. The CO stripping voltammogram obtained on an extended polycrystalline Pt electrode after CO annealing is also included for comparison. The current density was evaluated with respect to the real Pt surface area obtained by using CO oxidation charge, assuming the charge for oxidizing 1 ML of CO is 420 µC cm-2.32,36 The charge of CO oxidation for each Pt film thickness is evaluated individually by integrating the CO stripping peaks in the corresponding voltammograms. The double-layer charging and surface oxidation charges are corrected for by subtracting the charges contained in the subsequent anodic potential scan voltammogram in the same potential range as that used in evaluating the CO stripping charge. The obtained CO oxidation charge in general increases with the Pt film thickness, suggesting an increase in the Pt surface area. The more commonly used method for evaluating the real Pt surface area employs the H adsorption charge. This approach is not used here because the large double-layer charging current on ITO creates significant

uncertainties in measuring the H adsorption/desorption charge in cases where the Au particle density is low and the Pt layer is thin. The errors are smaller when the CO oxidation charge, which is theoretically twice as much as the H desorption charge on the same surface, is used. Figure 3 reveals a clear Pt film thickness-dependent CO oxidation activity. Starting from the high particle coverage electrode (Figure 3A), the CO oxidation peak appears at around 0.8 V (peak I) when the Au particles are covered with 1 ML of Pt. This peak potential is more than 300 mV higher than that obtained on the polycrystalline bulk Pt electrode undergoing the same treatment (cf. the dotted trace in Figure 3). With increasing Pt film thickness up to 3 ML, the magnitude of peak I decreases, and the peak potential shifts to more negative values. Meanwhile, a new peak appears at around 0.55 V (peak II). At 4 ML of Pt, a second new peak starts to appear at ca. 0.48 V (peak III) and overlaps with peak II, whereas peak I decreases significantly and is centered around 0.75 V. After the addition of another layer of Pt, the magnitude of peaks II and III grows whereas peak I nearly completely disappears. With 7 ML of Pt (not shown), peak III outgrows peak II and becomes the dominant feature. Upon further increasing the Pt film thickness to 10 ML, only peak III is present, and the peak width measured as the full width at half-maximum (∆E ) 65 mV) is much narrower than that of peak I (∆E > 150 mV) on 1 ML of Pt. Note that peak III is 30 mV more negative than that on the bulk Pt electrode. Very similar Pt film thickness-dependent CO oxidation voltammograms were also obtained on Au particle ensemble electrodes with a particle coverage of 0.28 (Figure 3B). Interestingly, when the Au particle coverage decreases to 0.08 (Figure 3C), the changes in CO oxidation voltammograms with Pt film thickness are significantly different from those shown in Figure 3A,B. When the Au particles are covered with 1 ML of Pt, a broad (∆E > 200 mV) CO oxidation peak is observed at 0.85 V, together with a shoulder at around 0.7 V. The main peak is similar to peak I described above for the higher θp electrodes, and the shoulder is similar to peak II. With increasing Pt film thickness up to 10 ML, peak I loses its intensity rapidly, and peak II gains intensity and shifts to 0.65 V. Peak III was not observed at all. A common observation for CO oxidation on these Pt-coated Au particle electrodes is that the CO oxidation charge density is larger at the thinner films, especially for the first three layers. The origin of this observation is not clear at present. On small Pt particles, Stimming and co-workers also observed high CO oxidation charges that are more than double the corresponding hydrogen UPD charges.32,37 To aid our understanding of the film thickness-dependent CO oxidation activity on the Pt-coated Au nanoparticles, we also examined CO oxidation on Pt films deposited on a polycrystalline macroscopic Au electrode using the same deposition method (Figure 3D). CO stripping voltammograms were again recorded after the electrode underwent CO annealing pretreatment as in the nanoparticle ensemble electrode cases. To show the peak potential shift clearly, the potential axis is expanded in Figure 3D. Drastically different CO oxidation voltammograms were observed on the Pt-coated bulk Au electrode. On 1 ML of Pt, a sharp (∆E ≈ 25 mV) CO stripping peak is observed at 0.58 V, which is more than 200 mV more negative than that observed on 1 ML of Pt-coated Au nanoparticles. Upon increasing the Pt film thickness to 2 ML, the CO oxidation peak potential moves to 0.54 V. Upon further increasing the Pt film thickness to 3 ML,

(36) Maillard, F.; Schreier, S.; Hanzlik, M.; Savinova, E. R.; Weinkauf, S.; Stimming, U. Phys. Chem. Chem. Phys. 2005, 7, 385-393.

(37) Friedrich, K. A.; Henglein, F.; Stimming, U.; Unkauf, W. Electrochim. Acta 2000, 45, 3283-3293.

Figure 3. Selective CO stripping voltammograms as a function of Pt film thickness obtained in 0.1 M HClO4 on Pt-coated Au nanoparticle ensemble electrodes (A-C) and a Pt-coated polycrystalline bulk Au electrode (D). Particle coverages: (A) 0.45, (B) 0.28, and (C) 0.08. The Pt film thickness is as indicated. The dotted trace was obtained on a bulk Pt electrode. Scan rate: 0.1 V s-1. The arrows show the scan direction. Note that the potential axis in panel D is expanded to show the peak potential shift clearly.

Electrooxidation of Carbon Monoxide and Methanol

the peak shifts to 0.53 V, and the peak width decreases to ca. 20 mV. From this thickness up to 10 ML, the CO stripping peak potential is largely unchanged and is 20 mV more positive than that on a bulk Pt electrode (cf. the dotted trace). A comparison of the results presented in Figure 3 unravels different factors responsible for the film thickness-dependent CO oxidation activity. As stated at the outset, the adsorption and reactivity of metal thin films can be greatly affected by the substrate, and this change can be explained by the shift of the d band center energy.9,10 Two factors can contribute to the change in the metal overlayer d band center position: the strain effect and the ligand effect.38,39 The strain effect is induced by the lattice mismatch between the overlayer and the substrate. Depending on the lattice constants of the two metals, the monolayer is under either tensile strain when the overlayer lattice constant is smaller than the substrate or compressed strain when the overlayer lattice is larger. For the tensile strained overlayer, the d band width is narrower than that of the corresponding bulk metal, and to maintain the overall d band filling, the d band center increases (moves toward the Fermi level). Conversely, the d band center decreases when the overlayer is under compressed strain. The ligand effect arises from the heterometallic bonding between the two metals. If the bonding between the overlayer and the substrate metal atoms is weaker than that between atoms of the overlayer metal, then the d band width is narrowed for the overlayer, and the d band center increases. If it is the opposite, then the d band center decreases. The strain effect and the ligand effect are believed to be cumulative.38,39 For the Pt monolayer deposited on the bulk Au electrodes, the lattice constant of Pt is 4% smaller than that of Au;40 therefore, the Pt film is under significant tensile strain. The ligand effect can be evaluated by using the characteristic length (rd) of the metals.38,39 This parameter describes the bonding between the d states of the two metals. The d band width is proportional to rd3/2 of each metal.38,39 The rd value for Au is about 3% smaller than that for Pt.41 Consequently, the d band width is narrower for Pt films deposited on Au as compared to that for bulk Pt. The ligand effect is, however, smaller than the strain effect because the d band width is inversely proportional to the fifth power of the Au-Pt bond length r5.38,39 From d band theory, the tensile strain and the ligand effect will increase the d band energy, resulting in the stronger adsorption of OH and CO and therefore a more positive CO oxidation peak potential. With the film growing thicker, the tensile strain will be gradually released, and the ligand effect will become weaker. The lattice constant returns to its bulk value typically at a film thickness of 3-5 ML.42 From the CO stripping voltammograms obtained on the Pt films deposited on the bulk Au electrode (Figure 3D), it is clear that these effects are dominant on the first three monolayers. Above 3 ML of Pt, the CO oxidation peak potential does not decrease further, suggesting that the effects have indeed largely disappeared. For the Pt overlayer on Au particles, the strain and ligand effects also exist but are likely different from those on films deposited on bulk Au electrodes. On one hand, compared with the Pt films on the bulk Au electrode, the tensile strain in the (38) Kitchin, J. R.; Norskov, J. K.; Barteau, M. A.; Chen, J. G. Phys. ReV. Lett. 2004, 93, 156801-156804. (39) Kitchin, J. R.; Norskov, J. K.; Barteau, M. A.; Chen, J. G. J. Chem. Phys. 2004, 120, 10240-10246. (40) Kittel, C. Introduction to Solid State Physics; John Wiley & Sons: New York, 1996. (41) Harrison, W. A. Electronic Structure and the Properties of Solids; Freeman: San Francisco, 1980. (42) Gu¨nther, C.; Vrijmoeth, J.; Hwang, R. Q.; Behm, R. J. Phys. ReV. Lett. 1995, 74, 754-757.

Langmuir, Vol. 23, No. 13, 2007 7369

Pt overlayer on the particles is expected to be smaller because the Au-Au bond length shortens with decreasing particle size. At 3 nm, the Au-Au bond length (r) is between 0.284 and 0.286 nm as predicted by density functional theory (DFT) calculations43 and measured by extended X-ray absorption fine structure spectroscopy44,45 and electron diffraction.46 This length is about 1% shorter than the bulk value of 0.288 nm.40 On the other hand, as shown by DFT calculations, smaller particles have a narrower d band width than the bulk.47 The d band width is proportional to rd3, and inversely proportional to the fifth power of the AuAu bond length r5.41 Therefore, rd is smaller for the 3 nm Au particles as compared to that for the bulk Au. The shortening of rd further decreases the d band width of the Pt film,38,39 resulting in an additional elevation of the d band energy as compared to that of the films on bulk Au. These effects together strengthen the adsorption of CO and OH on the Pt monolayer on Au nanoparticles, yielding a higher CO oxidation potential. As the film grows thicker, both the strain and the ligand effects become weaker as a result of lattice relaxation.48 Hence, the CO oxidation peak shifts to more negative potentials. The stronger CO and OH adsorptions manifest themselves in the broadness of the CO stripping peak on Pt films that are thinner than 3 ML (Figure 3). The broad width of the CO stripping peak can be attributed to the low mobility of the reactants, as demonstrated by the Monte Carlo simulations.49-51 The stronger adsorption of OH is also evident from the thickness-dependent oxidation-reduction potential of the surface hydroxide/oxide on Pt overlayers (Figure 1). The oxide reduction peak potential shifts from 0.32 V at 1 ML to 0.4 V at 5 ML of Pt. A similar broad CO stripping peak has also been observed on small Pt nanoparticles at slightly lower potential and was attributed to the lower mobility of CO arising from the stronger CO adsorption on the nanoparticle surface.32 A second factor that may also contribute to the thicknessdependent CO oxidation is the size increase of the Pt-coated Au particle as the Pt film becomes thicker. Assuming a close-packed arrangement of Pt atoms perpendicular to the particle surface, each monolayer deposition will augment the particle by 0.48 nm if the Au particles are encapsulated by the Pt overlayer. For the film thickness examined here, the nominal particle size can increase from 3 to 8 nm. It has been shown by different groups that CO oxidation and O2 reduction depend on the Pt particle size.32-34,52,53 The particle size-dependent reactivity is attributed to either the fraction of edge and corner atoms52,53 or the electronic properties of the particles33,34 changing with their size. Maillard et al. observed a broad CO stripping peak at 0.95 V (vs RHE) on 1.9 nm Pt particles.32 The peak position shifts to 0.85 V and the peak shape becomes much narrower on 3.1 nm particles. The results presented in Figure 3 for nanoparticle electrodes show some similarities to these observations, which suggest that a (43) Haberlen, O. D.; Chung, S. C.; Stener, M.; Rosch, N. J. Chem. Phys. 1997, 106, 5189-5201. (44) Balerna, A.; Bernieri, E.; Picozzi, P.; Reale, A.; Santucci, S.; Burattini, E.; Mobilio, S. Surf. Sci. 1985, 156, 206-213. (45) Miller, J. T.; Kropf, A. J.; Zha, Y.; Regalbuto, J. R.; Delannoy, L.; Louis, C.; Bus, E.; van Bokhoven, J. A. J. Catal. 2006, 240, 222-234. (46) Lamber, R.; Wetjen, S.; Schulzekloff, G.; Baalmann, A. J. Phys. Chem. 1995, 99, 13834-13838. (47) Wang, J. L.; Wang, G. H.; Zhao, J. J. Phys. ReV. B 2002, 66, 35418. (48) Pinheiro, A. L. N.; Zei, M. S.; Luo, M. F.; Ertl, G. Surf. Sci. 2006, 600, 641-650. (49) Saravanan, C.; Markovic, N. M.; Head-Gordon, M.; Ross, P. N. J. Chem. Phys. 2001, 114, 6404-6412. (50) Koper, M. T. M.; Jansen, A. P. J.; van Santen, R. A.; Lukkien, J. J.; Hilbers, P. A. J. J. Chem. Phys. 1998, 109, 6051-6062. (51) Korzeniewski, C.; Kardash, D. J. Phys. Chem. B 2001, 105, 8663-8671. (52) Kinoshita, K. J. Electrochem. Soc. 1990, 137, 845-848. (53) Kinoshita, K. Electrochemical Oxygen Technology; Wiley: New York, 1992.

7370 Langmuir, Vol. 23, No. 13, 2007

similar size-dependent reactivity may also contribute to the CO stripping peak changes. However, there is no consensus on how strongly the CO oxidation activity depends on the Pt particle size. Takasu et al. reported a much smaller size dependence of particle reactivity.54 The CO stripping peak potential on 3 nm Pt particles is about 50 mV more positive than that on 5.5 nm particles.54 This observation is in agreement with a more recent study by Arenz et al., who examined CO oxidation on Pt particles from 1 to 30 nm.35 On the basis of the results in Figure 3, it is difficult to separate the CO oxidation peak potential shift caused by the strain and ligand effects from that caused by the size effect of the nanoparticle electrodes. A more complete assessment of these entangling effects would require studies of CO oxidation on Pt thin films deposited on Au and Pt particles of a similar size. Preliminary results obtained from CO oxidation on uniform arrays of wellseparated 4 nm Pt55 and 1 ML of Pt-coated 4 nm Au particles suggest that the strain and ligand effects are responsible for about a 150-200 mV potential shift. On the Pt particle arrays with varying particle size, the CO oxidation peak potential shifts from about 0.7 to 0.65 V when the Pt particle size increases from 2 to 8 nm, in line with Takasu et al.54 and Arenz et al.35 From the above discussion, the CO oxidation activity difference between Pt films deposited on the Au particles and on the bulk Au electrodes can be accounted for by the particle size effect and the larger ligand effect for the former. Although the factors discussed above can largely explain the thickness-dependent CO oxidation peak potential on Pt-coated Au particle ensemble electrodes, it cannot account for the difference observed on the electrodes with different particle coverage. Were these effects the only reason for the CO oxidation peak potential shift, the Au particle ensemble electrodes coated with Pt films should show a similar thickness dependence of the CO stripping voltammogram regardless of the Au nanoparticle coverage. This is not what is seen in Figure 3. There is a clear difference between the electrode with low θp and the electrodes with medium and high θp. That is the presence of peak III. In our previous study, we showed by AFM images that when θp is below 0.1 the particles are well-separated.26 For the electrode with a small particle coverage (Figure 3C), when the Pt film thickness increases from 1 to 10 ML the particles are still wellseparated. However, for the electrodes with higher θp, the particles will eventually be connected when the Pt thickness increases (cf. Figure 2). In a recent study of CO oxidation on Pt nanoparticles, Savinova and co-workers showed that with increasing particle loading the degree of particle agglomeration increases.36 Concomitantly, a new CO stripping peak appears and is about 100 mV more negative than the original peak observed on the separated particles.36 This peak potential is even lower than that typically obtained on a polycrystalline Pt surface. They attributed this high activity to the high defect density on the particle aggregates.36 Given that the results presented in Figure 3A,B closely resemble those reported by Savinova et al., the low CO oxidation overpotential on Pt overlayers thicker than 5 ML deposited on the medium and high θp electrodes may originate from the same effect. Methanol Oxidation. Figure 4 shows the cyclic voltammograms of methanol oxidation on Pt-coated Au nanoparticle ensemble electrodes as a function of particle coverage and Pt film thickness. The results obtained on a Pt-coated bulk Au electrode are also included for comparison. The CVs were (54) Takasu, Y.; Iwazaki, T.; Sugimoto, W.; Murakami, Y. Electrochem. Commun. 2000, 2, 671-674. (55) Kumar, S.; Zou, S. Electrochem. Commun. 2006, 8, 1151-1157.

Kumar and Zou

Figure 4. Anodic segments of cyclic voltammograms obtained in 1 M MeOH + 0.1 M HClO4 on Pt-coated Au nanoparticle electrodes (A-C) and a Pt-coated bulk polycrystalline Au electrode (D). Particle coverages: (A) 0.45, (B) 0.28, and (C) 0.08. The Pt film thickness is indicated in each panel. The dotted trace was obtained on a bulk Pt electrode. Arrows mark the scan direction. Scan rate: 0.1 V s-1.

obtained after the data shown in Figure 3 in a separate cell containing deaerated 1 M MeOH + 0.1 M HClO4. The electrode was introduced into the methanol solution at the open circuit potential, and the potential scan typically started 1 min later. Again, the current density is evaluated using the real Pt area obtained from the CO oxidation charge. An inspection of cyclic voltammograms obtained on the particle electrodes reveals a systematic change in methanol oxidation upon increasing the Pt film thickness. Starting from the electrodes with the highest particle coverage, the current density intensifies rapidly as the film grows from 1 to 4 ML and the peak shifts slightly to a more negative position. Further increasing the film thickness does not change the current density much, except that a shoulder at ca. 0.45 V starts to appear at 5 ML and grows up to 10 ML. Note that the current density at 4 ML already reaches that observed on a bulk polycrystalline Pt electrode (cf. Figure 4A, dotted trace). The methanol oxidation voltammograms obtained on electrodes with medium and low particle coverages (Figure 4B,C) are largely similar to those obtained on the high-particle-coverage electrode (Figure 4A). Nonetheless, some subtle differences do exist. As the particle coverage decreases, the methanol oxidation current density becomes smaller at the same Pt film thickness. For the medium θp electrode, the current density reaches the same peak value as for the bulk electrode at 7 ML, whereas on the low θp electrode it is lower than that on the bulk electrode even after the deposition of 10 ML of Pt. In contrast to the above results on Pt-coated Au nanoparticles, methanol oxidation on the Pt films deposited on the bulk Au electrode shows only a weak film thickness dependence; the current density increases only about 30% from 1 to 7 ML, as compared to an increase of more than 5 times on the particle electrodes (Figure 4). The film thickness-dependent methanol oxidation voltammograms on Pt-coated Au nanoparticle electrodes can be largely explained by their CO oxidation activity. On medium and high θp electrodes, when Pt films are thinner than 3 ML, CO oxidation

Electrooxidation of Carbon Monoxide and Methanol

is much less active (Figure 3), and the surface is more vulnerable to CO poisoning, which arises from adsorbed CO formed in methanol dehydrogenation. Hence, methanol oxidation is hindered. When the film is more than 3 ML, CO oxidation is much more facile, and the surface is less prone to being poisoned by CO. Therefore, the methanol oxidation current density is higher. In addition to CO oxidation activity, methanol dissociation also seems to depend on the film thickness, as is evident on the low θp electrode. Although CO oxidation occurs on the same potential for films thicker than 3 ML (Figure 3C), the methanol oxidation current keeps increasing until it reaches 7 ML. Nevertheless, the voltammogram is very similar between films thicker than 7 ML and the bulk Pt electrode, especially on medium- and high-particlecoverage electrodes, suggesting the possibility of using Pt thin films as the anode catalyst in the direct methanol fuel cell.

Langmuir, Vol. 23, No. 13, 2007 7371

effect, and particle aggregation induced by Pt film growth. Without particle aggregation, the Pt films (up to 10 ML) are less active for CO oxidation than the bulk Pt electrode. However, on the aggregated particles, the Pt films can be more active than the bulk. Corresponding to the increasing CO oxidation activity, the current density of methanol oxidation grows with the Pt film thickness. The peak current density and potential reach those on a polycrystalline Pt electrode when more than 4 ML of Pt is deposited on the Au particle electrodes with a particle coverage higher than 0.25. These findings suggest that it is feasible to reduce the Pt loading in methanol fuel cell catalysts by using thin films deposited on a more economic substrate. The study also demonstrates that it is possible to tune the electrocatalytic properties of a metal by exploiting the strain effect and the ligand effect in the thin film, an approach that has just begun to be exploited in the field.7,8,15-17

Conclusions We demonstrated that on Pt-coated Au nanoparticle ensemble electrodes, CO and methanol electrooxidation depends strongly on the Pt film thickness and Au nanoparticle coverage. For electrodes with medium and high Au particle coverages, the CO stripping peak position shifts from ca. 0.8 V at 1 ML to 0.45 V at 10 ML. Accompanying with this peak potential shift is the sharpening of the peak width from more than 150 to 65 mV. For the electrode with a low Au particle coverage, a similar peak width change was also observed, but the peak potential shift is much smaller, from 0.85 V at 1 ML Pt to 0.65 V at 10 ML. These observations are explained by d band theory, the particle size

Acknowledgment. This work is supported by an award from the Research Corporation (RI1218). Acknowledgment is also made to the donors of the American Chemical Society Petroleum Research Fund and the National Science Foundation (CHM0616436) for partial support of this research. Supporting Information Available: Cyclic voltammograms of a Au nanoparticle electrode in 1 mM CuSO4 + 0.1 M H2SO4 and a polycrystalline bulk Au electrode with and without 1 ML of Pt coating in 0.1 M HClO4. This material is available free of charge via the Internet at http://pubs.acs.org. LA0637216