Enhanced Gas Sensing by Individual SnO2 Nanowires and

NANO LETTERS

Enhanced Gas Sensing by Individual SnO2 Nanowires and Nanobelts Functionalized with Pd Catalyst Particles

2005 Vol. 5, No. 4 667-673

A. Kolmakov,*,† D. O. Klenov,‡ Y. Lilach,† S. Stemmer,‡ and M. Moskovits† Department of Chemistry and Biochemistry and Materials Department, UniVersity of CaliforniasSanta Barbara, Santa Barbara, California 93106 Received January 14, 2005

ABSTRACT The sensing ability of individual SnO2 nanowires and nanobelts configured as gas sensors was measured before and after functionalization with Pd catalyst particles. In situ deposition of Pd in the same reaction chamber in which the sensing measurements were carried out ensured that the observed modification in behavior was due to the Pd functionalization rather than the variation in properties from one nanowire to another. Changes in the conductance in the early stages of metal deposition (i.e., before metal percolation) indicated that the Pd nanoparticles on the nanowire surface created Schottky barrier-type junctions resulting in the formation of electron depletion regions within the nanowire, constricting the effective conduction channel and reducing the conductance. Pd-functionalized nanostructures exhibited a dramatic improvement in sensitivity toward oxygen and hydrogen due to the enhanced catalytic dissociation of the molecular adsorbate on the Pd nanoparticle surfaces and the subsequent diffusion of the resultant atomic species to the oxide surface.

Introduction. The large surface-to-volume ratio of onedimensional (1-D) semiconducting metal oxide nanostructures and the congruence of the carrier screening length with their lateral dimensions make them highly sensitive and efficient transducers of surface chemical processes into electrical signals. Significant progress has been reported in the use of metal oxide nanowires and nanobelts1-7 as sensors and in other electronic applications. Although promising results of the gas sensing performance of metal oxide nanowires have been reported,8 the development of highly selective and controllably sensitized devices9 remains a future challenge for oxides. In practice, the selectivity of gas sensors and catalysts is usually achieved by functionalizing the material with catalytically active metals. In this communication we report the successful and controllable sensitization of the surface of individual quasi-1-D SnO2 nanostructures with Pd nanoparticles. In addition to the surface chemistry of these systems, we report the atomic-level structure of the metal nanoparticles and the oxide support, obtained using conventional high-resolution as well as scanning transmission electron microscopy. We also report the nucleation and growth dynamics of the deposited metal nanoparticles, providing a preliminary level of understanding of the metal * Corresponding author. E-mail: [email protected]. Current address: Physics Department, SIUC, IL 62901. † Department of Chemistry and Biochemistry. ‡ Materials Department. 10.1021/nl050082v CCC: $30.25 Published on Web 03/04/2005

© 2005 American Chemical Society

diffusion and island growth processes occurring in systems with reduced dimensionality. In these systems new phenomena might be expected when the diffusion length of the adsorbate becomes comparable with the lateral dimensions of the supporting nanostructure. The metal nanoparticles were also found to form electroactive elements on the surface of the semiconducting metal oxide nanostructure, as determined from the conductance decrease observed during the metal deposition process. Using the nanostructure as a chemiresistor, we compared the sensing performance of an individual nanostructure (nanowire or nanobelt) before and after it was sensitized with catalytically active Pd nanoparticles. We explain the observed dramatic enhancement in sensitivity in terms of the catalytic action of Pd nanoparticles, which pre-dissociate the adsorbing species delivering atomic (rather than molecular) species to the surface of the nanostructure where they become chemisorbed. Experimental Section. Pristine quasi-1-D, SnO2 nanostructures were synthesized using techniques described in refs 10 and 11. Briefly, single-crystal SnO2 nanowires and nanobelts were vapor-grown in a tube furnace by thermal evaporation of SnO at 1000 °C into an Ar carrier gas (50 sccm, 200 Torr) containing traces of oxygen. The morphology and structure of the 1-D nanostructures and dopant particles were characterized by conventional (phase contrast) high-resolution transmission electron microscopy (HRTEM)

Figure 1. (a) Transmission electron micrograph of the two types of tin oxide nanostructures (nanowire and nanobelt) produced using the vapor synthesis method employed in this study. The nanostructures are covered with vapor-deposited Pd clusters. (b) Schematic view of device used for the in situ conductometric measurements under gas exposure and metal deposition.

and by atomic resolution high-angle annular dark-field (HAADF or Z-contrast) imaging in scanning transmission electron microscopy using a Tecnai F30U TEM with ultratwin objective lens (Cs ) 0.52 mm), operated at 200 and 300 kV, respectively. For thin samples, the image contrast in HAADF is roughly proportional to Z2 (atomic number), and provides the additional benefit of a chemically sensitive image. The composition and stoichiometry of the resultant 1-D nanostructures were also verified with energy-dispersive X-ray analysis in scanning electron microscopy (SEM/EDX), micro-Raman spectroscopy and X-ray diffraction (XRD). Nanowires and nanobelts were collected from the ceramic crucible in which they were formed, and placed on the oxidized side of a Si/SiO2 (300 nm) wafer. The surface density of the deposited nanostructures was kept low to facilitate wiring of individual nanostructures. Metallic (Ti (20 nm)/Au (200 nm)) micro-pads, which functioned as (ohmic) source (S) and drain (D) electrodes (Figure 1a), were vapor deposited onto pre-selected nanostructures (which were a few micrometers in length) using a contact mask. The (pdoped) Si substrate (G) was kept at ground potential during the conductometric measurements. As previously reported,12 the transport and sensing properties of the SnO2 nanowires can be strongly influenced by a parasitic gating effect induced by charged impurities that may occasionally accumulate on the surfaces of the nanowire and/or on neighboring areas of the supporting Si oxide. To avoid the contamination of the surfaces of the nanostructures and the oxide layer, no wet processes (such as those used in resist-based lithography) were employed at any stage of the device fabrication. Catalyst deposition and the subsequent gas sensing measurements were conducted in situ on individual nanostructures in a variable-temperature probe station. Pd was used in most experiments, but Au was also used in some experiments to gauge the effect of metal decoration on the transport properties of the tin oxide nanowires. Before any measurement or catalyst deposition was carried out, the samples were UV cleaned and annealed in vacuum (∼10-5 668

Torr) for at least an hour at T ) 473 °K. This pretreatment was also found to partially reduce the extent of oxidation of the n-type nanowire, thereby increasing its conductance. After cleaning, the conductance of the nanowires drops slowly due to the readsorption of residual ambient gases, as previously reported 5. The rate of conductance change due to readsorption of (undetermined) ambient gases is some 2 orders of magnitude slower than the observed adsorption and desorption rates for oxygen or hydrogen. This effect can therefore be ignored when analyzing the rates of analyte adsorption/ desorption. However, it was taken into consideration when measuring the nanowire conductance during metal deposition. Gas sensing measurements were carried out by sequentially introducing oxygen and hydrogen pulses with peak pressures of 1 to 2 mTorr into the chamber using electronically pulsed leak valves. The resultant changes in source-drain current, IDS, were measured as a function of time at a bias VDS ) 2 V. After completing the electron transport and sensing measurements on the selected nanowire, Pd was vapordeposited onto the nanowire at a rate ∼2.5 × 10-3 ML/s in situ using a thoroughly degassed and calibrated metal-vapor source (Figure 1b). The effect of metal deposition on the nanowire conductance was monitored throughout the deposition process by continuously measuring IDS. A special care was taken to prove that any significant changes in IDS within at least the first 103 s of Pd deposition are solely determined by processes taking place in the nanowire itself rather than on the gate oxide layer. Finally, the same set of gas sensing experiments were carried out as those performed on the pristine (i.e., metal-free) nanostructure, thereby assessing the effect of Pd deposition on the same nanostructure. Results and Discussion. As in previous studies with tin oxide nanostructures,11 two general types of quasi-1-D SnO2 nanostructures were formed in our synthesis and used in these experiments: nanobelts and nanowires (Figure 1a). The diameters of the as-prepared nanowires and nanobelts typically varied from several tens to hundreds of nanometers. Nano Lett., Vol. 5, No. 4, 2005

Figure 3. Comparison of the HRTEM images of palladium particles (a) on the carbon support film and (b) on the surface of a SnO2 nanostructure. The orientation of the particles are indicated. In all cases the HRTEM images suggest that the palladium particles have bulk crystal structure (cubic close packed).

Figure 2. HRTEM images of a nanobelt covered with palladium particles. (Inset) Fourier transform of the HRTEM image confirms that the nanobelt is viewed along two mutually perpendicular directions [010] (a) and [101] (b), respectively. Analysis of the lattice plane spacings of the Pd particle shown in b indicates that the particle is viewed along [111].

SnO2 nanobelts and nanowires are high-quality single crystals with the rutile structure and occasionally an orthorhombic structure. HRTEM images of a nanobelt recorded along two different crystallographic directions are shown in Figure 2. Fourier transforms (FFTs) of the HRTEM images (insets) as well as macroscopic XRD and Raman data confirmed the rutile structure. The images also show that the facets of the nanobelts are parallel to the (101) and (010) planes, respectively. The Pd particles deposited on the nanowires or nanobelts were crystalline even before coalescence. HRTEM images of Pd-decorated tin oxide nanowires showed Moire´ patterns due to the overlapping, crystal lattices of the two materials, as can be seen in Figure 2 (arrows). A specific epitaxial relationship as claimed for Pd particles grown onto Nano Lett., Vol. 5, No. 4, 2005

the (110) surfaces of SnO2 “nanosticks”13 could not be determined unambiguously, The lattice parameters of the particles were determined by analyzing lattices plane spacings in the FFTs of the HRTEM images. To exclude interference effects due to the SnO2 lattice, additional images were recorded of particles on neighboring portions of the carbon support film (Figure 3a). The lattice spacings of the palladium nanoparticles closely matched those of the bulk metal with no evidence of bulk oxidation despite their small size and exposure to air during sample transfer. This corroborates previous HRTEM studies14 where also no bulk oxidation was reported. Melting, coalescence and encapsulation of the Pd particles was observed during prolonged electron beam exposure or under intense electron beams during HRTEM similar to what was reported previously.15 To reduce beam damage and to avoid particle coalescence under the electron beam, beam exposure and current were kept to a minimum by reducing the operation voltage in HRTEM to 200 kV and/or by operating the microscope in STEM mode. High-angle annular dark-field STEM micrographs of single nanowires with different Pd coverage are shown in Figure 4. Because of the strong atomic number contrast of HAADF, Pd particles appear bright in these images. Pd particles sizes and density were similar on both nanobelts and nanowires. However, the particles deposited onto the nanowires showed a greater distribution in size and density due to the variation of the deposition angle resulting from the curvature of the nanowire (Figure 4c and e). The average Pd particle size and coverage are found to increase with deposition time (Figure 4). For a given deposition rate, the initial nucleation stage proceeds very fast. Small Pd clusters with diameters ranging between 1.3 and 1.5 nm are already observed on the SnO2 surface after only 210 s of exposure at which point the cluster density reaches a value of ∼4.5 × 1012 cm-2, or ∼75% of its final value (Figure 5a). Subsequent metal deposition results mainly in an increase in the average size of the nanoparticles rather than in the formation of new nuclei. Nanostructures exposed to Pd vapor for 1000 s are covered with Pd particles with average sizes ranging between ∼2.5 and 3 nm and densities of ∼5.7 × 1012 cm-2. For Pd exposure times exceeding ∼3000 s, the nanoparticles begin to coalesce, resulting in 669

Figure 4. HAADF-STEM images of a nanowires (top row; a, c, e) and a nanobelts (bottom row; b, d, f) recorded after different palladium deposition times, as indicated: 210 s (a, b), 1000 s (c, d) and 5000 s (e, f).

the appearance wormlike Pd structures (Figures 4 and 5b). After 5000 s of Pd deposition, the average height of the Pd deposit is ∼3-3.5 nm, but their lateral dimensions are not well defined because of coalescence, and the average number density of the particles decreases. At room temperature, the observed morphology and evolution with Pd coverage of the Pd deposits on the SnO2 nanostructures follows a VolmerWeber growth mode in agreement with earlier data.16 The evolution of IDS with increasing Au deposition time is shown in Figure 5b. For exposure times exceeding ∼4 × 103 s a dramatic increase in conductance is observed. This manifests the onset of percolation between adjacent nanoparticles, essentially shorting out the SnO2 nanowire. Prior to reaching percolation and beginning immediately upon metal (Pd or Au) deposition one observes a significant reduction in conductance that plateaus (Figures 5b and 6a). This conductance reduction is almost certainly due to the creation of nanoscopic depletion regions (nano-Schottky barriers) surrounding the newly formed Pd (or Au) nanoparticles (see inset in Figure 6b). The origin of these regions is the small difference in work function between the nanoparticles and the n-type SnO2. The drop in conductance implies that a net electron transfer takes place from the semiconductor to the metal particles. It should be noted, however, that for a quantitative understanding a traditional picture of the Schottky junction applied to these “nanoSchottky junctions” may have limited value. This is due to the poorly determined depletion length established under conditions when only a few electrons are transferable to a small metal nanoparticle.17 However, the traditional picture and terminology may be applied for the qualitative approach, which suffices for the purposes of the current study. As the 670

Figure 5. (a) Pd cluster density measured on SnO2 nanostructures as a function of deposition time. (b) Source-drain current, IDS, through a nanobelt (VDS ) 2 V) measured during Au deposition. The onset of the large current increase beyond ∼4 × 103 s is due to the onset of Au particle percolation. Nano Lett., Vol. 5, No. 4, 2005

Figure 7. Response of a pristine (dashed line) and Pd-functionalized (solid line) nanostructure to sequential oxygen and hydrogen pulses at 473 K (top pane) and 543 K (bottom).

Figure 6. (a) Changes in the source drain during the early stage of Pd deposition onto a SnO2 nanobelt. The characteristic drop in IDS indicates the formation of electroactive elements (Schottkybarrier-like junctions) on the nanostructure’s surface, which locally deplete electrons from the support. (b) Schematic view of the formation of electron depleted regions beneath and in the immediate vicinity of two Pd nanoparticles.

metal particles nucleate and grow, IDS eventually plateaus to a saturation value when neighboring depletion regions begin to overlap. Previous studies involving metal particles on macroscopic SnO2 single crystals report an opposite behavior at early stages of Pd deposition;18 that is, a conductance increase was observed immediately upon the start of deposition followed by a slow decrease in conductance over several hours. The observed increase in conductance was ascribed to the rapid oxidation of the Pd by ionosorbed (ionically chemisorbed) oxygens with a resulting electron transfer into the SnO2 conduction band. The different result obtained in this study can thus be explained with the partial removal of as much ionosorbed oxygen from the tin oxide nanostructures during the annealing in a vacuum prior to Pd decoration in our experiments. In contrast, the studies reported in ref 18 were carried out on SnO2 surfaces which were purposefully enriched with ionosorbed oxygen. The sensing performance of a SnO2 nanowire toward sequential oxygen and hydrogen pulses at two different temperatures before (dashed curves) and after (solid curves) Pd deposition is shown in Figure 7. Functionalizing the nanowire surface with Pd leads to an enhancement in IDS response for both gases and to an increased speed of response. Raising the temperature raises the value of IDS and shortens the response time of the sensor to O2 and H2. This effect is most prominent for oxygen for which the nanostructure only becomes activated above 543 K (compare bottom and top panels of Figure 7). Nano Lett., Vol. 5, No. 4, 2005

The increased overall sensing performance of the nanostructure upon functionalization with Pd can be rationalized in terms of existing models developed for SnO2 thin film sensors.19-23 The receptor and transduction functions of the nanosensor can, in general terms, be divided into those operating with oxidizing species (in our case O2) and those operating with reducing gases. These two classes of interactions, although being different in the types of chemical reactions at the nanowire surface and in the resulting directions of charge transfer, are nevertheless similar in their principle transduction mechanism. We will concentrate our discussion on oxygen adsorption, since the interaction of hydrogen with pristine and Pd-functionalized SnO2 was thoroughly discussed in refs 24-27. The high conductance of pristine SnO2 nanostructures in vacuum at 400-600 K results from the presence of shallow donor states due to a high density of oxygen vacancies kT

V0 T Vγ+ 0 + γe , which can be totally (γ ) 2) or partially (γ ) 1) ionized, with the two species coexisting in various proportions according to (among other parameters) the temperature. The presence of Sn interstitials can also contribute to formation of the donor states.28 These defects render the oxide an n-type semiconductor. Depending on the degree of sample reduction, exposure to oxygen leads to some of the oxygen being dissociated at surface defects. These activated oxygen species passivate the surface vacancies, drawing electrons from the bulk and localizing them on the ionosorbed oxygens (process 1 in Figure 8a). This process is thermally activated as can be deduced by comparing the response of IDS for a pristine SnO2 nanostructure to oxygen (Figure 7) at lower and higher temperatures. In contrast to films with thicknesses D . λD (λD is the Debye length for n-SnO2 for a given temperature), in which oxygen chemisorption induces band bending near the surface, the effect of adsorption on small diameter nanowires or thin nanobelts is to change the location of the Fermi level within the band gap of the nanostructure so that when process 1 (Figure 8a) becomes efficient the accompanying electron depletion causes, in turn, an appreciable conductance drop (Figure 7, bottom panel).

671

Figure 8. (a) Schematic depiction of the three major process taking place at a SnO2 nanowire/nanobelt surface: (1) ionosorption of oxygen at defect sites of the pristine surface; (2) molecular oxygen dissociation on Pd nanoparticles followed by spillover of the atomic species onto the oxide surface; (3) capture by a Pd nanoparticle of weakly adsorbed molecular oxygen that has diffused along the tin oxide surface to the Pd nanoparticle’s vicinity (followed by process 2). RS is the effective radius of the spillover zone, and RC is the radius of the collection zone. (b) Band diagram of the pristine SnO2 nanostructure and in the vicinity (and beneath) a Pd nanoparticle. The radius of the depletion region is determined by the radius of the spillover zone.

Commonly, two concepts are invoked to explain the improvement of the nanowire’s sensing performance upon Pd deposition. The “electronic mechanism” proposes the formation of depletion zones around the particles (Figure 6b) and attributes the improved sensing to the modulation of the nano-Schottky barriers (and hence the width of the conduction channel) due to changes in the oxidation state of the Pd (and therefore its work function) accompanying oxygen adsorption and desorption. Although the conduction changes during Pd deposition (Figure 6) confirm the creation of electro-active elements on nanostructure’s surface, the observed magnitude of the changes in sensing properties, kinetics and temperature dependence brought about by Pd functionalization cannot be explained exclusively by this mechanism. An alternative “chemical” mechanism proposes that the Pd nanoparticles catalytically activate the dissociation of molecular oxygen, whose atomic products then diffuse to the metal oxide support. This process greatly increases both the quantity of oxygen that can repopulate vacancies on the SnO2 surface and the rate at which this repopulation occurs, resulting in a greater (and faster) degree of electron withdrawal from the SnO2 (and at a lower temperature) than for the pristine metal oxide nanowire. (Pd is a far better oxygen dissociation catalyst than tin oxide.) This mechanism (process 2 in Figure 8) is well-established in the catalysis literature and known as the “spillover effect”.29,30 The size of the spillover zone (denoted RS in Figure 8) for oxygen was directly evaluated for Pd particles on a TiO2 (110) surface using STM31 and reported to be ∼10 nm to several tenths of nanometers in diameter depending on the experimental conditions. Applying these dimensions to the results obtained from the TEM images (Figures 4 and 5), at particle coverages of ∼1012 cm-2 the spillover zones overlap essentially covering the entire support, leading to the observed 672

saturation of the effect of Pd deposition beyond a certain particle density that is nevertheless well below the percolation threshold. We note that for molecular oxygen to dissociate through this “chemical” mechanism it is not necessary for the oxygen molecules to impinge from the gas phase directly upon the Pd nanoparticle’s surface. Boudart et al.32 and later many other groups (see refs 33 and 34 and references therein) pointed out that oxygen molecules can reside briefly on an oxide support and diffuse to a catalyst particle before it has had an opportunity to desorb. Accordingly, the effective “capture radius” (RC in Figure 8) of a Pd nanoparticle can significantly exceed the nanoparticle’s radius. At the particle density cited, the effective oxygen “collection zones”32 overlap, making the entire surface of the nanostructure an effective oxygen delivery system for the Pd nanoparticles. This so-called “back-spillover effect” (process-3 in Figure 8a) further increases the probability of oxygen ionosorption on the SnO2 support. The net result of these two processes is a significant enhancement of the probability of oxygen ionosorption when an appropriate oxide (such as SnO2) is covered with Pd, which, in turn, is reflected in the observed increase in nanowire response to oxygen. Conclusions. HRTEM and STEM studies of Pd particles deposited on the surface of SnO2 nanowires and faceted nanobelts reveal that the metal nanoparticle nucleation and growth process follows a Volmer-Weber growth mode. For nanostructures with lateral dimension of ∼100 nm the nanoparticle growth and aggregation kinetics show no significant difference from processes on macroscopic, singlecrystal tin oxide surfaces. Pd metal particles appear to be metallic even after exposure to air at room temperature. However, HRTEM images alone cannot exclude the presence of a monolayer of PdO on nanoparticle surface. Au and Pd metal nanoparticles grown on reduced n-SnO2 nanowires and nanobelts were found to form Schottky-barrier like junctions, which locally deplete the nanostructure of electrons. Combining in situ deposition and sensing characterization, using the same nanodevice, provided a direct route to understanding the effect of surface functionalization by Pd, thereby avoiding concerns that some of the differences between the behavior of pristine and functionalized nanowires and nanobelts is due to the variations between individual nanostructures and their contacts to metal electrodes. The dramatic improvement in sensing performance observed upon sensitization with Pd was ascribed to the combined effect of spillover of atomic oxygen formed catalytically on the Pd particles then migrating onto the tin oxide, and the back spillover effect in which weakly bound molecular oxygens migrate to the Pd and are catalytically dissociated. As a result, both the delivery of activate species to, and the capture of precursors from the SnO2 nanostructure surface are promoted by catalytically active Pd nanoparticles. We show that a 1-D metal oxide nanostructure decorated with metal nanoparticles becomes a monitorable heterogeneous catalyst element on which the functions of the support and the catalytically active element can be decoupled and independently controlled while monitoring the processes using the electron transport properties of the nanowire Nano Lett., Vol. 5, No. 4, 2005

(nanobelt). By further reducing the lateral dimensions of the nanowire or nanobelt and carefully controlling the nucleation and growth of the catalytic particles one can envision the possibility of sensitively monitoring the surface chemistry taking place on a few or even on a single nanoparticle. Acknowledgment. We thank Professor Horia Metiu, Dr. Steeve Chretien and Professor Eric McFarland for many insightful discussions. This work was supported by AFOSR DURINT grant F49620-01-0459, by the Institute for Collaborative Biotechnologies through grant DAAD19-03-D0004 from the U.S. Army Research Office and made extensive use of the MRL Central Facilities at UCSB supported by the National Science Foundation under award nos. DMR-0080034 and DMR-0216466 for the HRTEM/ STEM microscopy. References (1) Cui, Y.; Wei, Q. Q.; Park, H. K.; Lieber, C. M. Science 2001, 293, 1289. (2) Law, M.; Kind, H.; Messer, B.; Kim, F.; Yang, P. D. Angew. Chem., Int. Ed. 2002, 41, 2405. (3) Comini, E.; Faglia, G.; Sberveglieri, G.; Pan, Z. W.; Wang, Z. L. Appl. Phys. Lett. 2002, 81, 1869. (4) Kolmakov, A.; Zhang, Y.; Cheng, G.; Moskovits, M. AdV. Mater. 2003, 15, 997. (5) Arnold, M. S.; Avouris, P.; Pan, Z. W.; Wang, Z. L. J. Phys. Chem. B 2003, 107, 659. (6) Li, C.; Zhang, D. H.; Liu, X. L.; Han, S.; Tang, T.; Han, J.; Zhou, C. W. Appl. Phys. Lett. 2003, 82, 1613. (7) Wang, Y. L.; Jiang, X. C.; Xia, Y. N. J. Am. Chem. Soc. 2003, 125, 16176. (8) Zhang, D. H.; Liu, Z. Q.; Li, C.; Tang, T.; Liu, X. L.; Han, S.; Lei, B.; Zhou, C. W. Nano Lett. 2004, 4, 1919. (9) Patolsky, F.; Zheng, G. F.; Hayden, O.; Lakadamyali, M.; Zhuang, X. W.; Lieber, C. M. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 14017. (10) Pan, Z. W.; Dai, Z. R.; Wang, Z. L. Science 2001, 291, 1947. (11) Dai, Z. R.; Gole, J. L.; Stout, J. D.; Wang, Z. L. J. Phys. Chem. B 2002, 106, 1274.

Nano Lett., Vol. 5, No. 4, 2005

(12) Kolmakov, A.; Moskovits, M. Annu. ReV. Mater. Res. 2004, 34, 151. (13) Arbiol, J.; Cirera, A.; Peiro, F.; Cornet, A.; Morante, J. R.; Delgado, J. J.; Calvino, J. J. Appl. Phys. Lett. 2002, 80, 329. (14) Nepijko, S. A.; Klimenkov, M.; Adelt, M.; Kuhlenbeck, H.; Schlogl, R.; Freund, H. J. Langmuir 1999, 15, 5309. (15) Sun, X. H.; Teo, B. K. J. Cluster Sci. 2004, 15, 199. (16) Poirier, G. E.; Cavicchi, R. E.; Semancik, S. J. Vac. Sci. Technol., A 1994, 12, 2149. (17) Zhdanov, V. P. Surf. Sci. 2002, 512, L331. (18) Cavicchi, R.; Sukharev, V.; Semancik, S. Surf. Sci. 1998, 418, L81. (19) Yamazoe, N. Denki Kagaku 1982, 50, 29. (20) A. Hilger Series on Sensors; Moseley P. T., Tofield B. C., Eds.; Philadelphia, 1997. (21) Kohl, D. J. Phys. D: Appl. Phys. 2001, 34, R125. (22) Gopel, W. Sens. Actuators 1989, 16, 167. (23) Barsan, N.; Weimar, U. J. Electroceram. 2001, 7, 143. (24) Fryberger, T. B.; Semancik, S. Sens. Actuators, B 1990, 2, 305. (25) Roessner, F.; Roland, U.; Dmitriev, R. V. Spillover Assisted Catalytic Effects (SPACE) - a New Reaction Model For Bifunctionally Catalyzed Hydrocarbon Conversions. In SpilloVer and Migration of Surface Species on Catalysts; Li, C., Xin, Q., Eds.; Elsevier: New York, 1997; Vol. 112, p 63. (26) Roland, U.; Roessner, F. A New Model On The Nature Of SpiltOver Hydrogen. In SpilloVer and Migration of Surface Species on Catalysts; Li, C., Xin, Q., Eds.; Elsevier: New York, 1997; Vol. 112, p 191. (27) Boudart, M. J. Mol. Catal. A: Chem. 1999, 138, 319. (28) Kilic, C.; Zunger, A. Phys. ReV. Lett. 2002, 88. (29) Khoobiar, S. J. Phys. Chem. 1964, 68, 411. (30) Boudart, M.; Vannice, M. A.; Benson, J. E. Z. Phys. Chem. (Frankfurt) 1969, 64, 171. (31) Bennett, R. A.; Stone, P.; Bowker, M. Faraday Discuss. 1999, 267. (32) Tsu, K.; Boudart, M. Can. J. Chem., ReV. Can. Chim. 1961, 39, 1239. (33) Henry, C. R.; Chapon, C.; Duriez, C. J. Chem. Phys. 1991, 95, 700. (34) Bowker, M. ‘Seeing’ the active site in catalysis. STM and molecular beam studies of surface reactions. In 11th International Congress on Catalysis - 40th AnniVersary, Pts a and B, 1996; Vol. 101, p 287.

NL050082V

673