New and Future Developments in Catalysis - Read Online

Book Preview

New and Future Developments in Catalysis

You've reached the end of this preview. Sign up to read more!
Page 1 of 1

1

Gold-Based Catalysts for CO Oxidation, the Water-Gas Shift, and Desulfurization Processes

Jose A. Rodriguez,     Chemistry Department, Brookhaven National Laboratory, Upton, NY 11973, USA, rodrigez@bnl.gov

Acknowledgments

Many of the studies described here were done in collaboration with Z. Chang, J. Hanson, J. Evans, J. Hanson, J. Hrbek, F. Illas, T. Jirsak, J. Graciani, G. Liu, P. Liu, S.Ma, A. Maiti, K. Nakamura, J.-B. Park, M. Pérez, J.F. Sanz, S. Senanayake, D. Stacchiola, Y. Takahashi, and X. Wang. I am very grateful to all of them. This research was supported by the US Department of Energy, Division of Chemical Sciences.

1.1 Introduction

Recently, gold has become the subject of a lot of attention due to its unusual catalytic properties when dispersed on some oxide and carbide supports [1–11]. Bulk metallic gold is a very poor agent for the activation of molecules typically used in catalysis (H2, O2, CO, C2H4, etc.) [12,13], but atomic or molecular species bonded to low-index single crystal gold surfaces are chemically active [14–16]. Among the transition metals, gold is by far the least reactive and is often referred to as a coinage metal. In valence photoemission spectra for metallic gold [17], states with Au 6s,p character appear from 0 to 2 eV, while the Au 5d states extend from 2 to 8 eV. The low reactivity of metallic Au is a consequence of combining a deep-lying valence 5d band and very diffuse valence 6s,p orbitals [12,17].

Many experimental and theoretical studies have been focused on understanding the high catalytic activity of gold nanoparticles supported on oxides and carbides [1–11,18]. Quantum effects related to the small size of the particles could be responsible for the enhancement in catalytic activity with respect to bulk gold, but it is becoming more and more clear that interactions between the gold nanoparticles and the oxide or carbide support play a very important role [10a,10d,11,18]. The edge and corner sites of a gold nanoparticle (i.e., sites which have three to four metal atom neighbors) can bond well adsorbates like CO, O2, and SO2. They can even perform the catalytic oxidation of CO, but for more demanding reactions the chemical activity of the isolated Au nanoparticles is not enough. A comparison of the DeSOx activity for the Au/TiO2(1 1 0), Au/MgO(1 0 0), Au/TiC(0 0 1) surfaces illustrates the important role played by gold↔substrate interactions. The TiO2 and TiC supports are not simple spectators [11].

The next section of the chapter will discuss fundamental studies examining the bonding interactions of gold with metal oxide and carbide surfaces. Then, we will focus on the use of gold-based catalysts in CO oxidation, the water-gas shift, and the destruction of SO2.

1.2 Bonding Interactions Between Gold and Metal Oxide or Carbide Surfaces

Results of scanning tunneling microscopy (STM) and transmission electron microscopy (TEM) indicate that Au grows on most metal oxide surfaces forming three-dimensional (3D) particles [2b,10d,19]. For example, at 300 K, Au clusters nucleate mainly on step sites or on defect sites present in the terraces of TiO2(1 1 0) [2b,20,21]. The interaction of Au with an ideally flat TiO2(1 1 0) surface is quite weak [22], and a substantial amount of particle sintering occurs when the system is annealed from 25 to 600 K [2d,19]. In general, density functional (DF) calculations for the adsorption of Au atoms on MgO(0 0 1) and TiO2(1 1 0) give bonding energies below 0.5 eV [4,6,8a,10]. Only in the presence of O vacancies strong bonding interactions are seen between Au and the oxide surfaces [8a,10,22]. On the O vacancies, an oxide→Au charge transfer takes place [8a,10]. Studies of scanning tunneling spectroscopy (STS) indicate that the Au clusters supported on TiO2(1 1 0) have a small band gap (0.2–0.6 V) and electronic properties different from those of bulk metallic Au [2b]. This is important, since such a difference could be responsible for the variation in chemical activity when going from the nanoparticles to bulk gold [2b].

What happens when Au is deposited on a substrate which has physical and chemical properties different from those of an oxide? The carbides of the early-transition metals exhibit, in many aspects, a chemical behavior similar to that of very expensive noble metals (Pt, Pd, Ru, or Rh) [23]. Transition metal carbides exhibit broad and amazing physical and chemical properties [23–25]. Their properties may be viewed as resulting from a combination of those of covalent solids, ionic crystals, and transition metals [23,25–27]. In recent studies [11,18], high-resolution photoemission, STM, and DF calculations were used to study the adsorption of gold on a TiC(0 0 1) surface. The Au↔TiC(0 0 1) interactions were much stronger than typical Au ↔ oxide interactions [18]. For example, the calculated binding energies for an Au atom on TiC(0 0 1) and TiO2(1 1 0) are −1.91 eV [18] and −0.38 eV [8a], respectively. For Au/TiC(0 0 1), the photoemission and DF results point to the formation of Au—C bonds [18]. Overall the bond between Au and a TiC(0 0 1) surface exhibits very little ionic character, but there is a substantial polarization of electrons around Au [18] which enhances the chemical activity of this metal [11,28,29].

Figure 1.1 displays electron-localization function (ELF) plots for Au4 on a series of MC(0 0 1) (M = Ti, Zr, V, and Mo) [18,30]. In all these systems, there is a substantial concentration of electrons in the region outside the Au4 unit. A similar phenomenon was observed for Au, Au2, and other small clusters containing one layer of gold in contact with the carbide substrates [18,30]. In the case of clusters with two layers of Au, the electron polarization for the second layer of atoms was much less pronounced than that seen for the Au4/MC(0 0 1) systems [18,30]. On the basis of the charge polarization induced by the carbide substrates, one can expect big differences between the chemical reactivity of 2D and 3D gold clusters. The Au4/MC(0 0 1) systems should be very active in the bonding of electron acceptor molecules (CO, O2, C2H2, C2H4, SO2, etc.).

Figure 1.1 Calculated electron-polarization plots for a series of Au4/MC(0 0 1) surfaces (M = Ti, VC, Zr, Mo). Taken from Ref. [30], Copyright 2010 American Chemical Society.

1.3 Oxidation of Carbon Monoxide on Au-Oxide and Au-Carbide Surfaces

The oxidation of carbon monoxide (2CO + O2 → 2CO2) over metal/oxide catalysts is significant for understanding fundamental processes associated with methanol synthesis, the water-gas shift reaction, the reforming of alcohols, the operation of fuel cells, and automotive exhaust controls, to mention a few major applications in the area of catalysis [2,3,6,10,13,22]. In automotive exhaust emission control, the complete oxidation of carbon monoxide is of prime importance to meet increasingly stringent environmental regulations in a practical way [13]. Furthermore, since the classic studies of Langmuir, CO oxidation on metal and oxide surfaces is often viewed as an ideal reaction for fundamental investigations in heterogeneous catalysis [2,3,6,10,13,22]. High surface area Au/TiO2 catalysts are very efficient for the oxidation of CO (6,23) [1,2b,31,32]. The bottom trace in Figure 1.2 shows how the CO oxidation activity of an Au/TiO2(1 1 0) surface changes as a function of Au coverage [33]. A maximum activity is found for an Au coverage of ∼0.3 monolayer (ML). When these activity data are put together with STM results [2b,32], one finds that there is a marked size effect on the catalytic activity, with Au clusters in the range of 3–4 nm exhibiting the maximum reactivity. For this size, most of the particles have a band gap of 0.2–0.6 V according to scanning tunneling spectroscopy (STS) [2b]. Particles with a larger band gap (>1 V) display a lower reactivity, and particles with metallic character (band gap ∼0 V) are the least active. Thus, there is a correlation between the electronic and chemical properties of the supported Au nanoparticles. Studies of STM indicate that exposure to CO has no effect on the morphology of the Au/TiO2(1 1 0) surface [2b]. On the other hand, significant morphological changes occur after exposure to O2 or CO:O2 mixtures. In these cases, the Au cluster density is greatly reduced as a result of sintering [2b]. This sintering eventually leads to a decrease in the CO oxidation activity of the Au/TiO2(1 1 0) systems.

Figure 1.2 Oxidation of CO on Au/TiO2(1 1 0) and Au/CeOx/TiO2(1 1 0) surfaces as a function of Au coverage. In the case of Au/CeOx/TiO2(1 1 0), ∼12% of the titania surface was covered with ceria. Taken from Ref. [33], Copyright 2009 National Academy of Sciences USA.

In recent studies, the CO oxidation on Au/TiO2 catalysts occurs on metal sites at the gold-oxide interface [32,34]. Infrared-kinetic measurements indicate that O—O bond scission is activated by the formation of a CO—O2 complex at dual Ti–Au sites at the Au/TiO2 interface. DF, which provides the activation barriers for the formation and bond scission of the CO—O2 complex, confirms this model. The observation of sequential delivery and reaction of CO first from TiO2 sites and then from Au sites indicates that catalytic activity occurs at the perimeter of Au nanoparticles [32,34].

The catalytic activity of Au–TiO2 can be improved by the addition of ceria [33,35]. The ceria helps with the dispersion on the gold enhancing the rate of CO oxidation. Figure 1.3 shows an STM image acquired after depositing Ce on TiO2(1 1 0) under an atmosphere of O2[33]. Most of the spots (∼70%, labeled a) have a height of 1.3 ± 0.2 Å and correspond to small wires of CeOx[33]. A minority of the spots (∼30%, labeled b) have a height of 1.9 ± 0.3 Å and probably correspond to (1 × 2) reconstructions of TiO2(1 1 0) induced by O2 chemisorption [33,36,37]. Figure 1.3b displays an STM image taken after depositing ∼0.25 ML of Au on the CeOx/TiO2(1 1 0) surface of Figure 1.3a. The deposition of Au was done at room temperature. One can see particles of Au that were simultaneously located on a and b sites. When STM images for the Au/CeOx/TiO2(1 1 0) system are compared to those collected for plain Au/TiO2(1 1 0) [2b], the presence of ceria favors the dispersion of the gold on the titania terraces and the metal particles are not located mainly at steps of the surface as it happens in the case of Au/TiO2(1 1 0). The close contact between Au and CeOx leads to a very high catalytic activity. In Figure 1.2, the rate for CO oxidation on Au/CeOx/TiO2(1 1 0) system is 3–5 times larger than on plain Au/TiO2(1 1 0). In addition to helping with the dispersion of the Au, ceria also helps with the binding and dissociation of O2[33].

Figure 1.3 STM study of the morphology of Au/CeOx/TiO2(1 1 0). (a) STM image of a CeOx/TiO2(1 1 0) surface. Ce was deposited at 327 °C under an atmosphere of O2 (∼1 × 10−7 Torr) and then the sample was annealed at 900 K in O2. (b) STM image obtained after depositing gold on the CeOx/TiO2(1 1 0) surface that gave image (a) at room temperature. Both STM images correspond to an area of 20 × 20 nm. Taken from Ref. [33], Copyright 2009 National Academy of Sciences USA.

Another system extensively studied for CO oxidation is Au/MgO [4a,9,10,38]. When CO and O2 are coadsorbed on Au/MgO(1 0 0) at 100 K, the evolution of CO2 is observed at temperatures around 200 K or lower depending on the coverage of Au [10,38]. A key aspect for the reactivity of Au clusters on MgO(1 0 0) is their structural fluxionality that allows the adsorption and activation of O2[39]. DF calculations indicate that the active sites for CO oxidation on Au/MgO(1 0 0) involve low-coordinated Au atoms and Mg cations [4,38,39]. The oxide stabilizes a peroxolike intermediate, CO•O2, and then the oxidation reaction proceeds in the metal/oxide interface, see Figure 1.4. Such a reaction pathway is consistent with studies for Au/TiO2 catalysts described above [8b,32,34], which show that a strong contact between the Au nanoparticles and oxide support is indispensable for high catalytic activity because the periphery sites probably carry out the oxidation of CO.

Figure 1.4 Catalytic cycle for the oxidation of CO at an Au/MgO interface. Equilibrium structures as well as the potential energies with respect to gas phase 2 CO and O2 are shown for each relevant stage of the reaction. Taken from Ref. [4a], Copyright 2005 Elsevier.

Au nanoparticles dispersed on TiC films and TiC(0 0 1) can oxidize carbon monoxide at temperatures below 200 K [29,40]. After coadsorbing CO and O2 at ∼100 K, the evolution of CO2 is observed at 160–180 K when the temperature is ramped up [29,40]. Photoemission and temperature programmed desorption were used to study the adsorption of O2 on Au/TiC(0 0 1) and the subsequent oxidation of CO [40]. After dosing molecular oxygen to a TiC(0 0 1) surface pre-covered with 0.2 ML of Au at 150 K, the corresponding O 1s spectrum was dominated by a peak which corresponded to adsorbed O2. There was also an O2-induced shift in the binding energy of the Au 4f7/2 peak that implied direct Au–O2 bonding interactions. Heating to higher temperatures induced the dissociation of the adsorbed O2 at higher temperatures than that found for the formation of CO2 during the 2CO + O2 reaction [40]. Thus, the oxygen species active during the low-temperature oxidation of carbon monoxide on Au/TiC(0 0 1) is chemisorbed O2, as occurs in the case of Au/oxide surfaces. A large elongation (∼0.3 Å) calculated for the O—O bond upon adsorption of molecular oxygen on 2D Au clusters supported on TiC(0 0 1) [40] facilitates interactions with CO and the eventual cleavage of the bond to produce CO2 (OC ↔ O–O ↔ CO → 2CO2). An important difference with respect to the Au/oxide systems is in the origin of the oxidized CO molecule. Due to weak CO ↔ oxide interactions, the CO oxidized on Au/oxide surfaces is bonded to Au nanoparticles before reacting with O2 on Au sites or at the Au-oxide interface, see Figure 1.4. In the case of Au/TiC(0 0 1), the CO could be bound to Au sites or to the TiC substrate before reacting with O2[40].

1.4 Water-Gas Shift Reaction on Au-Oxide Surfaces

Currently, the primary source of hydrogen for the chemical and petrochemical industries comes from the steam reforming of hydrocarbons: CnHm + nH2O → nCO + (n − m/2)H2[13]. The reformed fuel usually contains 1–10% of CO, an impurity that can be a serious problem for chemical processes which use H2 as a feedstock. The water-gas shift reaction (WGS, CO + H2O → H2 + CO2) is critical for providing clean hydrogen [13]. Common industrial catalysts for the WGS (mixtures of Fe–Cr or Zn–Al–Cu oxides) are pyrophoric and normally require lengthy and complex activation steps before usage [13]. Recent works report that Au nanoparticles supported on oxides such as CeO2 and TiO2 are very efficient catalysts for the WGS reaction [5c]. This is remarkable since neither bulk Au nor bulk ceria and titania are known as WGS catalysts.

The as-prepared Au–CeO2 catalysts contain nanoparticles of pure gold and gold oxides (AuOx) dispersed on a nanoceria support [41]. Each of these gold species could be in the active phase and the ceria support may not be a simple spectator in these systems [41]. The top panel in Figure 1.5 shows data for the production of H2 and CO2 during the WGS over a powder gold-ceria catalyst [41]. The catalyst was held at temperatures of 300, 400, and 500 °C. The experimental setup did not detect significant catalytic activity at temperatures below 250 °C [41]. The chemical state of gold during the WGS was determined by means of in situ time-resolved X-ray absorption near-edge spectroscopy (XANES). The bottom panel in Figure 1.5 displays Au L3-edge XANES spectra collected at room temperature for fresh catalysts with an Au content of 0.5 wt% (dashed trace) or 2.4 wt% (solid traces) [41]. The line shape of these two spectra is very similar and shows a clear feature at ∼2.5 eV above the edge that is not seen for metallic gold and is characteristic of gold oxides [41,42]. The intensity of this peak is higher than that observed for Au2O and closer to that seen in Au2O3[42]. Once the 2.4 wt%Au–CeO2 catalyst was exposed to a mixture of CO/H2O at elevated temperatures, the XANES features for gold oxide disappeared. At temperatures above 200 °C, when significant WGS activity was detected, the line shape of the Au L3-edge resembled that of pure gold. The XANES spectra in Figure 1.5 were obtained under a reaction mixture of 5% CO and 3% H2O in He (total flow ∼10 ml/min) [41]. Similar results were found when using a 1% CO and 3% H2O in He reaction mixture. Thus, the in situ time-resolved X-ray absorption spectroscopy (XAS) data indicate that cationic Auδ+ species cannot be the key sites responsible for the WGS activity in Figure 1.5, because they do not exist under reaction conditions [41]. An identical finding has been reported for AuOx/Ce1−xZrxO2−y powder catalysts [5c]. In these catalysts, the active phase consisted of small Au aggregates (<2 nm in size) dispersed on partially reduced ceria (CeO1.94–CeO1.98) [5c,41].

Figure 1.5 Top panel: Relative amounts of H2 and CO2 formed during the WGS over a 2.4 wt% Au-ceria catalyst. A mixture of 5% CO and 3% H2O in He (total flow ∼10 ml/min) was passed over the catalyst at 300, 400, or 500 °C. Bottom panel: Au L3-edge XANES spectra collected in situ during the WGS reaction over the 2.4 wt% Au–CeO2 catalyst. For comparison, we also include the spectra for a fresh 0.5 wt% Au–CeO2 catalyst, dashed trace, and a gold foil. The vertical lines indicate the main features for AuOx versus metallic Au. Taken from Ref. [41], Copyright 2005 American Institute of Physics.

The kinetics of the WGS reaction have been investigated in detail on model catalysts generated by vapor-depositing nanoparticles of gold on CeO2(1 1 1) [3,5b], ZnO(0 0 0 ı¯) [5b], TiO2(1 1 0) [43], and polycrystalline MoO2[44]. On these oxide substrates, gold grows forming three-dimensional (3D) particles. Figure 1.6 displays the behavior of the Au/CeO2(1 1 1) and Au/ZnO(0 0 0 ı¯) catalysts as a function of gold coverage [5b]. CeO2(1 1 1) and ZnO(0 0 0 ı¯) are both O-terminated surfaces. These oxide supports are inactive as catalysts for the WGS reaction. In Figure 1.6, the catalytic activity of the metal/oxide systems initially increases when Au is added, reaching a maximum at ∼0.4–0.5 monolayer (ML). Above these coverages, the overall catalytic activity decreases. Similar trends have been observed for the WGS on Au/TiO2(1 1 0) [43] and Au/MoO2[44]. STM images show that the particle size of Au on CeO2(1 1 1), ZnO(0 0 0 ı¯), or TiO2(1 1 0) raises above 4 nm and continuously grows when the admetal coverage is increased beyond 0.5 ML. The trends in Figure 1.6 probably reflect changes in the size of the gold particles: High catalytic activity is seen for small gold particles (size <4 nm), and decreases as the particle size increases. Although the optimum WGS activity in Figure 1.6 is for admetal coverages of 0.4–0.5 ML, metal/oxide catalysts with gold coverages near 1 ML are still substantially more active than Au(1 1 1) or polycrystalline gold, surfaces which are not catalytically active [5b].

Figure 1.6 WGS activity of model Au/CeO2(1 1 1) and Au/ZnO(0 0 0 ı¯) catalysts as a function of gold coverage. Each surface was exposed to a mixture of 20 Torr of CO and 10 Torr of H2O at 625 K for 5 min. Steady state was reached 2–3 min after introducing the gases in the batch reactor. Taken from Ref. [5b], Copyright 2007 Wiley.

The gold atoms in the Au/CeO2(1 1 1) and Au/ZnO(0 0 0 ı¯) catalysts were probably not oxidized during the WGS process [5b]. After reaction, XPS showed Au 4f positions that were almost identical to those seen upon deposition of gold on the oxides and very different from those typically seen for AuOx species [41,5b]. Post-reaction surface analysis also showed the presence of formate- and/or carbonate-like groups on the surface of the catalysts. Possible reaction paths for the formation of these groups are discussed in Ref. [5c]. It is not completely clear if they are key intermediates in the WGS process or simple spectators.

Why are extended surfaces of gold inactive as catalysts for the WGS reaction? Figure 1.7 shows the calculated energy profile for the WGS on periodic Cu(1 0 0) and Au(1 0 0) surfaces [45]. All the adsorbates bond more weakly on Au(1 0 0) than on Cu(1 0 0). Consequently, the rate-limiting dissociation of H2O on Au(1 0 0) is highly endothermic (ΔE3 = +0.74 eV) and the corresponding barrier is also high (ΔEa3 = +1.53 eV). These DFT results are in agreement with experimental measurements, which show that Cu is a good WGS catalyst while Au is an extremely poor one [5b,46,47]. The results in Figure 1.7 indicate that gold will be an excellent WGS catalyst if in some way is helped with the dissociation of water [45]. Assuming that OH can be formed on Au(1 0 0) or Au(1 1 1), subsequent steps for the WGS process should occur readily on the gold substrate [45]. Indeed, experimental and theoretical studies have shown that water dissociates on O/Au(1 1 1) yielding hydroxyls which react with CO to produce CO2 and hydrogen [14].

Figure 1.7 DF calculated reaction profile for the WGS on Cu(1 0 0) and Au(1 0 0). Taken from Ref. [45], Copyright 2007 American Institute of Physics.

One can obtain a stable catalyst for the WGS by adding CeOx nanoparticles to Au(1 1 1) [16,48]. Figure 1.8 displays an Arrhenius plot for the WGS activity of a CeOx/Au(1 1 1) surface in which 20% of the gold substrate was covered by ceria [48]. For comparison we also include results obtained for the WGS on Cu(1 0 0) [5b], Cu(1 1 1) [49], and Cu/ZnO(0 0 0 ı¯) [5b]surfaces. The results in Figure 1.8 indicate that the inverse CeOx/Au(1 1 1) catalyst exhibits a larger WGS activity than those of copper surfaces or even Cu nanoparticles dispersed on a ZnO(0 0 0 ı¯) substrate. On Cu(1 1 1) and Cu(1 0 0), the apparent activation energies for the WGS are 18.1 and 15.2 kcal/mol, respectively [5b,49]. The apparent activation energy decreases to 12.4 kcal/mol on Cu/ZnO(0 0 0 ı¯) [5b]and 10.3 kcal/mol on CeOx/Au(1 1 1) [48]. In the inverse CeOx/Au(1 1 1) catalyst, the reactants can interact with defect sites of ceria nanoparticles, metal sites of the support, or the metal-oxide interface [16,48]. One can gain activity due to the active participation of oxide in the catalytic reaction [50].

Figure 1.8 Arrhenius plot for the WGS reaction rate on Cu(1 1 1), Cu(1 0 0), Cu/ZnO(0 0 0 ı¯) and on an Au(1 1 1) surface approximately 20% covered by ceria. The data were acquired with a pressure of 20 Torr of CO and 10 Torr of H2O and temperatures of 575, 600, 625, and 650 K. Taken from Ref. [48], Copyright 2010 Elsevier.

The studies described above indicate that highly active WGS catalysts are bifunctional with the metal and oxide catalyzing different parts of the reaction. To optimize the performance of these systems, one must enhance the participation of the metal and oxide phases in the catalytic process. Figure 1.9 shows three different configurations in which a metal and an oxide can be combined in a catalyst. In a conventional metal/oxide configuration, one enhances the reactivity of the metal but this usually covers the defect sites of the oxide (nucleation centers for the metal particles) which have chemical activity. This is not the case in the inverse oxide/metal catalyst, which enhances the participation of the oxide in the catalytic reaction [50]. In the inverse catalyst, the reactants can interact with defect sites of oxide nanoparticles, metal sites of the support, or the metal-oxide interface [50]. In the quest to optimize the reactivity of the metal and oxide phases, the most complex and promising configuration is a mixed-metal oxide array in which nanoparticles of a metal and an oxide can interact with the reactants. Extremely active WGS catalysts have been found after coadsorbing nanoparticles of gold and ceria on a TiO2(1 1 0) substrate [33,35].

Figure 1.9 Different configurations for a metal-oxide catalyst.

As shown in Figure 1.3, CeOx nanoparticles drastically affect the growth mode of Au on TiO2(1 1 0) [33]. The deposition of Au on CeOx/TiO2(1 1 0) produces nanoparticles of the admetal anchored on the CeOx nanoparticles. The high dispersion of Au helps to make Au/CeOx/TiO2(1 1 0) surfaces as outstanding catalysts for the WGS as shown in Figure 1.10. Au/TiO2(1 1 0) surfaces are good WGS catalysts but they did not come close to match the activity of Au/CeOx/TiO2(1 1 0) [33]. The same is valid when comparing to the WGS activities of Au/CeO2(1 1 1) [5b], CeOx/Au(1 1 1) [16,48], Cu/ZnO(0 0 0 ı¯) [5b], and copper single crystals [5b,46,47]. For the Au/CeOx/TiO2(1 1 0) catalyst in Figure 1.10, one could assume that the concentration of active sites is proportional to the number of ceria regions in contact with gold nanoparticles [33]. Since only 12% of the titania support was covered by ceria, as measured by ion scattering spectroscopy (ISS), the Au/CeOx/TiO2(1 1 0) catalyst must be at least 300 times more active than a Cu(1 0 0) surface on a per active-site basis[33]. In Au/CeOx/TiO2(1 1 0), one has a bifunctional catalyst: The adsorption and dissociation of water takes place on the oxide, CO adsorbs on the gold nanoparticles, and all subsequent reaction steps occur at oxide-metal interfaces. Au nanoparticles do catalyze the reaction of OH with CO to yield a HOCO intermediate and then H2 and CO2[45]. The results in Figure 1.10 illustrate the tremendous impact that an optimization of the chemical properties of gold and ceria can have on the activity of a WGS catalyst [33,51]. The catalyst configuration shown at the bottom of Figure 1.9 points to new ways for preparing highly active gold-oxide WGS catalysts.

Figure 1.10 Top panel: Water-gas shift activity of Au/TiO2(1 1 0) [25] and Au/CeOx/TiO2(1 1 0) as a function of Au coverage. The area of TiO2(1 1 0) covered by CeOx was measured with ISS, before depositing gold, and found to be ∼12% of the clean substrate. The reported values for the production of H2 (blue curve) and CO2 (black curve) were obtained after exposing the catalysts to 20 Torr of CO and 10 Torr of H2O at 625 K for 5 min [25,29]. The number of H2 and CO2 molecules produced is normalized by the sample surface area. Bottom panel: Comparison of the water-gas shift activity of Cu(1 0 0), Au(1 1 1), and 0.5 ML of Au supported on TiO2(1 1 0), CeO2(1 1 1), or CeOx/TiO2 (1 1 0). Taken from Ref. [33], Copyright 2009 National Academy of Sciences USA.

1.5 Decomposition of Sulfur Dioxide on Au-Oxide and Au-Carbide Surfaces

The destruction of SO2 (DeSOx) is a very important problem in environmental chemistry [52,53]. SO2 is frequently formed during the combustion of fossil-derived fuels in factories, power plants, houses, and automobiles [52,53]. Every year the negative effects of acid rain (main product of the oxidation of SO2 in the atmosphere) on the ecology and corrosion of monuments or buildings are tremendous [52,53]. Thus, new environmental regulations emphasize the need for more efficient technologies to destroy the SO2 formed in combustion processes [52–54]. Titania is the most common catalyst used in the chemical industry and oil refineries for the removal of SO2 through the Claus reaction: SO2 + 2H2S → 2H2O + 3Ssolid[55]. Different approaches are being tested for improving the performance of titania in DeSOx operations [52,55]. The addition of gold to TiO2 produces desulfurization catalysts with a high efficiency for the cleavage of S—O bonds [56]. In this respect, the Au/TiO2 system is much more chemically active than either pure titania or gold. Surfaces of metallic gold interact very weakly with SO2 and the molecule desorbs intact at temperatures below 200 K [56–58].

Figure 1.11a shows S 2p photoemission spectra for the adsorption of SO2 on TiO2(1 1 0) and Au/TiO2(1 1 0) at 300 K [56]. The main product of the adsorption of SO2 on stoichiometric TiO2(1 1 0) are SO3 and SO4 species [56,57]. A substantial concentration of O vacancies on the oxide surface is necessary to induce the decomposition of SO2 at high temperatures (>400 K). In contrast, Au/TiO2(1 1 0) surfaces fully dissociate SO2 at room temperature [56]. Figure 1.11a shows S 2p photoemission spectra for the adsorption of SO2 on TiO2(1 1 0) and Au/TiO2(1 1 0) at 300 K. The S 2p spectra indicate that upon adsorption of SO2 on Au/TiO2(1 1 0), SO4 and atomic S (produced by the full dissociation of SO2) coexist on the surface. For the systems in Figure 1.11a, the larger the Au coverage on titania (0.05 → 0.5 ML range), the bigger the amount of atomic S deposited. This trend points to a direct involvement of gold in the dissociation of SO2. A large shift in the corresponding Au 4f core level spectra also supports this idea [46]. For the spectra in Figure 1.11b, the Au/TiO2(1 1 0) surfaces were prepared by dosing Au at 300 K, followed by brief annealing to 750 K, and in the final step SO2 was dosed at 300 K. Annealing at temperatures near 750 K is a step in the preparation of many Au/TiO2(1 1 0) surfaces and is known to induce the formation of O vacancies in the titania substrate. The presence of O vacancies seems to favor the breaking of S—O bonds since the pre-annealed surface displays a decrease in the intensity of the SO4 signal and an increase in the S signal with respect to Au/TiO2(1 1 0) surfaces that were not pre-annealed (Figure 1.11a) [56].

Figure 1.11 S 2p photoemission spectra for the adsorption of SO2 on TiO2(1 1 0) and Au/TiO2(1 1 0) at 300 K. The spectra in part a are for Au/TiO2(1 1 0) surfaces in which SO2 was dosed after depositing Au on TiO2(1 1 0) at 300 K. On the other hand, the Au/TiO2(1 1 0) surfaces in parts b and c were prepared by deposition of Au at 300 K, followed by annealing at 750 K before dosing SO2 at room temperature. In these experiments, all the surfaces were exposed to 20 L of SO2. Taken from Ref. [56], Copyright 2002 American Chemical Society.

In Figure 1.11c, one can see the effect of annealing a SO2/Au/TiO2(1 1 0) surface from 300 to 700 or 850 K. These elevated temperatures are also used in DeSOx operations that use titania as a catalyst/sorbent [55]. In Figure 1.11c, an increase in the sample temperature leads to desorption of SO2 and a SO4,ads → Sads transformation. At the end of the experiment one finds that the amount of S deposited on the Au/TiO2(1 1 0) is much larger than observed in the case of a plain TiO2(1 1 0) surface [56]. All the photoemission results in Figure 1.11 indicate that the Au/TiO2 system has a high DeSOx activity.

MgO is another oxide that is frequently used in DeSOx processes [55,58]. The DeSOx activity of MgO can be enhanced by doping the oxide lattice with another metal (CrxMg1−xO2, NixMg1−xO2, FexMg1−xO2) [55,56] or by adding a metal to the surface of MgO [58]. Experiments were carried out to investigate the adsorption and possible decomposition of SO2 on Au/MgO(1 1 0) surfaces [58]. At 100–150 K, SO2 adsorbs molecularly on the Au particles deposited on MgO(1 0 0) and desorbs, with negligible decomposition into atomic S, at temperatures below 300 K [60]. The bonding energy of SO2 on the gold clusters supported on MgO(1 0 0) is significantly larger than on an extended Au surface, top panel in Figure 1.12, but there is no dissociation of S—O bonds [60]. The graph in the bottom of Figure 1.12 compares S 2p areas measured for atomic S after dosing the same amount of SO2 to Au/TiO2(1 1 0) and Au/MgO(1 0 0) surfaces at 300 K [60]. Neither TiO2(1 1 0) nor MgO(1 0 0) are able to dissociate SO2 on their own. On both oxide supports the largest activity for the full dissociation of SO2 is found in systems that contain Au coverages smaller than 1 ML when the average diameter of the nanoparticles is below 5 nm [56,60]. Clearly the Au/TiO2(1 1 0) systems are much more chemically active than the Au/MgO(1 0 0) systems. Catalytic tests also show that Au/TiO2 is much more active than Au/MgO for the Claus reaction or the reduction of SO2 with CO [59]. These data indicate that titania either plays a direct active role in the dissociation of SO2 or modifies the chemical properties of the supported Au nanoparticles. Full decomposition of SO2 is observed on TiO2(1 1 0) only after the formation of O vacancies [56]; it is very difficult to form O vacancies on MgO(1 0 0) [58,60]. The presence or absence of O vacancies does condition the DeSOx activity of the Au–TiO2 and Au–MgO interfaces [56,58]. Au particles like to interact with O vacancies of titania and on these adsorption sites an oxide-gold charge transfer has been predicted from theoretical studies and XPS measurements [56,61]. The DeSOx active Au/TiO2(1 1 0) surfaces combine Au atoms electronically perturbed and an oxide substrate with a significant amount of defects.

Figure 1.12 Top panel: S 2p signal intensity for SO2 absorbed on a polycrystalline film of gold and on an Au/ MgO(1 0 0) surface (ΘAu _∼0.3 ML). The sample was annealed for 15 s at the indicated temperatures. Taken from Ref. [60], Copyright 2003 Elsevier. Bottom panel: Relative amounts of atomic S formed by the full dissociation of SO2 on Au/MgO(1 0 0), Au/TiO2(1 1 0), and Au/TiC(0 0 1) surfaces at 300 K. Each surface was exposed to the same amount of SO2. Taken from Ref. [11], Copyright 2008 Wiley-VCH.

What is the intrinsic DeSOx activity of isolated Au nanoparticles? This question is very difficult to address at an experimental level. The bonding of SO2 to Au(1 0 0) and a series of gold clusters (Au6, Au8, and Au14) was investigated using DF calculations [60]. Very weak bonding interactions were observed on the extended Au surface with adsorption energies smaller than 0.15 eV. On the other hand, the corner atoms in the Au clusters were able to interact reasonably well with SO2, giving adsorption energies of 0.43–0.65 eV. However, none of the isolated clusters were able to dissociate the SO2 molecule. The DF calculations indicate that such a process is very endothermic on Au6, Au8, and Au14[60]. For example, the adsorption of SO2 on a Au14 particle is an exothermic process but, upon heating, the molecule should desorb instead of dissociating. The chemistry seen experimentally for SO2 on Au/MgO(1 0 0), Figure 1.12, seems to reflect mainly the intrinsic reactivity of Au nanoparticles with the oxide support playing only a minor role [60]. Variations in the strength of metalsupport interactions are probably the key to the large difference in chemical activity seen in Figure 1.12 for Au/MgO and Au/TiO2. DF calculations for SO2/Au/TiO2−x(1 1 0) systems show several bonding conformations in which Au and O-vacancy sites work in a cooperative way to dissociate the SO2 molecule [56].

The large electron polarization seen in Figure 1.1 for Au atoms in direct contact with TiC(0 0 1) suggests that these adatoms should be very efficient for bonding electron-acceptor molecules like SO2[18]. Figure 1.13 shows S 2p XPS spectra collected after dosing the same amount of SO2 at 150 or 300 K to a plain TiC(0 0 1) surface and a carbide surface pre-covered with a small amount of gold [11]. Pure TiC(0 0 1) is not very active for S—O bond cleavage [64]. At 140 K, SO2 adsorbs molecularly on TiC(0 0 1). Ti ↔ C interactions prevent strong bonding between the Ti centers and the SO2 molecule. The heat of adsorption of SO2 is in the range of 13–16 kcal/mol and most of the molecules desorb below room temperature with a fraction of the molecules undergoing full dissociation into S and O adatoms. [64] At 300 K, the amount of SO2 that fully dissociates, SO2 → Sads + 2Oads, on TiC(0 0 1) is quite low. On the other hand, the Au/TiC(0 0 1) system is quite active for the dissociation of the SO2 molecule at 150 or 300 K [11]. The bottom panel in Figure 1.12 compares the amount of atomic sulfur adsorbed after exposing Au/MgO(1 0 0) [60], Au/TiO2(1 1 0) [56], and Au/TiC(0 0 1) [11] to 5 L of SO2 at 300 K. Under these conditions the amount of SO2 that dissociates on Au/MgO(1 0 0) is negligible as discussed above [60]. On the other hand, Au/TiC(0 0 1) displays a higher DeSOx activity than Au/TiO2(1 1 0) even at large Au loads [11]. Images of STM for Au/TiC(0 0 1) point to a very high DeSOx activity when the average particle height is smaller than 0.5 nm [11]. The DeSOx activity of Au/TiC(0 0 1) decreases substantially when the particle height goes above 1 nm at Au coverages higher than 1 ML. Small Au clusters are essential for a high DeSOx activity [11].

Figure 1.13 S 2p photoemission spectra collected after dosing 5 L of SO2 at 150 and 300 K to TiC(0 0 1) and to a surface pre-covered with 0.2 ML of gold. The position of the S 2p features denotes the presence of SO2 or atomic sulfur on the surfaces. Taken from Ref. [11], Copyright 2008 Wiley-VCH.

In the top panel of Figure 1.14, we can see the calculated energy changes associated with the bonding and dissociation of SO2 to a isolated Au14 cluster [60]. SO2 interacts well with corner atoms of the Au14 cluster, but eventually the decomposition into SO and O or S and O is an uphill process [60]. This is not the case for Au/TiC(0 0 1) surfaces [11]. The bottom panel in Figure 1.14 shows the calculated adsorption energy for SO2 on clean TiC(0 0 1) and on carbide surfaces with Au atoms, Au4 or Au13 clusters, a Au wire, and a flat Au monolayer [11]. All the Au/TiC(0 0 1) surfaces bond SO2 stronger than clean TiC(0 0 1) or the corresponding isolated Au system. Spontaneous dissociation was observed when the SO2 was set at gold-carbide interfaces [11]. Thus, supported Au atoms, Au4, Au13, and an Au wire worked in a cooperative way with the carbide and dissociated S—O bonds. Photoemission results for the SO2/Au/TiC(0 0 1) system also indicate the direct participation of Au, Ti, and C sites in S—O bond cleavage [11]. In Figure 1.14, an ideal flat monolayer of Au bonded to TiC(0 0 1) adsorbs SO2 much stronger than Au(1 1 1) or Au(1 0 0), but it is not able to dissociate the adsorbate due to the lack of a gold-carbide interface. The DF calculations corroborate that the size of the Au particle has a drastic effect on the reactivity of the system. Supported Au29 displayed a much lower DeSOx activity than supported Au4 or Au13 and no dissociation of SO2 was observed [11]. The effects of the Au ↔ TiC(0 0 1) interactions were significant only when one had small Au particles. A comparison of the DeSOx studies for SO2/Au/TiO2−x(1 1 0) [56] and SO2/Au/TiC(0 0 1) [11] shows that these systems operate in different ways but in both cases the support used to disperse the gold particles plays an essential role in the chemical activity of the system.

Figure 1.14 Top: Calculated change in energy for the dissociation of a SO2 on an Au14 cluster. The zero of energy corresponds to an initial state with SO2 and the Au14 separated. Then, the SO2 molecules are adsorbed in the configuration shown in the figure. This is followed by cleavage of a S—O bond, and at the end the molecule is fully dissociated into one S and two O atoms. Taken from Ref. [60], Copyright 2003 Elsevier. Bottom: Calculated adsorption energies and bonding configurations for SO2 on TiC(0 0 1) and Au/TiC(0 0 1). Au atoms, Au4 and Au13 clusters, an Au wire, and a flat Au monolayer were deposited on TiC(0 0 1). The SO2 was initially set at the gold/carbide interface and, in most cases, spontaneously dissociated during geometry optimization. Color code: Au is shown as big yellow spheres, Ti as big gray spheres, C as small gray spheres, S as small yellow spheres, and O as small red spheres. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this book.) Taken from Ref. [11], Copyright 2008 WILEY-VCH.

1.6 Conclusions

When going from bulk metallic gold to gold nanoparticles, there is a substantial enhancement in chemical reactivity. Quantum effects related to the small size of the particles could be responsible for this enhancement. The edge and corner sites of a gold nanoparticle have distinctive electronic properties and can bond well adsorbates like CO, O2, H2O, and SO2. However, small Au clusters or nanoparticles still have trouble dissociating the O—O bond in O2, the H—O bonds in H2O, or the S—O bonds in SO2. Very good catalytic activity for CO oxidation, the water-gas shift, and DeSOx is observed when depositing Au nanoparticles on surfaces of oxides or carbides. The Au/TiO2 and Au/CeO2 systems work as bifunctional catalysts where the oxide helps with the dissociation of the bonds in O2, H2O, and SO2[4a,10,16,33,56,58]. In the case of Au supported on TiC or other carbides, bonding to the carbide substrate induces an electron polarization in gold and the admetal is now able to perform the oxidation of CO and destruction of SO2 in highly efficient ways [11,28,29,40].

References

1. (a)Haruta M. Catal Today. 1997;36:153.

1. Hayashi T, Tanaka K, Haruta M. J Catal. 1998;178:566.

2. (a)Chen MS, Goodman DW. Science. 2004;306:252.

2. (b)Walden M, Lai X, Goodman DW. Science. 1998;281:1647.

3. Rodriguez JA. In: Dekker Encyclopedia of Nanoscience and Nanotechnology. New York: Dekker; 2004;1297–1304.

4. (a)Molina LM, Hammer B. Appl Catal A. 2005;291:21.

4. (b)Matthey D, Wang JG, Wendt S, et al. Science. 2007;315:1692.

4. (c)Rodriguez JA. In: Anderson JA, Fernández-García M, eds. Supported Metals in Catalysis. second ed. London: Imperial College Press; 2012.

5. (a)Shapovalov V, Metiu HJ. Catalysis. 2007;245:205.

5. (b)Rodriguez JA, Liu P, Hrbek J, Evans J, Pérez M. Angew Chem Int Ed. 2007;46:1329.

5. (c)Burch R. Phys Chem Phys. 2006;8:5483.

5. (d)Fu Q, Saltsburg H, Flytzani-Stephanopoulos M. Science. 2003;301:935.

6. (a)Remediakis IN, Lopez N, Nørskov JK. Appl Catal A. 2005;291:13.

6. (b)Lopez N, Nørskov JK. J Am Chem Soc. 2002;124:11262.

7. (a)Campbell CT, Parker SC, Starr D. Science. 2002;298:811.

7. (b)Bondzie VA, Parker SC, Campbell CT. Catal Lett. 1999;63:143.

8. (a)Graciani J, Nambu A, Evans J, Rodriguez JA, Sanz JF. J Am Chem Soc. 2008;130:12056.

8. (b)Cruz-Hernárdez N, Sanz JF, Rodriguez JA. J Am Chem Soc. 2006;128:15600.

9. (a)Abad A, Concepcion P, Corma A, Garcia H. Angew Chem Int Ed. 2005;44:4066.

9. (b)Boronat M, Concepcion P, Corma A, Gozalez S, Illas F, Serna P. Am Chem Soc. 2007;129:16230.

10. (a)Sanchez A, Abbet S, Heiz U, et al. Phys Chem A. 1999;103:9573.

10. (b)Giordano L, Carrasco J, Di Valentin C, Illas F, Pacchioni G. J Chem Phys. 2006;124:174709.

10. (c)Inntam C, Moskaleva LV, Neyman KM, Nasluzov VA, Rösch N. Appl Phys A. 2006;82:181.

10. (d)Di Valentin C, Scagnelli A, Pacchioni G, Risse T, Freund H-J. Surf Sci. 2006;600:2434.

11. Rodriguez JA, Viñes F, Illas F, Liu P, Takahashi Y, Nakamura K. Angew Chem Int Ed. 2008;47:6685.

12. Hammer B, Nørskov JK. Nature. 1995;376:238.

13. Thomas JM, Thomas WJ. Principles and Practice of Heterogeneous Catalysis. New York: VCH; 1997.

14. (a)Ojifinni RA, Froemming NS, Gong J, et al. J Am Chem Soc. 2008;130:6801.

14. (b)Kim TS, Gong J, Ojifinni RA, White JM, Mullins CB. J Am Chem Soc. 2006;128:6282.

14. (c)Gong J, Ojifinni RA, Kim TS, White JM, Mullins CB. J Am Chem Soc. 2006;128:9012.

15. Min BK, Friend CM. Chem Rev. 2007;107:2709.

16. Rodriguez JA, Ma S, Liu P, Hrbek J, Evans J, Pérez M. Science. 2007;318:1757.

17. Rodriguez JA, Kuhn M. Surf Sci. 1995;330:L657.

18. Rodriguez JA, Viñes F, Illas F, Liu P, Takahashi Y, Nakamura KJ. Chem Phys. 2007;127:211102.

19. Henry CR. Surf Sci Rep. 1998;31:231.

20. Park JB, Conner SF, Chen DAJ. Phys Chem C. 2008;112:5490.

21. Maeda Y, Fujitani T, Tsubota S, Haruta M. Surf Sci. 2004;562:1.

22. Remediakis IN, Lopez N, Nørskov JK. Angew Chem Int Ed. 1824;2005:44.

23. Hwu H, Chen JG. Chem Rev. 2005;105:185.

24. Toth LE. Transition Metal Carbides and Nitrides. New York: Academic; 1971.

25. Oyama ST, ed. Chemistry of Transition Metal Carbides and Nitrides. Berlin: Springer; 1996.

26. Johansson LI. Surf Sci Rep. 1995;21:177.

27. (a)Viñes F, Sousa C, Liu P, Rodriguez JA, Illas F. J Chem Phys. 2005;122:174709.

27. (b)Liu P, Rodriguez JA. J Chem Phys. 2004;120:5414.

27. (c)Viñes F, Sousa C, Illas F, Liu P, Rodriguez JA. J Phys Chem C. 2007;111:16982.

28. Rodriguez JA, Illas F. Phys Chem Phys. 2012;14:427.

29. (a)Ono LK, Sudfeld D, Roldán-Cuenya B. Surf Sci. 2006;600:5041.

29. (b)Ono LK, Roldán-Cuenya B. Catal Lett. 2007;113:86.

30. Gómez E, Florez E, Rodriguez JA, Illas FJ. Phys Chem C. 2010;114:1622.

31. Schumacher B, Plzak V, Kinne M, Behm RJ. Catal Lett. 2003;89:109.

32. Green IX, Tang W, McEntee M, Neurock M, Yates JT. J Am Chem Soc. 2012;134:12717.

33. Park JB, Graciani J, Evans J, et al. Proc Nat Acad Sci (PNAS). 2009;106:4975.

34. Green IX, Tang W, McEntee M, Neurock M, Yates JT. Science. 2011;333:736.

35. Si R, Tao J, Evans J, et al. J Phys Chem C. 2012;116:23547.

36. Diebold U. Surf Sci Rep. 2003;48:53.

37. Wendt S, Sprunger PT, Lira E, et al. Science. 2008;320:1755.

38. Häkkinen H, Abbet S, Sanchez A, Heiz U, Landman U. Angew Chem Int Ed. 2003;42:1297.

39. Molina LM, Hammer B. Phys Rev Lett. 2003;90:206102.

40. Rodriguez JA, Feria L, Jirsak T, Takahashi Y, Nakamura K, Illas F. J Am Chem Soc. 2010;132:3177.

41. Wang X, Rodriguez JA, Hanson JC, Perez M, Evans J. J Chem Phys. 2005;123:221101.

42. Schwartz V, Mullins DR, Yan W, Chen B, Dai S, Overbury SHJ. Phys Chem B. 2004;108:15782.

43. Rodríguez JA, Evans J, Graciani J, et al. J Phys Chem C. 2009;113:7364.

44. Rodriguez JA, Liu P, Hrbek J, Pérez M, Evans J. J Mol Catal A Chem. 2008;281:59.

45. Liu P, Rodriguez JAJ. Chem Phys. 2007;126:164705.

46. Campbell CT, Daube KJ. Catalysis. 1987;104:109.

47. Nakamura J, Campbell JM, Campbell CTJ. Chem Soc Faraday Trans. 1990;86:2725.

48. Senanayake SD, Stacchiola D, Evans J, et al. Catalysis. 2010;271:392.

49. Rodriguez JA, Graciani J, Evans J, et al. Angew Chem Int Ed. 2009;48:8047.

50. Rodriguez JA, Hrbek J. Surf Sci. 2010;604:241.

51. Rodriguez JA. Catal Today. 2011;160:3.

52. Slack AV, Holliden GA. Sulfur Dioxide Removal from Waste Gases. second ed. Park Ridge, NJ: Noyes Data Corporation; 1975.

53. Stern AC, Boubel RW, Turner DB, Fox DL. Fundamentals of Air Pollution. second ed. Orlando: Academic; 1984.

54. (a)Armor JN, ed. Environmental Catalysis, ACS Symposium Series No 552. Washington, DC: American Chemical Society; 1994.

54. (b)Rodriguez JA, Jirsak T, Gonzalez L, Evans J, Perez M, Maiti AJ. Chem Phys. 2001;115:10914.

54. (c)Shelef M, McCabe RW. Catal Today. 2000;62:35.

54. (d)Kauffmann TG, Kaldor A, Stuntz GF, Kerby MC, Ansell LL. Catal Today. 2000;62:77.

54. (e)Rodriguez JA, Maiti AJ. Phys Chem B. 2000;104:3630.

55. Piéplu A, Saur O, Lavalley J-C, Legendre O, Nédez C. Catal Rev Sci Eng. 1998;40:409.

56. Rodriguez JA, Liu G, Jirsak T, et al. J Am Chem Soc. 2002;124:5242.

57. Liu G, Rodriguez JA, Dvorak J, Hrbek J, Jirsak T. Surf Sci. 2002;505:295.

58. Rodriguez JA, Liu P, Perez M, Liu G, Hrbek J. J Phys Chem A.