Water Adsorption and Thermal Decomposition on FeAl(110)

H2O chemistry on FeAl(110) has been investigated with temperature-programmed desorption (TPD), photoelectron spectroscopy, and high-resolution electro...
2 downloads 0 Views 813KB Size
J. Phys. Chem. 1996, 100, 18829-18838

18829

Water Adsorption and Thermal Decomposition on FeAl(110) N. R. Gleason† and D. R. Strongin*,‡ Department of Chemistry, Temple UniVersity, Philadelphia, PennsylVania 19122, and Department of Chemistry, State UniVersity of New York, Stony Brook, New York 11794-3400 ReceiVed: August 13, 1996; In Final Form: September 30, 1996X

H2O chemistry on FeAl(110) has been investigated with temperature-programmed desorption (TPD), photoelectron spectroscopy, and high-resolution electron energy loss spectroscopy (EELS). Exposure of FeAl(110) to water at 123 K results in predominantly dissociative adsorption (i.e., H2O f OH(ad) + H(ad)), although molecularly adsorbed H2O is also detected. All, or at least the majority, of the adsorbed H2O converts to adsorbed OH(ad) by 173 K. Photoemission suggests that the atomic composition of the OH(ad) binding site is primarily Al. Substantial decomposition of the OH(ad) adlayer occurs by 300 K, and the majority of this species dissociates by 410 K. Atomic oxygen resulting from OH decomposition (i.e., OH(ad) f O(ad) + H(ad)) remains on the surface as aluminum oxide, and surface hydrogen combines and desorbs as H2 in the 200400 K temperature range. This later reaction step is thought to be strongly influenced by the Fe component of the alloy.

1. Introduction The reaction and chemisorption of water has been characterized extensively on many solid surfaces with modern surface science techniques.1 Much of this prior research has dealt with the chemisorption and reaction of water on monometallic surfaces,2-18 due to its fundamental importance in areas like corrosion and catalysis.19 This motivation for investigating monometallic surfaces also applies to the interaction of water on alloy surfaces,20 due to the prevalence and importance of these surfaces in similar technological arenas.21 Research presented in this paper investigates the chemisorption of water with the FeAl(110) alloy which crystallizes in a body-centered CsCl structure.22 FeAl is a potential structural material, and its reactions with water are known to have detrimental effects on its mechanical properties.23,24 Surface science techniques are used in the present study to develop a general description of the reactivity of water on FeAl(110). Before our results are presented, we feel that it is useful to review prior research that has investigated the chemisorption of water on the separate components. These prior results will prove to be useful in helping us to interpret our results for FeAl(110) that are presented later. Hung et al. have investigated the adsorption of H2O on Fe(100) with high-resolution electron energy loss spectroscopy (EELS) and temperature-programmed desorption (TPD).11 Water associatively adsorbs on Fe(100) at 100 K, but heating to 243 K produces a hydroxyl overlayer. Further heating decomposes this hydroxyl layer and results in the desorption of water and hydrogen near 310 K.11 The adsorption of water on metal surfaces is face specific, as is exemplified by the complete dissociation of water on Fe(110) at low coverage.10 Molecular adsorption is presumed to occur as the second and subsequent layers hydrogen bond to the surface and to oxygen. For each surface, annealing a multilayer dosed surface to 673 K shows that oxygen remains, indicating that water adsorption is not reversible.10,11 In contrast to Fe surfaces, Crowell et al. have shown that the adsorption of H2O on Al(111) is mixed molecular and dissocia†

State University of New York. ‡Temple University. * To whom correspondence should be addressed: Tel (215) 204-7119; FAX (215) 204-1532; E-mail [email protected]. X Abstract published in AdVance ACS Abstracts, November 1, 1996.

S0022-3654(96)02449-5 CCC: $12.00

tive at 130 K, as confirmed by EELS.13 TPD shows that water desorbs from a multilayer state at 160 K, and H2 desorption occurs in the 100-200 K temperature range.14 No water desorption, however, is detected from a chemisorbed layer or recombination state, as has been observed for Fe(100). After annealing the H2O/Al(111) surface to 250 K, EELS shows that adsorbed hydroxyl species and chemisorbed oxygen remain on the surface.13 Further heating decomposes the hydroxyl species completely to chemisorbed oxygen and eventually produces Al2O3.13 In contrast, Norton and co-workers have found that water adsorption on Al(100) is completely molecular at low temperature.25 The bond to Al is quite strong and water desorption is not detected at low coverages, while H2 is observed to desorb between 200 and 500 K.26 Infrared spectroscopy of D2O/Al(100) shows that the OD stretching region is characterized by modes at 2720 cm-1 for non-hydrogen-bonded D2O(ad) at low coverages and 2760 cm-1 for OD(ad) produced by annealing, which would not be resolvable by EELS (typical resolution ∼70 cm-1).25 Research presented in this contribution uses EELS, TPD, and photoelectron spectroscopy to investigate the chemisorption of water on FeAl(110). Chemisorption of water on FeAl(110) at 123 K results in predominantly dissociative adsorption at low exposures, while some molecular water is detected after all exposures. The vibrational spectrum of hydroxyl groups on FeAl(110) and Al(111) are similar, suggesting that OH groups on FeAl(110) strongly interact with the Al component. Heating to 250 K results in the onset of hydrogen desorption and the decomposition of some of the hydroxyl groups. Similar to the Al surfaces, no desorption of water from FeAl(110) is observed upon heating, except for a multilayer state near 150 K. Surface hydrogen, resulting from water decomposition, combines and desorbs from FeAl(110) and Fe(100) at similar temperatures (near 250 K), suggesting that hydrogen desorbs from the Fe component of FeAl. By 410 K most of the OH(ad) has decomposed, and the resulting oxygen is bound to the Al component. 2. Experimental Methods and Sample Characterization All experiments presented in this paper were performed in a bakeable stainless steel ultrahigh-vacuum (UHV) chamber with a working base pressure of 3 × 10-10 Torr. UHV was obtained © 1996 American Chemical Society

18830 J. Phys. Chem., Vol. 100, No. 48, 1996 by ion, titanium sublimation, and turbomolecular pumps. The chamber was equipped with a quadrupole mass spectrometer, low-energy electron diffraction (LEED) optics, double-pass cylindrical mirror analyzer (CMA) with an electron gun for Auger electron spectroscopy (AES), X-ray source for X-ray photoelectron spectroscopy (XPS), EEL spectrometer, and ion gun for sample cleaning. The FeAl(110) crystal was mounted on a liquid nitrogen cryostat with cooling capabilities down to 120 K. Tantalum support wire was spot-welded to the edges of the 1 cm2 sample, and a type K thermocouple was spot-welded to the top of the sample. Heating of the sample was accomplished by passing current through the support wires. TPD experiments were performed with a heating rate of 10 K/s. All desorbing species were monitored simultaneously by the multiplexed mass spectrometer. A 1/8 in. dosing-tube attached to a variable leak valve was used to admit water into the UHV chamber and was typically 1 cm away from the center of the sample. The distilled water was purified by several freeze-pump-thaw cycles. Commercially purchased D2O (Sigma Chemical Co., 99.96%) was purified in the same manner without any further treatment. Hydrogen (Matheson, 99.99%) was admitted by back-filling the chamber through a second variable leak valve. Exposures in langmuirs (1 langmuir ) 10-6 Torr‚s) in this paper are uncorrected for this line-of-sight setup or for the cracking efficiencies of molecules in the ionization gauge. The ionization region of the mass spectrometer was housed in an enclosure with a small aperture. During TPD experiments, the sample was translated to within 0.5 cm of the aperture hole, limiting the detection of gases desorbing from the sample supports. The FeAl single crystal ingot from which the FeAl(110) crystal was cut from had a bulk composition of 60 at. % Fe, 40 at. % Al, and 0.02 at. % Zr, as quoted by the supplier (General Electric). It is noted that the FeAl phase exists in the atomic percent Al range of ∼40-48.27 The FeAl(110) sample (1 cm2, 0.2 cm thick) was oriented to within 1.0° of the desired orientation and mechanically polished. Analysis of the Fe 2p3/2 and Al 2s XPS (parameters given below) peak areas and using the appropriate sensitivity factors yielded a near-surface composition of 46 at. % Al, different than the quoted concentration for the bulk of the material. After introduction into the UHV chamber and prior to cleaning, FeAl(110) showed carbon and oxygen contamination. Cleaning of this intermetallic consisted of repeated cycles of 500-1000 eV argon ion bombardment and annealing (1173 K for 8 min, 1223 K for 2 min).28 This procedure produced a clean surface, with no detectable zirconium, carbon, and oxygen in XPS and EELS measurements. XPS was performed by using unmonochromatized Mg KR radiation (1253.6 eV) as the excitation source. XPS measurements of Fe 2p3/2 and Al 2s core levels were obtained with a CMA pass energy of 25 eV. Measurements of the O(1s) binding energy were performed with a higher CMA pass energy of 50 eV. XPS spectra of the O(1s) level presented in this paper are obtained after exposing FeAl(110) to water vapor at 120 K and after heating to various temperatures that are indicated for the different spectra. Data after heating were obtained by raising the sample temperature to the specified temperature (10 K/s) momentarily and then cooling the sample back to 123 K. All XPS binding energies presented in this paper are referenced to the Fermi level by aligning the Fe 2p3/2 core level to 706.8 eV. EELS data were acquired using a 3.5 eV primary beam. Spectra presented are the sum of 10-40 scans over the indicated region, depending upon the surface coverage of water. The typical full width at half maximum (fwhm) of the elastic peak from a clean surface was 9 meV, giving a resolution of

Gleason and Strongin approximately 80 cm-1. The resolution of spectra obtained from FeAl(110) after exposure to water were typically 10 meV. We note that periodic control experiments were carried out to ascertain whether any contaminants appeared in the time required to obtain a spectrum from residual gases in the UHV chamber. In the time to acquire a typical spectrum for a lowdosed surface, EELS spectra for our control experiments showed the appearance of H2O vibrations in the low-frequency region, although their intensity was negligible compared to that detected after the surface is dosed. (No modes appeared in the δ(HOH) and ν(OH) regions.) All synchrotron data presented in this paper were acquired at the U7b beamline at the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory. The experimental chamber at the NSLS had a base pressure of 2 × 10-10 Torr and was pumped by a 300 L/s ion pump, turbomolecular pump, and titanium sublimation pump. Sample cleaning was accomplished by using the same conditions mentioned before to ensure consistency. Photoelectrons emitted from the sample during photoemission were energy analyzed by a VSW hemispherical (HA 100) analyzer mounted 90° to the beam port. All photoemission spectra are recorded using a pass energy of 10 eV. The photon energy used for core level spectra was 160 eV. By using this photon energy, the kinetic energy of the photoemitted electrons originating from the Al 2p or transition metal 3p core level could be kept within the range 85-110 eV. Electrons with such kinetic energies are estimated to have an escape depth of ∼6 Å.29 These photoemission experiments, therefore, in contrast to conventional XPS, are most sensitive to the electronic structure of the first few layers of the alloy surface. An ideal termination of the (110) plane of an alloy such as FeAl that adopts the CsCl structure will be composed of equal amounts of both alloy components. In the case of the alloy NiAl, which adopts the same bulk structure as FeAl, equal amounts of Ni and Al make up the outermost layer that is commensurate with the bulk structure. The surface and even composition of FeAl(110) is not so straightforward. A LEED pattern for a FeAl(110) crystal obtained in our laboratory has been presented in an earlier publication.28 In the present study a different crystal has been used (cut from the same FeAl ingot), and the LEED pattern is similar except that more severe streaking of the diffraction spots are observed, suggesting that the outermost layer experiences reconstruction. A more detailed analysis of FeAl(110) (without Zr) has been performed by Graupner et al.,30 and they conclude that this crystallographic plane of FeAl is incommensurate with the bulk and that there is preferential segregation of Al to the near surface region (under the conditions used to clean our sample). While a more detailed structural study needs to be performed to elucidate the exact structure of this surface, we have further characterized our sample with low-energy ion scattering (ISS) that is sensitive to the atomic composition of the outermost layer. Figure 1 exhibits ISS data obtained from a sputtered surface (spectrum A) that has not been annealed and the same surface after annealing (spectrum B, 8 min at 1173 K and 2 min 1223 K). These spectra were obtained with a primary beam of 1 keV He+ and a scattering angle of 120°. Energy distribution curves of the scattered He+ were obtained by the use of the same hemispherical analyzer used for all the synchrotron photoemission experiments. For the purpose of this paper, the most important result from each of the spectra is that both component metals are present in the surface layer, although from the figure it is evident that annealing causes the surface stoichiometry of Al to Fe (Al/Fe) to change relative to the

Water Adsorption on FeAl(110)

J. Phys. Chem., Vol. 100, No. 48, 1996 18831

Figure 1. ISS spectra of FeAl(110) after sputtering (spectrum A, lower curve) and annealing as described in the Experimental Section (spectrum B, upper curve). Features associated with O and Ta are due to the scattering of ions from the Ta support wire used to mount the sample.

sputtered surface. In addition to the Fe and Al peaks, features that we associate with O and Ta are also present. These latter features are due to the Ta support wires holding the sample. Unfortunately, the size of the sample and width of the ion beam used to obtain ISS data precluded the elimination of these features. We add, however, that photoemission of this same surface (some of those data are presented later) shows no O or Ta contamination. In order to estimate the relative ratio of Fe and Al in the outermost layer, comparisons to ISS of monometallic Fe and Al samples were performed. Al and Fe spectral features in each of the two curves in Figure 1 were fit with Gaussians and compared to the peak areas of spectra of the pure metals obtained in our laboratory. From this analysis we estimate that the sputtered sample is depleted of Al and is 35 at. % Al. This is not unexpected, as it has been found for NiAl alloys that sputtering preferentially removes the Al component.31 Analysis of the spectrum obtained by annealing and those of the pure metals leads us to estimate that the surface is approximately 55 ( 5 at. % Al upon annealing. It is to be stressed that the goal of the ISS experiment was to show that both components are present in the surface and that the surface concentrations quoted are rough estimates and are not quantitative.

Figure 2. TPD spectra for H2O/FeAl(110). The upper panel shows the desorption of water. The lower panel shows desorption traces of hydrogen product.

3. Results 3.1. Thermal Desorption. Figure 2 shows desorption spectra of water and hydrogen from FeAl(110) as a function of water exposure at 120 K. The upper panel shows desorption traces for water, and the lower panel contains the hydrogen spectra. (H2 and H2O traces are recorded in a single experiment.) Hydrogen desorption is observed in the temperature interval of 200 and 400 K, at all water exposures. At the lowest water exposure, the peak temperature of hydrogen desorption is near 300 K, but as the exposure is increased, the peak temperature progressively shifts down in temperature, ultimately to 220 K at the highest exposure. In contrast to the hydrogen desorption traces, water shows desorption only after exposures greater than 0.5 langmuir. The water desorption feature that appears near 150 K is thought to be associated with desorption from a condensed layer, since this states grows indefinitely with increasing water exposure. Also, the desorption temperature of this state is similar to that observed for the sublimation of water from other metals.1,11,14 TPD experiments were carried out that monitored hydrogen desorption from FeAl(110) after exposure to molecular hydrogen. No H2 (D2) was experimentally observed to desorb from

Figure 3. EELS spectra of a 10 langmuir dosed FeAl(110) surface as it is stepwise annealed to 695 K. After annealing to 173 K, all features associated with molecular H2O disappear.

FeAl(110) after exposure (1-3000 langmuirs) to H2 (D2). It is inferred from this result that this bimetallic surface cannot dissociate dihydrogen under conditions associated with UHV experiments. 3.2. Vibrational Spectroscopy. Figure 3 exhibits EELS data for FeAl(110), which has initially been exposed to 10 langmuirs of H2O at 123 K, and then annealed to various temperatures. The 10 langmuirs spectrum is similar to EELS spectra obtained for condensed water, and the spectral features can be readily assigned.1,13 Loss peaks at 1660 and 3460 cm-1 are assigned

18832 J. Phys. Chem., Vol. 100, No. 48, 1996

Gleason and Strongin

TABLE 1: Summary of Selected Vibrational Modes of H2O on FeAl, Al, and Fea ν(M‚‚‚OH2)

system FeAl(110) Al(111)c Fe(100)d

b

576 660 (660) 415 (390)

Fe(110)e,f a

δ(HOH)

ν(OH)

1640 (1190) 1655 (1225) 1590 (1190)

∼3680 (2680) 3445 (2590)g 3470 (2490) 3595 (2560) 3380 (2595)

1630 (1220)

TABLE 2: Summary of Vibrational Frequencies (cm-1) for OH(ad) on FeAl(110), NiAl(110), and Monometallic Surfacesa ν(M‚‚‚O)

system b

FeAl(110)

NiAl(110)c Al(111)d Fe(100)e

cm-1,

Frequencies are given in and values in parentheses are for D2O. b This work. c Reference 13. d Reference 11. e Reference 10. f Values quoted are for high exposure. H2O dissociates at low coverage. g Values quoted are for a multilayer.

Fe(110)f

370 (370) 790 (760) 790 (700) 765 (765) 475 (465) 470 (470) 960 (960)

δ(M‚‚‚O-H)

ν(OH) 3740 (2735)

785 (570) 1190

3760 (2750) 3745 (2720) 3595 3390 (2510) 3620 (2715) 3300 (2505)

a

c

Figure 4. Comparison of 10 langmuirs of H2O/FeAl(110) and D2O/ FeAl(110) after the surface is annealed to 173 K.

to the δ(HOH) and ν(OH) vibrational modes of water, respectively. Features at 240 and 770 cm-1 are associated with the hindered translation and rotation of condensed water, respectively. Finally, the high-frequency loss feature at 4150 cm-1 is believed to be a combination band of ν(OH) vibrations and hindered rotations.4,5,11 These assignments for the vibrational modes of water are listed in Table 1, along with complementary data for H2O on Al(111), Fe(100), and Fe(110). Also, vibrational frequencies for condensed D2O are given in parentheses in the table and were used to verify our assignments for H2O. Heating H2O/FeAl(110) to 173 K results in a spectrum that is markedly different than the multilayer spectrum; most noticeable is the elimination of the features at 240, 3460, and 4150 cm-1. In addition, the δ(HOH) mode at 1660 cm-1 is significantly reduced. Actually this mode may even be completely eliminated, since the remaining intensity in that region may be due overtones and/or combination bands of oxygen on aluminum that have been experimentally observed for O2/Al(111) at low temperature.17 Heating to 173 K also results in the growth of new modes at 370 and 3740 cm-1 (O-H stretching region) and in the reduction of spectral intensity at 770 cm-1. Given that the intensity of the δ(HOH) vibration, characteristic of molecular water, is reduced, the OH stretching mode at 3740 cm-1 is most likely due to the presence of adsorbed hydroxyl groups, which have been observed for H2O/ Al(111).13 Insight into the origin of the 370 and 770 cm-1 loss features is obtained by inspection of Figure 4, which compares EELS data for H2O/FeAl(110) and D2O/FeAl(110), obtained by individually dosing FeAl(110) with 10 langmuirs of D2O and H2O at 123 K and then heating to 173 K. The D2O spectrum shows that the 3740 cm-1 mode has shifted to 2735 cm-1 and that modes of H2O at 370 and 770 cm-1 show little movement upon isotopic substitution. The lack of a significant isotopic shift for the two low-frequency modes suggests that the 370 and 770 cm-1 modes are due to metal-oxygen stretches, ν(M-O). Consistent with this assignment is prior EELS research on H2O/Al(111),13 which shows that features develop at 365

Values in parentheses are frequencies for OD(ad). b This work. Reference 41. d Reference 13. e Reference 11. f Reference 10.

and 775 cm-1 due to OH(ad). This previous research, and the absence of an isotope effect for these modes in our experiments, suggest that the metal-oxygen stretches may be due to OH(ad) (or perhaps the complete decomposition of H2O to O(ad)). In either case, the spectrum obtained after heating the water-dosed surface to 173 K indicates that adsorbed hydroxyl is present on FeAl(110). Table 2 compiles the vibrational frequencies for the loss modes in the 173 K spectra, along with values for vibrations of OH(ad) on monometallic Al and Fe for comparison (discussed later in this paper). Further examination of Figure 3 shows that heating H2O/ FeAl(110) to 255 K results in spectral changes in the region near 770 cm-1. Perhaps the most noticeable change is that a relatively sharp feature begins to appear at 865 cm-1. This feature becomes more resolved in the 305 K spectrum. Another change in the 255 K data (relative to the 173 K spectrum) is the intensity decrease of the ν(OH) mode at 3740 cm-1. Heating to 410 K results in a further reduction of the ν(OH) mode intensity and in the appearance of three resolved modes at 425, 660, and 865 cm-1. These low-frequency modes, which persist after heating to 695 K, are attributed to the fundamental vibrations of Al2O3.13,32 The spectra in Figure 3 show that the ν(OH) mode at 3740 cm-1 in the 173 K spectrum remains at this frequency position even when molecular H2O is no longer present on the surface. This experimental result suggests that the 3740 cm-1 mode exhibited in the 173 K EELS spectrum is due to the O-H stretch of OH(ad) and not H2O. Figure 5 presents EELS data for FeAl(110) as a function of water exposure at 120 K. The 0.6 langmuir spectrum exhibits several vibrational modes at 355, 576, 775, 1640, and 3680 cm-1. Increasing the exposure to 2.2 langmuirs results in an increase of the intensity of the 1640 and 3700 cm-1 features. The 576 cm-1 mode also increases in intensity, relative to the vibrational modes at 355 and 775 cm-1. Focusing attention on the ν(OH) stretching feature in the 0.7 and 2.2 langmuir spectra, there appears to be an asymmetry on the low-frequency side of the peak centered at 3680 cm-1. We reserve discussing this asymmetry to later after we have presented all our spectroscopic results, but we will propose that the asymmetry is due to the presence of molecular water coadsorbed with hydroxyl groups. Nearly doubling the exposure to 5 langmuirs results in a significant increase in the 1640 and 3700 cm-1 modes. In addition to these intensity changes, a shoulder near 3500 cm-1 appears on the 3700 cm-1 loss, and the low-frequency modes between 200 and 1000 cm-1 coalesce into a broad feature centered at 775 cm-1. An increase in the water exposure to 10 langmuirs results in the shoulder on the 3700 cm-1 evolving into a relatively intense mode centered at 3460 cm-1, which is associated with the O-H stretch in a condensed layer of H2O. Assignment of the vibrational modes exhibited in Figure 5 is aided by a review of Figure 6 that shows EELS spectra after FeAl(110) is exposed individually to 2.5 langmuirs of H2O and

Water Adsorption on FeAl(110)

J. Phys. Chem., Vol. 100, No. 48, 1996 18833

Figure 7. XPS O(1s) core level spectra of 10 langmuir H2O/FeAl(110) 120 K and after annealing to 695 K. Figure 5. EELS spectra of FeAl(110) as a function of water exposure at 120 K. The 10 langmuir spectrum is representative of a multilayer adsorption.

Figure 6. Comparison of ∼2.5 langmuir H2O/FeAl(110) and D2O/ FeAl(110) at 120 K.

2.2 langmuirs of D2O. The 1640 and 1190 cm-1 modes in the low-coverage spectra of H2O and D2O, respectively, are assigned to the deformation mode of water, on the basis of the observed isotopic shift ratio of 1.38. Interpretation of the remaining features, however, is not as straightforward as in the water multilayer circumstance. The 3680 cm-1 mode, which shifts to 2680 cm-1 in the D2O spectrum, is due to a ν(OH) mode, but whether adsorbed H2O or OH is responsible for the vibration is not evident from these data. This same ambiguity exists in assigning the 355, 576, and 775 cm-1 modes in the 2.2 langmuir H2O/FeAl(110) spectrum. However, based on the annealing data presented in Figures 3 and 4, it is likely that the modes at 355, 775, and 3700 cm-1 are due to the presence of adsorbed hydroxyl species, since these modes appear at almost identical energy loss positions when molecular H2O is no longer present on the surface. We tentatively assign the mode at 576 cm-1 in the 0.7 and 2.2 langmuirs spectra exhibited in Figure 5 to the metal-water stretching mode [ν(M-H2O)]. Analogous modes are observed for associatively adsorbed H2O on Al(111) at 660 cm-1, and on Fe(100) at 425 cm-1.11,13 The assignment of the 576 cm-1 mode also is consistent with two experimental observations. First, the intensity of the 576 cm-1 mode increases, relative to the hydroxyl modes at 355 and 775 cm-1, when the exposure

is increased from 0.7 to 2.2 langmuirs. This spectral change occurs along with an increase of the δ(HOH) mode, suggesting that the mode depends on the surface concentration of associatively adsorbed water. Second, this mode is not present in the 173 K spectra, suggesting that this mode is not due to vibrations of hydroxyl groups. 3.3. XPS of H2O/FeAl(110). XPS measurements presented in this section add support to many of the conclusions obtained from EELS experiments. Figure 7 exhibits XPS O(1s) spectra of FeAl(110) that has been exposed to 10 langmuirs of H2O and heated to various temperatures. Various spectral changes occur in these XPS spectra as the temperature is increased. Heating the surface from 123 to 173 K results in a loss of a feature centered at 534.2 eV and the appearance of an O(1s) peak at 532.7 eV. The 534.2 eV feature is assigned to water in a condensed layer on FeAl(110). This contention is supported by noting that heating to 173 K leads to a significant reduction in the O(1s) photoelectron intensity, consistent with the desorption of a water multilayer (see Figure 2). The O(1s) peak centered at 532.7 eV in the 173 K XPS spectrum is assigned to the O(1s) binding energy of hydroxyl groups. This assignment is supported by research by Fuggle et al. that has obtained an O(1s) binding energy of 533.3 eV for (OH)ad on polycrystalline Al.18 This result from XPS also is consistent with our EELS results, which suggest that hydroxyl groups are in significant concentration on FeAl(110) at 173 K. Vibrational spectroscopy also has shown that decomposition of hydroxyl groups occurs by 250 K and is nearly complete by 410 K. This same temperature dependence of the hydroxyl decomposition step is inferred from the XPS data. XPS shows that a shoulder appears at 531.2 eV in the 250 K spectrum and that further heating to 300 K produces a well-resolved peak at this same energy position. XPS peak areas of the O(1s) core levels allow us to estimate that ∼35% of the OH(ad) has been converted to O(ad) at 250 K, and by 300 K, nearly half of OH(ad) originally present at 173 K has decomposed to chemisorbed oxygen. Additional heating to 400 K significantly reduces the remaining spectral weight at 532.7 eV due to OH species. Again, on the basis of XPS peak areas, we estimate that greater than 70% of the OH(ad) present at 173 K has decomposed to O(ad) by 400 K. Further annealing to 695 K leads to an O(1s) peak at 531.6 eV. We assign the 531.2 and 531.6 eV O(1s) features to aluminum oxide,33 consistent with our conclusion from the EELS data. Further analysis of the integrated peak

18834 J. Phys. Chem., Vol. 100, No. 48, 1996

Gleason and Strongin

Figure 8. XPS of the Fe 2p3/2 and Al 2s levels of clean FeAl(110) and after the surface is exposed to 10 langmuirs of H2O and annealed to 695 K.

areas of the O(1s) features shows that the concentration of oxygen-containing species does not change on the surface as the temperature is raised from 173 to 695 K. This result also is consistent with TPD experiments that show that the only species that desorbs from FeAl(110) at temperatures greater than 200 K is molecular hydrogen. We note that a previous study of H2O on Al(111) by Szalkowski15 assigned O(1s) binding energies of 533.3 and 534.5 eV to adsorbed H2O at low and high coverage, respectively. However, assignment of our 532.8 eV peak to chemisorbed H2O would not be consistent with our experimental results. Data presented in Figure 7 shows that the 532.7 eV feature persists up to 400 K. EELS shows conclusively that H2O is not present at these temperatures, as evidenced by the absence of the scissoring mode, δ(HOH), of H2O. The formation of aluminum oxide also is evidenced by examining the core levels of the Fe and Al substrate atoms. Figure 8 shows the clean XPS Fe 2p3/2 and Al 2s core levels and those same core levels after a 10 langmuirs dosed surface has been annealed to 695 K. The Al 2s core level of the water treated surface has been fit with two Gaussians, one centered at 117.3 eV and one at 119.4 eV. The high-binding energy feature at 119.4 eV, which appears in the Al 2s spectrum after water decomposition, is characteristic of aluminum oxide. It is also noted that the iron component remains metallic, although the signal has become attenuated, presumably due to Al2O3 formation on the surface. From the attenuation of the Fe 2p3/2 core level intensity, it is possible to calculate the Al2O3 film thickness. Using the expression from Briggs and Seah,34

I/I0 ) e-d/λ sin θ where λ is the mean free path of the photoelectron through the solid, approximated to be 7.5 Å,29 it is possible to estimate the Al2O3 film thickness, d, as 3.5 Å. Figure 9 exhibits XPS O(1s) measurements after FeAl(110) is exposed to varying amounts of H2O at 120 K. The goal of these experiments is to help determine whether the predominant species at low coverage is OH(ad) as the EELS results suggest. To interpret these experiments, we keep in mind the XPS results from above that indicate that the binding energy of OH(ad) is at least 1 eV lower than that for H2O(ad) . For the two lowest exposures, 0.5 and 2 langmuirs, the peak maxima are centered at ∼532.5 eV, although the full width at half-maximum of each peak is greater than 3 eV, suggesting that more than one oxygencontaining species, OH(ad) and H2O(ad), are contributing to the signal. The Gaussian fits of the data provided in the figure are

Figure 9. XPS O(1s) core level spectra as a function of exposure at 120 K. The peak centered at 532.5 eV at lowest exposures is due to OH(ad), and the spectral weight detected at higher binding energy is due to H2O(ad). As the exposure is increased, the feature due to H2O(ad) continues to increase in intensity. Gaussian fits are provided to highlight the changes in the O(1s) core level as the exposures are increased.

primarily to guide the reader to the development of the OH(ad) and H2O(ad) contributions to the signal as the exposure is increased, but they also allow an estimate of the relative amounts of each species. For the 0.5 and 2 langmuirs exposures, the predominant species on the surface at 120 K is OH(ad), with a binding energy of 532.6 eV, while about 30-40% of the total signal is due to the presence of H2O(ad). The binding energy of H2O(ad) varies as the exposure is increased, starting at 533.5 eV for a chemisorbed state and gradually increasing to greater that 534 eV for a condensed layer. The O(1s) core level of the 4.3 langmuir dosed surface shows significant broadening to higher binding energy. The fit of the data shows that the adlayer is a ∼50/50 mix of OH(ad) and H2O(ad). The 6 langmuir dose represents a multilayer, although the peak area analysis shows that there is approximately a 10% concentration of OH(ad). These data confirm the EELS results presented in the preceding section that suggest that the predominant species at low coverages is OH(ad). 3.4. Photoemission of H2O/FeAl(110) Using Synchrotron Radiation. Data presented in the preceding sections have helped to identify the adsorbed species resulting from the adsorption of H2O on FeAl(110) at 120 K and after heating to higher temperatures. Data present thus far, however, do not allow us to make strong statements concerning the binding site of the different adsorbates. To address this issue, a series of photoemission experiments were carried out at the National Synchrotron Light Source (NSLS) that probed the electronic structure of the outermost atoms that chemisorb the adsorbate. Figure 10 shows Al 2p core level data for FeAl(110) after H2O adsorption at 120 K and subsequent annealing to 700 K. For the clean alloy surface, the Al 2p peak is at 72.2 eV, which represents a shift of about 0.7 eV to lower binding energy compared to pure Al, and is consistent with previous studies in our laboratory28 for FeAl. Data in Figure 10 show that adsorption of water induces a new feature that is positioned 0.8 eV to higher binding energy of the main peak at 72.2 eV. EELS and XPS suggest that OH(ad) and H2O(ad) coexist on FeAl(110) under these experimental conditions. We suspect that the feature at 72.2 eV is most likely due to hydroxyl binding directly

Water Adsorption on FeAl(110)

J. Phys. Chem., Vol. 100, No. 48, 1996 18835

Figure 11. Fe 3p core level spectra of FeAl(110) after dosing with 0.7 langmuir of H20 at 120 K and annealing to 250 K. Figure 10. Al 2p core level spectra of FeAl(110) after dosing with 0.7L of H2O at 120 K and annealing to 700 K.

to the Al, but we briefly discuss this assignment later. Substantial spectral changes occur after the surface is annealed to 173 K; two features develop at 73.5 and 74.9 eV. Spectroscopic data presented previously suggest that annealing H2O/ FeAl(110) to 173 K creates a hydroxylated overlayer, although the presence of chemisorbed oxygen was not eliminated by EELS as a possibility. We contend that these photoemission experiments show that chemisorbed oxygen is indeed present at 173 K, since this species is associated with the 74.9 eV feature in the Al 2p spectrum. Prior studies of oxygen on Al have generally assigned features that are ∼2.4 eV to higher binding energy than the metallic feature to Al2O3.35 Other studies have postulated that the development of this feature can also be associated with the incorporation of chemisorbed oxygen into the near subsurface or between the first and second layers of pure Al.36 This latter assignment seems like a more likely scenario, since it would be consistent with our EELS data. Prior studies have shown that the relative intensities (as well as the stretching frequencies) of the Al-O stretches between 300 and 900 cm-1 for O2/Al(111) vary as a function of oxygen coverage and sample temperature.37,38 In particular, the vibration associated with subsurface oxygen at ∼800 cm-1 tends to be relatively weak at low temperature (below 300 K) and oxygen coverage38 and could therefore be suppressed by the intense vibrations of OH(ad) in the low-frequency region for H2O/FeAl(110). This circumstance is in contrast to EELS of AlxOy that exhibits intense vibrational modes below 1200 cm-1. These intense modes do not start to develop for H2O/FeAl(110) until the surface is annealed to temperatures above 300 K. The Al 2p feature at 74.9 eV in the 173 K spectrum progressively grows in intensity and gradually shifts to 75 eV as H2O/FeAl(110) is stepwise heated to 700 K. We associate these changes with the conversion of OH(ad) to O(ad) and the formation of aluminum oxide. Even this assignment is not definitive, since various forms of Al hydroxides and Al oxides show a wide range of values between 73.5 and 76.5 eV.39,40 Since the Al 2p core level in FeAl shifts to by 0.7 eV lower binding energy relative to pure Al (BE ) 72.9 eV) upon alloy formation, the situation becomes more complicated. The feature at 73.5 eV observed in the 173 K spectrum is most likely due to OH(ad), since the binding energy shift falls within the range observed for Al hydroxides and EELS confirms the existence of this species at 173 K. However, it is also possible to associate this feature with adsorbed oxygen, since surface oxygen on

aluminum shows a shift of 1.4 eV to higher binding energy relative to the pure metallic peak,35 and this peak persists to temperatures where OH(ad) are not detected by our EELS or O(1s) XPS data. It seems unlikely that any straightforward assignment of the features between 73.5 and 75 eV will be possible, since the composition of the surface, that is, the relative concentration of OH(ad) to O(ad), is changing significantly as the temperature is increased. It is noted that the feature at 72.2 eV, associated with metallic aluminum, appears to decrease at the expense of the higher binding energy peaks in the 300 and 450 K spectra. Annealing to 700 K, however, results in an increase of the 72.2 eV peak. We associate the increase in the intensity of the metallic aluminum peak as being due to surface segregation of aluminum. Figure 11 exhibits complementary Fe 3p core level data for adsorption of H2O at 120 K and after stepwise heating to 250 K. The most important observation from Figure 11 is that the “line shape” (and binding energy) of the Fe 3p feature remains relatively unaffected by the presence of water, hydroxyl groups, or chemisorbed oxygen on the surface. To emphasize the constancy of the line shape throughout the temperature range of 120-250 K, each spectrum in Figure 11 has been fit with an exponential Gaussian peak. In each fit the width and binding energy of the peak have been held constant (only the amplitude is allowed to vary), showing that the Fe 3p feature, within our experimental resolution, is unaffected by the decomposition fragments of water binding preferentially to the Al component. These data suggest that Fe remains in its metallic state within the near surface region during the decomposition of water. (This conclusion is consistent with the data presented in Figure 8 that showed that the formation of Al2O3 produces no change in the Fe 2p3/2 core level.) While the line shape is constant, there are changes in the Fe 3p data that deserve mention. First, the intensity of the Fe 3p feature is reduced by a factor of about 2 after H2O adsorption, ∼5 times less than the attenuation of the Al 2p level after adsorption. On the basis of the lack of significant changes in the line shape of the Fe 3p level during the stepwise heating, we suspect that this attenuation is due to H2O and or OH(ad) adsorbing to Al rather than to Fe. We would expect more spectral weight to appear on the high binding side of the Fe 3p feature if significant interaction of the water and decomposition fragments were occurring at these temperatures. Segregation of Al at 120 K on top of the Fe component after H2O adsorption also is a possibility, but we feel it is less likely than the explanation above. If Al segregation were the reason, it might

18836 J. Phys. Chem., Vol. 100, No. 48, 1996 be expected that the Fe 3p intensity would diminish upon heating, since the increase in temperature would induce more diffusion. This phenomenon, however, is not observed upon heating to 250 K. The second change in the Fe 3p data after H2O adsorption is in our opinion less striking. There appears to be a small shift (or slight broadening) to higher binding energy (0.2 eV) in the peak maximum of the Fe 3p feature after exposure to H2O at 120 K. It is suspected that this shift is due to the perturbation of the Fe-Al bond due to H2O adsorption (discussed in the next section) in the outermost surface. In short, it is suspected that H2O adsorption results in a heterogeneous chemical environment for Fe that results in a slight broadening or apparent shift of the Fe 3p binding energy. 4. Discussion From the data presented in the preceding section, the following reaction steps may be proposed for the adsorption and decomposition of H2O on FeAl(110):

H2O(g) f H2O(multilayer)

T < 150 K

H2O(multilayer) f H2O(g)

T ) 150 K

H2O(ad) f OH(ad) + H(ad)

T ) 123-173 K

OH(ad) f O(ad) + H(ad)

T ) 250-400 K

2H(ad) f H2,(g)

T ) 200-400 K

yO(ad) + xAl f AlxOy

T ) 250-700 K

4.1. Initial Interaction of H2O on FeAl(110) at 120 K. With the least hesitation we conclude from our results that at least a fraction of the water that interacts with FeAl(110) at 120 K chemisorbs in an associative manner. This conclusion is based largely on the experimental observation of the δ(HOH) mode in the EELS data of H2O/FeAl(110). Also consistent with this contention is the presence of spectral weight above 533 eV in the O(1s) spectrum of H2O/FeAl(110) at 120 K. Deciding from our EELS data whether some fraction of water dissociates to OH(ad) at 120 K is slightly more ambiguous due to the lack of a distinctive mode that can be assigned with this moiety. Heating the water multilayer to 173 K eliminates the δ(HOH) mode of associatively adsorbed H2O and leaves a high-frequency mode at 3740 cm-1, and in this circumstance we can assign this high-frequency mode to the ν(OH) stretch of hydroxyl. Dosing FeAl(110) with relatively low exposures of H2O at 120 K, however, results in loss features between 3680 and 3700 cm-1 which are asymmetric to lower stretching frequency. Unlike the mode at 3740 cm-1, the modes between 3680 and 3700 cm-1 are not easily assigned, since at 120 K associatively adsorbed H2O is present. Hence, it is difficult to determine whether OH(ad) is present at 120 K, since its O-H stretching mode is expected to be similar in frequency to the corresponding mode of associatively adsorbed H2O. As mentioned previously, the energy loss positions for H2O/ FeAl(110) are very similar to those experimentally observed for H2O/Al, and it is probably safe to inspect our EELS results in view of these prior studies. Prior research of water on Al by Szalkowski suggested that ν(OH) modes near or greater than 3680 cm-1 could be due to the asymmetric OH stretch of water.15 Crowell et al., however, have proposed that the mode near 3700 cm-1 may be associated with OH(ad),13 since the frequency is very similar to the loss position of OH(ad) (∼3740 cm-1) resulting from the decomposition of H2O at relatively high temperatures (300 K). More recent research by Norton

Gleason and Strongin and co-workers have detected a high-frequency ν(OH) stretch for low coverages of H2O on Al/(100) using infrared spectroscopy. They propose that the mode is due to non-hydrogenbonded water molecules, as either isolated molecules on the surface or the free OH groups at the edges of a water clusters on a surface.25 Their results also indicated that the loss frequency of the ν(OH) mode of associatively adsorbed H2O and OH(ad) can reside at similar frequencies. There appears to be a general way in which our results can be interpreted. The ν(OH) mode at ∼3700 cm-1 is most likely due to a combination of molecular H2O and OH(ad), consistent with the detected asymmetry of the peak. This picture is also consistent with the XPS results that show two O(1s) binding energies are detected at 120 K for lower coverages, with the primary peak at 532.5 eV due to OH(ad) and a smaller contribution near 533 eV due to molecular H2O. Since the mode at ∼3700 cm-1 is asymmetric to lower frequency and it is known that the ν(OH) mode of hydroxyl groups falls at a higher frequency than molecular H2O (at least 60 cm-1 higher), it is reasonable to suggest that the peak centered at ∼3700 cm-1 is due to OH(ad) and the shoulder at lower frequency is due to H2O(ad). This explanation is not entirely satisfactory since it does not explain why the ν(OH) mode of OH(ad) occurs at 3740 cm-1 when the hydroxyl overlayer is prepared by heating a water multilayer to 173 K. However, this difference may be due to differences in the OH(ad) coverage that result from the two preparation routes or that by heating to 173 K there is some restructuring of the FeAl surface. Besides providing insight into the atomic composition of the binding site of specific surface species, the results of the synchrotron-based experiments help understand the nature of the surface species on FeAl(110) upon exposure to H2O at 120 K. Photoemission shows that a new feature in the Al 2p region develops 0.8 eV higher in binding energy than the feature associated with Al in the clean alloy. Based on prior studies, this binding energy shift is consistent with the presence of hydroxyl groups,37 and supports the surface picture of a mixed layer of OH(ad) and H2O on FeAl(110) at 120 K. (The possibility that adsorbed H2O causes the 73.5 eV feature, however, cannot be ruled out.) The mode of adsorption of H2O on FeAl(110) is contrary to what has been experimentally observed for H2O/NiAl(110). Our results for this latter system suggest that H2O primarily associatively adsorbs on NiAl(110), consistent with the experimental observation that H2O, unlike on FeAl(110), desorbs from NiAl at submultilayer exposures.41 Interestingly, the primary contribution to the O(1s) binding energy feature of H2O/NiAl(110) at 120 K is at 533.3 eV and the minority species is at 532.5 eV, which implies that the predominant species is H2O(ad) with a small contribution from OH(ad). This is the opposite of what has been observed for H2O/FeAl(110). Also, the energy loss position of the ν(OH) mode of associatively adsorbed H2O on NiAl(110) is at 3620 cm-1, and upon heating to 173 K it shifts to a frequency near 3740 cm-1, which we associate with OH(ad). The similarity between the EELS spectra as a function of exposure obtained in this study for FeAl(110) and the spectra obtained by Crowell et al. for H2O/Al(111)13 lends support to a surface bonding picture that has the decomposition fragments of H2O (and possibly molecularly adsorbed H2O) bonding to the Al component of the alloy at low temperature. However, this similarity of the EELS spectra is perhaps surprising considering that the environment of the Al component of the FeAl alloy is expected to be markedly different than monometallic Al. We suspect that the similarity might be the result of

Water Adsorption on FeAl(110) a perturbation of the FeAl bond in the surface layer upon water adsorption. Water typically bonds to metal surfaces via the oxygen atom, with a bond energy that varies between 42 and 50 kJ/mol for smooth surfaces and is approximately 10 kJ/mol stronger for rough surfaces, based on first-order desorption kinetics.1 The Fe-Al bond strength is ∼25 kJ/mol,42 substantially lower than the metal-water bond strength. Based on these bond energy estimates, it is possible that water adsorption forms Al-H2O and/or Al-OH bonds at the expense of the FeAl bond. This scenario may partially explain why all of the XPS and EELS spectra resemble that observed on pure Al, since the high affinity of Al for oxygen43 may direct water to preferentially adsorb on that component. It is revealing to compare this behavior of H2O on FeAl(110) to that of NiAl(110). Unlike the H2O/FeAl(110) circumstance, the vibrational spectra of H2O/NiAl(110) at 120 K differs significantly from that of H2O on FeAl(110) and monometallic Al. We believe that this is due to the relatively high bond energy of NiAl (59 kJ/mol)42 compared to FeAl, enabling the Ni-Al bond to remain intact even following the chemisorption of H2O. 4.2. Annealing of H2O above 120 K. Both XPS and EELS data suggest that heating a submonolayer-dosed surface to 173 K, which is past the desorption temperature of water, decomposes molecular water to produce OH(ad) and atomic hydrogen. Furthermore, the synchrotron-based experiments suggest that conversion of some OH(ad) to O(ad) and H(ad) may even occur. In either case TPD shows that decomposition of the molecular adsorbate is an efficient reaction step, considering that no H2O is found to desorb from FeAl(110) after submultilayer exposure. Table 2 compiles the stretching frequency of the ν(OH) mode of OH(ad) on FeAl(110) and, as comparison, on NiAl(110) and monometallic Al(111). The similarity of the ν(OH) frequencies of OH(ad) on pure Al and the alloy surfaces strongly suggests that adsorbed hydroxyl interact preferentially with the Al component. Also, the frequency of the O-H stretching mode on FeAl(110) (and on NiAl(110)) is relatively constant throughout the temperature interval 173-600 K. This not only suggests that the fundamentals of the OH(ad) surface bonding are constant over this temperature interval but also suggests that atomic oxygen generated as a decomposition product does not significantly affect the interaction of OH(ad) with the surface. This strong similarity between the vibrations of OH(ad) on FeAl(110) and NiAl(110) to the same species on monometallic Al again supports the conclusion that the Al component is preferentially binding OH(ad). Annealing the surface to higher temperatures initiates the decomposition of adsorbed hydroxyl group as confirmed by TPD, XPS, and EELS. In particular, OH(ad) decomposition is observed when the surface is annealed to 250 K, and the decomposition process continues until OH(ad) is no longer detected above 400 K. For comparison, adsorbed hydroxyl groups are stable on Al(111) at temperatures up to 450 K.13 Hydrogen desorption is complete by 400 K, and its similarity to TPD spectra for pure iron suggests that the kinetics of hydrogen desorption are controlled to a large degree by the Fe component of FeAl(110). Concerning the interaction between Fe and hydrogen it is noted, that the maximum concentration (as inferred from EELS) of hydroxyl groups on FeAl(110) occurs near 173 K. (At higher temperatures this species starts to decrease). On Al(111) hydroxyl remains relatively stable up to a temperature of 250 K.13 This comparison might suggest that the FeAl alloy is more reactive than pure Al for the decomposition of OH(ad), possibly due to the reactivity of the Fe component for the resulting surface hydrogen product.

J. Phys. Chem., Vol. 100, No. 48, 1996 18837 The presence of chemisorbed oxygen on the aluminum component is confirmed by EELS, XPS, and photoemission obtained with synchrotron radiation. As is the case for the binding of OH(ad), the high affinity of Al for oxygen and the high heat of formation of Al2O3 causes the O(ad) to preferentially bond to and eventually oxidize the Al component. The EELS spectrum after annealing a multilayer dosed surface to 695 K is nearly identical to that previously observed for Al2O3,14 and the XPS metal core level spectra confirms that the aluminum component is preferentially oxidized. This same behavior is observed for OH(ad) and O(ad) on the NiAl(110) system, again indicating that the chemistry of these oxygen-containing species is controlled by the Al component.41 5. Summary Spectroscopic data presented in this paper suggests that H2O undergoes both associative and dissociative adsorption on FeAl(110) at 120 K. Any associatively adsorbed H2O dissociates to OH(ad) upon heating to 173 K. By 250 K, the decomposition of OH(ad) and the desorption of molecular hydrogen have started to occur. Continued annealing further decomposes the adsorbed hydroxyl species to chemisorbed oxygen, and the atomic hydrogen product continues to recombine and desorb. Oxygen resulting from the decomposition reactions of water preferentially bonds to the Al component and in the case where FeAl(110) is exposed to large amounts of water eventually results in the formation of an aluminum oxide overlayer. The similarity between the vibrational data for H2O/FeAl(110) and H2O/Al(111) as well as the synchrotron photoemission data leads us to the conclusion that both the associatively adsorbed H2O and OH(ad) interact most strongly with the Al component. References and Notes (1) Thiel, P. A.; Madey, T. E. Surf. Sci. Rep. 1987, 7, 456. (2) Fisher, G. B.; Sexton, B. A. Phys. ReV. Lett. 1980, 44, 683. (3) Ibach, H.; Lehwald, S. Surf. Sci. 1980, 91, 187. (4) Thiel, P. A.; Hoffmann, F. M.; Weinberg, W. H. J. Chem. Phys. 1981, 75, 5556. (5) Thiel, P. A.; DePaola, R. A.; Hoffmann, F. M. J. Chem. Phys. 1984, 80, 5326. (6) Heras, J. M.; Papp, H.; Spiess, W. Surf. Sci. 1982, 117, 590. (7) Olle`, L.; Salmero`n, M.; Baro` A. M. J. Vac. Sci. Technol. A. 1985, 3, 1866. (8) Benndorf, C.; Nobl, C.; Madey, T. E. Surf. Sci. 1984, 138, 292. (9) Brosseau, R.; Brustein, M. R.; Ellis, T. H. Surf. Sci. 1993, 280, 23. (10) Baro`; A. M.; Erley, W. J. Vac. Sci. Technol. 1982, 20, 580. (11) Hung, W.-H.; Schwartz, J.; Bernasek, S. L. Surf. Sci. 1991, 248, 332. (12) Hung, W.-H.; Schwartz, J.; Bernasek, S. L. Surf. Sci. 1993, 294, 21. (13) Crowell, J. E.; Chen, J. G.; Hercules, D. M.;Yates, Jr., J. T. J. Chem. Phys. 1987, 86, 5804. (14) Chen, J. G.; Basu, P.; Ng, L.; Yates, Jr., J. T. Surf. Sci. 1988, 194, 397. (15) Szalkowski, F. J. J. Chem. Phys. 1982, 77, 5224. (16) Paul, J.; Hoffmann, F. M. J. Phys. Chem. 1986, 90, 5321. (17) Chen, J. G.; Crowell, J. E.; Yates, Jr., J. T. J. Chem. Phys. 1986, 84, 5906. (18) Fuggle, J. C.; Watson, L. M.; Fabian, D. J.; Affrossman, S. Surf. Sci. 1975, 49, 61. (19) Nieuwenhuys, B. E. In The Chemical Physics of Solid Surfaces: Coadsorption, Promoters, and Poisons; King, D. A.; Woodruff, D. P., Eds.; Elsevier: Amsterdam, 1993; Chapter 6. (20) Chia, W. J.; Chung, Y. W. J. Vac. Sci. Technol. A 1995, 13 (3), 1687. (21) George, E. P.; Liu, C. T.; Pope, D. P. Scr. Metall. 1994, 30, 37. (22) Chang, K.-M.; Darolia, R.; Lippsitt, H. A. In High-Temperature Ordered Intermetallic Alloys IV; Johnson, L. A., Pope, D. P., Stiegler, J. O., Eds.; Materials Research Society Pittsburgh, PA, 1991; Vol. 213. (23) Lui, C. T.; George, E. P. In Alloy Phase Stability and Design; Stocks, G. M., Pope, D. P., Stiegler, J. O., Eds.; Material Research Society: Pittsburgh, PA, 1991; Vol. 186. (24) Fu, C. L.; Painter, G. S. J. Mater. Res. 1991, 6, 719.

18838 J. Phys. Chem., Vol. 100, No. 48, 1996 (25) Bushby, S. J.; Callen, B. W.; Griffiths, K.; Esposto, F. J.; Timsit, R. S.; Norton, P. R. Surf. Sci. 1993, 298, L181. (26) Memmert, U.; Bushby, S. J.; Norton, P. R. Surf. Sci. 1989, 219, 327. (27) Ko¨ster, W.; Go¨decke, T. Z. Metallkd. 1980, 71, 765. (28) Gleason, N. R.; Strongin, D. R. Surf. Sci. 1993, 295, 306. (29) Seah, M. P.; Dench, W. A. Surf. Interface Anal. 1979, 1, 2. (30) Graupner, H.; Hammer, L.; Mueller, K.; Zehner, D. M. Surf. Sci. 1995, 322, 103. (31) Lui, S.-C.; Davenport, J. W.; Plummer, E. W.; Zehner, D. M.; Fernando, G. W. Phys. ReV. B 1990, 42, 1582. (32) Chen, P. J.; Colaianni, M. L.; Yates, Jr., J. T. Phys. ReV. B 1990, 41, 8025. (33) Pashutski, A.; Hoffman, A.; Folman, M. Surf. Sci. 1989, 208, L91. (34) Briggs, D.; Seah, M. P. Practical Surface Analysis by Auger and X-Ray Photoelectron Spectroscopy; John Wiley & Sons: New York, 1983.

Gleason and Strongin (35) Flodstro¨m, S. A.; Martinsson, C. W. B.; Bachrach, R. Z.; Hagstro¨m, S. B. M.; Bauer, R. S. Phys. ReV. Lett. 1978, 40, 907. (36) Norman, D.; Brennan, S.; Jaeger, R.; Stohr, J. Surf. Sci. 1981, 105, L297. (37) Chen, J. G.; Crowell, J. E.; Yates, J. T. Jr., Phys. ReV. B 1986, 33, 1436. (38) Crowell, J. E.; Chen, J. G.; Hercules, D. M.; Yates, Jr., J. T. J. Chem. Phys. 1987, 86, 5804. (39) Halverson, D. E.; Cocke, D. L. J. Vac. Sci. Technol. A 1989, 7, 40. (40) Thomas, S.; Sherwood, P. M. A. Anal. Chem. 1992, 64, 2488. (41) Gleason, N. R.; Chaturvedi, S.; Strongin, D. R. Surf. Sci. 1995, 326, 27. (42) Pasturel, A.; Manh, D. N.; Mayou, D. J. Phys. Chem. Solids 1986, 47, 325. (43) Netzer, F. P.; Madey, T. E. Surf. Sci. 1983, 127, L102.

JP962449D