9336
J. Phys. Chem. C 2008, 112, 9336–9345
Growth and Characterization of Rh and Pd Nanoparticles on Oxidized and Reduced CeOx(111) Thin Films by Scanning Tunneling Microscopy Jing Zhou, A. P. Baddorf, D. R. Mullins, and S. H. Overbury* Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6201 ReceiVed: NoVember 26, 2007; ReVised Manuscript ReceiVed: April 3, 2008
The growth and structure of Rh and Pd nanoparticles on vapor-deposited ceria thin films grown epitaxially on Ru(0001) were investigated by scanning tunneling microscopy as a function of coverage, postdeposition annealing temperatures, as well as ceria oxidation states. Both metals grow as three-dimensional nanoparticles on the fully oxidized CeO2(111) thin films at room temperature. At low coverage, Rh particles preferentially decorate step edges on ceria, forming particles with smaller average size than those on the terraces. With increasing coverage, the number of Rh particles increases until near 2.3 monolayers, where the Rh particles cover most of the ceria surface and the particle size becomes relatively uniform. Larger Rh or Pd particles can be prepared by annealing the surface after deposition at 300 K; however, the particle size distribution becomes broader during this process. Similar growth behavior was observed on the reduced ceria surfaces. It is concluded that metal particle size or morphology is not responsible for previously reported differences in surface chemistry observed when Rh is deposited on reduced CeOx compared to fully oxidized CeO2. Instead, it is proposed that the enhanced dissociation occurs at the interface between the metal and the reduced support, which, coupled with rapid O diffusion, may lead to high dissociation fractions. 1. Introduction Ceria-supported precious metal catalysts such as Rh, Pd, and Pt are widely used in many important applications, including three-way automobile emission-control catalysis and fuel cells, owing to the peculiar redox properties and oxygen storage capacity of ceria as well as the synergistic effect between ceria supports and metals.1–4 It is generally known that metal particles supported on oxides can have very high surface area and highly active sites, the oxide may enter into reaction pathways, and the metal particles can exhibit different chemical reactivity depending upon their size or structure.5,6 To understand metal nanoparticles supported on single crystal or highly oriented oxide films, detailed structure can be achieved with scanning probe techniques and correlated with chemical properties using temperature-programmed reaction (TPR), desorption (TPD), and other ultrahigh vacuum (UHV)-based techniques. For CeO2 oxide surfaces, only Au growth on (111) surfaces has been studied by scanning probe microscopy,7,8 although there has been considerable work on metal-free CeO2 surfaces.9–15 Following our previous work on surface chemistry by Rh and Pd supported on CeO2(111) surfaces,16–22 the current work aims to achieve a fundamental understanding of the nucleation and growth of Rh and Pd metal particles on ceria using scanning tunneling microscopy (STM). Research has shown that ceria-supported metal particles can exhibit unique reactivity dependent on the extent of reduction of the ceria.16–31 For example, CO adsorbs nondissociatively on Rh nanoparticles deposited on fully oxidized ceria films. However, if the ceria is partially reduced, CO dissociates on the supported Rh particles and produces atomic C that can be observed by core-level photoemission.19,20 Similarly, NO dissociation is also enhanced when Rh particles are supported on reduced CeO2(111).18 * Corresponding author. E-mail:
[email protected].
The mechanism for the enhanced dissociation that occurs when Rh is supported on highly reduced CeO2 is unclear.19,20 Several hypotheses have been advanced, including the effects of electron transfer from the reduced ceria to the Rh particles, structural variations in the Rh particles on the reduced ceria, variation in activity at the metal-support interface, or O spillover from the Rh to the ceria.16,18 Coverage-dependent studies in NO adsorption have suggested that additional oxygen spillover from Rh onto reduced ceria cannot explain the enhancement. Nucleation, growth, morphology, and stability of Rh particles might be expected to depend upon the presence of oxygen defects or other structural variation due to the presence of surface Ce3+ cations. Resulting variations in structure or morphology of the Rh particles may in turn affect the reactivity toward adsorbate dissociation. For example, it has been established that NO dissociation on Rh occurs preferentially on rough, high-index facets compared to smoother low-index faces, 32 the density of which will vary on supported particles depending on size, shape, and orientation. Similarly, although CO does not dissociate on low-index Rh surfaces,33,34 there is evidence that some dissociation may occur on stepped surfaces.35,36 These results suggest that variation in crystallographic facets or of low coordinate sites on supported particles may lead to variation in dissociation activity. This explanation has been invoked to explain the particle size dependence of CO dissociation on Rh particles supported on aluminum oxide37 and for Pt particles supported on TiO2(110).6 To date, the mechanism by which the reduction of cerium oxide may affect the reactivity of supported particles is still unresolved, partly because of the lack of structural information about the CeO2 surface and of the supported particles. In this paper, we report our studies of the growth of Rh and Pd metal particles deposited on CeO2(111) and reduced CeOx(111) thin films. Coverage-dependent particle sizes are measured after deposition at 300 K and after annealing to higher temperatures. We report that, for comparable coverages and
10.1021/jp711198c CCC: $40.75 2008 American Chemical Society Published on Web 05/29/2008
Rh and PD Nanoparticles on CeOx(111) Thin Films deposition temperatures, the Rh particle size distributions are nearly independent of the oxidation state of the CeO2 support. The results allow us to eliminate variation in the Rh particle size or morphology as a factor responsible for the pronounced difference in surface chemistry observed for Rh supported on reduced CeOx compared to fully oxidized CeO2. Alternative explanations are offered. 2. Experimental Section Experiments were performed in a multichamber UHV system that consisted of a central transfer chamber connecting a load lock, a surface analysis and preparation chamber, and a variabletemperature Omicron STM chamber.38 Sample preparation was conducted in the analysis chamber at a pressure of less than 2 × 10-10 Torr. The analysis chamber was equipped with a sputter gun, Ce, Rh, and Pd electron beam evaporation sources with flux monitors, low-energy electron diffraction (LEED) optics, and Auger electron spectroscopy (AES) using a cylindrical mirror analyzer detector. A Ru(0001) single crystal was used as a substrate for the growth of ceria thin films. Ru(0001) with a diameter of 10 mm and a thickness of 1.5 mm was held in thermal contact to a Omicron style tantalum single plate using Ta straps. The Ru crystal was heated by electron bombardment from a tungsten filament. The reported temperatures were based on temperature calibrations of the sample heater at various power settings with a thermal couple (type K) spot-welded at the center of the solid Ta plate without the crystal. The crystal was cleaned using procedures described in detail elsewhere.39 Since deposition of ceria was performed in ambient oxygen, the surface used for the growth of films contained adsorbed oxygen and exhibited a Ru(0001)-p(2 × 2)O LEED pattern.8 High-resolution STM images also confirm the formation of the Ru(0001)-p(2 × 2)O surface (data not shown). Fully oxidized CeO2 thin films were grown on Ru(0001) by deposition of Ce with a flux of ∼0.1 monolayers/min from an electron-beam evaporator in 2 × 10-7 Torr O2 at 700 K and subsequently annealing the surface to 1000 K for 3 min.28,40 The CeO2(111) thin films were estimated to be 2-3 nm thick.22,29,40 Reduced CeOx(111) thin films can be grown by decreasing the oxygen pressure during Ce deposition. The degree of ceria reduction was calibrated by growing ceria thin films under the same conditions in a separate UHV apparatus, where cerium oxidation states were determined by X-ray photoelectron spectroscopy (XPS).22,28 After film growth, LEED patterns were always checked to assess the long-range order of the films. AES spectra using a primary electron beam energy of 3 KeV were recorded to check the coverage and the cleanliness of the films. The atomic structures of the ceria films were imaged using STM. Rh and Pd were vapor-deposited onto the ceria films at 300 K using water-cooled, e-beam heated evaporators at rates of 0.5 and 0.2 ML/min, respectively. The metal coverage is expressed in monolayers (ML) where one monolayer Rh or Pd are defined as 1.6 × 1015 and 1.5 × 1015 atoms/cm2, respectively. During the metal deposition, the flux readout from the evaporator was monitored to ensure reproducible results. The total Rh or Pd coverage was calculated in the basis of the STM measurements of the diameter and height of the metal particles, as described elsewhere,41,42 which can overestimate the metal coverage as a result of tip convolution effects. The size and density of Rh and Pd particles were examined as a function of metal coverage, postdeposition annealing temperatures, as well as ceria oxidation states using STM. When postannealing, the surface was heated to the desired temperature, but not held at
J. Phys. Chem. C, Vol. 112, No. 25, 2008 9337 a temperature for more than a few seconds. W tips were used for all the STM experiments, which were etched in 2 M NaOH solution and subsequently annealed by electron-bombardment in vacuum. All the STM images were collected using a constant current mode (sample bias: 3.0-4.0 V; tunneling current: 0.5-1.0 nA) at room temperature. 3. Results 3.1. Growth of CeO2 Films. STM images of ceria films epitaxially grown on Ru(0001) show flat and well-ordered films (Figure 1a). The image reveals that the film completely covers the Ru substrate. The displayed diffraction spots in the LEED pattern (inset, Figure 1a) are from the ceria p(1.4 × 1.4) LEED pattern. After deposition of 2-3 nm films, no LEED pattern from the Ru substrate was observed. The indexing employed for the observed spots is based on comparison with those from the Ru(0001), as described previously.8,9,22 This comparison also indicates that the crystallographic direction of the epitaxial ceria thin film is in alignment with the principle azimuth direction of the Ru(0001). The large-scale STM image in Figure 1a shows that the CeO2(111) film consists of atomically flat terraces that are 50-100 nm wide. The measured height between neighboring terraces on the ceria surface is about 0.3 nm, consistent with the bulk 0.313 nm spacing of O-Ce-O trilayers in the CeO2(111) fluorite bulk structure.8,43 A higher magnification image is shown in Figure 1b with atomic resolution. With the tunnelling conditions (3-4 V, 0.5-1.0 nA) used in the study, biased to image unoccupied states, it is likely that the Ce atoms are the bright features seen in the image.8 The measured distance between two adjacent Ce atoms is about 0.4 nm, agreeing with the 0.383 nm expected surface spacing. Besides the Ce cations, Figure 1b also reveals the presence of a low density of surface point defects. Although surface defects have been previously attributed to missing oxygen atoms on the surface of the film,8,10–12 the possibility of the defects formed due to missing Ce atoms or surface adsorbates such as O/O2 and OH/H2O should not be ruled out. Reduced ceria films also grow as flat films with moderately wide terraces. An STM image of a reduced film is shown in Figure 1c. This film was estimated to have 50% Ce4+ (designated CeO1.75) according to the XPS measurements from the ceria film grown under similar conditions in a different UHV apparatus.22,29 The LEED pattern and the STM images at the scale of Figure 1a (data not shown) do not clearly distinguish the reduced ceria film from the fully oxidized one. Specifically, the appearance of steps and terraces are comparable. At the higher resolution of Figure 1c, the ordered CeO2 lattice is still observed as in Figure 1b, indicating a comparable structure. In some regions, the bright Ce atoms are visible and show clear lattice structure, but there are dark regions where the Ce atoms are less bright or not visible. We associate these regions with oxygen vacancies or disruption of the surface oxygen caused by the reduction, which modulates the brightness of the underlying Ce . The reduced ceria films are more defective, exhibiting point and line defects. The number of the surface defects and the surface roughness increase with the degree of ceria film reduction, with root-mean-square (rms) roughness of 0.024, 0.029, and 0.033 nm for surfaces of CeO2, CeO1.75, and CeO1.60, respectively. For greater reduction (data not shown) traces of the CeO2 lattice are still visible but with greater disorder. 3.2. Rh Growth on CeO2(111): Dependence on Rh Coverage. The growth of Rh on the fully oxidized CeO2(111) surface was first studied at 300 K as a function of Rh coverage. STM
9338 J. Phys. Chem. C, Vol. 112, No. 25, 2008
Figure 1. (a) Large-scale and (b) atomic-scale resolution STM images of a 3-nm-thick CeO2(111) film; Inset in panel (a) is a LEED pattern taken from CeO2(111). (c) Atomic-scale resolution STM image of a 3-nm-thick reduced CeO1.75(111). Images were taken with a constantcurrent STM mode (3-4 V and 0.5-1.0 nA).
images are shown for three different coverages in Figure 2a-c. Results from analysis of these and other STM images are presented in Table 1. For each Rh deposition, a new ceria film was prepared. At the lowest Rh coverage studied (0.7 ML), Rh shows a clear tendency to remain at the step edges. Although the ceria film has a large terrace surface area, in the region imaged in Figure 2a, 65% of Rh particles are found anchored at the step edges. Care should be used in interpreting the fraction of particles at steps versus terraces since this fraction will obviously depend upon the number of steps in the imaged
Zhou et al. region, but clearly the step edges are very densely covered by the Rh particles. The observed total Rh particle density (both steps and terraces) in Figure 2a is 3.7 × 1012/cm2. Furthermore, the Rh particles exhibit a bimodal diameter distribution, as is shown by the histograms in Figure 3. Rh particles at the step edges are smaller than those formed on the terraces with mean diameters of 2.5 and 3.8 nm, respectively (Table 1). Deposition of 1.1 ML Rh on CeO2(111) at 300 K increases the Rh particle density to 4.3 × 1012/cm2 (Figure 2b). The increased Rh coverage primarily increases the density of particles on the terraces, since the step edges are already essentially saturated by 0.7 ML. At 1.1 ML Rh coverage, the particles at the step edge maintain about the same size and density as at the lower coverage, as can be seen from comparisons of Figure 2a,b. This shows that the particles first grow at the steps during deposition and saturate. Additional Rh nucleates and forms particles located on terraces. Further Rh deposition from 1.1 to 2.3 ML doubles the Rh particle density to 9.0 × 1012/cm2 (Figure 2c). At this higher exposure, Rh particles cover most of the ceria surface with a uniform spatial distribution. There is no longer a clear distinction between terrace and step sites or obvious bimodality of particle sizes. Although the average Rh particle size does not change appreciably, the size distribution narrows because of the lack of distinction of the smaller step-edge particles. 3.3. Rh Growth on CeO2(111): Dependence on Postannealing Temperature. Oxide-supported metal particles can exhibit unique size-related reactivity; therefore it is very important to understand how the sizes of ceria-supported Rh nanoparticles can be controlled. For this reason, Rh particle size was investigated for several annealing temperatures. Images of Rh particles were recorded for the same three coverages after annealing the surfaces in Figure 2a-c to 800 K, and these are shown in Figure 2d-f. For all three Rh coverages, the Rh particle sizes increase after heating the surfaces to 800 K at the expense of particle densities. With the increase of Rh coverage, the average Rh particle size does not change significantly, but the particle density increases. Similar to the Rh particles after deposition at 300 K, Rh particles still preferentially decorate the step edges at low Rh coverage of 0.7 and 1.1 ML after they are annealed to 800 K. Furthermore, the Rh size distribution is broadened at the lower coverages of 0.7 and 1.1 ML as a result of the differences in mean diameter between particles at steps and particles on terrace sites. At a coverage of 2.3 ML, the Rh particles again have a more uniform size distribution. Figure 4 demonstrates that various sizes of Rh nanoparticles can be prepared by varying the postannealing temperatures. First, 2.3 ML Rh was deposited on CeO2(111) at 300 K (Figure 4a), producing Rh particles with mean diameters of 3.9 nm with a particle density of 9 × 1012/cm2. Rh particles grow slightly in size above 500 K, but heating to 800 K causes a significant growth of Rh particles, almost double in both diameter and height at the expense of Rh particle density (Table 1). Subsequently annealing this surface to 900 K causes a further increase in the Rh particle size and a corresponding decrease of Rh particle density (Figure 4c). At this temperature, most Rh particles tend to align along the step edges, and the ceria substrate was clearly observable. Further heating the surface to 1000 K causes a significant broadening in Rh particle size distribution. Both very large particles up to 9.0 nm in diameter and small particles, 3.5 nm in diameter, are formed on the ceria (Figure 4d). 3.4. Rh growth dependence on ceria oxidation states. An important issue in the study of Rh growth on ceria is to
Rh and PD Nanoparticles on CeOx(111) Thin Films
J. Phys. Chem. C, Vol. 112, No. 25, 2008 9339
Figure 2. STM images of (a) 0.7 ML Rh, (b) 1.1 ML Rh, and (c) 2.3 ML Rh deposited on CeO2(111) at 300 K. (d-f) STM images of Rh/ CeO2(111) in panels a-c annealed to 800 K, respectively. All images are 100 × 100 nm2 taken with a constant-current STM mode (3-4 V and 0.5-1.0 nA).
address whether Rh particles exhibit any morphological dependence on the degree of reduction of the ceria films, a possible explanation for the ceria oxidation-state-dependent reactivity of Rh particles.16,18,20,22,26 To address this question, Rh deposited on a reduced ceria film was compared with deposition on a fully oxidized film. Nearly equal coverages of Rh were deposited at 300 K on CeO2(111) and CeO1.75(111) and then annealed to 800 K before recording STM images. These conditions were chosen to match surfaces prepared for previous surface chemistry studies.16,18,20,22,26 Results are compared in Figure 5a,b. This comparison
suggests that Rh grows similarly on both surfaces. In both cases, there is a tendency for the particles to locate at steps compared to terraces. This may indicate that steps are more important nucleation sites than point defects, presumably present in greater numbers on the reduced surface. The Rh particles have the same appearance and comparable heightto-diameter ratios on both the oxidized and the reduced surfaces. However, facets and atomic structure of the islands could not be resolved by STM, so differences in crystallographic orientation, defect populations or facet sizes could not be ruled out. Further increase of ceria film reduction from
9340 J. Phys. Chem. C, Vol. 112, No. 25, 2008
Zhou et al.
TABLE 1: Rh Particle Sizes and Density As a Function of Rh Coverage, Annealing Temperature, and Estimated Average Surface Cerium Oxidation State calculated. coveragea T anneal (K) (ML) x in CeOxb
site
0.7
2.0
300
0.7 1.1 1.1 2.3 2.3 2.3 2.3
2.0 2.0 2.0 2.00 2.0 2.0 2.0
800 300 800 300 800 900 1000
0.7
1.75
300
0.7 1.1 0.7
1.75 1.75 1.60
800 800 800
terrace step average total total total total total total “small” “large” total terrace step total total total total
0.7 0.7
2.00 2.00
300 800
total total
mean particle diameter mean particle heightc particle density (nm) (nm) (1012 cm-2) height/ diameter ratio Rh Growth 3.8 ( 0.6 2.5 ( 0.5 3.1 ( 0.9 5.5 ( 1.2 3.9 ( 0.7 5.7 ( 1.0 3.9 ( 0.6 6.1 ( 0.7 6.7 ( 1.0 3.5 9 6.8 ( 1.5 4.0 ( 0.7 3.4 ( 0.4 3.7 ( 0.6 5.0 ( 0.6 5.3 ( 0.6 5.7 ( 0.9 Pd Growth 4.4 ( 1.0 8.4 ( 1.0
0.9 ( 0.2 0.5 ( 0.2 0.7 ( 0.3 1.2 ( 0.4 0.8 ( 0.2 1.2 ( 0.3 0.8 ( 0.1 1.3 ( 0.2 1.5 ( 0.3 0.6 2.1 1.7 ( 0.5 0.7 ( 0.2 0.6 ( 0.1 0.7 ( 0.2 1.1 ( 0.2 1.0 ( 0.2 1.0 ( 0.3
1.3 2.4 3.7 0.9 4.3 1.3 9.0 2.9 1.3 0.9 1.4 1.5 2.9 1.2 2.0 1.1
0.24 ( 0.06 0.20 ( 0.09 0.23 ( 0.12 0.22 ( 0.09 0.21 ( 0.06 0.21 ( 0.06 0.21 ( 0.04 0.21 ( 0.04 0.22 ( 0.05 0.17 0.23 0.25 ( 0.09 0.18 ( 0.06 0.18 ( 0.04 0.18 ( 0.06 0.22 ( 0.05 0.19 ( 0.04 0.18 ( 0.06
0.8 ( 0.2 1.6 ( 0.2
1.9 0.3
0.18 ( 0.06 0.19 ( 0.03
a Calculated from observed mean particle volumes and number densities as observed by STM. b Estimated from preparation parameters and prior XPS results.22,29 c Measured from STM line profiles.
1.1 ML (Figure 5c) only increases the Rh particle density but does not change the average Rh particle size. 3.5. Pd Growth on CeO2(111). The growth of Rh and Pd were directly compared using STM. For comparison to Rh particles (0.7 ML) in Figure 2a,d, 0.7 ML Pd was deposited on CeO2(111) at 300 K and annealed to 800 K (Figure 6a,b). Like Rh, deposition of Pd on CeO2 at room temperature also produces three-dimensional (3-D) particles, and the Pd particles anchor selectively at the step edges. However, the same coverage produces larger Pd particles compared to Rh particles, with diameters of 4.4 and 3.1 nm, respectively. The Pd particle density (1.9 × 1012/cm2) at this coverage is only half of that from Rh (3.7 × 1012/cm2). At 800 K, the sizes of both Pd and Rh particles are both nearly doubled compared to those at 300 K. The Pd particle sizes increase to 8.4 nm compared to 5.5 nm for Rh, and the Pd particle density drops to only one-third of that from the Rh surface. Pd growth on reduced ceria surfaces was also investigated by STM. However, no apparent difference in growth behavior between Pd on fully oxidized ceria and reduced ceria was observed. 4. Discussion
Figure 3. Histograms of the particle diameters (a) and the particle heights (b) are shown for 0.7 ML of Rh deposited on fully oxidized CeO2.
CeO1.75(111) to CeO1.6(111) did not cause any significant changes in Rh particle size or particle density (data not shown). Similar to the Rh growth on CeO2(111), increasing Rh coverage on CeO1.75(111) from 0.7 ML (Figure 5b) to
Although the oxidation state of the particles observed by STM can not be directly determined by STM, previous XPS data showed that Rh and Pd remain metallic when deposited onto fully oxidized or reduced CeOx and do not cause reduction of the cerium oxide.16,19,20,25,44 The appearance of individual 3-D particles in the STM images is at least consistent with nonwetting, unmixed growth of metallic Rh or Pd on top of the relatively undisturbed oxide support. This result is expected from the energetics of oxide formation. For example, partial reduction of CeO2 by Rh to create the most stable form of Rh oxide, as described by eq 1 is unfavorable by about 400 kJ/mol Rh (based upon ∆Hf298).45 Similar oxidation of Pd by CeO2 to form PdO is also unfavorable by 297 kJ/mol.
2Rh + 6CeO2 f 3Ce2O3 + Rh2O3
(1)
This consideration is based upon bulk energetics and does not
Rh and PD Nanoparticles on CeOx(111) Thin Films discount the possibility that individual oxygen vacancies or Ce3+ cations might be created, e.g., at the Rh-oxide interface or around the perimeter of the Rh particles. STM images of both Rh and Pd deposition obviously indicate that the metal particles grow as isolated particles with thicknesses greater than a single monolayer. This growth mode is usually referred to as Volmer-Weber or 3-D growth.46,47 Previously we have shown from comparison of CO desorption and quantitative measurement of Rh dose that Rh grows as nanoparticles 2-3 layers thick for submonolayer coverage.22 A 3-D growth model has previously been concluded also for Au on CeO2.7 For Pt on CeO2, previous measurements of AES intensity versus deposition time are consistent with a steady buildup of small 3-D Pt particles, 48 while, on the basis of CO uptake, Mullins and Zhang concluded that Pt grows at 300 K as islands 1-2 ML thick on CeO2(111).24 Having demonstrated by STM that the growth mode for Rh and Pd on CeO2 is 3-D, it is of interest to compare this with the growth mode that is predicted theoretically. The growth mode of metals on oxides may be predicted based upon thermodynamic parameters. Campbell46 has described conditions under which 3-D growth is thermodynamically favored over the growth of a monolayer or of flat two-dimensional structures that wet the surface (2-D growth). The criterion for 3-D growth, given in eq 2, is that the metal-oxide interfacial energy, γm,o must be greater than the difference between the oxide-vacuum surface energy, γv,o, and the metal-vacuum surface energy, γv,o.
J. Phys. Chem. C, Vol. 112, No. 25, 2008 9341
γm,o > γv,o - γv,m
(2)
If eq 2 is satisfied, then the metal does not wet the oxide/ vacuum interface, leaving regions of clean oxide surface between 3-D crystallites. The criterion can also be expressed in terms of a metal-oxide adhesion energy, Eadh, stated in eq 3, which provides an alternative form of the conditions required for 3-D growth.46
Eadh < 2γv,m
(3)
Although the surface free energies of CeO2(111) (γv,o ∼ 0.7 J/m2),49 Rh (γv,m∼2.5 J/m2) and Pd (γv,m ∼ 1.9 J/m2)50 have been reported, neither the metal-oxide interfacial energy nor the adhesion energy are well-known for a Rh-CeO2 or a Pd-CeO2 interface. Without knowing γm,o or Eadh, it is impossible to apply this criterion to predict the growth mode. Hu et al.51 have provided an alternative approach to predicting growth mode. As demonstrated in their correlation of metal growth data on TiO2(110) single crystal surfaces,51 the growth mode of a particular metal can be simply predicted by comparison of the metal’s heat of sublimation -∆sH298 and the heat of formation of its oxide -∆fH298 oxide, either per mole of metal or per mole of oxygen. A metal shows 3-D island growth if -∆sH0metal /-∆fH0oxide > 1. These enthalpies are well-known, making it easy to apply this criterion. As shown in Table 2, this ratio has the values of 3.2 and 5.9 for Rh and Pd, respectively. Therefore, this criterion predicts 3-D growth for both Rh and Pd, in agreement with the STM observation.
Figure 4. STM image of 2.3 ML Rh deposited on CeO2(111) at 300 K (a) and then subsequently annealed to 800 K (b), 900 K (c), and 1000 K (d), respectively. All images are 100 × 100 nm2 taken with a constant-current STM mode (3-4 V and 0.5-1.0 nA).
9342 J. Phys. Chem. C, Vol. 112, No. 25, 2008
Zhou et al.
Figure 6. STM images of (a) 0.7 ML Pd deposited on CeO2(111) at 300 K and (b) STM image of Pd/CeO2(111) in (a) annealed to 800 K. All images are 100 × 100 nm2 taken with a constant-current STM mode (3-4 V and 0.5-1.0 nA).
TABLE 2: Literature Values for Predicting Growth Mode Using Criterion of Hu et al.51 ∆sH298 kJ/mola Rh Pd PdO Rh2O3 CeO2 Ce2O3
∆fH298 kJ/molb
556 506
∆sH298/∆fH298 per mole metal
∆sH298/∆fH298 per mole oxygen
3.2 5.9
4.9 5.9
85.4 343 1089 1796
a Heat of sublimation at 298 K per mole of metal atoms.74 b Heat of formation at 298 K per mole of oxide.45
Figure 5. STM images of 0.7 ML Rh deposited on CeO2(111) (a) and CeO1.75(111) (b). Rh was deposited at 300 K and subsequently annealed to 800 K. An STM image of 1.1 ML Rh deposited on CeO1.75(111) at 300 K and subsequently annealed to 800 K is shown in panel c. All images are 100 × 100 nm2 taken with a constant-current STM mode (3-4 V and 0.5-1.0 nA).
Although Hu et al. developed this criterion for TiO2, they argue that it should be appropriate for other reducible oxides. The present results indicate that it correctly predicts the result for CeO2, at least in the case of Pd and Rh. It is possible to use STM data to estimate adhesion energies. One method is described by Worren et al.52 and is based upon the hypothesis that the equilibrium geometry of adsorbed metal particles are determined by surface energy relationships similar to the Wulff criteria.53 The key point is that the height-to-width ratio of the supported clusters is determined by the relationship
between the interface energy and the substrate surface energy. We note that the height/diameter ratio of the particles, provided in Table 1, are reasonably constant for Rh particles, as expected from the model. Worren et al. showed that, for a (111) terminated particle, the adhesion energy, Eadh, can be related to the surface free energies for different crystallographic planes and the height-to-width ratio of the top (111) facet, H/W111, through eq 4.
Eadh ) 2γ111 - H/W111[- (3 ⁄ √2)γv,m(111) + √3γv,m(110) +
√(3/2)γv,m(100)] (4) Using this method requires that the orientation of the particles is known at the interface and the width of the surface facet, W111, be measured. Although we were unable to resolve the
Rh and PD Nanoparticles on CeOx(111) Thin Films
J. Phys. Chem. C, Vol. 112, No. 25, 2008 9343
TABLE 3: Prediction of Adhesion Energy from STM and Criterion for 3-D Growth metal
γv,m(111) (J/m2)a
γv,m(100) (J/m2)a
γv,m(110) (J/m2)a
height/diameter ratiob
Eadh (J/m2)c
Eadh/2γv,m(111)d
Rh Pd
2.47 1.92
2.80 2.33
2.90 2.22
0.21 ( 0.04 0.185 ( 0.01
4.26 3.35
0.86 0.87
a Vacuum-metal surface free energy for different crystallographic planes by density functional theory (DFT) calculations.50 b Average from Table 1. c Upper limit calculated from eq 4 assuming metal surface has (111) orientation. d Criterion for 3-D growth mode. Values less than unity imply 3-D growth.
atomic structure of the top of the particles or see faceting, it is instructive to consider the value that results from assuming a (111) orientation. This orientation is reasonably expected in view of the six-fold symmetry of the CeO2(111) surface and the lower energy of the Rh(111) surface compared to other terminations (Table 3). The actual width of the facet, W111, must be smaller than the measured particle diameter, and so the actual H/W111 must be larger than the measured height/diameter ratio. Since the expression in brackets in eq 4 is positive, the value of Eadh obtained from the measured height/diameter ratio is an upper limit for the actual Eadh. Using eq 4, we estimate the upper limit of Eadh for Rh and Pd on CeO2 (111) shown in Table 3. The resulting values satisfy the criteria for 3-D growth given in eq 3, i.e., for both Rh and Pd, the ratio Eadh/2γv,m is less than unity. The STM observations indicate that Pd deposition gives rise to larger particles with a lower density compared to Rh deposition at comparable deposition temperatures. This result matches previous CO adsorption studies where CO TPD was used to titrate surface sites to obtain metal dispersions.22,44 For a similar deposition of Rh and Pd at 300 K and total coverage in the range of 0.5 to 1 ML, dispersions were found to be about 16-20% and 40-50% for Pd and Rh particles, respectively. This growth trend observed for Rh and Pd on CeO2 is similar to that observed previously on other oxides. On Al2O3, for instance, slightly larger particles were obtained for Pd growth than for Rh when deposited under similar conditions,54 in agreement with the trend observed here on CeO2. Furthermore, on TiO2 the growth mode and the measured particles sizes (diameters and thickness) were roughly comparable for Rh55 and Pd56,57 as in the present study. However, simple arguments to compare the metal particle sizes based upon thermodynamic properties such as heats of sublimation are not expected to work well because the growth at 300 K is determined largely by kinetic effects. The observed preferential growth at step edges and the difference in particle sizes at terrace versus step sites are indications that growth kinetics are a dominant factor in determining particle size, as discussed in a later section. The present results confirm that steps play an important role in the metal nucleation and growth process. During growth on CeO2 at 300 K, both Rh and Pd predominantly locate at the step edges rather than on terraces, at least for low coverage (0.7 ML). Preferential growth at step edges has been observed previously for 0.4 monolayer Au growth on CeO28 and for Pd and Cu on TiO2(110).58,59 Possible reasons are either a stronger bonding with step edges compared with that on flat terraces or a high barrier for atoms to move away from the step edges compared to that of terrace diffusion.58 Furthermore, although the reduced ceria contains many more oxygen vacancies and point defects compared with the fully oxidized ceria, the particle sizes and their distribution between steps and terraces are comparable for both surfaces (compare panels a and b in Figure 5). This result implies that terrace point defects are not particularly effective nucleation sites at 300 K. Similar results are found for growth of Rh and Pd on thin alumina films at 300 K.54,60–63 On these films, step edges and domain boundaries
are also the primary nucleation sites at 300 K, although point defects are the dominant nucleation sites at 90 K. At low coverages, the decreased mobility of Rh atoms at the step edges compared with those at the flat terraces results in smaller particle sizes at the step edge, apparent in Figure 2a and summarized Table 1. It is interesting to note that, with increasing coverage, Rh continues to nucleate on the flat terraces after saturating the step sites and covers most of the surface. The bimodal size distribution of Rh particles (step vs terrace) disappears, and a narrow size distribution of Rh particles is observed. This may be described as self-limiting growth and has been observed previously.59,64 It is attributed to the fact that the growth rate of an existing particle is inhibited by the presence of new particles while nucleation of new particles continues. This balance between growth inhibition and continued nucleation is typified by the behavior of Cu on TiO2, which has been discussed in detail previously and analyzed by lattice gas models.59 One result is that the particle sizes vary slowly with coverage, at least above total coverages of about 0.5 ML. This result is seen for Au/CeO27 and Cu/TiO2.59 The implication is that it is not possible to grow very small particles with the current approach. To better understand the chemistry on particles smaller than about 2 nm (or larger than 8 nm), different deposition conditions or a different method of depositing the metal on the surface is required. Heating metal particles to elevated temperatures on oxide surfaces can lead to encapsulation by suboxides, which in turn can alter catalytic behavior.65–70 Encapsulation of Rh supported on single crystal TiO2 has been observed after heating the surface in vacuum to 900 K, on the basis of the total disappearance of the Rh low energy ion scattering peak.71 Pt, Rh, and mixed Pt-Rh particles on TiO2 become encapsulated by heating in vacuum above 700 K.72 On CeO2, encapsulation has also been reported. Encapsulation of Pt by the reduced ceria occurs upon heating, as indicated by total suppression of CO adsorption.24,44 Similarly, encapsulation has been observed by high-resolution electron microscopy on highly dispersed CeO2, but only after high temperature reduction.73 Our STM images do not show evidence for encapsulation of Rh by ceria upon heating in vacuum up to 800 K (Figure 4), although the sizes increase as described above. Previous STM work has shown that encapsulation of Rh on reduced TiO2 can occur without obvious changes in the STM image.72 Therefore, the lack of pronounced changes in the STM in the present case does not by itself eliminate the possibility of encapsulation. However, previous CO TPD and XPS data suggest that encapsulation of Rh does not occur on either oxidized or reduced thin ceria films.20 In that work it was shown that the Rh 3d core-level spectra on oxidized and reduced ceria as well as the amount of CO that adsorbs from all Rh surfaces are very similar. Likewise, for Pd on ceria, no encapsulation was observed in CO TPD or from XPS measurements.44 As described in the introduction, XPS and TPD studies have revealed that the surface chemistry of Rh particles deposited on reduced ceria are dramatically different from that on fully
9344 J. Phys. Chem. C, Vol. 112, No. 25, 2008 oxidized ceria.18–20 The present STM results do not find observable differences in the Rh particle morphology, which could explain these differences. Instead, STM shows that, on either oxidized or reduced ceria, deposition of 0.7 ML of Rh deposited at 300 K followed by heating to 800 K yields Rh particles with similar mean particle size, surface density, and height-to-width ratio as indicated by comparison of Figure 5a,b and the results in Table 1. In both cases (oxidized or reduced ceria), the particle diameter and heights grow comparably upon annealing. Previously, it was demonstrated that for Rh deposited on the reduced CeOx support, enhanced CO dissociation occurs regardless of whether the Rh particles were made bigger by subsequently annealing to 700 K.20 In addition, for Rh deposited at 300 K on the fully oxidized CeO2 support, CO dissociation does not occur, regardless of whether the surface was subsequently annealed to 800 K.20,44 It is concluded from these STM results that it is properties of the CeO2 substrate and not the Rh morphology or particles sizes (at least in the size range of 3-7 nm) that determine dissociation activity. Although the diameter and height of the particles are observed, STM did not permit imaging of the atomic structure of the particles, and so the relative amounts of steps, kinks, and edge sites on the Rh particles could not be assessed. Variations in the distribution of site types could lead to variation in their activity and ability to dissociate adsorbate molecules. However, it seems improbable that Rh particles with comparable diameters and thicknesses could have substantially different average atomic arrangements. So what is the explanation for the observed enhanced dissociation probability of NO and CO on Rh particles supported on reduced ceria?18–20 One likely explanation is that highly reactive sites exist at the interface between the metal particle and the reduced ceria support. These sites could correspond to oxygen vacancies in the ceria surface or to step edge sites where the metal particles tend to reside. Since only a small fraction of the adsorption sites on the metal particles are at the interface, molecular dissociation would seem to be limited to a small fraction of the adsorbed molecules. The fact that very high dissociation fractions are actually observed therefore requires facile exchange of the molecules and dissociated atoms between the interface sites and the other sites on the particle. The dissociated O atoms must diffuse rapidly away from the dissociation site, disappearing into the reduced ceria. Upon recombination at high temperatures, O is returned from the lattice. This diffusion is observed when labeled C18O is adsorbed (and dissociated) on Rh supported on reduced, unlabeled Ce16O2.19,20 This process would be mostly independent of the detailed structure of the Rh particle. 5. Conclusions We used STM to investigate the growth of Rh and Pd on reducible ceria thin films grown on Ru(0001). Our study shows that deposition of Rh and Pd leads to 3-D metal particles on CeO2(111) and CeOx(111), consistent with thermodynamic data. Pd particles are, on average, slightly larger than Rh particles under the same conditions. Step edges play an important role in the nucleation and growth. Metal particles preferentially anchor at the step edges at a low coverage with a smaller particle size than those that nucleate on terraces. Uniform particle sizes appear at higher coverage. Annealing after deposition at 300 K increases the metal particle size without encapsulation, but also broadens the particle size distribution. Our study demonstrates that varying the oxidation state of ceria does not cause an observable change in growth of Rh or Pd, and we conclude
Zhou et al. that particle morphology is not the explanation for differences in surface chemistry on variously oxidized ceria surfaces. Instead, it is proposed that dissociation occurs at the interface between the metal and the reduced support, and this effect is coupled with rapid O diffusion away from the reaction site to yield high dissociation fractions. Acknowledgment. J.Z. would like to acknowledge the technical support from Gary W. Ownby in the Materials Science and Technology Division at the Oak Ridge National Laboratory. Research sponsored by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, U.S. Department of Energy, under Contract DE-AC0500OR22725 with Oak Ridge National Laboratory, managed and operated by UT-Battelle, LLC. References and Notes (1) Trovarelli, A. Catalysis by Ceria and Related Materials; Imperial College Press: London, 2002. (2) Park, S. D.; Vohs, J. M.; Gorte, R. J. Nature 2000, 404, 265. (3) Deluga, G. A.; Salge, J. R.; Schmidt, L. D.; Verykios, X. E. Science 2004, 303, 993. (4) Trovarelli, A. Catal. ReV.: Sci. Eng. 1996, 38, 439. (5) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647. (6) Gan, S.; Liang, Y.; Baer, D. R.; Sievers, M. R.; Herman, G. S.; Peden, C. H. F. J. Phys. Chem. B 2001, 105, 2412. (7) Lu, J. L.; Gao, H. J.; Shaikhutdinov, S.; Freund, H. J. Catal. Lett. 2007, 114, 8. (8) Lu, J. L.; Gao, H. J.; Shaikhutdinov, S.; Freund, H. J. Surf. Sci. 2006, 600, 5004. (9) Norenberg, H.; Briggs, G. A. D. Phys. ReV. Lett. 1997, 79, 4222. (10) Norenberg, H.; Briggs, G. A. D. Surf. Sci. 1998, 404, 734. (11) Norenberg, H.; Briggs, G. A. D. Surf. Sci. 1999, 435, 127. (12) Norenberg, H.; Briggs, G. A. D. Surf. Sci. 1999, 424, L352. (13) Fukui, K.; Takakusagi, S.; Tero, R.; Aizawa, M.; Namai, Y.; Iwasawa, Y. Phys. Chem. Chem. Phys. 2003, 5, 5349. (14) Namai, Y.; Fukui, K.; Iwasawa, Y. J. Phys. Chem. B 2003, 107, 11666. (15) Namai, Y.; Fukui, K. I.; Iwasawa, Y. Catal. Today 2003, 85, 79. (16) Mullins, D. R.; Overbury, S. H. Surf. Sci. 2002, 511, L293. (17) Mullins, D. R.; Zhang, K. J. Phys. Chem. B 2001, 105, 1374. (18) Overbury, S. H.; Mullins, D. R.; Kundakovic, L. Surf. Sci. 2001, 470, 243. (19) Mullins, D. R.; Kundakovic, L.; Overbury, S. H. J. Catal. 2000, 195, 169. (20) Mullins, D. R.; Overbury, S. H. J. Catal. 1999, 188, 340. (21) Overbury, S. H.; Huntley, D. R.; Mullins, D. R.; Ailey, K. S.; Radulovic, P. V. J. Vac. Sci. Technol. A 1997, 15, 1647. (22) Zhou, J.; Mullins, D. R. J. Phys. Chem. B 2006, 110, 15994. (23) Mullins, D. R. Surf. Sci. 2004, 556, 159. (24) Mullins, D. R.; Zhang, K. Z. Surf. Sci. 2002, 513, 163. (25) Kundakovic, L.; Mullins, D. R.; Overbury, S. H. Surf. Sci. 2000, 457, 51. (26) Overbury, S. H.; Mullins, D. R.; Huntley, D. R.; Kundakovic, L. J. Catal. 1999, 186, 296. (27) Overbury, S. H.; Mullins, D. R. Abstr. Pap. Am. Chem. Soc. 1999, 217, U628. (28) Zhou, J.; Mullins, D. R. Surf. Sci. 2006, 600, 1540. (29) Mullins, D. R.; Robbins, M. D.; Zhou, J. Surf. Sci. 2006, 600, 1547. (30) Putna, E. S.; Gorte, R. J.; Vohs, J. M.; Graham, G. W. J. Catal. 1998, 178, 598. (31) Stubenrauch, J.; Vohs, J. M. J. Catal. 1996, 159, 50. (32) Hendrickx, H. A. C. M.; Nieuwenhuys, B. E. Surf. Sci. 1986, 175, 185. (33) de Jong, A. M.; Niemantsverdriet, J. W. J. Chem. Phys. 1994, 101, 10126. (34) Linke, R.; Curulla, D.; Hopstaken, M. J. P.; Niemantsverdriet, J. W. J. Chem. Phys. 2001, 115, 8209. (35) Rebholz, M.; Prins, R.; Kruse, N. Surf. Sci. Lett. 1991, 259, L797. (36) Castner, D. G.; Somorjai, G. A. Surf. Sci. 1979, 83, 60. (37) Andersson, S.; Frank, M.; Sandell, A.; Giertz, A.; Brena, B.; Bruhwiler, P. A.; Mårtensson, N.; Libuda, J.; Bau¨mer, M.; Freund, H.-J. J. Chem. Phys. 1998, 108, 2967. (38) Zhou, J.; Dag, S.; Senanayake, S. D.; Hathorn, B. C.; Kalinin, S. V.; Meunier, V.; Mullins, D. R.; Overbury, S. H.; Baddorf, A. P. Phys. ReV. B 2006, 74. (39) Baddorf, A. P.; Jahns, V.; Zehner, D. M.; Zajonz, H.; Gibbs, D. Surf. Sci. 2002, 498, 74.
Rh and PD Nanoparticles on CeOx(111) Thin Films (40) Mullins, D. R.; Radulovic, P. V.; Overbury, S. H. Surf. Sci. 1999, 429, 186. (41) Reddic, J. E.; Zhou, J.; Chen, D. A. Surf. Sci. 2001, 494, L767. (42) Frank, M.; Kuhnemuth, R.; Ba¨umer, M.; Freund, H. J. Surf. Sci. 1999, 428, 288. (43) Esch, F.; Fabris, S.; Zhou, L.; Montini, T.; Africh, C.; Fornasiero, P.; Comelli, G.; Rosei, R. Science 2005, 309, 752. (44) Senanayake, S. D.; Zhou, J.; Baddorf, A. P.; Mullins, D. R. Surf. Sci. 2007, 601, 3215. (45) CRC Handbook of Chemistry and Physics, 88th ed.; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 2007. (46) Campbell, C. T. Surf. Sci. Rep. 1997, 27, 1. (47) Vook, R. W. Int. Met. ReV. 1982, 27, 209. (48) Zafiris, G. S.; Gorte, R. J. Surf. Sci. 1992, 276, 86. (49) Nolan, M.; Grigoleit, S.; Sayle, D. C.; Parker, S. C.; Watson, G. W. Surf. Sci. 2005, 576, 217. (50) Vitos, L.; Ruban, A. V.; Skriver, H. L.; Kollar, J. Surf. Sci. 1998, 411, 186. (51) Hu, M.; Noda, S.; Komiyama, H. Surf. Sci. 2002, 513, 530. (52) Worren, T.; Hansen, K. H.; Laegsgaard, E.; Besenbacher, F.; Stensgaard, I. Surf. Sci. 2001, 477, 8. (53) Wulff, G. Z. Kristallogr. 1901, 34, 449. (54) Ba¨umer, M.; Frank, M.; Heemeier, M.; Ku¨hnemuth, R.; Stempel, S.; Freund, H. J. Surf. Sci. 2000, 454, 957. (55) Park, J. B.; Ratliff, J. S.; Ma, S.; Chen, D. A. Surf. Sci. Rep. 2006, 600, 2913. (56) Lai, X.; St. Clair, T. P.; Valden, M.; Goodman, D. W. Prog. Surf. Sci. 1998, 59, 25. (57) Stone, P.; Poulston, S.; Bennett, R. A.; Bowker, M. Chem. Commun. 1998, 1369.
J. Phys. Chem. C, Vol. 112, No. 25, 2008 9345 (58) Xu, C.; Lai, X.; Zajac, G. W.; Goodman, D. W. Phys. ReV. B 1997, 56, 13464. (59) Chen, D. A.; Bartelt, M. C.; Hwang, R. Q.; McCarty, K. F. Surf. Sci. 2000, 450, 78. (60) Ba¨umer, M.; Frank, M.; Libuda, J.; Stempel, S.; Freund, H. J. Surf. Sci. 1997, 391, 204. (61) Frank, M.; Ku¨hnemuth, R.; Ba¨umer, M.; Freund, H. J. Surf. Sci. 2000, 454-456, 968. (62) Frank, M.; Baumer, M. Phys. Chem. Chem. Phys. 2000, 2, 4265. (63) Baumer, M.; Freund, H. J. Prog. Surf. Sci. 1999, 61, 127. (64) Zhou, J.; Chen, D. A. Surf. Sci. 2003, 527, 183. (65) Haller, G. L.; Resasco, D. E. In AdVances in Catalysis; Eley, D. D., Pines, H., Weisz, P. B., Eds.; Academic Press: New York, 1989; Vol. 36; p 173. (66) Tauster, S. J. Acc. Chem. Res. 1987, 20, 389. (67) Bowker, M.; Stone, P.; Morrall, P.; Smith, R.; Bennett, R.; Perkins, N.; Kvon, R.; Pang, C.; Fourre, E.; Hall, M. J. Catal. 2005, 234, 172. (68) Bernal, S.; Calvino, J. J.; Cauqui, M. A.; Cifredo, G. A.; Jobacho, A.; Rodriguez-Izquierdo, J. M. Appl. Catal. A: Gen. 1993, 99, 1. (69) Miessner, H.; Naito, S.; Tamaru, K. J. Catal. 1985, 94, 300. (70) Sadeghi, H. R.; Henrich, V. E. J. Catal. 1984, 87, 279. (71) Ovari, L.; Kiss, J. Appl. Surf. Sci. 2006, 252, 8624. (72) Ozturk, O.; Park, J. B.; Ma, S.; Ratliff, J. S.; Zhou, J.; Mullins, D. R.; Chen, D. A. Surf. Sci. 2007, 601, 3099. (73) Bernal, S.; Botana, F. J.; Calvino, J. J.; Cifredo, G. A.; Perez-Omil, J. A.; Pintado, J. M. Catal. Today 1995, 28, 219. (74) Langes Handbook of Chemistry; 15th ed.; Dean, J. A., Ed.; McGraw Hill Handbooks: New York, 1999.
JP711198C