J. Phys. Chem. B 2001, 105, 8369-8374
8369
The Kinetics of the Reaction of Gaseous Hydrogen Atoms with Oxygen on Cu(111) Surfaces toward Water Th. Kammler† and J. Ku1 ppers*,†,‡ Experimentalphysik III, UniVersita¨ t Bayreuth, 95440 Bayreuth, Germany, Max-Planck-Institut fu¨ r Plasmaphysik (EURATOM Association), 85748 Garching, Germany ReceiVed: March 30, 2001; In Final Form: June 22, 2001
The reactions of H atoms with 18O adsorbed on Cu(111) surfaces were studied as a function of the oxygen coverage between 80 and 600 K with Auger electron and thermal desorption spectroscopies. The formation of gaseous products was monitored simultaneously with admission of the atom flux. Adsorbed and gas-phase water were observed as products with temperature-dependent yields and oxygen was completely consumed during the reactions. At reaction temperatures below the onset of water desorption on Cu(111) near 145 K, adsorbed water was the prevailing product, monitored in postreaction desorption spectra. Above 150 K the kinetics of formation of gaseous water are in accordance with two consecutive reaction stepssformation of adsorbed OH followed by hydrogenation of OH toward water which then desorbs isothermally. The ratedetermining step for the water reaction is the formation of OH from gaseous H and adsorbed oxygen with a phenomenological cross-section of 0.26 Å2 at 200 K. At this temperature the OH hydrogenation cross-section is 5.8 Å2. Above 300 K a significant acceleration of the water rate was observed. This is interpreted through the action of an apparent activation barrier for O hydrogenation caused by the temperature-dependent availability of reactive O atoms.
1. Introduction Water formation from adsorbed oxygen through impact of gaseous H atoms involves two hydrogenation steps: (1) Hg + Oad f OHad, (2) Hg + OHad f H2Oad, H2Og. Since the sum of the dissociation energies O-H and H-OH amounts to 9.6 eV, the overall reaction is substantially exothermal on 3d metal surfaces, which exhibit oxygen binding energies of about 5 eV. The energy released in the hydrogenation of OHad and the small water adsorption energy might even suffice to cause energetic desorption of the water product upon its formation. Irrespective of the acting mechanism, Eley-Rideal,1 or hot atom,2 crosssection type quantities σO and σOH control the kinetics of water formation.3 The magnitude of these cross-sections are at most of the size of a surface unit cell, i.e., in the order of Å2, and the smaller one identifies the rate-determining step. Previous studies of the Hg/Oad interaction on Ru4,5 and Ni6 surfaces revealed step (1) as rate-determining. On Ni(100), the measured kinetics of gas-phase water formation above the water desorption temperature could be very well described by two consecutive direct reactions with cross-sections σO ) 0.3 Å2 and σOH ) 4.5 Å2. Below the water desorption temperature 90% of the formed water molecules remained adsorbed on the surface and 10% were ejected into the gas phase upon reaction. The present study investigates water formation on Cu(111) surfaces and is complementary to a similar study on Cu(100) surfaces.7 2. Experimental Section The experiments were performed in a UHV system equipped with instrumentation for Auger electron and thermal desorption * Corresponding author. E-mail:
[email protected]. † Universita ¨ t Bayreuth. ‡ Max-Planck-Institut fu ¨ r Plasmaphysik.
spectroscopies and two heated tube type atom sources for generation of H and D. The H source is located on the axis of a small differentially pumped vacuum chamber which sticks into the main system, very similar to the experimental arrangement described in detail in a recent publication from this group.8 During reaction and desorption experiments the sample was placed close to an orifice located at the front of the small chamber. The H flux can be applied to the surface in a stepfunction-like fashion by opening a shutter which separates the small and main vacuum systems. The flux from the atom sources was calculated from the gas throughput, tube temperature, and the 2H/H2 thermodynamic equilibrium data. It is given below in units of Cu(111) monolayers per second, 1 ML s-1 ) 1.77 × 1015 H cm-2 s-1. A multiplexed quadrupole mass spectrometer in the differentially pumped small chamber allows us to monitor gaseous product formation during H admission to the Cu(111) surface and the recorded partial pressures are proportional to desorption or formation rates of the respective species. The disk-shaped Cu(111) crystal was clamped between two W wires which were inserted into spark-eroded grooves on opposite sides of the crystal periphery. These W wires were fixed at Cu rods which are attached via sapphire isolation plates to a cryogenic Cu container located at the lower end of a XYZΘΦ precision manipulator. A NiCr-Ni thermocouple inserted into a hole eroded into the sample served for temperature control. Cleanliness of the Cu(111) surface after sputtering and annealing was monitored with AES and hydrogen TD spectra.9 Oxygen 18O was admitted to the sample via a leak valve from a pumped 2 gas manifold. Water formation was never observed in TD spectra of coadsorbed Had and Oad layers on Cu(111) surfaces. In view of the results to be discussed below, this is due to the fact that the Langmuir-Hinshelwood reaction between Oad and Had toward OHad does not occur on Cu(111).
10.1021/jp0112222 CCC: $20.00 © 2001 American Chemical Society Published on Web 08/14/2001
8370 J. Phys. Chem. B, Vol. 105, No. 35, 2001
Figure 1. Thermal desorption spectra recorded after admission of water to clean Cu(111) surfaces. The dotted line represents a spectrum measured at a surface saturated with H prior to a 3 L water dose. The inset provides a plot of the logarithmic desorption rates as a function of inverse temperature.
3. Results The reactions of H with adsorbed oxygen can lead to gaseous as well as adsorbed water. To identify the role of the water desorption process in the formation of gaseous water via impact of Hg on Oad/Cu(111), thermal desorption spectra were recorded after admission of water to Cu(111) surfaces. The spectra shown in Figure 1 and analyzed by the logarithmic desorption rates shown in the inset illustrate that in a wide coverage regime water desorbs between 145 and 180 K according to a zero-order rate law with a desorption energy of 47 kJ/mol. A separation of the mono- and multilayer regimes for exposures below and above about 3 L, respectively, is not apparent from Figure 1. Also shown in Figure 1 as a dotted line is a sample desorption trace measured after admitting 3 L water to a H-saturated Cu(111) surface. It is seen that the effect of preadsorbed H on water uptake is small and marginal on the kinetics of water desorption. According to these data, in reaction measurements performed above 180 K the desorption step is not rate limiting, but below 145 K adsorbed water might be the prevailing reaction product. Between these limiting temperatures the desorption step of water will affect the kinetics of gaseous water formation if adsorbed water is a product species. A first series of reaction measurements was performed at fixed oxygen coverage and varied reaction temperatures. Prior to reaction, clean Cu(111) surfaces were exposed to 600 L of 18O2 oxygen at 600 K in order to establish a reference oxygen coverage of a half monolayer (vide infra), controlled by AES spectroscopy. Oxygen uptake curves measured separately (not shown) confirmed that with this procedure the oxide regime was avoided. After installation of a required substrate temperature, the sample was placed in front of the above-mentioned orifice and a H flux of 0.37 ML s-1 was directed at the surface by opening of the shutter. The partial pressures of hydrogenic species and water (amu 20, 19, 18) were monitored until the accumulated H fluence was several hundreds monolayers of H
Kammler and Ku¨ppers
Figure 2. Kinetics of water formation measured at various temperatures during admission of H to oxygen-covered Cu(111) surfaces. At t ) 0 a H flux of 0.37 ML s-1 was directed at the surface. Note the different reaction time scales in the upper and lower panels.
and the recorded water partial pressure had assumed a constant level close to the background level. Subsequently, a thermal desorption measurement was performed while monitoring the water and hydrogen signals. Finally, an Auger electron spectrum was recorded in order to determine whether oxygen was still remaining on the surface. Figure 2 displays the kinetics of gaseous water formation for increasing sample temperatures. At 78 and 140 K, only very small amounts of water are released into the gas phase. At and above 150 K, from Figure 2, gaseous water formation is apparent with increasing rates. The kinetics have in common that the water rates grow from zero to maximum values and decrease subsequently. With increasing temperature the initial rise of the rates gets steeper and the rates achieve higher maxima at earlier times. Note that the time scale in the lower panel of Figure 2 is expanded 7.4-fold with respect to that in the upper panel. The kinetics measured at 200 K is included in both panels in order to emphasize how fast the reaction proceeds at elevated temperatures. Postreaction water desorption spectra shown in Figure 3 confirm water formation at 78 K. The desorption spectrum peaks around 160 K, as expected from the data in Figure 1 for ordinary water desorption. Accordingly, desorption apparent from Figure 3 occurred from adsorbed water and not through a reactive step toward water from water precursors. Between 140 and 150 K the yield of water in postreaction desorption spectra drops significantly. The inset in Figure 3 illustrates that in this temperature range the reaction yields toward gaseous and adsorbed water exchange their role. The yield of gaseous water, calculated by integration of the rates displayed in Figure 2, assumes a constant value above 200 K. This is in accordance with the observation that with AES no oxygen was detected after the reactions and it can be concluded that all adsorbed O was consumed for water formation. The details of water formation through application of H at 80 K was measured at the above specified reference oxygen
Reaction of Gaseous H Atoms with Oxygen on Cu(111)
Figure 3. Thermal desorption spectra measured after completion of reaction measurements of H with oxygen on Cu(111) at various temperatures. The inset provides the yields of water collected during reactions and in postreaction desorption spectra.
Figure 4. Thermal desorption spectra measured after application of increasing fluences of H at oxygen-covered Cu(111) surfaces at 80 K. The inset displays the desorption yields as a function of applied H fluence.
adlayers (600 L at 600 K) with subsequent admission of varied H fluences followed by collection of TD spectra. The series of spectra shown in Figure 4 illustrates that the yield of adsorbed water increases with increasing applied H fluence, as expected. The shape of the desorption traces are not what one would expect from Figure 1 for water desorption. However, one has to consider that the traces shown in Figure 4 are desorption spectra of coadsorbed Oad/OHad/H2Oad adlayers with decreasing contribution from Oad, increasing contribution from H2Oad, and
J. Phys. Chem. B, Vol. 105, No. 35, 2001 8371
Figure 5. Rate of water formation in the HDO and H2O channels measured at 200 K during directing a H flux at oxygen-covered surfaces which were already exposed to varied D fluences prior to admission of H. Top panel: HDO kinetics; bottom panel: H2O kinetics. The inset displays logarithmic plots of the water rates after preexposure of 6.3 ML of D.
varying coverage of OHad. Below applied H fluences of 10 ML, reaction limited desorption of water from OHad groups formed during H exposure is apparent around 225 K, either via HadOHad recombination or via OHad-OHad disproportionation. Since the latter reaction is of second order with respect to the OHad coverage and the reaction limited features in Figure 4 do not exhibit a temperature shift, they are attributed to Had-OHad recombination. Above H fluences of 10 ML, the water signals in Figure 4 exhibit a desorption limited peak at 155 K and a smaller reaction limited peak at 225 K. Above H fluences of 100 ML, the desorption spectra are dominated by a strong water peak at 155 K, as expected from Figure 1. The pronounced hightemperature tail of this peak is probably due to the fact that the reaction was not completed and some OHad still remained on the surface. Close inspection of the bottom spectra in Figure 4 reveals that the recombination feature at 225 K is smaller than the desorption signal of preexisting water at 155 K, if the tailing of the latter peak is considered. This illustrates that there is more H2Oad on the surface than OHad which suggests that the step Hg + Oad f OHad is slower than the step Hg + OHad f H2Oad. A proof of this conjecture is provided by the results shown in Figure 5. To collect these data, oxygen-covered surfaces were exposed to D fluences prior to performing the reaction with H atoms, thus providing ODad groups for reaction with Hg. A reaction temperature of 200 K was chosen because at that temperature water desorption is so fast that is not rate limiting and for the kinetics of gaseous water only the two hydrogenation reactions need to be considered. As expected from the experimental procedure, HDO and H2O evolve and the amount of HDO grows initially with the applied Dg fluence, which is required. It is also seen that the HDO rates assume their maximum values right at reaction start, in accordance with the one-step reaction Hg + ODad f HDOgas. In contrast to this, the
8372 J. Phys. Chem. B, Vol. 105, No. 35, 2001
Figure 6. Kinetics of water formation measured at 300 K during admitting H atoms to oxygen-covered Cu(111) surfaces at various O precoverages.
H2O rates grow initially linear from zero, as expected for a sequence of two hydrogenation reactions with OHad as the intermediate. The logarithmic plots of the water rates illustrate that at 200 K the H2Og and HDOg kinetics later on the reaction time scale are very well described by an exponential decay. Accordingly, the rates are proportional to exp(-σφt) with appropriate cross-sections. The HDO signal decays much faster than the H2O signal. Since ODad groups were already present on the surface at start of the Hg admission, the slope of the logarithmic HDO rate provides the hydrogenation cross-section of ODad. Neglecting isotope effects, one obtains from the slope of the logarithmic rate σOH ) 1.1 Å2. Likewise, the slope of the logarithmic H2O rate provides the cross-section of the ratelimiting reaction step in the sequence of the stepwise hydrogenation of Oad to OHad and OHad to H2O. Since this value, 0.2 Å2, is significantly smaller than 1.1 Å2, one can approximate the cross-section for hydrogenation of Oad by σO ) 0.2 Å2. These σ values confirm that the formation of OHad through the interaction of Oad with Hg is much slower than the hydrogenation by Hg of OHad toward H2O, in accordance with the speculation based on the spectra shown in Figure 4. This conclusion is not affected by the fact that the σ’s are phenomenologically defined quantities and do not necessarily represent simple physical crosssections.3 To investigate whether the oxygen coverage somehow modifies the mechanism of water formation, reactions were performed at 300 K sample temperature after preparation of oxygen adlayers at 600 K. The water kinetics shown in Figure 6 suggest that at 300 K the mechanism is not affected by the initial oxygen coverage: the cross-sections extracted from the logarithmic rates do not depend on the initial O coverage. Exposure of H to a O covered Cu(111) surface below the hydrogen desorption temperature leads to the buildup of an adsorbed layer of H. Through abstraction reactions Hg + Had f H2,g and adsorption Hg f Had a reaction/adsorption cycle proceeds in parallel to water formation. To investigate this, surfaces were prepared with different oxygen precoverages and subsequently exposed to a 30 ML D fluence. The HD kinetics
Kammler and Ku¨ppers
Figure 7. Kinetics of HD formation during admission of H to oxygencovered Cu(111) surfaces which were exposed to a fluence of 30 ML of D atoms prior to H admission.
measured during admission of Hg to these adlayers are shown in Figure 7. It is seen that the decrease of the HDg rate on the reaction time scale proceeds slower with higher oxygen precoverage. Accordingly, the apparent abstraction cross-section of Dad by Hg is affected by the presence of coadsorbed Oad and it is expected that simultaneously Dad (or Had) abstraction affects the H2O kinetics. 4. Discussion The results presented above have shown that in the temperature range 80 K to 600 K, adsorbed oxygen on Cu(111) surfaces is readily transformed into water by impact of atomic H. The first step, hydrogenation of Oad, has been identified as rate determining, as was previously found for reaction of H atoms with Oad on Ni6 and Ru surfaces,4,5 and recently on Pt(111).10 Well below the water desorption temperature, below 140 K, only a small fraction (6%) of the formed product molecules are ejected into the gas phase and the majority of products remain adsorbed. On Ni(100) surfaces this fraction was reported as 10%. If isothermal water desorption occurs on the time scale of the present experiments, around 150 K, the gaseous water rate is affected by a more or less rate-limiting desorption step. If desorption is not rate limiting, above 200 K, the gas-phase water kinetics exhibits a complicated temperature dependence, with a significant rate acceleration above 300 K. The spectra shown in Figure 1 illustrate that at 200 K the water desorption step can be neglected for the gaseous water rate. Therefore, as a starting point for the analysis of the water kinetics this temperature is selected. For two consecutive direct (Eley-Rideal or hot-atom type) reaction steps, H + Oad f OHad, controlled by a cross-section σO, and H + OHad f H2Ogas, controlled by σOH, the rate of gaseous water formation R(H2Ogas) can be expressed as6
R ) σO × σOH/(σOH - σO) × Φ × [Oad]0 × (exp(-σO × Φ × t) - exp(-σOH × Φ × t)) (1)
Reaction of Gaseous H Atoms with Oxygen on Cu(111) with Φ as atom flux, [Oad]0 as initial O coverage, and t as time. A fit of eq 1 to the 200 K water kinetics shown in Figure 2 reveals that the measured kinetics can be excellently reproduced by choosing σO ) 0.26 Å2 and σOH ) 5.8 Å2. In the limit of short reaction time the above rate R is given by σO × σOH × Φ2 × [Oad]0 × t, a linear increase of R with time, which applies to the kinetics at 200 K seen in Figure 2. In the limit of long reaction time and σO , σOH, R is approximated by the exponential function σO × Φ × [Oad]0 × exp(-σO × Φ × t), which is confirmed through a logarithmic plot of the 200 K rates. These cross-sections should be the same as those deduced from the H/D/Oad reaction measurement shown in Figure 5. Whereas for σO this is fulfilled, the two σOH cross-sections differ considerably and it is impossible to describe the decrease of the HDO product rate in Figure 5 by a cross-section of 5.8 Å2. This would imply a much faster rate decrease than the measured one. Likewise, it is impossible to fit the 200 K water kinetics shown in Figure 2 with a σOH ) 1.1 Å2. There is not obvious explanation for this contradiction. The measurements were performed at the same temperature with the same atom fluxes. The only difference is that Figure 2 refers to OHad hydrogenation whereas Figure 5 applies to ODad hydrogenation. It is hardly conceivable that an isotope effect is the origin of these significantly different cross-sections. The cross-sections extracted from fitting eq 1 to the 200 K data in Figure 2, σO ) 0.26 Å2 and σOH ) 5.8 Å2, are close to those determined for the water reaction on Ni(100) surfaces: σO ) 0.3 Å2 and σOH ) 4.5 Å2. The magnitude of σOH is about the size of the unit cell on Cu(111), A ) 5.66 Å2, which sets an upper limit of a cross-section since σ ) A is equivalent to a reaction probability of unity for an incoming atom. It is apparent from Figure 2 that between 300 and 600 K the water kinetics cannot be described by a rate expression as that given by eq 1. Two features are present: the slope of the initial linear rate increases between 200 and 450 K and the rate later in the reaction is not a simple exponential, clearly seen in the kinetics at 450 and 500 K. There are several possibilities to account for these phenomena. At first there might contribute the recombination pathway Had + OHad f H2Ogas. The desorption measurements prior to reaction completion shown in Figure 4 confirms that this step occurs at 225 K. This step cannot contribute above 400 K since the equilibrium H coverage is negligible at these temperatures.9 Second, the disproportionation reaction OHad + OHad f H2Ogas + Oad might contribute. However, the assumption that water is formed through a sequence of two hydrogenation reactions of Oad followed by OHad/OHad disproportionation contradicts an initial linear increase of the gaseous water rate. Furthermore, as seen in Figure 2, the fast water increase in the initial reaction period spans about a second, during which only a fraction of a monolayer of H was admitted to the surface. The rather inefficient use of OHad by disproportionation to form water is not consistent with the fast kinetics apparent from Figure 2. A third possibility is the introduction of activation barriers Eσ for the cross-sections σO and σOH, according to an expression like σ ) σ0 × exp(-Eσ/kT). Since at 200 K the magnitude of σOH is already close to its upper limit, only the O hydrogenation cross-section σO might increase with temperature. An increase of σO provides two features which were observed experimentally: a steeper linear increase of the initial water rate and a faster decay of the rate in the late reaction regime. To simplify the interpretation of the temperature dependence of the water kinetics it is assumed that one effect, and not a
J. Phys. Chem. B, Vol. 105, No. 35, 2001 8373 mixture of several effects, is its origin. Since Had/OHad recombination and OHad/OHad disproportionation have been ruled out as applicable above 400 K, an increase of the crosssection σO remains as a plausible alternative. Small activation barriers for gas-phase atom or radical reactions are not uncommon.11 In the present case of a surface reaction a phenomenological activation barrier might originate from a feature which is absent in gas-phase reactions. The barrier might not stem from an activation of the reaction itself, but from the supply of reactive atoms on the surface. Applied to the reaction in question, Hg + Oad f OHad, this could, for example, be caused by a diffusion of O atoms from the borders of oxygen covered islands to the empty Cu(111) surface or by a steric effect. It is known that the Cu(111) surface reconstructs upon oxygen adsorption 12 to achieve a structure close to that of Cu2O. Although in the present study no LEED observations were made, previous studies have shown that the procedure used here for preparation of the oxygen adlayers causes ordered 2 × 2 superstructures with an oxygen coverage of about 0.5.13 The position of the O atom in the 3-fold hollow site14,15 has been determined with low-energy ion scattering16 and SEXAFS12 as about 0.3 Å above the outer surface plane. This position is only assumed by O during annealing of oxygen-exposed Cu(111) surfaces at elevated temperatures. An oxygen atom adsorbed almost in the plane of the Cu atoms is difficult to attack by impinging H atoms, in accordance with a small hydrogenation cross-section of Oad. Only during a vibrational movement of Oad toward the surface the H might have a chance to accomplish bonding. A freshly formed OHad group might move to a top site in order to achieve the favorable adsorption geometry. Accessibility of this OHad group for reaction with Hg is not hampered by geometric effects and the corresponding cross-section σOH is expected to be large. In this speculative reaction scenario the small and temperature-dependent cross-section σO is rationalized through the thermally activated vibration of the O atom in its hollow site and the high cross-section σOH through its exposed adsorption geometry. More studies of Hg/Oad water reactions are needed to clarify this issue. It is seen in Figure 2 that the water kinetics at 450 and 500 K exhibit a specific feature. After a reaction time of 10 s and 20 s, respectively, the rates drop faster than before. The same characteristics seem to be present in the kinetics at 550 K, even earlier on the reaction time scale. It was found that the logarithms of the rates at 450, 500, and 550 K do not exhibit linear segments. Expressed by the water rate R ) σ × Φ × ΘO this indicates the action of a coverage dependent cross-section for hydrogenation of Oad or OHad: σ ) σ(ΘO, ΘOH). It might indicate a structural effect, or it might occur because mobile O atoms diffuse around on the surface and undergo collisions with hot H atoms. At small O coverages, the probability of reaction of O with hot H atoms decreases if these do not collide with an O atom within their range. More studies and additional experimental techniques are needed for a better understanding of these phenomena. The data shown in Figure 6 illustrate that at 300 K the kinetics is not affected by the oxygen precoverage. Within the scatter of the data the rates scale to a common curve and can be fitted by a sum of two exponentials according to eq 1. This is a further indication for the above speculation that the phenomena observed at and above 450 K are connected with a new route to water. During the reactions of Oad with Hg, simultaneously H adsorption and abstraction occur. It was expected that the Had
8374 J. Phys. Chem. B, Vol. 105, No. 35, 2001 by Hg abstraction reaction kinetics would be substantially affected by the ongoing water reaction and would not exhibit any similarity with that observed without the presence of Oad. However, as Figure 7 documents, this expectation was unjustified. The abstraction kinetics of the reaction Hg + Dadf HDg on a surface without adsorbed O present was found as reported previously.9 With increasing Oad coverage the only new effects are a smaller HD rate, which is required, and a small deviation from a linear relation between the logarithmic HD rate and reaction time. One has to consider that during the reaction period shown in Figure 7 strongly adsorbed Oad is replaced by weakly adsorbed H2Oad because at 78 K reaction temperature most of the formed water products remain adsorbed at the surface. The Dad abstraction reaction proceeds almost unaffected by the ongoing water formation. It seems as if Dad abstraction and water formation are to a great extent decoupled from each other although they proceed on the very same surface. This feature, in retrospect, is the reason for the possibility to interpret the kinetic data shown in Figure 2 without considering the processes connected with H adsorption and abstraction. There is an additional complication of the water kinetics on Cu: on the (111) and (100) surface orientations the phenomenology of the water kinetics is completely different.7 In the limit of the “simple” low-temperature kinetics, T < 100 K, on Cu(100) about 50% of the formed water molecules are ejected into the gas phase, contrasted by only 6% on Cu(111). This significant influence of the surface orientation might stem from the geometric effects discussed above. 5. Conclusions Oxygen adsorbed on Cu(111) is readily abstracted toward water by gaseous H atoms. Below the water desorption
Kammler and Ku¨ppers temperature near 150 K only a small fraction (6%) of the formed water products appear through energetic desorption in the gas phase. Around 200 K the kinetics of gas-phase water can be described by a sequence of two direct reactions. The crosssections of these direct atom-adsorbate (Eley-Rideal or hot-atom type) reactions via Oad (σO ) 0.26 Å2) and OHad (σOH ) 5.7 Å2) hydrogenation are similar to those measured on Ni(100) surfaces. Above 300 K the reaction phenomenology suggests that σO increases with temperature, resulting in a significant acceleration of the gaseous water rate. The simultaneously proceeding H adsorption/abstraction reaction is almost decoupled from the water reaction. References and Notes (1) Eley, D. D.; Rideal, E. K. Nature 1940, 146, 401. (2) Harris, J.; Kasemo, B. Surf. Sci. 1981, 105, L281. (3) Kammler, Th.; Kolovos-Vellianitis, D.; Ku¨ppers, J. Surf. Sci. 2000, 460, 91. (4) Jachimowski, T. A.; Weinberg, W. H. J. Chem. Phys. 1994, 101, 10997. (5) Schick, M.; Xie, J.; Mitchell, J.; Weinberg, W. H. J. Chem. Phys. 1996, 104, 7713. (6) Kammler, Th.; Scherl, M.; Ku¨ppers, J. Surf. Sci. 1997, 382, 116. (7) Kolovos-Vellianitis, D.; Kammler, Th.; Ku¨ppers, J. Surf. Sci. 2000, 460, 91. (8) Dinger, A.; Lutterloh, C.; Ku¨ppers, J. J. Chem. Phys. 2001, 114, 5338. (9) Kammler, Th.; Ku¨ppers, J. J. Chem. Phys. 1999, 111, 811. (10) Lang, E.; Biener, J.; Ku¨ppers, J., submitted. (11) Baulch, D. L.;Cobos, C. J.; Cox, R. A.; Esser, C.; Frank, P.; Just, Th.; Kerr, J. A.; Pilling, M. J.; Troe, J.; Walker, R. W.; Warnatz, J. J. Phys. Chem. Ref. Data 1992, 21, 411. (12) Haase, J.; Kuhr, H.-J. Surf. Sci. 1988, 203, L695. (13) Ertl, G. Surf. Sci. 1967, 6, 208. (14) Dubois, L. H. Surf. Sci. 1982, 119, 399. (15) Sueyoshi, T.; Sasaki, T.; Iwasawa, Y. Surf. Sci. 1996, 365, 310. (16) Niehus, H. Surf. Sci. 1983, 130, 41.