Angle-resolved photoemission of xenon adsorbed on Pt(111) : commensurate and incommensurate monolayers

HAL Id: jpa-00210856

https://hal.archives-ouvertes.fr/jpa-00210856

Submitted on 1 Jan 1988

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Angle-resolved photoemission of xenon adsorbed on Pt(111) : commensurate and incommensurate

monolayers

A. Cassuto, J.J. Ehrhardt

To cite this version:

A. Cassuto, J.J. Ehrhardt. Angle-resolved photoemission of xenon adsorbed on Pt(111) : com- mensurate and incommensurate monolayers. Journal de Physique, 1988, 49 (10), pp.1753-1759.

�10.1051/jphys:0198800490100175300�. �jpa-00210856�

Angle-resolved photoemission ^{of xenon adsorbed on Pt(111) :}

commensurate and incommensurate monolayers

A. Cassuto and J. J. Ehrhardt

CNRS, Laboratoire Maurice Letort associé à l’Université de Nancy I, ^BP 104, 54600, Villers les Nancy, ^France (Requ le 12 avril 1988, révisé le 9 juin 1988, accepté ^{le 13} juin 1988)

Résumé.

Nous avons tracé les courbes de dispersion ^{de la structure} commensurable (~3 ^x ~3) ^{R30° et}

hexagonale compacte incommensurable tournée du xénon adsorbé sur Pt(111). Grâce à la comparaison ^avec

d’autres résultats de la littérature obtenus ^sur cuivre, palladium, aluminium, argent, platine ^et graphite, ^nous

avons pu ^montrer que le dédoublement des niveaux 5p3/2 ^était principalement ^{dû au} recouvrement des orbitales 5p au-dessous d’une distance minimale. 11 augmente faiblement lorsque la distance xénon-xénon décroît et ne dépend pas de la ^nature du substrat.

Abstract.

The dispersion ^{curves of the} commensurate (~3 ^x ~3) ^{R30° and the} hexagonal compact incommensurate rotated xenon overlayers ^on Pt(111) ^are given. Comparison with other angle-resolved photoemission ^{results on} copper, palladium, aluminum, silver, platinum ^and graphite show that the splitting ^of

the 5p3/2 ^{levels is} mainly ^{due to the} overlap ^{of the} 5p orbitals from neighbouring ^{Xe atoms below a minimum}

distance. It increases only slightly ^with decreasing ^distances independently ^on the substrate ^nature.

Classification

Physics ^Abstracts

71.25T

73.90

I. Introduction.

Physisorbed layers ^{of rare} gases and particularly

xenon layers ^{form a} hexagonal (or nearly hexagonal)

compact ^{structure on most} single crystals [1-17]. ^In

some cases, e.g. ^on Cu(110) [10, 11], Cu(lll) [3, 12], Pd(lll) [15] ^and Pt(lll) [17], ^less dense solid

layers ^can also be formed at lower coverages. The

Xe/Pt(lll) system ^{has been} extensively studied in recent years ^{on an} almost defect free sample by ^Kern

et al. [18-21] using high resolution helium beam

scattering. It appears ^to be one of the most interest-

ing ^of quasi-2D systems ^{obtained sofar} by physisorp-

tion of rare gases ^on single crystal metal surfaces.

Indeed, depending ^on the coverage and temperature,

a variety ^of phases may be formed : ^a commensurate one (J3 ^x B/3) ^R300 (C) stable above 63 K ; ^a striped incommensurate one (SI) ; ^a hexagonal ^in-

commensurate one (HI) and, ^a hexagonal ^incom-

mensurate rotated phase (HIR).

The band structure of xenon layers ^{has been} scarcely studied. The 5p levels of the solid phases ^are

the 5pl/2 ^{and two} 5p3/2 ^levels (mj = ^{± 1/2,}

mj = ^± 3/2). ^{It has} been demonstrated that the 5p3/2 splitting ^is mainly ^{due to lateral} interactions and direct overlap ^{of the} 5p ^wave functions from

neighbouring ^{xenon atoms when the} adlayer ^is sufficiently close-packed ^{in an} ordered array [9, 22, 23]. The energy order of the 5p3/2 ^levels (identified by ^their angular dependence [9] ^and using polarized

electrons [24, 25]) is such that the lateral interaction model and the crystal field model [26, 27] ^can only explain the results. However, ^as pointed ^out by

Antoniewicz [28], the latter requires unreasonably large positive charges ^{on surface atoms to} produce

the observed splitting. Moreover, ^{there is no} splitting

for a 2D gas ^on W(100) [29], Ag(lll), Cu(lll), Ru(001) [30, 31] ^{and at} large ^{distances between}

xenon atoms in the repulsive Xe/Pd(100) ^system [22]. ^The dispersion ^{curves of the xenon incommen-}

surate hcp layer have been determined on Pd(100) [9], ^on Al(lll) [16], ^{on the} c(2 ^x 2) phase ^{and on}

the almost hexagonal phase ^on Cu(110) [11] ^{and on}

the (J3 ^x J3) R300 phase ^on Pt(lll) [32]. ^The general trend of all these results, compared ^{in the}

last reference, ^{is in} agreement ^with theoretical calculations [33, 34]. One observes an increase in the

dispersion for all bands and in the 5p3/2 splitting

with decreasing interatomic distances in the xenon

layer. At normal emission (T point), ^the 5pl/2 binding energy referenced ^to the vacuum level is almost independent ^on the substrate nature (for Cu,

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0198800490100175300

1754

Pd, ^{Al and} Pt) ^and equals ^{11.85 eV,} supporting ^a major effect of the initial state in the photoemission

process ^as demonstrated by Wandelt for several Xe/metal surfaces [35]. ^{As a} consequence, for vari-

ous metal surfaces, photoemission of adsorbed xenon

(PAX) ^{in one} particular direction of detection and

preferably ^near the normal [32], ^{has been used as a}

« local work function probe » ^{in order to} sample

structural defects [36-38], ^adsorbed species [39-41],

and alloys [42-43]. ^{However, it must} be mentioned that on Pb, ^{Ga and Ba} overlayers [44], ^distance- dependent ^{shifts are not due to} work-function

changes ^induced by the Xe adatom itself but to

changes in the final state screening [45].

In this paper, ^a comparison will be made between the dispersion ^{curves of the} commensurate (C) ^and hexagonal incommensurate rotated (HIR) layer (new results) ^{in the context of the} previous angle-

resolved photoemission ^{data on} other substrates.

We will demonstrate that the splitting ^{of the} 5p3/2

levels is moderately ^{sensitive to} the interatomic distance and relatively insensitive to the nature of the substrate.

2. Experimental.

Most of the experiments ^{were made in an} apparatus

(Atomic Energy Commission

Service de Physique Atomique ^{des Surfaces,} Saclay, France) consisting

of a home made ultrahigh-vacuum moveable electron spectrometer (hemispherical, ^radius

³⁰ mm)

mounted in ^a two-level chamber equipped ^{with a}

rare-gas discharge lamp (Leybold), ^a low-energy

diffraction (LEED) optics (Riber), ^an ion gun

(Riber) and electron bombardment for crystal prep- aration. Pumping is ensured through ^a cryopump

(Air Liquide), ^an ion pump and/or ^a trapped ^oil

diffusion pump (Edwards). Base pressures below

10- g Pa are obtained after bakeout. Additional results at normal emission were obtained at LURE

(Laboratoire d’Utilisation du Rayonnement ^Electro- magn6tique, Orsay, France) using synchrotron ^radi-

ation light between 19.5 and 30 eV and in Nancy

with our ESCA equipment (Leybold-Heraeus) ^{and a}

rare-gas discharge lamp.

The Pt crystal ^{was cut from a} single crystal ^{rod of}

5N purity (Materials Research) after orientation to within 0.5° and mechanically polished. ^{The Pt} sample

was preoriented ^{under X-ray back} reflection before

mounting ^{with the} light ^incidence plane along ^the

rM orientation of the Pt(lll) ^{crystal. It is} clamped

onto a tantalum reservoir containing circulating liquid ^nitrogen. The minimum temperature ^obtain- able on the sample ^{was estimated to 80-85 K} (see later). After ion bombardment and annealing

around 1 000 K, Auger ^electron spectroscopy (AES)

revealed no impurity (C, Si, Ca, Al) ^at the detection limit [46, 47] ^{and a} sharp diffraction pattern ^{of the}

clean (1 ^x 1) ^{surface was} observed. The LEED pattern ^{showed a} slight misalignment (3-4°) ^with

respect ^{to the FM} orientation.

Photoelectron spectra ^were taken with unpolarized

He I radiation at an incidence angle equal ^{to 45° with}

15 meV steps for the electron kinetic energy. Absol- ute work function was determined ^as well ^as the instrumental resolution (150 meV) ^{at the Fermi} edge. ^{At LURE,} the resolution was only ^{220 meV}

and in Nancy around 300 meV.

3. Results.

3.1 LEED. - At the lowest temperature ^{of the} sample, adsorption ^{of xenon at} 10- 6 Pa pressure

(gauge reading uncorrected) produces ^the

(J3 ^x B/3) ^{R300 pattern} (C phase). ^{This structure is} easily observed and stable under vacuum. When the pressure is increased up ^to 10- 4 _Pa, an intermediate incommensurate hcp phase (HI) (which like the C

phase is rotated with respect ^to the substrate by 30°)

appears transiently ^{and converts} readily ^{into an}

incommensurate rotated phase (HIR) consisting ^of

domains rotated from the common orientation of the C and HI phase by ^{± 3°. Under vacuum, the reverse}

sequence of patterns is observed and the C phase ^is

recovered. From the heats of adsorption ^{of the}

various phases [17], the minimum temperature ^of the sample ^can be estimated to be 80-85 K. At this temperature, ^the phase diagram given by ^{Kern et al.}

is not clear [18, 21]. However, ^{at 88} K, ^{it was shown} in a preceeding paper [17] that the HIR phase

existed at saturation of the xenon overlayer.

3.2 ARUPS. - We will consider here the C and the HIR overlayers. Dispersion curves were made along

the rKMK direction of the C phase. ^{The LEED} pattern of the C layer ^and the Surface Brillouin Zone (SBZ) ^are schematically ^{shown in} figure ^1.

The Xe-Xe distances were estimated to be respect- ively around 4.8 A and 4.4 A for the C and HIR

phase. Along ^{the rKMK} direction, the kll ^{values for}

the C phase ^{are 0.878} A-1 ¹ at the K point ^and

1.317 A-1 at the M point. ^{Due to the} slight ^misorien-

Fig. ^1.

^Low electron energy diffraction pattern ^{of the}

(J3 ^x B/3) ^{R30° xenon} overlayer. Surface Brillouin zone

(SBZ) ^{of xenon.}

tation of the crystal, ^{we note also} kp ^{at the H} point,

viz 0.814 A-1. The corresponding ^{values for the}

HIR phase ^{are 0.9558} A-1 ¹ and 1.4378 Å - 1. Two domains in the HIR phase coexist in such a way that

one domain is almost strictly in the TKMK direction while the other one is misoriented by ^{about 6°-7°. As}

a consequence, dispersion ^curves of the HIR layer

are expected ^{to be less accurate than} for the C

phase.

The direction of the normal was determined

precisely using various methods [32], including ^the

symmetry ^{of the} dispersion ^curve itself. Photoelec- tron spectra ^were taken every 4 degrees ^{in order to}

obtain roughly ^0.1 A-1 variation in kll , according ^to

where E is the kinetic energy (eV) of the electrons

leaving ^the crystal, ^{and 0 the} angle ^of analysis ^with respect ^to the surface normal. Knowing ^{the work}

function 4>, ^{E is} given by

The experimental ^work function of the clean metal ^was found to be 5.85 eV. A 0.60 eV decrease

was obtained with the C overlayer ^and slightly ^more (0.65 eV) with the HIR layer. ^Values of kll ^were computed taking ^{into account the account the work}

function of the covered surface.

Figure ² gives ^the photoelectron spectra ^{of the C} and HIR phases ^{at the T} point. Several features should be noted.

(a) The three 5p ^levels of adsorbed xenon are

clearly ^{seen. Their} positions ^{are almost the same for}

the two phases. ^{A similar} ksylt ^{was obtained at}

LURE in the 19.5-20 eV photon energy range and in

Nancy using ^{He I radiation} (21.21 eV).

Fig. ^2.

Photoemission spectra ^at normal emission of the C and HIR overlayers ^{of xenon on} Pt(lll) starting ^from

the Fermi edge (h v

^21.21 eV).

(b) ^{On the} high binding energy side of the 5p1/2

and 5p3/2 (mj = ^{± 1/2) levels extra} transitions exist.

They ^{do not} disperse ^{and were} attributed to indirect transitions due to defects in the crystal [32].

(c) After removal of the base line the spectra

were decomposed using ^an optimization routine into five peaks (Gaussian-Lorentzian) [48] to account for the extra transitions. Allowing ^{for an} experimental

resolution of 150 meV deduced from the Fermi cut-

off, the natural width of the lines is around 200 meV at the r point ^{for the C} layer. ^The 5p3/2, _{mj =} ^{± 3/2}

level is broader (600 meV) because of the remaining

influence of a Pt 5d induced transition. This influence

disappears upon analysis away from the surface normal because of a different dispersion ^behaviour.

An example ^of decomposition ^of spectra ^{of the C} and HIR layers ^{at the r} point ^is given ⁱⁿ figure ^3.

Fig. ^3.

Deconvolution into sublevels at normal emis- sion : (a) ^the commensurate layer (C) ; (b) ^the hexagonal

incommensurate rotated layer (HIR). Binding energies

are referenced to the Fermi level.

The same decomposition procedure ^{was followed}

to determine the dispersion ^{curves. Note that off}

normal direction, ^the 5p levels of the HIR phase

were broader than for the C phase, ^a fact which is not surprising because of the coexistence, ^{of two}

rotated domains. As expected, ^the dispersion ^curves

show more scatter for the HIR phase than for the C

phase ; they ^are given ⁱⁿ figure ^4.

4. Discussion.

Qualitative agreement exists between the dispersion

curves of figure 4 and other experimental ^{results on}

different substrates and layers [9, 11, 16, 32].

Despite different accuracies for the two layers ^exami-

ned here, ^{we note the} following points :

(a) ^{at the r} point, ^the binding energy of all levels

are only slightly different for the C and HIR layers ⁱⁿ

the three apparatus ^{that we} have used (see ^also Fig. 2). This result is at variance with those obtained

by Schonhense [24, 25] ^{for the} 5p3/2 ^levels using ^a

1756

Fig. ^4.

Dispersion ^{curves of the} 5p ^{xenon levels} along

rKMK direction ; binding energies ^are referenced to the Fermi level ; ^incidence angle

^{45° :} (o, 0) ^C layer ; (A, A) ^HIR layer.

photon energy of about 11 eV. This author found ^a shift of 0.5 eV towards lower binding energies ^be-

tween the C and the HIR layer. ^{Kern et al.} [21]

attributed such a large ^{shift to a} change ^{in the}

distance between the layer ^{and the substrate of}

about 0.3 A. They ^{used the} theoretical work of Lang [49] ^to support ^this hypothesis. However, Lang

derived a shift of 0.2 eV only for the 4d levels of

xenon adsorbed on Al(lll) ^{for a one} Bohr radius

(0.52 A) variation in vertical distance. The rs par- ameter equal ^{to 2} chosen in this case (which express-

es the density ^{of the} positive background) ^{is reason-}

able for Pt [50]. ^{Moreover, the} experimental ^results

of Mariani et al. [11] ^{did not} show any difference in

binding energy of the ^xenon 5p levels between the

c(2 ^x 2) layer ^{and the} uniaxially compressed phase

where such a change in vertical distances is expected.

We do not see any clear explanation ^{for the} discrep-

ancy between the results of Schonhense et al. and

ours apart ^{from the} possibility ^of strong ^{final state} effects due to the low photon energy used in their case ;

(b) ^{a maximum} binding energy is observed for all bands near point ^K and the band structure is nearly

flat for the mjl ^{± 1/2} levels between points ^{K and} M ;

(c) ^the 5p3/2, ml/ ± 3/2 level passes through ^a

maximum in binding energy ^near point ^{K and} through ^{a minimum near} point M ;

(d) apart from small variations in the width of the bands, ^{of the} spin-orbit splitting ^{and the} 5p3/2 levels,

the qualitative behaviour is the same for the C layer

and the HIR layer.

In order to assess more clearly the influence of the Xe-Xe lateral interactions on dispersion ^and split- ting, ^we have recalled in table I all the photoemission

results that we know along ^{with the} xenon-xenon

distance in the layer. Some of the results have been determined after enlargement ^{of the} original figures

and may be relatively inaccurate. For Cu(110), ^the

two structures are not strictly hexagonal ^{but unia-} xially compressed (along ^the {110} direction). ^Con-

trary ^to table II in reference [32] ^and following ^a suggestion ^{of Mandel} [51]

^effective

^{Xe-Xe dis-}

tances calculated from the values of the areas for adatom under the assumption ^{of a} hexagonal ^close- packed ^{Xe structure} have been used. Some _very

recent results for xenon monolayers ^and bilayers ^on graphite (001) [52] ^{are also} included. The corre-

sponding ^values obtained from differences in threshold positions in the work of Schonhense [24]

are such that the separation ^{of the} 5p3/2 ^{bands at}

T point ^{is smaller} for the first layer than for the

bilayer. ^On graphite, depending ^on temperature

(below ^{and above 63} K) ^{the first} layer ^{is commensur-}

ate or incommensurate with the substrate [53]. ^In

the second case, the xenon-xenon distance varies

continuously between 4.51 A and 4.44 A. It explains

the question ^mark.

Table I deserves more comments :

(a) ^the general ^{trend is in} agreement with theoreti- cal calculations [33, 34]. ^With decreasing ^Xe-Xe

distances one observes an increase in the dispersion

for all bands and in the 5p3/2 splitting. The variation of the spin-orbit splitting is less obvious. For com-

parison, the theoretical values of Hermann et al. [33]

for an hcp layer (4.48 A) ^{are 0.38,} 0.48 and 0.6 eV for the maximum dispersion of the three bands, ^0.43

and 1.32 eV for the 5p3/2 splitting ^{and the} spin-orbit splitting. If the interatomic spacings in the adsorbed Xe monolayer ^are taken into account [51], ^{a rela-} tively good agreement is found for the dispersion ^of

the Sp112 level ;

(b) a parameter, quite ^{sensitive to the distance, is}

the 5p3/2 splitting ^{at the r} point. ^{In the case of the} repulsive Xe/Pd(100) ^{system, Bradshaw et al.} [22]

have observed a splitting of these levels at coverages

as low as 0.30. This corresponds ^{to a} Xe-Xe distance of around 5.4 A. This system presents ^the advantage

of a continuous variation of the Xe-Xe distance.

Unfortunately, the variation of this splitting ^versus

coverage ^{was not} recorded in this last work. Results show nevertheless that the 5p ^{orbitals start to} overlap ^{when the Xe-Xe distance is close to 5.4} A ; (c) between the Xe (B/3 x /3-) R30’/Pt(lll) layer ^{and the} hcp layer ^this splitting ^varies only

between 0.4 and 0.65 eV. For the hcp Xe/Cu(lll) layer ^{and the} hcp Xe/Ru(001) layer ^{Eder et al.} [30, 31] have found splittings respectively equal ^{to 0.3}

and 0.2 eV, values much lower than for the hcp

Table I.

Dispersion ^results.

[TW1] : This work - Apparatus ⁱⁿ Saclay.

[TW2] : This work - Leybold apparatus ⁱⁿ Nancy.

(*) : ^{at M or K} point.

(**) : bilayer.

Fig. ^5.

Separation ^{of the} 5p3/2 ^{levels at the r} point

versus the xenon-xenon distance.

Xe/Ag(lll) layer (0.5 eV). Accordingly they pro-

posed ^{that the} splitting ^was « mainly ^caused by

lateral interactions and 2D band structure formation and that the size of the splitting ^{could be} regarded ^as

a measure of the lateral Xe-Xe attraction ^». They

added that

this is compatible with the sequence of

dipole ^moments

^However, this attractive expla-

nation is contradicted by the data of table I.

It must be noticed that in Eder’s experiments [54]

the photoemission spectra ^{are recorded at an} angle

of about 45° and not in the normal direction.

Therefore the emission point in the SBZ depends

therefore on the azimuthal orientation of the sub- strate. Indeed, ^{as it can be seen in} figure ^{6 of}

reference [11], ^for Cu(110) ^the splitting ^{of the} 5p3/2

levels depends ^on the azimuthal angle ^{for a} given

detection angle. For instance at 45°, it is smaller in the rKMK direction than in the TMT direction

(0.25 ^eV compared ^{to 0.5} eV). ^A similar result is found for Xe/graphite (001) [52]. ^The splitting ^{of the} 5p3/2 ^{levels at 45°} is 0.34 in the FKMK direction and 0.65 in the FMF direction. The Ag ^{and Cu films}

used by ^{Eder et al. were} epitaxially grown ^on

Ru(001) [54]. ^{On bulk} Cu(lll) [12] ^and Ru(001) [55] ^{the xenon saturated} layer is rotated with respect

to the substrate ; ^{it is} aligned ^to the substrate on

Ag(lll) [5,6]. ^{The 30°} difference in angle ^{of the} overlayer corresponds exactly ^{to a} change ^{from the}

FKMK direction (Cu(lll) ^and Ru(001)) ^{to the}

FMF direction (Ag(lll)). ^{As a} consequence, the results of Eder et al., ^when interpreted self-consist-

ently, ^{do not} contadict those presented in table I and do not require ^a special interpretation.

Therefore it must be concluded that despite ^differ-

1758

ent adsorption energies ^and dipole ^{moments for}

xenon adsorbed on Cu, Al, Pd, Pt, Ag ^and graphite

there is no significant change ^{of the xenon} 5p3/2 splitting ^{on different} substrates, apart ^{from small} variations due to the different xenon-xenon dis- tances.

5. Conclusion.

Like on other metals xenon adsorbed on Pt(lll) ⁱⁿ

2D dense phases shows marked dispersion ^{effects on}

its 5p levels. Due to lateral interactions and to the

overlap ^of orbitals, ^the 5p3/2 ^{levels are} split. ^This splitting appears ^to be mainly characteristic of ^the

xenon-xenon distance. It varies between approxi- mately 0.45 eV and 0.65 eV for distances between, roughly, 4.8 and 4.3 A.

Acknowledgements.

The authors would like to thank J. Cousty ^{and R.}

Riwan (SPAS-Saclay) for fruitful discussions and

appreciate ^the opportunity ^{to have} used their angle-

resolved equipment.

References

[1] PALMBERG, ^P. W., Surf. ^{Sci. 25} (1971) ^598.

[2] CHESTERS, ^{M. A. and} PRITCHARD, J., Surf. ^{Sci. 28} (1971) ^460.

[3] CHESTERS, ^M. A., HUSSAIN, ^{M. and} PRITCHARD, J., Surf. ^{Sci. 87} (1973) ^415.

[4] IGNATIEV, A., JONES, ^{A. V. and} RHODIN, ^T. N., Surf. ^{Sci. 30} (1972) ^573.

[5] ^{COHEN, P.} I., UNGURIS, J. and WEBB, ^M. B., Surf.

Sci. 58 (1976) ^429.

[6] UNGURIS, J., BRUCH, ^L. W., MOOG, ^{E. R. and} WEBB, ^M. B., Surf. ^{Sci. 87} (1979) ^415.

[7] ^TAPPING, R. L. and NORTON, ^P. R., Chem. Phys.

Lett. 41 (1976) ^252.

[8] KÜPPERS, J., NITSCHKE, F., WANDELT, ^{K. and} ERTL, G., Surf. ^{Sci. 87} (1979) ^295.

[9] SCHEFFLER, M., HORN, K., BRADSHAW, ^{A. M. and} KAMBE, K., Surf. ^{Sci. 80} (1979) ^69.

[10] GLACHANT, A., JAUBERT, M., BIENFAIT, ^{M. and} BOATO, G., Surf. ^{Sci. 115} (1981) ^219.

[11] ^MARIANI, C., HORN, ^{K. and} BRADSHAW, ^A. M., Phys. ^{Rev. B 25} (1982) ^7798.

[12] LARSSON, ^C. G., PAUL, ^{J. and} WALLDEN, L., Phys.

Scr. 4 (1983) ^44.

[13] GIBSON, ^{K. D. and} SIBENER, S. J., Faraday ^Disc.

Chem. Soc. 80 (1985) ^203.

[14] CHRISTMAN, ^{K. and} DEMUTH, J. E., Surf. ^{Sci. 120} (1982) ^291.

[15] WANDELT, ^{K. and} HULSE, J., ^{J. Chem.} Phys. ⁸⁰ (1984) ^1340.

[16] ^MANDEL, T., KAINDL, G., DOMKE, M., FISHER, ^W.

and SCHNEIDER, ^N. D., Phys. Rev. Lett. 55

(1985) ^1638.

[17] ^POELSMA, B., VERHEIJ, ^L. K. and COMSA, G., Surf.

Sci. 152/153 (1985) ^851.

[18] KERN, K., DAVID, R., PALMER, R. L. and COMSA, G., Appl. Phys. ^{A 41} (1986) ^91.

[19] KERN, K., Phys. ^{Rev. B 35} (1987) ^8265.

[20] KERN, K., DAVID, R., ZEPPENFELD, P., PALMER, ^R.

L. and COMSA, G., Solid State Commun. 62

(1987) ^391.

[21] KERN, K., DAVID, R., ZEPPENFELD, ^P. and COMSA, G., Surf. ^{Sci. 195} (1988) ^353.

[22] HORN, K., SCHEFFER, ^{M. and} BRADSHAW, ^A. M., Phys. ^{Rev. Lett. 41} (1978) ^822.

[23] ^{BRADSHAW, A. M. and} SCHEFFER, ^{J. Vac. Sci.}

Technol. 16 (1979) ^447.

[24] SCHONHENSE, G., EYERS, A., FRIESS, U., SCHAFERS, F. and HEINZMANN, U., Phys. ^{Rev. Lett. 54} (1985) ^547.

[25] SCHÖNHENSE, G., Appl. Phys. ^{A 41} (1986) ^39.

[26] WACLAWSKI, ^{B. J. and} HERBST, ^J. F., Phys. ^Rev.

Lett. 35 (1975) ^1594.

[27] ^{HERBST, J.} F., Phys. ^{Rev. B 15} (1977) ^3720.

[28] ANTONIEWICZ, ^P. R., Phys. Rev. Lett. 88 (1977) ^374.

[29] ^{ERSKINE, J.} L., Phys. ^{Rev. B 24} (1981) ^2236.

[30] EDER, S., MARKERT, K., JABLONSKI, ^{A. and WAN-}

DELT, K., ^Ber. Bunsenges. Phys. ^{Chem. 90} (1986) ^225.

[31] JABLONSKI, A., EDER, S., MARKERT, ^{K. and WAN-}

DELT, K., J. Vac. Sci. Technol. A 4 (1986) ^1510.

[32] CASSUTO, A., EHRHARDT, ^J. J., COUSTY, ^{J. and} RIWAN, R., Surf. ^{Sci. 194} (1988) ^579.

[33] HERMANN, K., NOFFKE, ^{J. and} HORN, K., Phys.

Rev. B 22 (1980) ^1022.

[34] KAMBE, K., Surf. ^{Sci. 123} (1982) ^95.

[35] WANDELT, K., J. Vac. Sci. Technol. A 2 (1984) ^82.

[36] WANDELT, K., HULSE, ^{J. and} KUPPERS, J., Surf. ^Sci.

104 (1981) ^212.

[37] DAISER, ^{S. and} WANDELT, K., Surf. ^{Sci. 128} (1983)

61.

[38] MIRANDA, R., DAISER, S., WANDELT, ^{K. and} ERTL, G., Surf. ^{Sci. 131} (1983) ^61.

[39] JABLONSKI, A., EDER, S. and WANDELT, K., Appl.

Surf. ^{Sci. 22} (1985) ^309.

[40] ALNOT, M., GORODETSKII, V., CASSUTO, ^{A. and} EHRHARDT, J. J., Surf. ^{Sci. 162} (1985) ^886.

[41] DOLLE, P., ALNOT, M., EHRHARDT, J. J., THOMY,

A. and CASSUTO, A., Surf. Sci. 152/153 (1985)

886.

[42] MARKERT, K., DOLLE, P., NIEMANTSVERDRIET, ^J.

W. and WANDELT, K., J. Vac. Sci. Technol. A 5

(1987) ^2849.

[43] ALNOT, M., GORODETSKII, V., CASSUTO, ^{A. and} EHRHARDT, ^J. J., Thin Solid films ¹⁵¹ (1987)

251.

[44] ^{JACOBI, K. and} ASTALDI, C., Activity Report ^{of the}

Fritz-Haber Institut-Berlin (1987) ; Surf. ^Sci.

192 (1987) ^499.

[45] ^{CHIANG, T.} C., KAINDL, ^{G. and} MANDEL, T., Phys.

Rev. B 33 (1986) ^695.

[46] BONZEL, ^M. P., FRANKEN, ^{A. M. and} PIRUG, C., Surf. ^{Sci. 104} (1981) ^625.

[47] JUPILLE, J., Surf. ^{Sci. 123} (1982) ^L674.

[48] GADZUK, ^J. W., HOLLOWAY, S., MARIANI, ^{C. and} HORN, K., Phys. Rev. Lett. 48 (1982) ^1288.

[49] ^{LANG, N. D.,} Phys. Rev. Lett. 46 (1981) ^842.

[50] ^LANG, N. D. and KOHN, W., Phys. ^{Rev. B 7} (1973)

3541.

[51] MANDEL, T., ^{Private commun.}

[52] MANDEL, T., DOMKE, ^{M. and} KAINDL, G., Surf. ^Sci.

197 (1988) ^81.

[53] ^{MOWFORTH, C.} W., RAYMENT, T., THOMAS, ^R. K., Soc. Faraday Trans. 2 82 (1986) ^1621.

[54] WANDELT, K., Private commun.

[55] ^{HARTMUT, S., DPG} meeting, ^Karlsruhe (March ^14-

18, 1988).