- Email: [email protected]
Orientation of water adsorbed on Cu(110)
279
Surface Science 126 (1983) 279-285 North-Holland Publishing Company
ORIENTATION
OF WATER ADSORBED ON Cu(ll0)
C. MARIANI * and K. HORN Fritz- Haber - fnstitur der Max - Planck - Gesellschaft,
Faradayweg 4 - 6, D - 1 OOt?Berlin 33, Fed. Rep.
of Germany Received
24 August
1982; accepted
for publication
22 September
1982
The method of determining the orientation of an adsorbed molecule by angle-resolved phot~~ssion with polarised light has been applied to the case of water adsorbed on clean copper (110). By using dipole selection rules and symmetry arguments, it is found that in the full monolayer as well as at submonolayer coverages, the water molecules are adsorbed in a tilted geometry. Only small shifts in the valence photoemission peak relative to the gas phase spectrum are found, indicative of a weak chemisorption bond. These results are explained, by recourse to LEED observations, in terms of recent models of water layers on metal surfaces.
1. Introduction Water adsorption on metal single crystals has recently been studied with a number of surface-sensitive techniques [l-9]. A central question has been the o~entation of the water molecules on the surface. Netzer and Madey [S], using the ESDIAD technique, concluded that the array of H,O molecules on Ni( 111) contains a distribution of tilt angles with respect to the surface normal. The evidence from vibrational spectroscopy indicates that, even at very low coverages, H,O forms hydrogen-bonded clusters on Ru(OOl) and Pt(lOO) [1,2,6], with the hydrogen bonding leading to a large broadening and intensity enhancement of the O-H stretching vibration. For H,O on Cu( 1lo), however, Spitzer and Ltith [8] conclude from UPS and ELS experiments that, at least up to 0.5 monolayer coverage, cluster formation is not a dominant effect, and that the water molecules are bonded to the surface in a configuration with the molecular plane normal to the surface. In view of this situation, it is interesting to investigate water adsorption on metals in angle-resolved photoemission, and to interpret the results obtained with polarised light by using dipole selection rules, a method which has been successfully applied to the study of the
* Present address: Dipartimento ione Cosentino Scaio, Italy.
di Fisica,
0039-6028/83/ooO0-0000/$03.00
Universita
degli studi della Calabria,
0 1983 North-Holland
I-87030 Castigl-
280
C. Mariani, K. Horn / Orientation of water on Cu(ll0)
electronic and geometric structures of several adsorbed molecules present study seems to be the first application of this method adsorption.
[lo]. The to water
2. Experimental A commercial angle-resolving photoelectron spectrometer (Vacuum Generators ADES 400) equipped with a triple-reflection polariser [ 1 l] was used in all experiments. Specimen preparation, work function determination, etc. have been previously described [12]. Triply distilled water, further purified by freezing/thawing cycles under vacuum, was used for the exposures.
3. Results and discussion Adsorption of water on Cu( 110) at 90 K leads to the appearance of three extra peaks in the He I photoelectron spectrum in normal electron emission, as shown in fig. 1. These peaks at 6.8, 8.8, and 12.6 eV below E, are assigned to the 1b,, 3a ,, and lb, valence orbitals of H,O by comparison with the gas phase spectrum, which is also shown (ref. [ 131). Using the work function of clean Cu( 1 IO) (4.5 eV) it is found that all peaks are shifted to higher binding energy by about 1.4 eV compared with the gas phase. We ascribe this shift to the effect of extramolecular relaxation. Any differential shift which might be indicative of a bonding shift on one of the orbitals is certainly smaller than 0.2 eV, although it must be stressed that, particularly for the 3a, and 1b, peaks, a determination of exact peak position is rather difficult due to the large half width and overlap. Upon heating the water-exposed surface, it is found that all water-induced peaks disappear above a temperature of 170 K; no evidence for dissociation was found, the photoelectron spectrum after desorption being identical to that of the clean surface recorded before exposure, in disagreement with results of Spitzer and Ltith [8] who observe the formation of a hydroxyl species. The origin of this disagreement is difficult to determine at present; it is known that small traces of co-adsorbed oxygen can induce water dissociation and hydroxyl formation. The low desorption temperature and the lack of strong bonding shifts observed in the photoelectron spectra both show that the chemisorption bond is weak; from the desorption temperature the heat of adsorption is estimated to be 50-60 kJ mol- ‘. Water causes a large change in work function, which was monitored by recording the low energy cutoff in the spectra with the crystal biased at - 10 V. It is found that the work function decreases by about 0.95 f 0.03 eV after an exposure of 0.8 L. At higher exposures multilayer formation occurs, which results in the photoemission peaks appearing at higher binding energy.
C. Mariani, K. Horn / Orientation of water on Cu(ll0)
281
‘bl
3a1
lb2 1,1!,
I
21
19 Energy
I
17 below
I
I
13
15 E,ac
b)
surface
l
c 1 s - polarmd
hght
I s- polarised
hght
d
ii parallel 06L H20
18
I 16
1 14
I 12
I 10
8I
I 6
Energy
Fig. 1. Photoelectron surface. unpolarised
hI below
2I EF
I
EC (eV) ’
spectra of (a) water in the gas phase. from ref.
[ 13):
light; (c) water absorbed on Cu( I IO) at a temperature
recorded with s-polarised He
I
to (001)
(b) spectrum of clean
of 93 K. exposure 0.6 L.
light. with the electric vector along the (I IO) azimuth: (d) same hut
electric vector oriented along the (001) azimuth. Spectra (b)-(d) emission.
were recorded in normal electron
282
C. Mariani, K. Horn / Orientation of water on Cu(Il0)
The value for A&,,,, is in good agreement with that measured by other groups [8,15]. The work function shows a further decrease of about 200 mV beyond the monolayer coverage, the completion of which was judged from the onset of a shift in photoemission peak positions. We base our determination of the site symmetry of the adsorbed water molecules on the spectra of fig. 1, recorded with polarised He I radiation, with the electrons detected in a narrow cone around the surface normal. It has previously been shown [lo] that, in such a geometry, only emission from those orbitals which form a basis for the same irreducible representations as those of the Cartesian components of the momentum operator is allowed. Emission from an AZ-type orbital in C Zv symmetry [14] is always forbidden in normal emission, for example. Furthermore, when s-polarised light is used, emission from orbitals that belong to the totally symmetric irreducible representation is forbidden, being allowed only when the electric field at the surface has a normal component [lo]. This polarisation selection rule enables one, starting from a correlation of peaks in the spectrum of the adsorbate with that of the gas phase, to determine the local symmetry of the adsorbed species, or rather to place certain restrictions on the choice of adsorbate point groups. The present case of Hz0 on Cu(ll0) serves to explain this procedure. We have already assigned the peaks in the spectra of fig. 1 by comparison with the gas phase spectrum. Since the highest site symmetry on a fee metal (110) surface is CZv, it is appropriate to start our analysis with the assumption that the water molecule occupies such a site in which its gas phase symmetry is not lowered, for example an “on top” site either in the trough along the [liO] azimuth or perpendicular to it. Bonding is most likely to occur through the oxygen lone pair (the lb,), which is also closest in energy to the metal d levels. For a single adsorbed water molecule, a likely configuration is one in which the water is adsorbed with the oxygen end nearest to the metal, the hydrogen atoms The spectra shown in pointing away from the surface in a C,, configuration. fig. 1 prove that this configuration is not taken up by the water molecules under our experimental conditions, i.e. at 90 K and coverages at around 0.6 maximum coverage. Here, we observe appreciable emission from the 3a, orbital in normal emission and with s-polarised light irrespective of the orientation of the electric vector with respect to the substrate azimuths; this peak would be forbidden if the H,O molecules were adsorbed in a C,, configuration. The presence of several domains of differently oriented molecules, or of two kinds of differently oriented molecules in a larger unit cell, does not affect this conclusion as long as local C,, symmetry is maintained. We thus conclude that, already at coverages below a full monolayer, the molecules do not occupy C,, sites, i.e. they might be tilted such that the C, axis is no longer a valid symmmetry element. This is not the only information on the geometry that one can derive from the spectra of fig. 1, however. A number of different “tilted” geometries may
C. Mariani,
K. Horn / Orientation
of water on Cu(lI0)
283
be visualised, for example one in which the molecular plane is still normal to the surface, but one hydrogen atom closer to the metal than the other. By examining this geometry, we find that any kind of configuration in which the molecular plane coincides with a mirror plane of the crystal, i.e. where C, site symmetry is maintained, would still cause some peaks to be forbidden in
a
b Fig. 2. (a) Model of a single sheet of hexagonal ice, taken from ref. [6) (with permission); (b) ball model of the ice layer in geometry (a) on an fee (110) surface; white spheres indicate top layer copper atoms, small black spheres indicate oxygen atoms, small grey spheres indicate hydrogen atoms.
284
C. Mariani, K. Horn / Orientation of water on Cu(Il0)
spectra recorded with s-polarised light. If the molecular plane were coincident with the (170) mirror plane of the substrate, for example, emission from the lB, orbital would not be allowed in s-polarised light with the electric vector parallel to the [ Ii01 direction, while emission from the lB, and 3A, orbitals would be forbidden if the electric vector were oriented parallel to the [OOl] direction. (The case for a water molecule oriented with its molecular plane parallel to the (001) mirror plane can be treated similarly.) This is, however, contrary to the experimental observations. Only for a geometry where all molecules occupy sites such that no symmetry element is maintained, for example if the molecular plane does not contain the surface normal, or if the adsorbed water molecules are arranged in several different geometries, would all peaks appear in the spectra, irrespective of the orientation of the electric vector. These findings can be correlated with LEED results by Spitzer and Ltith [8] and Grider et al. [ 151 who observe, already at coverages below a complete monolayer, the formation of a c(2 x 2) superstructure. Since the driving force for a tilting of the water molecules is the interaction of neighbouring admolecules through the hydrogen bond, it was proposed that adsorption of water on metals proceeds through island growth [ 11, and that the geometry of the water molecules in these islands closely resembles that in a single sheet of hexagonal ice [6]: this interpretation was found to be consistent with LEED and ESDIAD results as well as vibrational spectra of H,O on Ru(OO1) [3,4,6]. Now a superposition of this ice structure, a drawing of which is shown in fig. 2a, on a Cu(ll0) surface yields a c(2 X 2) structure with only very little lattice distortion in the ice layer. It is very likely that this structure, which is arranged on an fee (110) surface in a ball model in fig. 2b, is formed because we can exclude other c(2 x 2) models with sites of high symmetry, and because we expect hydrogenbonded clusters to occur also on this metal surface. In this structure, there are four differently oriented water molecules. This explains why all bands were present in the photoelectron spectra of fig. 1, because not only do some molecules occupy low symmetry sites, but there are also four different water molecules in this structure, and emission from these causes all peaks to appear in the spectra. The correlation of the above photoemission results with the LEED observations yields a consistent picture of the structure of adsorbed water on Cu( 1lo), and shows the possibility of obtaining information on adsorption site symmetries through angle-resolved photoemission using polarised light. Acknowledgements We gratefully acknowledge discussions with S. Holloway, D. Grider, Richardson and T. Madey. This work was supported by the Deutsche schungsgemeinschaft through Sonderforschungsbereich 6 Project A 5.
N.V. For-
C. Mariani, K. Horn / Orientation of water on Cu(ll0)
285
References [1] H. Ibach and S. Lehwald, Surface Sci. 91 (1980) 187. [2] B.A. Sexton, Surface Sci. 94 (1980) 435. [3] P.A. Thiel, F.M. Hoffmann and W.H. Weinberg, Le Vide, Les Couches Minces 201 Suppl. (1980) 307. [4] T. Madey and J.T. Yates, Chem. Phys. Letters 51 (1977) 77. [5] F.P. Netzer and T.E. Madey, Phys. Rev. Letters 47 (1981) 928. [6] K. Kretzschmar, J.K. Sass, A.M. Bradshaw and S. Holloway, Surface Sci. 115 (1982) 183. [7] C. Benndorf, C. Nobl, M. Rtisenberg and F. Thieme, Surface Sci. 111 (1981) 87. (81 A. Spitzer and H. Ltith, to be published. [9] D. Schmeisser, F.J. Himpsel, G. Hollinger, B. Reihl and K. Jacobi, to be published. [lo] M. Scheffler and A.M. Bradshaw, J. Vacuum Sci. Technol. 16 (1979) 447; E.W. Plummer and W. Eberhardt, Advan. Chem. Phys. 49 (1982) 533. [ 1 l] K. Jacobi, P. Geng and W. Ranke, J. Phys. El 1 (1978) 928. [12] C. Mariani, K. Horn and A.M. Bradshaw, Phys. Rev. B, in press. [13] D.W. Turner, A.D. Baker, C. Baker and C.R. Brundle, Molecular Photoelectron Spectroscopy (Wiley, London, 1971). (141 K. Horn, K. Jacobi and A.M. Bradshaw, J. Vacuum Sci. Technol. 15 (1978) 575. [15] K. Bange, D. Grider and J.K. Sass, Surface Sci. 126 (1983) 437.
Surface Science 126 (1983) 279-285 North-Holland Publishing Company
ORIENTATION
OF WATER ADSORBED ON Cu(ll0)
C. MARIANI * and K. HORN Fritz- Haber - fnstitur der Max - Planck - Gesellschaft,
Faradayweg 4 - 6, D - 1 OOt?Berlin 33, Fed. Rep.
of Germany Received
24 August
1982; accepted
for publication
22 September
1982
The method of determining the orientation of an adsorbed molecule by angle-resolved phot~~ssion with polarised light has been applied to the case of water adsorbed on clean copper (110). By using dipole selection rules and symmetry arguments, it is found that in the full monolayer as well as at submonolayer coverages, the water molecules are adsorbed in a tilted geometry. Only small shifts in the valence photoemission peak relative to the gas phase spectrum are found, indicative of a weak chemisorption bond. These results are explained, by recourse to LEED observations, in terms of recent models of water layers on metal surfaces.
1. Introduction Water adsorption on metal single crystals has recently been studied with a number of surface-sensitive techniques [l-9]. A central question has been the o~entation of the water molecules on the surface. Netzer and Madey [S], using the ESDIAD technique, concluded that the array of H,O molecules on Ni( 111) contains a distribution of tilt angles with respect to the surface normal. The evidence from vibrational spectroscopy indicates that, even at very low coverages, H,O forms hydrogen-bonded clusters on Ru(OOl) and Pt(lOO) [1,2,6], with the hydrogen bonding leading to a large broadening and intensity enhancement of the O-H stretching vibration. For H,O on Cu( 1lo), however, Spitzer and Ltith [8] conclude from UPS and ELS experiments that, at least up to 0.5 monolayer coverage, cluster formation is not a dominant effect, and that the water molecules are bonded to the surface in a configuration with the molecular plane normal to the surface. In view of this situation, it is interesting to investigate water adsorption on metals in angle-resolved photoemission, and to interpret the results obtained with polarised light by using dipole selection rules, a method which has been successfully applied to the study of the
* Present address: Dipartimento ione Cosentino Scaio, Italy.
di Fisica,
0039-6028/83/ooO0-0000/$03.00
Universita
degli studi della Calabria,
0 1983 North-Holland
I-87030 Castigl-
280
C. Mariani, K. Horn / Orientation of water on Cu(ll0)
electronic and geometric structures of several adsorbed molecules present study seems to be the first application of this method adsorption.
[lo]. The to water
2. Experimental A commercial angle-resolving photoelectron spectrometer (Vacuum Generators ADES 400) equipped with a triple-reflection polariser [ 1 l] was used in all experiments. Specimen preparation, work function determination, etc. have been previously described [12]. Triply distilled water, further purified by freezing/thawing cycles under vacuum, was used for the exposures.
3. Results and discussion Adsorption of water on Cu( 110) at 90 K leads to the appearance of three extra peaks in the He I photoelectron spectrum in normal electron emission, as shown in fig. 1. These peaks at 6.8, 8.8, and 12.6 eV below E, are assigned to the 1b,, 3a ,, and lb, valence orbitals of H,O by comparison with the gas phase spectrum, which is also shown (ref. [ 131). Using the work function of clean Cu( 1 IO) (4.5 eV) it is found that all peaks are shifted to higher binding energy by about 1.4 eV compared with the gas phase. We ascribe this shift to the effect of extramolecular relaxation. Any differential shift which might be indicative of a bonding shift on one of the orbitals is certainly smaller than 0.2 eV, although it must be stressed that, particularly for the 3a, and 1b, peaks, a determination of exact peak position is rather difficult due to the large half width and overlap. Upon heating the water-exposed surface, it is found that all water-induced peaks disappear above a temperature of 170 K; no evidence for dissociation was found, the photoelectron spectrum after desorption being identical to that of the clean surface recorded before exposure, in disagreement with results of Spitzer and Ltith [8] who observe the formation of a hydroxyl species. The origin of this disagreement is difficult to determine at present; it is known that small traces of co-adsorbed oxygen can induce water dissociation and hydroxyl formation. The low desorption temperature and the lack of strong bonding shifts observed in the photoelectron spectra both show that the chemisorption bond is weak; from the desorption temperature the heat of adsorption is estimated to be 50-60 kJ mol- ‘. Water causes a large change in work function, which was monitored by recording the low energy cutoff in the spectra with the crystal biased at - 10 V. It is found that the work function decreases by about 0.95 f 0.03 eV after an exposure of 0.8 L. At higher exposures multilayer formation occurs, which results in the photoemission peaks appearing at higher binding energy.
C. Mariani, K. Horn / Orientation of water on Cu(ll0)
281
‘bl
3a1
lb2 1,1!,
I
21
19 Energy
I
17 below
I
I
13
15 E,ac
b)
surface
l
c 1 s - polarmd
hght
I s- polarised
hght
d
ii parallel 06L H20
18
I 16
1 14
I 12
I 10
8I
I 6
Energy
Fig. 1. Photoelectron surface. unpolarised
hI below
2I EF
I
EC (eV) ’
spectra of (a) water in the gas phase. from ref.
[ 13):
light; (c) water absorbed on Cu( I IO) at a temperature
recorded with s-polarised He
I
to (001)
(b) spectrum of clean
of 93 K. exposure 0.6 L.
light. with the electric vector along the (I IO) azimuth: (d) same hut
electric vector oriented along the (001) azimuth. Spectra (b)-(d) emission.
were recorded in normal electron
282
C. Mariani, K. Horn / Orientation of water on Cu(Il0)
The value for A&,,,, is in good agreement with that measured by other groups [8,15]. The work function shows a further decrease of about 200 mV beyond the monolayer coverage, the completion of which was judged from the onset of a shift in photoemission peak positions. We base our determination of the site symmetry of the adsorbed water molecules on the spectra of fig. 1, recorded with polarised He I radiation, with the electrons detected in a narrow cone around the surface normal. It has previously been shown [lo] that, in such a geometry, only emission from those orbitals which form a basis for the same irreducible representations as those of the Cartesian components of the momentum operator is allowed. Emission from an AZ-type orbital in C Zv symmetry [14] is always forbidden in normal emission, for example. Furthermore, when s-polarised light is used, emission from orbitals that belong to the totally symmetric irreducible representation is forbidden, being allowed only when the electric field at the surface has a normal component [lo]. This polarisation selection rule enables one, starting from a correlation of peaks in the spectrum of the adsorbate with that of the gas phase, to determine the local symmetry of the adsorbed species, or rather to place certain restrictions on the choice of adsorbate point groups. The present case of Hz0 on Cu(ll0) serves to explain this procedure. We have already assigned the peaks in the spectra of fig. 1 by comparison with the gas phase spectrum. Since the highest site symmetry on a fee metal (110) surface is CZv, it is appropriate to start our analysis with the assumption that the water molecule occupies such a site in which its gas phase symmetry is not lowered, for example an “on top” site either in the trough along the [liO] azimuth or perpendicular to it. Bonding is most likely to occur through the oxygen lone pair (the lb,), which is also closest in energy to the metal d levels. For a single adsorbed water molecule, a likely configuration is one in which the water is adsorbed with the oxygen end nearest to the metal, the hydrogen atoms The spectra shown in pointing away from the surface in a C,, configuration. fig. 1 prove that this configuration is not taken up by the water molecules under our experimental conditions, i.e. at 90 K and coverages at around 0.6 maximum coverage. Here, we observe appreciable emission from the 3a, orbital in normal emission and with s-polarised light irrespective of the orientation of the electric vector with respect to the substrate azimuths; this peak would be forbidden if the H,O molecules were adsorbed in a C,, configuration. The presence of several domains of differently oriented molecules, or of two kinds of differently oriented molecules in a larger unit cell, does not affect this conclusion as long as local C,, symmetry is maintained. We thus conclude that, already at coverages below a full monolayer, the molecules do not occupy C,, sites, i.e. they might be tilted such that the C, axis is no longer a valid symmmetry element. This is not the only information on the geometry that one can derive from the spectra of fig. 1, however. A number of different “tilted” geometries may
C. Mariani,
K. Horn / Orientation
of water on Cu(lI0)
283
be visualised, for example one in which the molecular plane is still normal to the surface, but one hydrogen atom closer to the metal than the other. By examining this geometry, we find that any kind of configuration in which the molecular plane coincides with a mirror plane of the crystal, i.e. where C, site symmetry is maintained, would still cause some peaks to be forbidden in
a
b Fig. 2. (a) Model of a single sheet of hexagonal ice, taken from ref. [6) (with permission); (b) ball model of the ice layer in geometry (a) on an fee (110) surface; white spheres indicate top layer copper atoms, small black spheres indicate oxygen atoms, small grey spheres indicate hydrogen atoms.
284
C. Mariani, K. Horn / Orientation of water on Cu(Il0)
spectra recorded with s-polarised light. If the molecular plane were coincident with the (170) mirror plane of the substrate, for example, emission from the lB, orbital would not be allowed in s-polarised light with the electric vector parallel to the [ Ii01 direction, while emission from the lB, and 3A, orbitals would be forbidden if the electric vector were oriented parallel to the [OOl] direction. (The case for a water molecule oriented with its molecular plane parallel to the (001) mirror plane can be treated similarly.) This is, however, contrary to the experimental observations. Only for a geometry where all molecules occupy sites such that no symmetry element is maintained, for example if the molecular plane does not contain the surface normal, or if the adsorbed water molecules are arranged in several different geometries, would all peaks appear in the spectra, irrespective of the orientation of the electric vector. These findings can be correlated with LEED results by Spitzer and Ltith [8] and Grider et al. [ 151 who observe, already at coverages below a complete monolayer, the formation of a c(2 x 2) superstructure. Since the driving force for a tilting of the water molecules is the interaction of neighbouring admolecules through the hydrogen bond, it was proposed that adsorption of water on metals proceeds through island growth [ 11, and that the geometry of the water molecules in these islands closely resembles that in a single sheet of hexagonal ice [6]: this interpretation was found to be consistent with LEED and ESDIAD results as well as vibrational spectra of H,O on Ru(OO1) [3,4,6]. Now a superposition of this ice structure, a drawing of which is shown in fig. 2a, on a Cu(ll0) surface yields a c(2 X 2) structure with only very little lattice distortion in the ice layer. It is very likely that this structure, which is arranged on an fee (110) surface in a ball model in fig. 2b, is formed because we can exclude other c(2 x 2) models with sites of high symmetry, and because we expect hydrogenbonded clusters to occur also on this metal surface. In this structure, there are four differently oriented water molecules. This explains why all bands were present in the photoelectron spectra of fig. 1, because not only do some molecules occupy low symmetry sites, but there are also four different water molecules in this structure, and emission from these causes all peaks to appear in the spectra. The correlation of the above photoemission results with the LEED observations yields a consistent picture of the structure of adsorbed water on Cu( 1lo), and shows the possibility of obtaining information on adsorption site symmetries through angle-resolved photoemission using polarised light. Acknowledgements We gratefully acknowledge discussions with S. Holloway, D. Grider, Richardson and T. Madey. This work was supported by the Deutsche schungsgemeinschaft through Sonderforschungsbereich 6 Project A 5.
N.V. For-
C. Mariani, K. Horn / Orientation of water on Cu(ll0)
285
References [1] H. Ibach and S. Lehwald, Surface Sci. 91 (1980) 187. [2] B.A. Sexton, Surface Sci. 94 (1980) 435. [3] P.A. Thiel, F.M. Hoffmann and W.H. Weinberg, Le Vide, Les Couches Minces 201 Suppl. (1980) 307. [4] T. Madey and J.T. Yates, Chem. Phys. Letters 51 (1977) 77. [5] F.P. Netzer and T.E. Madey, Phys. Rev. Letters 47 (1981) 928. [6] K. Kretzschmar, J.K. Sass, A.M. Bradshaw and S. Holloway, Surface Sci. 115 (1982) 183. [7] C. Benndorf, C. Nobl, M. Rtisenberg and F. Thieme, Surface Sci. 111 (1981) 87. (81 A. Spitzer and H. Ltith, to be published. [9] D. Schmeisser, F.J. Himpsel, G. Hollinger, B. Reihl and K. Jacobi, to be published. [lo] M. Scheffler and A.M. Bradshaw, J. Vacuum Sci. Technol. 16 (1979) 447; E.W. Plummer and W. Eberhardt, Advan. Chem. Phys. 49 (1982) 533. [ 1 l] K. Jacobi, P. Geng and W. Ranke, J. Phys. El 1 (1978) 928. [12] C. Mariani, K. Horn and A.M. Bradshaw, Phys. Rev. B, in press. [13] D.W. Turner, A.D. Baker, C. Baker and C.R. Brundle, Molecular Photoelectron Spectroscopy (Wiley, London, 1971). (141 K. Horn, K. Jacobi and A.M. Bradshaw, J. Vacuum Sci. Technol. 15 (1978) 575. [15] K. Bange, D. Grider and J.K. Sass, Surface Sci. 126 (1983) 437.