Ru3(CO)12 Adsorption and Decomposition on TiO2 - TAMU Chemistry

with threaded rods. The doser is isolated from the main analysis chamber by a gate valve and can be evacuated independently of the main chamber and th...
0 downloads 7 Views 299KB Size
Download PDF
698

Langmuir 2002, 18, 698-705

Ru3(CO)12 Adsorption and Decomposition on TiO2 D. C. Meier,† G. A. Rizzi,‡ G. Granozzi,‡ X. Lai,† and D. W. Goodman*,† Department of Chemistry, Texas A&M University, P.O. Box 30012, College Station, Texas 77842-3012, and Dipartimento di Chimica Inorganica, Metallorganica ed Analitica, University of Padova, Unita` di Ricerca INFM, Via Loredan 4, I-35131 Padova, Italy Received July 9, 2001. In Final Form: October 19, 2001 Triruthenium dodecacarbonyl (Ru3(CO)12) adsorption and decomposition on titania (TiO2) were studied using infrared reflection absorption spectroscopy, temperature-programmed desorption, Auger electron spectroscopy, and scanning tunneling microscopy. It was found that the vapor-deposited cluster adsorbs with the Ru3 plane oriented normal to the surface. The clusters form an amorphous layer at 90 K, order at 195 K, and begin to decompose in a vacuum at ca. 250 K. The clusters decompose in two stages, with complete decomposition occurring above 600 K. The clusters are highly dispersed on the surface and decompose cleanly to form small ruthenium clusters up to 3 nm across and single atom rows up to 20 nm long.

Introduction Metal carbonyls are important as precursors in catalysis and thin film technologies; thus, their surface chemistry is of fundamental and practical interest. The adsorption and decomposition in ultrahigh vacuum (UHV) of metal carbonyls on various surfaces have been the subject of several recent investigations (refs 1-4 and references therein). While their decomposition to metal clusters for heterogeneous catalysts (both mono- and bimetallic) has been the subject of numerous studies (see for example ref 5), it is only recently that metal carbonyls have been used as sources for the deposition of ultrathin oxide and/or metal films on semiconductors and metals.6 Some metal carbonyls are sufficiently volatile7 such that introduction into a UHV chamber is practical via an effusion cell. The pressure during the deposition (10-8-10-9 Torr) favors decarbonylation for certain systems, allowing the deposition of virtually carbon-free, highly dispersed metal clusters. This approach has recently been used to produce epitaxial films of RuO2 by dosing Ru3(CO)12 onto a heated TiO2(110) substrate under oxygen ambient.8 Ruthenium and ruthenia/titania composites are of interest for a variety of fundamental and practical reasons. Ruthenium is known to be active for low-temperature methane coupling9 and is active for CO and CO2 methanation, particularly when partially oxidized on titania supports.10 Ruthenia/titania mixed metal oxides are used in dimensionally stable anodes (DSA).11 Mixed oxides are * To whom correspondence should be addressed. † Texas A&M University. ‡ University of Padova. (1) Clowes, S. K.; Seddon, E. A.; McCash, E. M. Surf. Sci. 2000, 464, L667. (2) Fachini, E. R.; Cabrera, C. R. Langmuir 1999, 15, 717. (3) Sun, L.; McCash, E. M. Surf. Sci. 1999, 422, 77. (4) Huang, W.; Zybill, C. E.; Luo, L.; Hieringer, W.; Huang, H. H. Organometallics 1998, 17, 5825. (5) Psaro, R.; Recchia, S. Catal. Today 1998, 41, 139. (6) Rizzi, G. A.; Zanoni, R.; Di Siro, S.; Perriello, L.; Granozzi, G. Surf. Sci. 2000, 462, 187. (7) Chandra, D.; Garner, M. L.; Lau, K. H. J. Phase Equilib. 1999, 20, 565. (8) Rizzi, G. A.; Magrin, A.; Granozzi, G. Surf. Sci. 1999, 443, 277. (9) Lenz-Solomun, P.; Wu, M. C.; Goodman, D. W. Catal. Lett. 1994, 25, 75. (10) Thampi, K. R.; Kiwi, J.; Gratzel, M. Nature 1987, 327, 506. (11) Beer, H. B. J. Electrochem. Soc. 1980, 127, C303.

also used to chemically fine-tune catalysts. For example, by varying the composition ratio of silica/alumina, the acidity of the composite oxide can be adjusted.12,13 In this regard, ruthenia/titania has shown promise as a binary metal oxide support, particularly since ruthenia and titania have similar crystalline structures. It has been shown that ruthenium can be incorporated into the titania crystal lattice via decomposition of Ru3(CO)12.14 Furthermore, while the two oxides are similar in structure, their electronic properties are quite different as a result of varying degrees of d-shell filling. Titania is an electronpoor semiconductor; ruthenia is relatively electron-rich. Since the activity of supported metal catalysts can depend profoundly on the precise nature of the oxide support, a selection of electronically tunable substrates should allow these effects to be quantified and optimized. At temperatures as low as 573 K in an oxygen atmosphere (10-6 Torr), metallic Ru particles, formed after the loss of the CO ligands from the parent carbonyl, are easily oxidized to obtain a carbon-free RuO2 epitaxial film. Such behavior has been demonstrated by in situ angleresolved photoelectron diffraction (XPD) studies.15 At higher substrate temperatures, the formation of a Ti(1-x)RuxO2 binary surface oxide is favored.14-16 X-ray photoelectron spectroscopy (XPS) results indicate that Ru3(CO)12 is adsorbed on TiO2(110) at room temperature, resulting in a saturated surface of a partially decarbonylated species. Annealing this surface at 473 K leads to the evolution of CO, suggesting some further decomposition of the precursor. Eventually at 573 K, metallic Ru nanoparticles are obtained. In a previous study14 in which polycrystalline TiO2 was impregnated at room temperature with Ru3(CO)12, the formation of subcarbonyl species bonded via Ru-O-Ti bridges was reported. Temperature-programmed desorption (TPD) indicated that the parent cluster is physisorbed (12) Busca, G. Phys. Chem. Chem. Phys. 1999, 1, 723. (13) Daniell, W.; Schubert, U.; Glockler, R.; Meyer, A.; Noweck, K.; Knozinger, H. Appl. Catal., A 2000, 196, 247. (14) Rizzi, G. A.; Magrin, A.; Granozzi, G. Phys. Chem. Chem. Phys. 1999, 1, 709. (15) Rizzi, G. A.; Sambi, M.; Magrin, A.; Granozzi, G. Surf. Sci. 2000, 454, 30. (16) Kim, Y. J.; Gao, Y.; Chambers, S. A. Appl. Surf. Sci. 1997, 120, 250.

10.1021/la011041r CCC: $22.00 © 2002 American Chemical Society Published on Web 01/10/2002

Ru3(CO)12 Adsorption and Decomposition on TiO2

Figure 1. Selected IRAS spectra of 3 langmuir Ru3(CO)12/ TiO2/Mo(110) at a series of temperatures in UHV. After heating to 295 K, no IRAS features are visible. Recooling the sample to 90 K and exposing to CO yields the bottom spectrum, also devoid of IRAS features (see text).

at room temperature and begins to decompose via CO evolution at 363 K. Extended X-ray absorption fine structure (EXAFS) and infrared data suggest the formation of [Ru3(CO)6(µ-OTi)3] following an anneal to 473 K.17 This system serves as the benchmark for comparison of the present study with previous work. Here, we report the use of infrared reflection absorption spectroscopy (IRAS) and TPD to study the adsorption and decomposition of Ru3(CO)12 on TiO2 and the potential of this carbonyl precursor as a metal source for synthesizing thin films. Scanning tunneling microscopy (STM) of a TiO2(110) single crystal exposed to Ru3(CO)12 vapor was used to acquire images before and after heat treatment at 573 K in an effort to develop a more complete picture of the decomposition process. Experimental Section The spectroscopic portion of this study was performed in a UHV chamber (base pressure less than 2 × 10-10 Torr) equipped with an Auger electron gun and cylindrical mirror analyzer (Perkin-Elmer), a quadrupole mass analyzer (QMS, UTI Instruments), a low-energy electron diffraction (LEED, Perkin-Elmer) spectrometer, and an infrared cell separated from the main chamber by a two-stage sliding seal. This apparatus has been described in detail elsewhere.18,19 The main chamber is also fitted with an effusion cell doser for the introduction of metal carbonyls or other solids with suitable vapor pressures. The doser is constructed of a perforated tantalum foil (H. Cross Co.) pocket mounted inside a UHV stainless steel tube and held in place (17) Asakura, K.; Bando, K. K.; Iwasawa, Y. J. Chem. Soc., Faraday Trans. 1990, 86, 2645. (18) Campbell, R. A.; Goodman, D. W. Rev. Sci. Instrum. 1992, 63, 172. (19) Leung, L. W. H.; He, J. W.; Goodman, D. W. J. Chem. Phys. 1990, 93, 8328.

Langmuir, Vol. 18, No. 3, 2002 699

Figure 2. Expanded IRAS spectrum from Figure 1 at 195 K. The line spectrum overlay is taken from Battiston et al. (ref 24); dashed lines denote equatorial stretching modes, and the solid line is the pure axial mode. with threaded rods. The doser is isolated from the main analysis chamber by a gate valve and can be evacuated independently of the main chamber and the infrared cell through an attached needle valve. This construction allows for the evacuation of ambient atmosphere without significant contamination of the UHV system. The doser’s position on the main chamber also allows for the sample to be protected from contamination by insertion into the IR cell when the gate valve is initally opened. Using this configuration, outgassing of the doser could be accomplished independent of the sample. The IRAS cell is fitted with flange-mounted CaF2 windows in the infrared beam path. The optics and windows are aligned such that the angle of incidence is 85° from surface normal, an ideal angle for IRAS.20 The infrared spectrometer used for data collection is a Mattson Cygnus 100 in single-beam mode using an MCT detector. All spectra collected in this study are averages of 512 scans at 4 cm-1 resolution. The sample consists of a Mo(110) single crystal (Single Crystal Limited) that is mounted on the sample probe with a tantalum wire loop spot-welded around the edge of the sample and a type C thermocouple (wire supplied by H. Cross Co.) spot-welded to the back. This mount allows for resistive heating to 1500 K and liquid nitrogen cooling to 90 K. The sample was cleaned by alternating annealing at 800 K in 5 × 10-6 Torr O2 and electron beam heating to 2000 K until no carbon or oxygen was observed using Auger electron spectroscopy (AES). The Ti doser was constructed by wrapping titanium wire (99.99+%, Goodfellow) around a tungsten wire loop (H. Cross Co.) that could be heated resistively. The doser was outgassed thoroughly by resistive heating prior to use. Oxygen (99.99%, Botco, Bryan, TX) and carbon monoxide (99.99%, Matheson Gas Products) were further purified by fractional condensation and transferred to glass bulbs. The CO bulb was left immersed in liquid nitrogen during experiments in order to condense volatile metal carbonyls that are typically formed when CO comes in contact with the steel (20) Hoffmann, F. M. Surf. Sci. Rep. 1983, 3, 107.

700

Langmuir, Vol. 18, No. 3, 2002

Figure 3. An illustration of the two possible orientations of the Ru3(CO)12 molecule with respect to the surface. Since only modes perpendicular to the surface plane are visible in IRAS, it can be used to determine molecular orientation. gas manifold. Ru3(CO)12 (99%, Strem Chemicals, Inc.) was introduced into the effusion doser as received. The titanium oxide thin films were produced by evaporation of titanium metal onto the Mo(110) crystal at 650 K in a background of 5 × 10-6 Torr O2. The films were annealed to 800 K in this oxygen background in order to ensure full oxidation and to eliminate adventitious carbon that was occasionally found in the films prior to annealing. Details of the film synthesis can be found elsewhere;21 essentially, these films are the rutile phase with a preponderance of (110) facets. Ru3(CO)12 was then dosed on the films at 90 K at a number of exposures; the exposure pressure for the carbonyl species was typically 1.8 × 10-9 Torr. IRAS and TPD were performed on the samples prior to decarbonylation, and AES was carried out afterward. TPD data were collected from 100 to 800 K at 28 atomic mass units and 5 K s-1. AES was collected from 20 to 600 eV at 3 kV beam voltage. Because Ru remained on the surface after the decomposition of Ru3(CO)12, the resulting composite films were evaporated from the sample by an e-beam flash to 2000 K. This procedure produces a clean Mo(110) substrate upon which a new titania film could be synthesized. The imaging portion of this study was performed in a combined high-pressure cell/multitechnique UHV analysis chamber (base pressure less than 2 × 10-10 Torr) that is equipped with a doublepass cylindrical mirror analyzer for AES and XPS, LEED optics (Perkin-Elmer), a QMS (UTI Instruments), and a UHV scanning tunneling microscope (Omicron). The apparatus consists of four main sections: a high-pressure cell, a surface analysis chamber, a sample preparation chamber, and a STM chamber; details are available elsewhere.22 A Type C thermocouple (wire provided by H. Cross Co.) was attached to the edge of a TiO2(110) crystal (Commercial Crystal Laboratories) using high-temperature ceramic adhesive (AREMCO 571). The thermocouple was used to measure the surface temperature and to calibrate a pyrometer (OMEGA OS3700) prior to its disconnection for the STM measurements. The (21) Lai, X.; Guo, Q.; Min, B. K.; Goodman, D. W. Surf. Sci. 2001, 487, 1. (22) Xu, C.; Oh, W. S.; Liu, G.; Kim, D. Y.; Goodman, D. W. J. Vac. Sci. Technol., A 1997, 15, 1261.

Meier et al.

Figure 4. Expanded IRAS spectrum from Figure 1 at 195 K. The line spectrum overlay is taken from Asakura et al. (ref 17). Note the substantial difference in position and relative intensity of the low-energy peaks. pyrometer was then used to measure the temperature in subsequent annealing experiments. The crystal was prepared for experiments by alternating argon ion sputtering with hightemperature (1000 K and higher) annealing. This cleaning procedure has been described in detail elsewhere.23 Following cleaning, the sample was introduced into the high-pressure cell to which the Ru3(CO)12 effusion cell was attached. After exposure to the cluster vapor, the sample was removed to the vacuum chamber in order to conduct the annealing and STM experiments.

Results and Discussion Figure 1 shows the carbonyl stretching region of the IRAS spectrum of Ru3(CO)12 (exposure equal to 3 langmuir, where 1 langmuir ) 1 × 10-6 Torr s) adsorbed on TiO2 at 90 K, as well as spectra acquired as the sample was heated. At 90 K, two broad features are apparent at 2084 and 2047 cm-1. Both features exhibit a tail sloping to the lowenergy side. As the temperature is increased, the feature at 2047 cm-1 decreases slightly in intensity while the 2084 cm-1 feature increases slightly in intensity, with the lowenergy tail diminishing. Between 180 and 195 K, the spectrum transforms sharply in that a symmetric feature appears at 2029 cm-1 with a small shoulder at 2041 cm-1. The strong feature at 2084 cm-1 also becomes more symmetric at this temperature. As the temperature is increased further, the entire spectrum red-shifts slightly and becomes broader and more asymmetric until no features are apparent at 295 K. This sample was then cooled to 90 K, and 1 × 10-5 Torr CO was introduced; no absorption features were apparent at this temperature and pressure. (23) Xu, C.; Lai, X.; Zajac, G. W.; Goodman, D. W. Phys. Rev. B 1997, 56, 13464.

Ru3(CO)12 Adsorption and Decomposition on TiO2

Figure 5. CO TPD of Ru3(CO)12/TiO2/Mo(110) at a series of exposures. The temperature at which IRAS features are no longer visible (295 K) is marked for reference. Note the substantial CO desorption above 295 K.

Figure 2 is an expanded view of the spectra taken at 195 K, where a transmission infrared overlay collected from Ru3(CO)12 dissolved in hexane24 is also shown for comparison. The peaks are presented as-acquired with their relative peak intensities; each is labeled with its assigned symmetry, either A2′′ (for the vibrational mode fully axial with respect to the Ru3 plane) or E′ (for the equatorial vibrational modes). The spectral features for the adsorbed species are blue-shifted from those of the dissolved species; however, they are quite similar in amplitude and relative position to those lines assigned to E′ modes. This phenomenon can be explained in light of the IRAS selection rule, which forbids absorption by any vibrational mode that has no surface normal dynamic dipole moment.20 The consequences of this selection rule are illustrated in Figure 3, which shows a simplified diagram of the Ru3(CO)12 molecule adsorbed on titania with the Ru3 plane either parallel or perpendicular to the surface plane. For the parallel case, all axial carbonyls are aligned normal to the surface, allowing IRAS excitation of only the axial vibrational modes; for the perpendicular case, only equatorial dynamic dipole moments are normal to the surface, resulting in IRAS excitation of only the equatorial modes. To explain these experimental findings, a comparison with infrared experimental data on similar systems is useful. It is unlikely that the observed peaks can be attributed to species arising from the complete decomposition of the original trinuclear cluster since these features do not correspond to any of those measured for CO adsorbed on an analogous Ru/TiO2 powdered system. (24) Battiston, G. A.; Sbrignadello, G.; Bor, G. Inorg. Chem. 1980, 19, 1973.

Langmuir, Vol. 18, No. 3, 2002 701

These powdered samples show transmission IR features at 2140 cm-1 (w) and 2085 cm-1 (w) and a broad heterogeneous band with its maximum located around 2060 cm-1 (s).25 There is also no agreement between the spectra of this study and that of CO adsorbed on metallic single crystals, such as CO/Ru(001), which have infrared absorption features at 2080 cm-1 (w) and 2048 cm-1 (s).26 Furthermore, a significant breakdown in symmetry due to strong adsorbate-support interactions is unlikely, as the increased spectral complexity expected to result from such interactions does not appear. See, for example, the infrared spectra of Ru2Os(CO)12, RuOs2(CO)12,27 FeRu2(CO)12, and Fe2Ru(CO)12,28 systems in which the molecular symmetry has been reduced from D3h to C2v. A similar symmetry reduction would be consistent for Ru3(CO)12 oriented with the Ru3 plane normal to the substrate, assuming there is a sufficiently strong adsorbatesubstrate interaction. Considering these points, the IRAS spectra are most consistent with random initial adsorption (indicated by the broad heterogeneous features at low temperature) with subsequent ordering between 180 and 195 K into a geometry where the Ru3 plane is perpendicular to the surface. As previously discussed, only equatorial vibrational modes should be observed when the Ru3 plane is perpendicular to the surface. Comparison of the surface spectra to the solution spectra yields similar peak amplitude and relative position, with the notable exception that the axial mode observed in the solution phase is invisible in IRAS. IRAS data cannot be used to determine how individual ruthenium atoms are oriented with respect to the surface, nor can IRAS show whether there is a preferred molecular orientation with respect to the substrate atomic rows; however, the orientation of the Ru3 plane with respect to the surface plane is clear. These results contrast with those in which Ru3(CO)12 was adsorbed onto a titania powder by wet impregnation.17 The bands that Asakura et al. assign to physisorbed Ru3(CO)12 nearly match the two most intense bands of the sample dissolved in hexane (2065 cm-1 (s) and 2030 cm-1 (s)), while those features observed following a heat treatment at 473 K (2139 cm-1 (w), 2081 cm-1 (s), and 2009 cm-1 (s), as shown in the line spectrum overlay in Figure 4) are somewhat different than those observed in the thin film sample of this study (2085 cm-1 (s), 2041 cm-1 (w), and 2029 cm-1 (w)). It is evident from the IRAS data alone that the thin film sample and the powdered sample are not entirely analogous. It is unclear whether these differences are due to the phase of the substrate (the powder is expected to contain substantially more anatase than the thin film), the nature of the deposition method, or both. In any case, since the infrared spectra of the sample formed by wet impregnation is isotropic, no direct molecular orientation comparison can be made; the variations in peak positions and relative intensities between the two methods are the primary evidence that the sample formed by vapor deposition differs from that formed by wet impregnation. Thus, the evidence of this work, which indicates a perpendicular molecular orientation, need not be considered contradictory to that of Asakura et al., which suggests a parallel orientation. Finally, following elimination of the IRAS signal at 295 K, no IRAS features are seen upon introduction of CO at (25) Robbins, J. L. J. Catal. 1989, 115, 120. (26) Peden, C. H. F.; Goodman, D. W.; Weisel, M. D.; Hoffmann, F. M. Surf. Sci. 1991, 253, 44. (27) Battiston, G. A.; Bor, G.; Dietler, U. K.; Kettle, S. F. A.; Rossetti, R.; Sbrignadello, G.; Stanghellini, P. L. Inorg. Chem. 1980, 19, 1961. (28) Yawney, D. B. W.; Stone, F. G. A. J. Chem. Soc. A 1969, 502.

702

Langmuir, Vol. 18, No. 3, 2002

Meier et al.

Figure 6. AES spectra of the TiO2/Mo(110) sample prior to (top) and following (bottom) Ru3(CO)12 exposure (5 langmuir) and TPD to 800 K.

90 K. Two possible reasons for this result are either complete cluster desorption or decomposition of the cluster in such a way that few ruthenium sites are available for CO adsorption (due to insertion and oxidation of ruthenium into the titania lattice). The TPD and AES results reported here together with the results of Rizzi et al.14 suggest that the latter is the case. Figure 5 shows CO TPD data for several Ru3(CO)12 exposures. Most notably, CO evolution occurs in two stages at all coverages. For the 0.5 langmuir exposure, the first decomposition feature appears at 300 K, followed by a second at 500 K. At this low coverage, the peak areas are approximately identical. As coverage is increased, a number of things occur. First, the second desorption peak increases substantially with respect to the first while shifting to lower temperature (360 K at 5 langmuir). Also, at the higher coverages two new features emerge: the first, centered at about 150 K, is attributed to multilayer desorption, while the second appears around 610 K. The assignment of the TPD decomposition features is aided substantially by the temperature-dependent IRAS spectra discussed above. Note that at 295 K there are no features in the IRAS spectrum that correspond to the TPD desorption feature at 300 K (as marked in Figure 5).

However, the TPD data show that the CO does not fully desorb from the surface until 500 K or higher; a substantial amount of CO remains on the surface after IRAS no longer shows any evidence of it. Since the IRAS features have been assigned to the equatorial carbonyl vibrational modes of the cluster with the Ru3 plane oriented perpendicular to the surface, it follows that these modes disappear with the initial carbonyl elimination from the cluster. This does not necessarily indicate that exclusively equatorial carbonyl groups are desorbing; the equatorial vibrational modes would also disappear upon dissociation of Ru-Ru bonds, resulting in the demolition of the original cluster. Furthermore, since no other bands are visible after the first stage of the decomposition, it follows that the remaining surface CO is bound tangential to the surface plane, much like the axial CO molecules in the adsorbed cluster. It has been suggested previously by Asakura et al.17 that the initial desorption leaves a ruthenium polynuclear subcarbonyl species bound to the surface, that is, Ru3(CO)6(µ-OTi)3. While this could be a plausible interpretation of the current data, other possibilities must be considered in light of the previously discussed discrepancies between the vapor-deposited and wet-impregnated samples. One such possibility could be the formation

Ru3(CO)12 Adsorption and Decomposition on TiO2

Langmuir, Vol. 18, No. 3, 2002 703

Figure 7. STM of Ru3(CO)12/TiO2(110) after annealing in a vacuum at 600 K for 60 s. Note the highly dispersed clusters and the numerous stripes of adatoms along the [001] surface vector.

of single-metal subcarbonyl species, such as Ru(CO)2 and Ru(CO)3. These species have been reported on TiO2 previously25 and could account for the increased intensity of the high-temperature decomposition peak with respect to the low-temperature peak as the cluster coverage is increased. In effect, initial decomposition of the adsorbed cluster at low coverage would produce 3Ru(CO)2 + 6CO, resulting in the nearly equal CO TPD peak areas during cluster decomposition. As coverage increases, greater interaction between Ru(CO)2 and desorbing carbonyls could result in the formation of some Ru(CO)3 with a stoichiometry of XRu(CO)3 + (3 - X)Ru(CO)2 + (6 - X)CO, where X ranges from 0 to 3 and increases with coverage. This explanation allows for up to a 3:1 peak area ratio between the second and first decomposition features at high coverage, which is consistent with the TPD results. As both of these single-metal ruthenium subcarbonyl species could be planar, their formation would be consistent with the IRAS data assuming that their molecular planes are mostly parallel to the surface. AES verifies that the cluster decomposition is complete and clean at 800 K. Figure 6 shows the Auger of the TiO2/ Mo(110) system before and after 5 langmuir Ru3(CO)12 exposure and TPD to 800 K. The adsorbed ruthenium is plainly seen (259 eV), but an unfortunate coincidence of the ruthenium and carbon features requires comparison to the published peak shapes.29 The ratio of the positive amplitude of the derivative feature to the total amplitude is approximated as a linear combination of the ratios found (29) Davis, L. E.; MacDonald, N. C.; Palmberg, P. W.; Riach, G. E.; Weber, R. E. In Handbook of Auger Electron Spectroscopy, 2nd ed.; Physical Electronics Division, Perkin-Elmer Corp.: Eden Prairie, MN, 1976.

for the pure elements. The relative concentration of ruthenium can then be found by using the usual elemental composition equation, thereby factoring in the Auger sensitivities for the respective features. This computation indicates that approximately 90% clean ruthenium remains on the surface following Ru3(CO)12 decomposition, assuming no corrections are necessary to compensate for slight variations in the electronic structures of the adsorbed species. Regardless of the differences between the decomposition characteristics of the vapor-deposited sample and those of the wet-impregnated sample, the vapor deposition method clearly results in highly dispersed ruthenium clusters on the rutile surface. Figure 7 shows a large field STM image of the clean TiO2(110) after exposure to Ru3(CO)12 and vacuum annealing at 600 K for 60 s. The surface is covered with clusters of dimensions up to 3 nm in diameter. While there is some agglomeration at step edges, it is not noticeably more abundant than clustering on terraces. Also visible in the image are a number of bright “stripes” of atoms parallel to the [001] surface vector with lengths approaching 20 nm. Finally, there are single bright atomic spots, a phenomenon previously assigned to interstitial titanium ions moving from the bulk to the surface upon annealing, albeit at somewhat higher temperatures than here.23 Due to the chemical insensitivity of STM, it cannot be conclusively determined whether the stripes are due entirely to interstitial transport, ruthenium atoms, or some combination thereof. An expanded image of the same surface is shown in Figure 8. In the center of the image, a cluster of 1.3 nm diameter and 0.2 nm thickness is outlined (A), a size corresponding to a ruthenium cluster numbering on the

704

Langmuir, Vol. 18, No. 3, 2002

Meier et al.

Figure 8. Expanded view of Figure 7. Two features (A and B) discussed in the text are circled. Feature A is accompanied by a cross section showing a cluster thickness of 0.2 nm, which is on the order of a single atomic layer.

order of 14-20 atoms. Just below that cluster, there is a structure (B) which appears as three bright spots that are collinear with the TiO2 [001] vector directly atop features previously assigned to exposed rows of titanium ions in the rutile structure. These spots are flanked on each side by slightly dimmer spots that are atop the dark rows previously assigned to the TiO2 bridging oxygen.30 While it is tempting to assign this structure to three adjacent Ru(CO)2 units arising from the decomposition of a single trinuclear cluster, it must be stressed that structures such as these are not common; longer stripes or larger multicluster features make up the preponderance of the surface features (as shown in Figure 7). Further study is required to determine whether these are in fact the result (30) Onishi, H.; Fukui, K.; Iwasawa, Y. Bull. Chem. Soc. Jpn. 1995, 68, 2447.

of decomposed single trinuclear clusters or simply a fortuitous collection of interstitials; however, such features are consistent with the decomposition scheme in which perpendicularly adsorbed molecular Ru3(CO)12 decomposes into IRAS-invisible Ru(CO)2 collinear with the [001] surface vector, which at higher temperatures further decomposes to form three ruthenium atoms on the surface (Figure 9). The STM evidence provides another indirect means to compare vapor-deposited samples to wet-impregnated ones. While the clusters of this study are at most a few tens of atoms in size, EXAFS studies performed on wetimpregnated samples indicate high Ru-Ru coordination and therefore larger three-dimensional particles than those shown in the present work.31

Ru3(CO)12 Adsorption and Decomposition on TiO2

Langmuir, Vol. 18, No. 3, 2002 705

which vapor-deposited Ru3(CO)12 decomposes on titania surfaces. It is found by IRAS that deposition at low temperature gives rise to a disordered film that orders as the temperature increases into an adsorbed system in which largely unperturbed clusters are oriented with the Ru3 plane perpendicular to the surface plane. These clusters begin to decompose at 250 K and are invisible in IRAS starting from 295 K. TPD results confirm that decomposition occurs in two stages and is complete at 600 K. This result is consistent with the formation of planar single-metal ruthenium subcarbonyls adsorbed parallel to the surface which then further decompose to form ruthenium aggregates. STM results show clusters up to 3 nm in diameter located on both terraces and step edges as well as rows of atoms oriented atop the [001] rows of titanium ions.

Figure 9. Proposed mechanism for Ru3(CO)12 decomposition including an adsorbed Ru(CO)2 subcarbonyl species.

Conclusions We have developed, through a combination of techniques, a comprehensive proposal for the mechanism by

Acknowledgment. We acknowledge with pleasure the support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences. G.G. and G.A.R. gratefully acknowledge funding to this project coming from “Progetto Finalizzato Materiali Speciali per Tecnologie Avanzate II” of the CNR, Rome, and from Ministero della Ricerca Scientifica e Tecnologica (MURST) through the National Program “Strati ultrasottili di ossidi e solfuri inorganici: crescita, caratterizzazione e reattivita` superficiale”. D.C.M. gratefully acknowledges Dr. Timothy R. Hughbanks for providing helpful insights into the symmetry elements and vibrational characteristics of these clusters. LA011041R (31) Mizushima, T.; Tohji, K.; Udagawa, Y.; Ueno, A. J. Am. Chem. Soc. 1990, 112, 7887.