J. Phys. Chem. C 2007, 111, 2165-2176
2165
Understanding the Reactivity of Oxide-Supported Bimetallic Clusters: Reaction of NO with CO on TiO2(110)-Supported Pt-Rh Clusters J. B. Park, J. S. Ratliff, S. Ma,† and D. A. Chen* Department of Chemistry and Biochemistry, UniVersity of South Carolina, Columbia, South Carolina 29208 ReceiVed: July 10, 2006; In Final Form: October 10, 2006
The reactions of CO, NO, and NO with CO have been studied on Pt, Rh, and bimetallic Pt-Rh clusters deposited on TiO2(110). The following four cluster surfaces were investigated: 4 ML of Rh, 4 ML of Pt, 2 ML of Rh + 2 ML of Pt (Rh + Pt), and 2 ML of Pt + 2 ML of Rh (Pt + Rh). Scanning tunneling microscopy studies demonstrated that the surfaces exhibited similar cluster sizes and densities, and low-energy ion scattering experiments showed that the surfaces of the bimetallic clusters were Pt-rich (20-30% Rh) regardless of the order of metal deposition; therefore, both Pt and Rh atoms are capable of diffusing to the cluster surface at room temperature. Notably, heating the surface caused substantial encapsulation of the metal clusters by titania at 700 K and complete encapsulation at 800 K. In temperature programmed desorption experiments, the activities of the Pt and Rh clusters for CO and NO dissociation were found to be higher than those of the (111) surfaces of the corresponding single crystals. For both reactions, the activities of the Rh + Pt and Pt + Rh clusters were identical to each other and intermediate between that of pure Rh and pure Pt. For the reaction of NO with CO, the bimetallic clusters exhibited the greatest production of CO2 and the highest fraction of NO dissociation. On pure Rh clusters, CO2 production is inhibited by the preferential adsorption of NO over CO, whereas on the pure Pt clusters, CO adsorption is favored over NO. Only the Pt-Rh surfaces can provide sites for both NO dissociation and CO adsorption that are necessary for facilitating CO2 formation.
Introduction Bimetallic catalysts are known to exhibit superior properties compared to the individual pure metal components.1,2 The PtRh bimetallic system is a topic of great interest because Pt and Rh are important components of the automobile catalytic converter, which transforms toxic NO and CO to N2 and CO2; the catalyst typically consists of metal clusters supported on an oxide material.3 In order to direct the design and development of improved bimetallic catalysts, it is necessary to have a fundamental understanding of how the presence of the second metal promotes the activity of the catalyst. To this end, it is useful to study model systems consisting of vapor-deposited clusters on single-crystal oxide surfaces under highly controlled conditions, such as ultrahigh vacuum (UHV). These investigations provide a means of extracting basic information regarding changes in the Pt-Rh surface composition, morphology, and interactions between the support and the clusters. Although there have been a number of investigations regarding characterization of bulk Pt-Rh bimetallic surfaces using UHV surface science techniques,4-10 there are no studies of oxide-supported Pt-Rh clusters that address surface composition, morphology, and reaction chemistry. We are specifically interested in understanding if synergistic or additive chemistry is observed on the Pt-Rh bimetallic clusters and exploring the effects of metal-substrate interactions on the surface reactivity. The TiO2(110) surface was chosen as a support since titania is one of the few metal oxides that can be made sufficiently * Corresponding author. Phone: 803-777-1050. Fax: 803-777-9521. E-mail:
[email protected]. † Current address: Chemistry Department, Brookhaven National Laboratory, Upton, NY 11973.
conductive for scanning tunneling microscopy and low energy ion scattering experiments simply by heating the crystal in UHV. This process selectively removes oxygen from the crystal, resulting in an n-type semiconductor. The (110) surface of rutile titania has a well-defined surface structure that is stable against reconstruction up to ∼1000 K,11 and titania is also reported to have excellent potential as a catalytic support material.12-15 Due to their high degree of reducibility, titania surfaces are known to participate in reactions occurring on the supported metal clusters by providing a source of oxygen.16,17 Furthermore, interactions between Au clusters and the titania support are believed to contribute to the superior activity of titania-supported Au clusters in oxidation reactions.14,18-24 The reaction of NO with CO has been extensively studied on single-crystal Rh,25-33 Pt,34-40 and Pt-Rh alloy surfaces41-46 as well as supported clusters.41,47-52 Since the TiO2 supported clusters in this work exhibit very similar reactivity to the bulk metal surfaces, this comprehensive body of research can be used to aid in understanding reaction mechanisms on the supported clusters. For reactions carried out in UHV, N2 and CO2 are the major desorption products on both Rh28,53 and Pt,35,40,54,55 and Rh is considered to be the more active surface due to its increased activity for NO dissociation.28 In some cases, N2O is observed either as a minor product38,53 or not at all,28,40 but N2O production is a more important reaction pathway at higher pressures25,27,32,41,45,48,49,56,57 since high coverages of NO are needed for the NO + N reaction that produces N2O.48 Although NO dimers have been reported on metal surfaces,58-60 there is no experimental evidence for (NO)261 on Rh(111) or Pt(111).58 In general, the accepted mechanisms for CO2, N2,
10.1021/jp064333f CCC: $37.00 © 2007 American Chemical Society Published on Web 01/18/2007
2166 J. Phys. Chem. C, Vol. 111, No. 5, 2007 and N2O product formation on these surface are as follows:25,39,42,55,62
NO(g) f NO(a) NO(a) f N(a) + O(a) N(a) +N(a) f N2(g) high temperature NO(a) + N(a) f N2(g) + O(a) low temperature NO(a) + N(a) f N2O(g) CO(g) f CO(a) CO(a) + O(a) f CO2(g) O(a) + O(a) f O2(g) The ability of the surface to dissociate NO plays an important role in facilitating the CO + NO reaction because NO dissociation provides the source of adsorbed oxygen for CO oxidation. On single-crystal Rh28,29,53 and Pt35,40,54,55,63 surfaces, N2 and O2 are observed as the major desorption products during temperature programmed desorption (TPD) experiments in UHV. N-O bond scission is facile on Rh(111) with ∼50% dissociation at saturation coverages and complete dissociation at lower coverage.33,53 On Rh single-crystal surfaces, TPD experiments show that saturation coverages of NO result in two N2 desorption peaks: at 400-470 K coincident with NO desorption and at 600-800 K from atom recombination.33,53 NO dissociation is highly sensitive to surface structure, and dissociation is faster on the (100) surface compared to (111).33 Furthermore, the N2 desorption temperature is dependent on the strength of the metal-nitrogen bond. For example, the higher desorption temperature for recombinant N2 on Rh(100) compared to Rh(111) is attributed to a higher heat of adsorption for nitrogen atoms on Rh(100).33 In contrast, Pt surfaces are much less active for NO dissociation compared to Rh.28,42,48 No dissociation occurs on Pt(111) other than the ∼2% attributed to reaction at defects,63 whereas 15% dissociates on Pt(110)64 and ∼50% on Pt(100). A single N2 desorption peak at ∼500 K is observed for TPD experiments of NO decomposition on bulk Pt surfaces.35,63,64 In the work reported here, we have investigated CO, NO, and NO + CO reactions on pure Pt, pure Rh, and bimetallic Pt-Rh clusters supported on TiO2(110) under UHV conditions. Pt-Rh bimetallic clusters prepared by sequential deposition of the two metals exhibit surface composition and reactivities that are not dependent on the order of metal deposition, indicating that both Pt and Rh atoms diffuse within the cluster at room temperature. The surfaces of the clusters are in general more active for CO and NO dissociation compared to corresponding close-packed (111) faces of Pt and Rh. However, only the bimetallic Pt-Rh clusters generated appreciable CO2 from NO + CO reaction even though bulk Rh is reported to be the most active for the reduction of NO with CO.28 The higher activity of the Pt-Rh clusters compared to pure Pt and pure Rh is attributed to the availability of sites for both CO adsorption and NO dissociation on the bimetallic surfaces. We propose that the loss of activity for NO reduction on the pure Rh clusters is related to interactions between the Rh clusters and the TiO2 support. Experimental Section The experiments described here were carried out in a customdesigned Omicron UHV system with a base pressure e7.0 ×
Park et al. 10-11 Torr. This system is equipped with a quadrupole mass spectrometer (Inficon, Transpector 2) for temperature programmed desorption (TPD) studies as well as a hemispherical energy analyzer (Omicron, EA125) for low energy ion scattering (LEIS) and X-ray photoelectron spectroscopy (XPS) experiments. In addition, the vacuum system has facilities for variabletemperature scanning tunneling microscopy (Omicron, VTSTM 25) and low energy electron diffraction/Auger electron spectroscopy (Omicron, SPEC 3) and is equipped with a fourpocket metal evaporator (Oxford Applied Research, EGCO4), quartz crystal microbalance (Inficon, XTM/2), and ion gun (Omicron, ISE 10). A more detailed description of the chamber can be found elsewhere.65-67 The rutile TiO2 (110) single crystal (1 cm × 1 cm × 0.1 cm) was purchased from Princeton Scientific Corporation and was heated by electron bombardment using a home-built sample stage, which has been previously described in detail.65,68 The sample temperature was measured using a Type K thermocouple spotwelded onto the Ta foil on which the crystal was mounted. The absolute temperature of the sample was checked with an infrared pyrometer (Heitronics, model KT19.81 II, 8-10 µm) and was found to be identical within (5 K to that measured by the thermocouple at all temperatures. The emissivity value used in these measurements was 0.55, which was calculated from the measured reflectance and transmittance of the TiO2 crystal after correcting for the transmittance of the ZnSe window on the chamber. The 0.55 emissivity was also in perfect agreement with the high temperature water desorption peak at 490 K observed in water TPD experiments, as compared with data from the literature, in which the thermocouple has been carefully calibrated.69 Unreduced, partially reduced and fully reduced TiO2 crystals were found to have the same reflectance and transmittance at 8-10 µm, and the emissitivities of metal oxides are believed to be independent of temperature in this wavelength region. The TiO2 surface was cleaned by several cycles of Ar+ sputtering (1 keV, 40 min) and annealing (950 K, 5 min), and XPS studies confirmed that there were no surface contaminants after this treatment. Furthermore, the crystal had a sharp low energy electron diffraction pattern corresponding to the unreconstructed (1 × 1) surface, and high resolution STM images of the surface exhibited bright Ti rows separated by 6.5 Å, as typically observed for TiO2(110)-(1 × 1).11,70 STM experiments were carried out in constant current mode with electrochemically etched tungsten tips67 using tunneling currents of 0.1 nA at a sample bias of +2.0 V. All gases were used as supplied by the manufacturer. Ar (99.9995%) was provided by Matheson Tri-Gas, and CO (99.99%) and NO (99.5%) were obtained from National Specialty Gases. 13CO (99% 13C) and 15NO (98% 15N) were supplied by Cambridge Isotope Laboratories, Inc., whereas 18O2 (97% 18O) was provided by Isotec. The Pt and Rh clusters were deposited onto the TiO2 surface at room temperature at a rate of 0.4 ML/min using a commercial metal evaporator, which heats the pure metal rods by electron bombardment. A coverage of one monolayer is defined as 1.50 × 1015 atoms/cm2 for Pt and 1.60 × 1015 atoms/cm2 for Rh, according to the packing densities of the (111) surfaces. Coverages were estimated from cluster densities and average cluster sizes observed in the STM images, assuming that the clusters have paraboloid shapes.66 The metal flux was also monitored with a quartz crystal microbalance to ensure consistent coverages for each deposition. Low energy ion scattering spectra were collected using a 1 kV He+ beam at fixed detection angle of 130° and a chamber
Oxide-Supported Bimetallic Clusters pressure of 5 × 10-8 Torr, which corresponded to a beam current of 9.2 µA/cm2. Each LEIS spectrum was collected within 1 min to minimize surface damage from the He+ beam. In order to achieve reproducible beam currents without damaging the surface, the beam current was first stabilized and monitored at one position on the surface, and then moved to a different position for the collection of LEIS data. The integrated peak intensities from spectra acquired from 6 successive scans of the same surface had relative standard deviations of e2.3%. For surfaces containing both Pt and Rh, the peak integration of the first metal was carried out by initially subtracting out the contribution of the second metal using the appropriately scaled peak shape from the spectrum of pure 4 ML clusters. For the TPD experiments, the surface was exposed at room temperature to a saturation dose of all of the reactant gases via a stainless steel directed dosing tube positioned 2 mm from the sample surface. Dosing times and pressures were chosen to achieve the maximum integrated intensities for all of the desorption products. The CO/NO and 13CO/15NO gas mixtures were prepared in a 1:1 ratio by measuring the partial pressure of each gas (corrected for convectron gauge sensitivity) prior to mixing. The coadsorption of this 1:1 gas mixture is denoted as “CO/NO” in this work, and all experiments involving the adsorption of both NO and CO on the same surface are coadsorption experiments unless otherwise stated. The mass spectrometer was covered by a cylindrical stainless steel shroud with a 6 mm diameter aperture cut in the center of the end cap. The crystal was placed 3 mm away from the aperture during TPD experiments in order to minimize detection of desorption from the sample holder. An in-house LabVIEW program (National Instruments) was interfaced to a programmable power supply (Kepco, ABC 10-10DM), which generated a temperature ramp of 2 K/s. Wide mass scans from 0 to 100 amu did not detect any masses other than those of the reported products and their associated cracking fragments. All TPD experiments were carried out on freshly deposited, unannealed cluster surfaces. Results Surface Chemistry. Before studying the coadsorption of CO and NO on the bimetallic clusters, the chemistry of CO and NO alone was studied on the following cluster surfaces on TiO2: 4 ML of pure Rh, 4 ML of pure Pt, 2 ML Pt + 2 ML Rh (Pt + Rh) and 2 ML Rh + 2ML Pt (Rh + Pt). Notably, no CO or NO desorption from the TiO2(110) itself was observed after exposing the surface to the respective gases. On the 4 ML Rh clusters, CO desorbs in a broad peak centered around 490 K and in two high-temperature peaks at 810 and 870 K (Figure 1). Based on studies of CO on single-crystal Rh surfaces, the low temperature feature is attributed to the desorption of molecular CO, whereas the high temperature peaks are assigned to recombination of atomic carbon and oxygen after CO dissociation.71,72 In contrast, on the 4 ML Pt clusters, the main desorption feature is centered around 530 K, and only trace desorption can be detected in the high temperature region (830 K). The two bimetallic cluster surfaces prepared by reversing the order of Pt and Rh deposition exhibit identical surface chemistry. A molecular desorption peak is observed at 520 K as well as a less intense peak at 825 K from recombination of atomic carbon and oxygen. The intensity of the high temperature feature is smaller than the peak observed on pure Rh, indicating that the extent of CO dissociation on the bimetallic clusters is intermediate between that on Rh and Pt. The total CO desorption yields on the four surfaces are all within 10% of each other. However, 25% of the total CO desorption occurs at high
J. Phys. Chem. C, Vol. 111, No. 5, 2007 2167
Figure 1. Temperature programmed desorption data for a saturation exposure of CO dosed on the following surfaces at room temperature: (a) 4 ML of Rh; (b) 4 ML of Pt; (c) 2 ML Pt + 2 ML Rh; and (d) 2 ML Rh + 2 ML Pt on TiO2(110).
temperature on Rh clusters compared to ∼10% on the two bimetallic surfaces and only 1% on the Pt clusters. The titania crystal was reoxidized by exposure to 18O2 at 600 K for 15 min prior to deposition of 4 ML of Rh in order to address the role of lattice oxygen in CO desorption. The resulting TPD spectra (not shown) exhibited C18O (30 amu) peaks at 810 and 870 K only, demonstrating that at high temperature, lattice oxygen participates in oxidation of the atomic carbon produced by CO dissociation. The chemistry of NO on the pure and bimetallic clusters was investigated by adsorption of 15NO, which was used to distinguish between 15N2O (46 amu) and CO2 (44 amu) produced from trace CO contamination. Figure 2 shows that on the 4 ML Rh clusters molecular 15NO (31 amu) desorbs at 500 K. The N-O bond is also broken to form 15N2 (30 amu) in a small peak at 538 K and a much more intense peak at 960 K. On the 4 ML Pt clusters, the 15NO desorption peak is smaller than on Rh and occurs at a temperature 30 K lower. The 15N2 peak at low temperature is three times greater than on Rh, whereas the high temperature production is 70% lower. Again, the two Pt-Rh bimetallic clusters exhibit chemistry nearly identical to each other and intermediate between those of the pure cluster surfaces; 15NO desorption occurs at 485 K, and 15N production at high temperature is 80-86% of that on Rh. 2 The initial 15N2 desorption peak appears at a temperature 15 K lower on the Pt clusters compared to Rh, following the same trend as the 15NO desorption. A small amount of 44 amu signal is also detected at 500 K from all surfaces except the 4 ML Rh clusters and is attributed to the formation of CO2 from the reaction of surface oxygen with adsorbed CO, which originates from residual CO in the chamber. No other products were detected from NO reaction on the four cluster surfaces, with the exception of trace quantities of 15N2O (46 amu) from the 4 ML Pt clusters as indicated by a small signal that was barely detectable above the baseline. Assuming equal ionization probability for NO and N2, the ratio of peak intensities for 15NO and 15N show that ∼70% NO dissociation occurs on both 2
2168 J. Phys. Chem. C, Vol. 111, No. 5, 2007
Figure 2. Temperature programmed desorption data for a saturation exposure of 15NO dosed on the following surfaces at room temperature: (a) 4 ML of Rh; (b) 4 ML of Pt; (c) 2 ML Pt + 2 ML Rh; and (d) 2 ML Rh + 2 ML Pt on TiO2(110).
4 ML Rh and 4 ML Pt clusters with ∼78% dissociation on the two bimetallic cluster surfaces. Although more N2 is produced on the Rh clusters, the NO desorption is also the greatest on Rh. The experiments with unlabeled 14NO confirmed all of our product assignments, demonstrating that 14NO at 30 amu and 14N at 28 amu are the main desorption products along with 2 trace amounts of CO2 at 44 amu. On Rh surfaces, the temperature of the recombinant N2 desorption peak is known to shift with coverage,29,53,58,73,74 but the ∼400 K temperature difference between the two N2 peaks on the cluster surfaces is too large to be accounted for by coverage effects. Recombinant N2 production shifts from ∼500 K at high coverages to ∼700 K at low coverages, but the 700 K feature is still present as a shoulder at high coverage on Rh(111).28,75 In one report, appearance of the low temperature N2 desorption state was proposed to be an artifact caused by NO decomposition on the hot filament of the mass spectrometer.33 However, this is not the case in our study since the low temperature N2 desorption occurs at temperatures 40-50 K higher than NO evolution, and the intensities of the N2 peaks on the cluster surfaces do not scale with the respective NO desorption intensities. Figure 3 shows the TPD spectra from the reaction of CO with NO on the pure and bimetallic clusters. Isotopically labeled 13CO and 15NO were used for these studies in order to distinguish between production of 15N2 (30 amu) and 13CO (29 amu) and between 13CO2 (45 amu) and 15N2O (46 amu). On
Park et al. the 4 ML Rh clusters, the major desorption products are those associated with 15NO. Specifically, a large 15NO peak is observed at 485 K, and 15N2 desorption occurs primarily at 930 K as well as in a smaller peak at 520 K. The desorption of 13CO is detected in a low temperature peak at 400 K and small high temperature feature at 810 K. The 930 K desorption peak in the 29 amu signal is attributed to 14N15N rather than 13CO, based on the lack of 13 amu signal and the fact that this peak is coincident with 15N2 production; moreover, the major contaminant in the 98% purity 15NO is 14NO. Traces of 45 amu desorption attributed to 13CO2 are detected around 450 K. In contrast, on the 4 ML Pt clusters, the main desorption features are associated with 13CO rather than 15NO adsorption. A 13CO molecular desorption peak at 540 K is the dominant feature in the TPD spectrum with a smaller recombinant peak observed at 830 K. 13CO2 production on the Pt clusters is three times higher than on the Rh clusters and occurs at approximately the same temperature. Only a small 31 amu signal at 400 K from 15NO desorption is detected, and the 940 K 15N peak at 30 2 amu is 20% of that on the Rh clusters. No desorption of 15N2 is detected at lower temperatures. For 15NO/13CO coadsorption on both Rh and Pt clusters, the desorption of the more weakly bound adsorbate is shifted to lower temperature. On the 4 ML Rh clusters, the presence of coadsorbed 15NO appears to lower the temperature for 13CO desorption since the peak maximum is shifted from ∼500 to 410 K. However, it should be noted that the broadness of the desorption peak for 13CO in the absence of 15NO indicates that there is more than one desorption state, and this apparent shift to lower temperature is due to the loss of the higher temperature desorption states, which are present for 13CO in the absence of 15NO. The lower temperature states are attributed to repulsive interactions between 15NO and 13CO, as reported on Rh(111) and Pt(111).28 Similarly, the lower apparent 15NO desorption peak temperature observed on the 4 ML Pt clusters compared to 15NO adsorption only is due to the loss of the higher temperature desorption states that contribute to 15NO desorption in the absence of 13CO. Repulsive interactions between 15NO and 13CO appear to be stronger than between 15NO molecules. Reaction on the two bimetallic Pt-Rh cluster surfaces resulted in significant 13CO2 yields at 500 K, in contrast with what was observed on the pure Pt and pure Rh clusters (Figure 3). The peak temperature for CO2 production on Pt is lower than that on the Rh and Pt-Rh surfaces and corresponds to the low temperature shoulder observed for 13CO2 desorption on the Pt-Rh clusters. The production of 15N2 on the bimetallic clusters is similar to that on the Rh surface except that the 520 and 940 K peaks are slightly less intense. In terms of desorption of the reactants, the small 15NO signal is comparable in intensity to desorption on the 4 ML Pt clusters, but two distinct features at 405 and 490 K were observed, roughly corresponding to the desorption temperatures on pure Pt and Rh clusters, respectively. Similarly, the molecular desorption of 13CO occurs at two distinct temperatures, with the 450 K feature corresponding to desorption from pure Rh clusters and the 560 K feature corresponding to desorption on pure Pt. A recombinant 13CO peak at 820 K is also detected. As previously explained, the 29 amu peak at 940 K is assigned to 15N14N instead of 13CO on allsurfaces. All product assignments are consistent with the masses observed in the coadsorption TPD of unlabeled 14NO and 12CO on the four different cluster surfaces. The yields for desorbing species in the 15NO/13CO coadsorption experiments on the four different cluster surfaces are summarized in Table 1, and all integrated intensities are
Oxide-Supported Bimetallic Clusters
J. Phys. Chem. C, Vol. 111, No. 5, 2007 2169
Figure 3. Temperature programmed desorption data for a saturation exposure of 15NO and 13CO mixed in a 1:1 ratio and dosed on the following surfaces at room temperature: (a) 4 ML of Rh; (b) 4 ML of Pt; (c) 2 ML Pt + 2 ML Rh; and (d) 2 ML Rh + 2 ML Pt on TiO2(110).
TABLE 1: Product Yields for the Reaction of 13CO with 15NO on the Various Cluster Surfacesa surface 4 ML Rh 4 ML Pt 2 ML Pt + 2 ML Rh 2 ML Rh + 2ML Pt a
13CO 2
0.15 0.42 1.0 0.94
15N 2
1.0 0.21 0.69 0.66
13CO
0.20 1.0 0.64 0.83
TABLE 2: Integrated Intensities from TPD Studies of the Experiments on the 4 ML Rh Clustersa
15NO
1.0 0.08 0.13 0.09
All yields are normalized against the highest observed value.
normalized to the highest observed value. The Pt + Rh and Rh + Pt bimetallic clusters have nearly identical reactivities and evolve 13CO2 as a major reaction product, in contrast to the pure metal clusters. 15N2 production from the bimetallic clusters is slightly lower but still comparable to that on pure Rh clusters. On the pure Rh clusters, desorption of 15NO and dissociation of 15NO to produce 15N2 are the dominant surface processes in 13CO/15NO coadsorption, whereas 13CO and 13CO represent 2 minor desorption species. On the pure Pt clusters, the situation is reversed: 13CO is the major desorption species, more 13CO2 is produced than on Rh, and the 15NO and 15N2 associated with 15NO adsorption and reaction are minor desorption species. Using the uncorrected intensities of the 15N2 peaks as a measure of total NO dissociation, the TPD results also indicate that pure Rh is more active for NO dissociation than pure Pt, whereas the bimetallic Pt-Rh clusters exhibit intermediate activity. The fraction of adsorbed NO that dissociates is 80% on 4 ML of Rh, 90% on 4 ML of Pt, and ∼95% on the two Pt-Rh cluster surfaces. In order to understand the nature of competitive CO and NO adsorption on Pt and Rh, the following sequential dosing experiments were carried out on the 4 ML Pt and 4 ML Rh clusters: a saturation exposure of NO followed by a saturation
N2 (28 amu) at ∼930 K
NO (30 amu)
CO2 (44 amu)
0.15
1.00 0.86
1.00 0.70
0.05 0.09
0.31
0.50
0.46
0.15
0.27
0.70
0.72
0.07
CO (28 amu) CO only NO only coadsorbed CO and NO (1:1 ratio) sequential CO + NO sequential NO + CO
1.00
a The intensities for 28 and 30 amu are normalized to the values for CO and NO adsorption, respectively, in the absence of coadsorbate. The 44 amu signal is normalized to the value for CO2 production on the Pt-Rh bimetallic clusters.
exposure of CO and a saturation exposure of CO followed by a saturation exposure of NO. These results are compared with the coadsorption experiments as well as with CO and NO dosed in the absence of the coadsorbate in Tables 2 and 3. On the 4 ML Rh clusters, the two sequential dosing experiments produce approximately the same amount of CO (28 amu), whereas the coadsorption experiment produces ∼50% less. However, all three experiments generate no more than 30% of the CO produced from the adsorption of CO in the absence of NO. In terms of NO (30 amu) desorption from the 4 ML Rh clusters, both the coadsorption experiment and the sequential adsorption with NO first evolved ∼30% less NO than for NO adsorption alone on Rh; the adsorption of CO before NO results in ∼50% less NO. The amount of N2 produced from NO dissociation follows the same trend as NO desorption. Thus, NO is
2170 J. Phys. Chem. C, Vol. 111, No. 5, 2007
Park et al.
TABLE 3: Integrated Intensities from TPD Studies of the Experiments on the 4 ML Pt Clustersa N2 (28 amu) at ∼930 K
NO (30 amu)
CO2 (44 amu)
0.97
1.00 0.53
1.00 0.24
0.17 0.39
0.85
0
0.01
0.12
0.38
0
0.24
0.54
CO (28 amu) CO only NO only coadsorbed CO and NO (1:1 ratio) sequential CO + NO sequential NO + CO
1.00
a The intensities for 28 and 30 amu are normalized to the values for CO and NO adsorption, respectively, in the absence of coadsorbate. The 44 amu signal is normalized to the value for CO2 production on the Pt-Rh bimetallic clusters.
preferentially adsorbed over CO on the Rh surface, and NO is also capable of easily displacing preadsorbed CO, whereas only a small amount of NO can be displaced by subsequent exposure to CO. The greatest production of CO2 (44 amu) is observed when CO is preadsorbed on the surface before NO, suggesting that the production of CO2 on Rh is limited by the ability of CO to adsorb in the presence of NO. However, the CO2 production for all of the experiments on the Rh clusters was no more than ∼15% of that on the bimetallic Pt-Rh surface. On the Pt clusters, the amount of CO evolved in the coadsorption studies is approximately the same as in the CO only experiment. Although the sequential adsorption of CO followed by NO decreases the CO desorption intensity by ∼15%, the adsorption of NO before CO drops the CO yield to only ∼38% of the value in the absence of NO. The adsorption of CO is therefore favored over NO on Pt so that the presence of NO does not significantly decrease the amount of coadsorbed CO. Furthermore, only a small fraction of CO can be displaced by NO. For all of the experiments, the intensity of the NO desorption peak is significantly diminished in the presence of CO. Coadsorption and sequential adsorption decrease the NO intensity by 76%, whereas the adsorption of CO before NO drops the NO signal by 99%. Similarly, the high temperature N2 production from NO dissociation is decreased significantly for the coadsorption experiment, and no N2 is detected for the sequential adsorption studies. In terms of CO2 production, adsorption of NO before CO generates the greatest CO2 yield, followed by coadsorption of NO and CO, but both of these yields are only 40-50% of that observed for coadsorption on the bimetallic Pt-Rh clusters. When CO is adsorbed prior to NO, the level of CO2 production is comparable to that for NO alone on the 4 ML Pt clusters, where the source of CO is from background contamination. In conclusion, CO adsorption is favored over NO adsorption on the 4 ML Pt clusters, and although almost no desorption of molecular NO is detected, some NO dissociation must occur in the presence of CO since CO2 is formed. Surface Characterization. In order to understand changes in metal surface area during heating in TPD experiments, STM images of the pure and bimetallic surfaces were collected at room temperature and after heating with a linear ramp of 2 K/s to 500, 600, and 800 K. The surfaces with 4 ML Rh, 4 ML Pt, 2 ML Rh + 2 ML Pt, and 2 ML Pt + 2 ML Rh clusters all look similar to each other: average cluster sizes range from 8.5 to 9.9 Å in height and 44-51 Å in diameter with cluster densities of 8-9 × 1012/cm2. STM images for the first three surfaces are shown in Figure 4, but the Pt+Rh surface is not
included since its surface morphology appeared identical to the Rh+Pt surface. The average cluster dimensions reported in Table 4 are based on STM line profile measurements from a set of 20 clusters for each surface. After heating to 800 K, larger clusters are present on all four surfaces due to sintering, with diameters of ∼60-65 Å, and heights of 12-13 Å (Table 4). Slightly larger cluster diameters and smaller densities are observed on the 4 ML Rh surface, suggesting that Rh clusters sinter more easily than Pt and Pt-Rh on TiO2. The surface areas of the clusters after heating to different temperatures are summarized in Figure 5 for all four surfaces and are normalized against the value for the 4 ML Rh clusters. These areas were calculated by numerical integration of the STM images, followed by subtracting out the area contribution from the uncovered substrate; this procedure has been described in more detail elsewhere.16 Although tip convolution effects could result in slightly overestimated surface areas for the clusters, all of the surfaces had similar clusters sizes and densities, and therefore, contributions from tip convolution should be comparable. At room temperature, the four surfaces have nearly equivalent cluster surface areas that are within 4% of each other. The surface area of the Rh clusters decreases the most rapidly after annealing, which is consistent with more extensive sintering for the Rh clusters compared to Pt and Pt-Rh. At 500 K, the surface area of the Rh clusters drops by 5%, but the areas for the Pt and two bimetallic cluster surfaces are essentially unchanged. After heating to 800 K, the surface area for the Rh clusters decreases by 30%, whereas surface areas of the Pt and Pt-Rh clusters are reduced by less than 10%. Low energy ion scattering signals for Pt and Rh are shown in Figure 6 for the Rh, Pt, Pt + Rh, and Rh + Pt cluster surfaces heated to various temperatures at a rate of 2 K/s. At room temperature, both bimetallic cluster surfaces are Pt-rich, but the Pt + Rh clusters have a greater fraction of surface Rh. The Rh compositions of the cluster surfaces are calculated to be 30% for Pt + Rh and 20% for Rh + Pt, based on intensities of the Pt and Rh signals for the pure 4 ML clusters and their respective surfaces areas.76 Thus, both Pt and Rh atoms are capable of diffusing to the surface at room temperature to achieve energetically favorable surface compositions. After heating to 800 K, the Pt and Rh signals decrease to zero for all four surfaces. Given that clusters can still be imaged on the surface at this temperature and that no elements other than titanium and oxygen are detected, the complete loss of metal signal suggests that the metal clusters become encapsulated by titania. Indeed, there are a number of other studies in the literature that have reported the encapsulation of Rh77,78 and Pt79-81 clusters by TiOx after annealing these metals on TiO2(110) in ultrahigh vacuum. A comparison of the signals for 4 ML of Rh and 4 ML of Pt shows that the Rh signal decreases more rapidly with temperature than Pt, particularly between 500 and 600 K. This behavior cannot be completely explained by faster sintering for Rh because the cluster surface area decreases by only 30% at 600 K, whereas the Rh LEIS signal drops by 70%. Therefore, we conclude that the Rh clusters become encapsulated by titania at a lower temperature than Pt. The two types of Pt-Rh clusters both exhibit a 85-90% reduction in Pt signal at 700 K, similar to what is observed for the pure Pt clusters, whereas the Rh signal drops almost to zero, as observed for the pure Rh clusters. This also supports the idea that migrating titania from the substrate is more likely to cover Rh than Pt on bimetallic cluster surfaces. Between 600 and 800 K, the sharp decrease in metal signal for all of the cluster surfaces is attributed predominantly to encapsulation. However, between 500-600 K, the decrease in
Oxide-Supported Bimetallic Clusters
J. Phys. Chem. C, Vol. 111, No. 5, 2007 2171
Figure 4. Scanning tunneling microscopy images (1000Å × 1000 Å) of the following metal clusters at room temperature: (a) 4 ML Rh; (b) 4 ML Pt; (c) 2 ML Pt + 2 ML Rh; and after heating to 800 K at 2 K/s: (d) 4 ML Rh; (e) 4 ML of Pt; and (f) 2 ML Pt+2 ML Rh.
TABLE 4: Average Cluster Sizes (in Å) for the Four Pure and Bimetallic Surfaces as Measured from a Set of 20 Clustersa
surface
average diameter at 295 K
average height at 295 K
average diameter at 800 K
average height at 800 K
4 ML Rh 4 ML Pt 2 ML Pt + 2 ML Rh 2 ML Rh + 2ML Pt
49 ( 10 47 ( 6 51 ( 7 44 ( 8
9.9 ( 2.9 9.2 ( 1.7 8.5 ( 1.5 9.5 ( 2.2
66 ( 13 59 ( 10 57 ( 10 59 ( 9
12 ( 3 12 ( 2 13 ( 3 12 ( 2
a
Error bars were calculated from the standard deviations.
Figure 6. Low energy ion scattering signals for Pt (filled symbols) and Rh (open symbols) for 4 ML Rh, 4 ML Pt, 2 ML Pt + 2 ML Rh, and 2 ML Rh + 2 ML Pt clusters on TiO2(110) after heating to various temperatures at 2 K/s. Error bars were determined from the standard deviations of the peak areas from 6 successive spectra.
Figure 5. Cluster surface areas as calculated from STM images for 4 ML Rh, 4 ML Pt, 2 ML Pt + 2ML Rh, and 2 ML Rh + 2 ML Pt clusters on TiO2(110) after heating to various temperatures at 2 K/s. All values are normalized to that of the 4 ML Rh clusters at room temperature.
metal LEIS signal could be due to the onset of encapsulation as well as a slight loss in surface area from sintering. The Pt signals for the bimetallic clusters follow the same trend as the
signal from the pure Pt clusters, but the Rh signals remain almost constant for the bimetallic clusters, compared to a substantial decrease for pure Rh. The fact that the Rh signal for the Rh + Pt clusters increases after annealing to 600 K suggests that Rh diffuses to the cluster surface when Pt is deposited on top of Rh. At 600 K, the Pt + Rh and Rh + Pt have identical surface compositions (28% Rh) due to the slight increase in Rh signal for Rh + Pt and corresponding decrease in signal for Pt + Rh after heating. Although studies of Rh on TiO2 have reported that the Rh clusters de-encapsulate after heating above 900 K,77 the LEIS experiments show that no metal was exposed even after heating to 1000 K. LEIS studies of the Pt-Rh clusters upon adsorption of NO and CO were also carried out to understand if the presence of adsorbates changes the surface composition of the bimetallic clusters. Previous studies of Pt-Rh alloy surfaces have reported that oxygen induces preferential segregation of Rh to the surface.
2172 J. Phys. Chem. C, Vol. 111, No. 5, 2007 Specifically, on the Pt0.25Rh0.75(100) alloy surface, the presence of adsorbed NO was found to induce Rh segregation to the surface due to the stronger binding of oxygen atoms to Rh compared to Pt;82 another study of Pt films on Rh single-crystal surfaces demonstrated that Rh atoms segregate to the surface when the surface is exposed to NO or O2.46 For CO adsorption, both the Pt and Rh signals are significantly attenuated at 300 K due to adsorbed molecular CO and remain lower as a result of screening from carbon and oxygen from CO dissociation. However, the Pt:Rh ratio is the same as in the absence of CO. In the case of NO adsorption, the metals signals at 300 K are also greatly attenuated due to molecular NO on the surface, but after heating to 600 K, the Pt signal increases almost to the same intensity as for the surface without NO, whereas the Rh signal remains smaller. Consequently, the surface at 600 K appears to contain less Rh (16% vs 28% Rh) after exposure to NO. This behavior was observed for both Rh + Pt and Rh + Pt and suggests that either the byproducts of NO dissociation primarily adsorb on Rh sites or that N and O adsorbates inhibit diffusion of Rh to the cluster surface. In any case, the surface composition of the Pt-Rh clusters is not dramatically changed by NO and CO adsorption, and there is no evidence for Rh enrichment in the presence of adsorbed oxygen. Notably, a nitrogen signal was not detected after heating NO on the Rh clusters to 600 K, suggesting that the nitrogen atoms are subsurface. Discussion NO and CO Adsorption and Dissociation. The surface chemistry observed on Rh, Pt, and Pt-Rh clusters supported on TiO2(110) is qualitatively similar to that on the corresponding single-crystal surfaces in terms of the products that are formed. However, the clusters in general show greater activity than the single-crystal (111) surfaces. For example, 25% of the adsorbed CO undergoes dissociation on the 4 ML Rh clusters, as demonstrated by the high temperature desorption peaks between 800 and 900 K. On Rh(111) surfaces, CO dissociation does not occur,71 but CO does dissociate on stepped Rh surfaces such as (755), (331), and (210).71,72 It is not surprising that the Rh clusters are more defective than close-packed Rh surfaces since metal clusters deposited at room temperature are known to have a higher fraction of step and kink sites compared to the bulk surfaces.83 Another difference between the single-crystal surfaces and the Rh clusters is that oxygen from the titania support participates in the oxidation of atomic carbon to CO, which is observed in high temperature desorption states between 800 and 900 K. Previous studies also show that highly reducible oxide supports such as titania16,17 and ceria84 are capable of oxidizing atomic carbon on metal clusters. However, no oxidation to CO2 was observed here, in contrast to Rh clusters on ceria thin films.85 On the 4 ML Pt clusters, CO does not dissociate, which is consistent with the lack of C-O bond breaking either on Pt(111)86 or stepped Pt surfaces such as (211).87 For NO reaction on 4 ML Rh and 4 ML Pt clusters, ∼70% of the adsorbed NO is dissociated, indicating that the clusters are slightly more active than the single-crystal metal surfaces. However, the Pt clusters are significantly more active for NO dissociation than Pt(111), suggesting that the cluster surfaces are more like the open or stepped Pt faces. On Pt(111), the 2% of NO that dissociates is attributed to reaction at defect sites.35 In contrast, 50% dissociation occurs on Pt(100),35 compared to 15% on (110)35 and 30% on stepped (111) surfaces.88 On Rh(111)53 and (100),33 approximately 50% of the NO is dissociated at saturation coverage. Since complete dissociation is observed
Park et al. on Rh at lower coverages, it has been proposed that open sites are required for NO dissociation, which is inhibited at higher coverages by adsorbed nitrogen and oxygen atoms as well as by molecular NO.33,53,58 NO dissociation is also inhibited at high coverage on Pt(100)38 and Pt-Rh surfaces.41 Reaction of NO with CO. In the reaction of NO with CO, the highest yields for the desired CO2 and N2 products are achieved on the bimetallic Pt-Rh clusters. The CO2 production is double that on pure Pt and ∼7 times higher than on pure Rh. Although the total nitrogen production on the 4 ML Rh clusters is ∼35% higher, ∼95% of the adsorbed NO dissociates on the bimetallic surfaces. Notably, the two Pt-Rh bimetallic cluster surfaces (Pt+Rh and Rh+Pt) exhibit nearly identical reactivity to each other for CO, NO, and CO/NO desorption and reaction. This result should be expected given that the LEIS experiments show that the Pt-Rh clusters have very similar surface compositions of 20-30% Rh, regardless of the order of metal deposition. The favorable reactivity on the Pt-Rh surfaces can be explained by the ability of the bimetallic surfaces to provide sites for both NO dissociation and CO adsorption. In comparison, NO is strongly adsorbed on the pure 4 ML Rh clusters to the near exclusion of CO, and on the pure 4 ML Pt clusters, CO adsorption is favored over NO. Thus, only the Pt-Rh surface provides the Rh sites needed for NO dissociation to form atomic oxygen, as well as the Pt sites necessary for CO adsorption so that the CO(a) + O(a) f CO2 reaction can occur. Infrared spectroscopy studies of Pt50Rh50 clusters on SiO2 have shown that CO prefers to adsorb on Pt, whereas NO preferentially adsorbs on Rh,89 and the same preferential adsorption behavior has been proposed for Pt-Rh clusters on Al2O3 based on kinetic modeling.50 The activity of the 4 ML Rh clusters in the reaction of NO with CO is distinctly different from that on single-crystal Rh surfaces since almost no CO2 is produced from the cluster surfaces. The lack of CO2 production could be partly attributed to the fact that NO is strongly bound to Rh and remains on the surface at 500 K, which is the temperature at which CO2 is evolved from the bimetallic surfaces. Due to the preferential adsorption of NO over CO on Rh, the fraction of CO desorbing from the 4 ML Rh clusters is only 9% of that CO yield in the absence of NO, and therefore, it is not surprising that the surface produces virtually no CO2. Even on Rh(111), CO2 production at 440 K from the reaction of NO with CO is inhibited by high coverages of NO since CO2 and NO also desorb at approximately the same temperature.28 In contrast, for the Pt and Pt-Rh clusters, the CO desorption is still 30-40% of the yield in the absence of NO, indicating that a significant amount of CO can still adsorb in the presence of NO, particularly since some of the adsorbed CO reacts to produce CO2. However, the Rh clusters still adsorb a high concentration of NO in the presence of CO because NO competes effectively for surface sites. On the single-crystal Rh surfaces NO adsorption is favored over CO,28,31 but enough CO is present on the surface to allow production of CO2, which is not observed on the Rh clusters. Therefore, it is possible that the clusters bind NO more strongly than the (111) and (110) surfaces due to the greater number of defect sites such as steps and kinks on the clusters. Indeed, the ∼480 K NO desorption temperature from NO/CO overlayers on the 4 ML Rh clusters is in general higher than the 410-450 K observed on the single-crystal surfaces.28,29,53 We also propose that the migration of titania to the surface of the Rh clusters is shutting down activity for the reaction of NO with CO; the topic of encapsulation is addressed in greater detail in the last section of the discussion. There is evidence from the LEIS studies that
Oxide-Supported Bimetallic Clusters the onset of encapsulation for the Rh clusters occurs at 500 K, which is the temperature at which CO2 is produced from the Pt-Rh bimetallic clusters and is also a lower temperature than that of encapsulation on Pt and Pt-Rh clusters. The reaction of NO with CO on the Rh clusters and bulk Rh surfaces are similar in terms of NO dissociation. The 80% NO dissociation on the 4 ML Rh clusters on TiO2 observed in this work is consistent with the reported 60-80% on Rh(111), Rh(110), and Rh clusters supported on R-Al2O3(0001).47 Although NO dissociation is reported to be inhibited by adsorbed nitrogen and oxygen atoms53 as well as adsorbed NO itself, the presence of adsorbed CO does not inhibit dissociation on Rh(111)27,90 or on the 4 ML Rh clusters. However, the adsorbed NO decreases the desorption temperature of CO, resulting in decreased CO2 production because less surface CO is available to react with adsorbed oxygen from NO dissociation. The lack of CO2 production on the Pt clusters compared to Pt-Rh is partially attributed to the decreased NO adsorption on the Pt surface in the presence of CO. For NO coadsorbed with CO on the Pt clusters, almost no NO desorption is observed due to the preferential adsorption of CO. This result is consistent with a number of studies on bulk Pt surfaces that have demonstrated that CO binds more strongly to Pt than NO.28,31,39,89,91 For example, on Pt(100) surfaces, IRAS and TPD studies have shown that at temperatures below that of NO dissociation (∼400 K), surface adsorption is dominated by nonreactive displacement of NO by CO, which eventually poisons the surface toward CO2 production.39 At ambient gas pressures on Pt(100), preadsorbed NO can be displaced by CO, whereas NO cannot be adsorbed on preadsorbed CO due to the higher heat of adsorption for CO compared to NO.39 This is also true under UHV conditions at room temperature for various Pt surfaces.92,93 On silicasupported Pt clusters, CO is adsorbed to the exclusion of NO and is also capable of displacing adsorbed NO; consequently, the reaction of NO with CO cannot occur.89 Furthermore, it has been demonstrated that CO interacts with NO to decrease the binding energy of NO,91 which explains the lower temperature NO desorption state on the 4 ML Pt clusters in the presence of CO. The decreased ability for Pt to dissociate NO compared to Rh also contributes to the lower CO2 yield for the reaction of NO with CO reaction since the concentration of O(a) is limited. On the Rh clusters, the dissociation of NO in the absence of CO is twice as high as on the Pt clusters. Furthermore, the activity for CO + NO reaction on Pt surfaces is directly correlated with the ability of the surface to dissociate NO. For example, Pt(111) has no activity91 whereas the more open Pt(100) and (410) surfaces are active for both CO + NO reaction and NO dissociation.36,38 In fact, NO and CO undergo nearly complete reaction to N2 and CO2 after heating to ∼410 K on Pt(100).40 The activity for the reaction of NO with CO should be considered higher on the Pt clusters, even though the production of N2 is nearly ten times greater on the Rh clusters. The amount of CO2 produced on the Rh clusters is barely detectable above the baseline, and the CO2 yield is three times higher on the Pt clusters. In contrast, the activity for the CO + NO reaction has been reported to be greater on bulk Rh surfaces compared to Pt.28,53,91 One report has suggested that the activity is higher because NO competes more effectively with CO for adsorption sites on Rh;28,31 the strong affinity of CO for Pt prevents NO from adsorbing on the surface, resulting in the lower activity on the Pt surface. However, our studies of the Pt clusters have shown that although little NO desorption is detected compared
J. Phys. Chem. C, Vol. 111, No. 5, 2007 2173 to CO desorption, most of the adsorbed NO (90%) dissociates to produce N2 at 950 K, and the recombinant N2 production in the presence of CO is still ∼70% of that for NO alone. Despite the greater activity for NO dissociation on Rh, the Pt clusters produce more N2 at low temperature. Low temperature N2 formation is attributed to the NO + N(a) f N2 and O(a) reaction, which is the mechanism that has been proposed for low temperature N2 desorption on Rh surfaces.28,53,94 On bulk Pt surfaces, only one N2 desorption state at approximately 500 K is detected and is assigned to recombination of diffusing nitrogen atoms.35,40,55,63 In our studies, two N2 desorption peaks are observed for both the Pt and Rh clusters. The lowtemperature peak has greater intensity and occurs at lower temperature on the Pt clusters compared to Rh clusters. Since adsorbed NO itself is known to inhibit NO dissociation on Rh,33,53,58 the lower temperature for NO desorption observed on Pt, which is presumably due to the stronger bonding of NO to Rh, may provide open sites for the dissociation of NO at lower temperature and subsequent reaction with NO to produce N2. The greater intensity of this peak on Pt suggests that the stronger metal-N bonding on Rh inhibits reaction of nitrogen atoms with adsorbed NO. Although N2O is a reasonable intermediate for NO + N(a) reaction, N2O desorption was not observed on the Rh or Pt-Rh clusters; similarly, N2O is not reported from NO reaction on Pt(111), (100), and (110)35 as well as Rh(111)53 and (100)33 under UHV conditions. However, on the Pt clusters, trace amounts of N2O desorption were detected, which is consistent with a transient N2O intermediate in the production of N2 at low temperature and the greater intensity of this peak on the Pt clusters. The TPD studies for the coadsorption experiments suggest that for the pure and bimetallic clusters, CO and NO adsorption occur at the same type of surface sites, whereas NO dissociation takes place at different sites. On the 4 ML Rh clusters, the intensity of the NO desorption peak in the presence of CO is ∼70% of that for NO alone despite the fact that NO is preferentially adsorbed on Rh. Furthermore, the NO peak intensity for the other three cluster surfaces is only ∼10% of that for NO alone. The amount of N2 produced at high temperature from NO dissociation on the 4 ML Rh clusters is the same with or without CO, whereas coadsorption of NO and CO decreases the N2 production by only 30% on Pt and ∼20% on Pt-Rh clusters even though the NO desorption is significantly diminished. Similarly, coadsorbed CO does not inhibit NO dissociation on Rh(111).90 Studies have reported that NO adsorbs initially on bridge sites and at top sites at higher coverages on Pt(111)61,63,91 and Pt(110),64 whereas CO initially adsorbs on top sites on Pt(111).95 For Pt(100), NO/CO overlayers have CO in both bridge and top sites;39 although the adsorption site of NO is more difficult to identify experimentally, density functional theory calculations suggest bridge sites.96 On Rh(111), CO occupies the bridge sites at high coverage and top sites at low coverage,90 but CO is preferentially adsorbed at top sites in the presence of NO.90 NO is found at bridge sites at all coverages on Rh(111).74,97 In general, NO adsorption at bridge sites and CO adsorption at top sites are favored on both bulk Pt and Rh surfaces, particularly in NO/CO overlayers. Therefore, it is possible CO adsorption does not interfere with NO dissociation because CO is primarily found at top sites whereas NO is primarily found at bridge sites, which are also the most active for NO dissociation.90 Average Properties of Pt-Rh Clusters. The dissociation of NO and CO on Pt-Rh clusters exhibits activity that is intermediate between pure Pt and pure Rh, suggesting that Pt
2174 J. Phys. Chem. C, Vol. 111, No. 5, 2007 and Rh atoms in the bimetallic clusters retain their individual properties so that the bimetallic surface shows average properties. On the Pt clusters, virtually no CO dissociation occurs, whereas on Rh, 25% of the CO dissociates; consequently the Pt-Rh clusters exhibit C-O bond breaking that is intermediate between that of Pt and Rh, dissociating 10% of the CO. In the reaction of NO, the Rh clusters are more effective at breaking N-O bonds than the Pt clusters, which produce only 30% of the recombinant N2 that desorbs from the more active Rh clusters. The Pt-Rh clusters produce 80-85% of the recombinant N2 yield for the Rh clusters. Since this yield is higher than the 65% that would be expected from simply averaging the Pt and Rh properties, we attribute the greater dissociation activity on Pt-Rh to the fact that NO dissociation on the Rh clusters at a saturation coverage of NO is limited by the number of unoccupied surface sites. On the Pt-Rh surfaces, NO is effectively dissociated at Rh sites, and the atomic nitrogen and oxygen byproducts could spill over to neighboring Pt sites. A model proposed by Leclercq and co-workers suggests that on Pt-Rh/Al2O3 clusters NO dissociation at a Rh site requires a vacant neighboring Pt site.57 For CO and NO coadsorption on the Pt-Rh clusters, the bimetallic clusters also show characteristics intermediate between Pt and Rh in terms of the peak shapes for CO and NO desorption. This is particularly evident for CO desorption on bimetallic clusters with varying metal compositions since CO desorbs at different temperatures from pure Pt and pure Rh. The low temperature peak at 470 K appears to be associated with desorption on Rh, whereas the 570 K peak is from desorption on Pt. Although desorption on the pure Rh and pure Pt clusters do not occur at exactly these temperatures, CO desorbs from Rh at a lower temperature than from Pt, and the ratio of the higher and lower temperature desorption peaks changes accordingly with the composition of the bimetallic clusters. Similarly, for NO desorption, both the 410 K peak associated with desorption from Pt and the 485 K peak from desorption from Rh are observed on the Pt-Rh clusters. In addition, previous studies of Pt-Rh clusters and singlecrystal alloy surfaces have reported that the bimetallic surfaces exhibit properties intermediate between that of pure Pt and pure Rh. For example, the main effect of Pt in the Pt10Rh90(111) alloy surface is to dilute the surface concentration of Rh atoms. The resulting surface is enriched in Pt with respect to the bulk (30% Pt) and exhibits O2 desorption and NO dissociation properties that are average between that of Pt(111) and Rh(111).41,98,99 A study of Pt-Rh clusters on alumina reports that NO adsorption occurs on Rh sites, whereas CO adsorption occurs on Pt, and that these Pt and Rh sites retain their individual adsorption characteristics.57 In this case, the average activity of the Pt-Rh catalyst is decreased because the less active Pt atoms dilute the activity of the more active Rh.57 However, on the supported Rh clusters studied here, the Rh surface has almost no activity for CO oxidation with NO, whereas the Pt-Rh bimetallic surface shows the greatest activity. As discussed in the previous section, the higher activity can be explained by the ability of the bimetallic surface to both adsorb CO and dissociate NO on pure Pt and pure Rh sites, respectively. Thus, the greater observed activity is still consistent with additive rather than synergistic properties of the Pt-Rh surfaces. Some studies have suggested that electronic modifications of Pt and Rh atoms could occur due to the presence of the second metal;41,49 although synergistic bimetallic interactions do not
Park et al. need to be invoked to interpret the data presented in this work, such interactions are not inconsistent with our results. Encapsulation Effects. A major difference between reaction on the clusters and bulk metal surfaces is that encapsulation of the clusters by titania occurs upon heating. For Rh clusters, significant encapsulation has occurred at 600 K, and for the Pt and Pt-Rh clusters, the temperature for encapsulation is approximately 100 K higher. In all cases, the dissociation of NO and production of CO2 as well as NO and CO desorption occur below 600 K. It is therefore not surprising that the lowtemperature reaction and desorption events are similar to those on the bulk surfaces. However, the recombinant desorption of N2 and CO is observed at temperatures at which cluster encapsulation has occurred, and temperatures are significantly higher than those on the single-crystal surfaces. For example, recombinant CO desorption from the 4 ML Rh cluster is observed at 800 and 870 K compared to 600-625 K on Rh(210)72 and ceria-supported Rh clusters.85 Recombinant N2 desorption is detected around 930 K on all four cluster surfaces, whereas the desorption temperature is much lower on bulk Rh and Pt surfaces at 700-800 and 500 K, respectively33,53,63 The higher desorption temperatures are not due to spillover and desorption from the titania surface. It is unlikely that atomic carbon will migrate from the clusters to the titania surface since carbon forms stronger bonds with Pt and Rh compared to Ti.100 Nitrogen migration from the clusters to the titania surface is thermodynamically favorable, given that TPD experiments show that nitrogen desorbs from Rh and Pt at temperatures as low as 500 K, whereas nitrogen does not desorb from TiN until ∼1300 K;101 However, the absence of a nitrogen signal in the LEIS experiments for NO on Rh clusters heated to 600 K suggests that the nitrogen atoms are subsurface. Furthermore, the detection of a nitrogen signal by XPS indicates that the nitrogen atoms are not covered by a thick titania layer. We propose that the higher desorption temperatures on the cluster surfaces are related to cluster encapsulation by titania, which could cover the nitrogen and carbon atoms on the cluster surface, resulting in more energy required for atom diffusion and recombination. XPS studies of NO dissociation on the 4 ML Rh clusters demonstrate that at 600 K and above the N(1s) binding energy of 396.85 eV is consistent with both atomic nitrogen on a metal surface102 and atomic nitrogen bound to the metal site on a metal oxide surface.103 The role of lattice oxygen in the oxidation of carbon to CO might also be enhanced by cluster encapsulation, which would generate more lattice oxygen on the cluster surfaces. However, the participation of lattice oxygen in the oxidation of surface species such as CO has been observed for Cu clusters on TiO2 even though Cu does not become encapsulated.17 In the reaction of NO on bulk Pt surfaces, O2 desorption is observed around 850 K,63 but is not observed at all on the Pt clusters. The lack of oxygen desorption could be explained by the formation of the reduced titania species on the Pt clusters which might be expected to bind oxygen as strongly as the TiO2 surface itself. Alternatively, the absence of O2 desorption could be attributed to migration of atomic oxygen from the cluster surfaces to the titania support, where oxygen vacancies in this reduced surface can be filled. O2 desorption is not expected to be observed on the Rh clusters since it occurs at a much higher temperature (1200-1300 K) on Rh(111);28 the surface was not heated to these elevated temperatures because TiO2(110)
Oxide-Supported Bimetallic Clusters undergoes significant oxygen loss and is no longer stable toward reconstruction.11 Summary The higher activity of pure Rh and Pt clusters for CO and NO dissociation compared to the (111) surfaces is attributed to the greater number of active sites, such as steps and kinks, on the cluster surfaces. Specifically, Rh(111) does not dissociate CO whereas the Rh clusters surfaces are capable of C-O bond dissociation. Likewise, the Pt clusters surfaces dissociate NO, whereas the Pt(111) surface does not. Pt-Rh bimetallic clusters prepared by sequential deposition of the two metals exhibited identical surface compositions and reaction chemistry regardless of the order of metal deposition. Thus, both Pt and Rh atoms have the ability to diffuse to the cluster surface at room temperature in order to achieve thermodynamically favorable surface compositions. The higher CO2 yields in the reaction of NO with CO on the Pt-Rh bimetallic clusters are attributed to the ability to readily dissociate NO on the Rh sites as well as to adsorb CO at Pt sites. On pure Rh clusters, almost no CO2 is produced because the strong adsorption of NO blocks sites for CO adsorption, and encapsulation by TiOx may also diminish the activity of the Rh clusters. On the pure Pt clusters, the CO2 yield is limited both by the preferential adsorption of CO over NO and by the decreased ability of Pt to dissociate NO. Acknowledgment. We gratefully acknowledge financial support from the National Science Foundation under a CAREER Award (CHE 0133926), the Department of Energy, Office of Basic Energy Sciences under a DOE/EPSCOR grant (DE-FG0201ER45892), and the U.S. Army Research Office (W911NF05-1-0184). References and Notes (1) Sinfelt, J. H. Acc. Chem. Res. 1977, 10, 15. (2) Sinfelt, J. H. Bimetallic Catalysts. DiscoVeries, Concepts, and Applications; John Wiley and Sons: New York, 1983. (3) Taylor, K. C. Catal. ReV.-Sci. Eng. 1993, 35, 457. (4) Wouda, P. T.; Nieuwenhuys, B. E.; Schmid, M.; Varga, P. Surf. Sci. 1996, 359, 17. (5) Wouda, P. T.; Schmid, M.; Hebenstreit, W.; Varga, P. Surf. Sci. 1997, 388, 63. (6) Hebenstreit, E. L. D.; Hebenstreit, W.; Schmid, M.; Varga, P. Surf. Sci. 1999, 441, 441. (7) Platzgummer, E.; Sporn, M.; Koller, R.; Forsthuber, S.; Schmid, M.; Hofer, W.; Varga, P. Surf. Sci. 1999, 419, 236. (8) Moest, B.; Helfensteyn, S.; Deurinck, P.; Nelis, M.; van der Gon, A. W. D.; Brongersma, H. H.; Creemers, C.; Nieuwenhuys, B. E. Surf. Sci. 2003, 536, 177. (9) Moest, B.; Wouda, P. T.; van der Gon, A. W. D.; Langelaar, M. C.; Brongersma, H. H.; Nieuwenhuys, B. E.; Boerma, D. O. Surf. Sci. 2001, 473, 159. (10) Baraldi, A.; Giacomello, D.; Rumiz, L.; Moretuzzo, M.; Lizzit, S.; de Mongeot, F. B.; Paolucci, G.; Comelli, G.; Rosei, R.; Nieuwenhuys, B. E.; Valbusa, U.; Kiskinova, M. P. J. Am. Chem. Soc. 2005, 127, 5671. (11) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (12) Kudo, A.; Steinberg, M.; Bard, A. J.; Campion, A.; Fox, M. A.; Mallouk, T. E.; Webber, S. C.; White, J. M. J. Catal. 1990, 125, 565. (13) Aritani, H.; Tanaka, T.; Akasaka, N.; Funabiki, T.; Yoshida, S.; Gotoh, H.; Okamoto, Y. J. Catal. 1997, 168, 412. (14) Chen, M. S.; Goodman, D. W. Science. 2004, 306, 252. (15) Choudhary, T. V.; Goodman, D. W. Top. Catal. 2002, 21, 25. (16) Zhou, J.; Ma, S.; Kang, Y. C.; Chen, D. A. J. Phys. Chem. B 2004, 108, 11633. (17) Varazo, K.; Parsons, F. W.; Ma, S.; Chen, D. A. J. Phys. Chem. B 2004, 108, 18274. (18) Haruta, A. Chem. Rec. 2003, 3, 75. (19) Goodman, D. W. Catal. Lett. 2005, 99, 1.
J. Phys. Chem. C, Vol. 111, No. 5, 2007 2175 (20) Chen, M.; Cai, Y.; Yan, Z.; Goodman, D. W. J. Am. Chem. Soc. 2006, 128, 6341. (21) Min, B. K.; Wallace, W. T.; Goodman, D. W. Surf. Sci. 2006, 600, L7. (22) Hutchings, G. J.; Haruta, M. Appl. Catal. A-Gen. 2005, 291, 2. (23) Beutler, A.; Lundgren, E.; Nyholm, R.; Andersen, J. N.; Setlik, B.; Heskett, D. Surf. Sci. 1997, 371, 381. (24) Schubert, M. M.; Hackenberg, S.; van Veen, A. C.; Muhler, M.; Plzak, V.; Behm, R. J. J. Catal. 2001, 197, 113. (25) Permana, H.; Ng, K. Y. S.; Peden, C. H. F.; Schmeig, S. J.; Lambert, D. K.; Belton, D. N. J. Catal. 1996, 164, 194. (26) Rienks, E. D. L.; Bakker, J. W.; Baraldi, A.; Carabineiro, S. A. C.; Lizzit, S.; Weststrate, C. J.; Nieuwenhuys, B. E. Surf. Sci. 2002, 516, 109. (27) Hopstaken, M. J. P.; Van Gennip, W. J. H.; Niemantsverdriet, J. W. Surf. Sci. 1999, 69, 433-435. (28) Root, T. W.; Schmidt, L. D.; Fisher, G. B. Surf. Sci. 1985, 150, 173. (29) Schmatloch, V.; Kruse, N. Surf. Sci. 1992, 269/270, 488. (30) Peden, C. H. F.; Goodman, D. W.; Blair, D. S.; Berlowitz, P. J.; Fisher, G. B.; Oh, S. H. J. Phys. Chem. 1988, 92, 1563. (31) Shelef, M.; Graham, G. W. Catal. ReV.-Sci. Eng. 1994, 36, 433. (32) Belton, D. N.; Schmieg, S. J. J. Catal. 1993, 144, 9. (33) Hopstaken, M. J. P.; Niemantsverdriet, J. W. J. Phys. Chem. B 2000, 104, 3058. (34) Banholzer, W. F.; Masel, R. I. Surf. Sci. 1984, 137, 339. (35) Gorte, R. J.; Schmidt, L. D.; Gland, J. L. Surf. Sci. 1981, 109, 367. (36) Fink, T.; Dath, J. P.; Bassett, M. R.; Imbihl, R.; Ertl, G. Surf. Sci. 1991, 245, 96. (37) Brandt, M.; Zagatta, G.; Boewering, N.; Heinzmann, U. Surf. Sci. 1997, 385, 346. (38) Zagatta, G.; Mueller, H.; Wehmeyer, O.; Brandt, M.; Boewering, N.; Heinzmann, U. Surf. Sci. 1994, 307-309, 199. (39) Miners, J. H.; Gerdner, P.; Woodruff, D. P. Surf. Sci. 2003, 547, 355. (40) Lesley, M. W.; Schmidt, L. D. Surf. Sci. 1985, 155, 215. (41) Ng, K. Y.; Belton, D. N.; Schmieg, S. J.; Fisher, G. B. J. Catal. 1994, 146, 394. (42) Wolf, R. M.; Siera, J.; van Delft, F. C. M.; Nieuwenhuys, B. E. Farady. Discuss. Chem. Soc. 1989, 87, 275. (43) Siera, J.; Rutten, F.; Nieuwenhuys, B. E. Catal. Today. 1991, 10, 353. (44) Rutten, F. J. M.; Nieuwenhuys, B. E.; McCoustra, M. R. S.; Chesters, M. A.; Hollins, P. J. Vac. Sci. Technol. A 1997, 15, 1619. (45) Hirano, H.; Yamada, T.; Tanaka, K. I. Surf. Sci. 1992, 262, 97. (46) Tanaka, K. I.; Sasahara, A. J. Mol. Catal. A 2000, 155, 13. (47) Altman, E. I.; Gorte, R. J. J. Catal. 1988, 113, 185. (48) Granger, P.; Malfoy, P.; Leclercq, G. J. Catal. 2004, 223, 142. (49) Parvulescu, V. I.; Grange, P.; Delmon, B. Catal. Today. 1998, 46, 233. (50) Granger, P.; Lecomte, J. J.; Dathy, C.; Leclercq, L.; Mabilon, G.; Prigent, M.; Leclercq, G. Investigation on the Role of Rhodium on the Kinetics of the Oxidation of CO by NO over Pt-Rh Catalysts. In Catalysis and AutomotiVe Pollution Control IV; Elsevier: New York, 1998; Vol. 116, pp 419. (51) Granger, P.; Dujardin, C.; Paul, J. F.; Leclercq, G. J. Mol. Catal. A-Chem. 2005, 228, 241. (52) Oh, S. H.; Eickel, C. C. J. Catal. 1991, 128, 526. (53) Root, T. W.; Schmidt, L. D.; Fisher, G. B. Surf. Sci. 1983, 134, 30. (54) Park, Y. O.; Banholzer, W. F.; Masel, R. I. Surf. Sci. 1985, 155, 341. (55) Lambert, R. M.; Comrie, C. M. Surf. Sci. 1974, 46, 61. (56) Peden, C. H. F.; Belton, D. N.; Schmieg, S. J. J. Catal. 1995, 155, 204. (57) Granger, P.; Lecomte, J. J.; Dathy, C.; Leclercq, L.; Leclercq, G. J. Catal. 1998, 175, 194. (58) Brown, W. A.; King, D. A. J. Phys. Chem. B 2000, 104, 2578. (59) Brown, W. A.; Sharma, R. K.; King, D. A.; Haq, S. J. Phys. Chem. 1996, 100, 12559. (60) Nelin, C. J.; Bagus, P. S.; Behm, J.; Brundle, C. R. Chem. Phys. Lett. 1984, 105, 58. (61) Hayden, B. E. Surf. Sci. 1983, 131, 419. (62) Granger, P.; Delannoy, L.; Leclercq, L.; Leclercq, G. J. Catal. 1998, 177, 147. (63) Gland, J. L.; Sexton, B. A. Surf. Sci. 1980, 94, 355. (64) Gorte, R. J.; Gland, J. L. Surf. Sci. 1981, 102, 348. (65) Zhou, J.; Ma, S.; Kang, Y. C.; Chen, D. A. Surf. Sci. 2004, 562, 113. (66) Zhou, J.; Chen, D. A. Surf. Sci. 2003, 527, 183. (67) Reddic, J. E.; Zhou, J.; Chen, D. A. Surf. Sci. 2001, 494, L767. (68) Illingworth, A.; Zhou, J.; Ozturk, O.; Chen, D. A. J. Vac. Sci. Technol. B 2004, 22, 2552.
2176 J. Phys. Chem. C, Vol. 111, No. 5, 2007 (69) Henderson, M. A.; Epling, W. S.; Perkins, C. L.; Peden, C. H. F.; Diebold, U. J. Phys. Chem. B 1999, 103, 5328. (70) Diebold, U.; Anderson, J. F.; Ng, K. O.; Vanderbilt, D. Phys. ReV. Lett. 1996, 77, 1322. (71) Castner, D. G.; Somorjai, G. A. Surf. Sci. 1979, 83, 60. (72) Rebholz, M.; Prins, R.; Kruse, N. Surf. Sci. 1991, 259, L797. (73) Schmatloch, V.; Jirka, I.; Kruse, N. J. Chem. Phys. 1994, 100, 8471. (74) Borg, H. J.; Reijerse, J.; Vansanten, R. A.; Niemantsverdriet, J. W. J. Chem. Phys. 1994, 101, 10052. (75) Belton, D. N.; Dimaggio, C. L.; Ng, K. Y. S. J. Catal. 1993, 144, 273. (76) Park, J. B.; Ratliff, J. S.; Ma, S.; Chen, D. A. Surf. Sci. 2006, 600, 2913. (77) Berko´, A.; Mensesi, G.; Solymosi, F. Surf. Sci. 1997, 372, 202. (78) Berko´, A.; Ulrych, I.; Prince, K. C. J. Phys. Chem. B 1998, 102, 3379. (79) Pesty, F.; Steinru¨ck, H. P.; Madey, T. E. Surf. Sci. 1995, 339, 83. (80) Dulub, O.; Hebenstreit, W.; Diebold, U. Phys. ReV. Lett. 2000, 84, 3646. (81) Jennison, D. R.; Dulub, O.; Hebenstreit, W.; Diebold, U. Surf. Sci. 2001, 492, L677. (82) Hirano, H.; Yamada, T.; Tanaka, K.; Siera, J.; Nieuwenhuys, B. E. Surf. Sci. 1989, 222, L804. (83) Yudanov, I. V.; Sahnoun, R.; Neyman, K. M.; Rosch, N.; Hoffmann, J.; Schauermann, S.; Johanek, V.; Unterhalt, H.; Rupprechter, G.; Libuda, J.; Freund, H. J. J. Phys. Chem. B 2003, 107, 255. (84) Bunluesin, T.; Gorte, R. J.; Graham, G. W. Appl. Catal. B 1998, 15, 107. (85) Putna, E. S.; Gorte, R. J.; Vohs, J. M.; Graham, G. W. J. Catal. 1998, 178, 598.
Park et al. (86) Hopster, H.; Ibach, H. Surf. Sci. 1978, 77, 109. (87) Tokuhisa, H.; Crooks, R. M. Langmuir 1996, 13, 5608. (88) Gland, J. L. Surf. Sci. 1978, 71, 327. (89) van Slooten, R. F.; Nieuwenhuys, B. E. J. Catal. 1990, 122, 429. (90) Root, T. W.; Fisher, G. B.; Schmidt, L. D. J. Chem. Phys. 1986, 85, 4687. (91) Gorte, R. J.; Schmidt, L. D. Surf. Sci. 1981, 111, 260. (92) Gardner, P.; Martin, R.; Tushaus, M.; Bradshaw, A. M. Surf. Sci. 1992, 270, 405. (93) Imbihl, R.; Fink, T.; Krischer, K. J. Chem. Phys. 1992, 96, 6236. (94) Campbell, C. T.; White, J. M. Appl. Surf. Sci. 1978, 1, 347. (95) Hayden, B. E.; Bradshaw, A. M. Surf. Sci. 1983, 125, 787. (96) Eichler, A.; Hafner, J. J. Catal. 2001, 204, 118. (97) Root, T. W.; Fisher, G. B.; Schmidt, L. D. J. Chem. Phys. 1986, 85, 4679. (98) Beck, D. D.; Dimaggio, C. L.; Fisher, G. B. Surf. Sci. 1993, 297, 293. (99) Fisher, G. B.; Dimaggio, C. L.; Beck, D. D.; King, D. A.; Huang, C. G.; Wang, D.; Bartholomew, C. H.; Joyner, R. W.; Guo, X. X.; Kaspar, J.; Tanaka, K. I.; Nieuwenhuys, B. E.; Bertolini, J. C. Stud. Surf. Sci. Catal. 1993, 75, 383. (100) CRC Handbook of Chemistry and Physics; 78th ed.; Lide, D. R., Ed.; CRC Press: New York, 1997. (101) Schulberg, M. T.; Allendorf, M. D.; Outka, D. A. J. Vac. Sci. Technol. A. 1996, 14, 3228. (102) Saito, T.; Imamura, M.; Matsubayashi, N.; Furuya, K.; Kikuchi, T.; Shimada, H. J. Electron Spectrosc. 2001, 119, 95. (103) Overbury, S. H.; Mullins, D. R.; Huntley, D. R.; Kundakovic, L. J. Catal. 1999, 186, 296.