Electronic transitions and dissociation in low energy grazing collisions of O+2 X 2Πgand a4Πu with W(110) partially covered by alkali atoms

Surface Science 318 (1994~ 403-412

Electronic transitions and dissociation in low energy grazing collisions of 0; X 2I& and a 4& with W( 110) partially covered by alkali atoms I-I. Miller, D. Gador, V. Kempter * Physikalisches

In&tat der Technischen Uttiversitiit Clausthal, Leibnizstrasse

4, 38678 Clausthai-ZeIlerfelrf,

Germany

Received 26 April 1994; accepted for publication 6 July 1994

Abstract Electron energy spectra from slow (50 eV) 0: X21T, and a4H, molecular ions colliding under grazing incidence with W(110) surfaces are reported. The surface work function was varied by the exposure to alkali (Na, K, Cs> atoms. The electron energy spectra are interpreted in terms of various inter- and ~~a-atomic Auger processes: For collisions with clean

W(110) and alkali coverages below 0.3 monolayers (ML) (in terms of the first completed adlayer at room temperature> Auger capture is observed only. At least for 0: X2& dissociative neut~~ation does not occur. Beyond this coverage core excited molecular states (molecular Rydberg states), formed by the addition of a 3s vg electron to the respective ion core, become populated. Their decay manifests itself by Auger deexcitation. In addition we see inter- and in&a-atomic Auger processes involving core excited states of atomic oxygen, most probably O-*(4S)3s2. These states originate from the collision induced dissociation of the core excited molecular states. At least for 0: a411, the dissociation via core excited molecular states appears to be efficient. From the energy spectra of the O- ions scattered into 90” with respect to the ion beam direction we conclude that under the chosen conditions the 0: projectile mainly interacts with the alkali adatoms.

1. In~ducti~n The scattering and dissociation of slow (typically below 500 eV energy) molecular ions incident under grazing angles onto single crystal surfaces has found considerable attention recently (see Refs. [l-4] for reviews). So far, mostly translational spectroscopy techniques have been applied which essentially give information on the charge state and the fragmentation pattern of the scattered projectiles after the

* Corresponding author. ~39-~28/94/$07.#

collision. Very recently, the electron energy spectra induced by slow collisions of molecular ions (Hl [5], CO’ 161,and Nz 171)under grazing incidence at clean and partially alkalated W(110) surfaces were reported. Under these conditions the electron emission is due to various Auger processes occurring as a consequence of charge transfer processes in front of the surface. Normally they take place along the ingoing part of the projectile’s trajectory, in any case, however, within about lo-l3 s after the electronic transition process (when the projectile is still in close contact with the surface). Thus, the study of the

0 1994 Elsevier Science B.V. All rig&s reserved

S%WOO39-6028(94)00416-l

404

H. MiiEIer

et al./Surface Science318

electron emission gives insight into what happens during the collision, in particular during the early stage of the projectile-surface interaction. Moreover, the electron energy spectra may provide a clear signature about particular charge transfer channels, such as resonant electron transfer and Auger capture. In the present paper we report the electron energy spectra for scattering of slow oxygen molecular ions in the X 2II, ground- and a 411U excited state at grazing incidence with clean and partially alkalated (Na, K, Cs) W(110) surfaces; the adsorption of alkali atoms leads to a strong decrease of the work function of the surface. The inte~retation of the electron spectra is facilitated by the availability of the electron spectra obtained for O+ collisions under similar conditions [8]. The motivation for the present study was the fact that the electron energy spectra can give information on a number of open issues as for instance: (1) Auger capture into the relatively tightly bound states X, a, and b of neutral 0, is assumed to be the key process which leads to the formation of undissociated backscattered 0, projectiles at clean surfaces [l-4]. The occurrence of the Auger capture process during a collision will be seen in the electron energy spectra provided one of the two involved surface electrons is emitted into the vacuum. (2) The role of an electron transfer either resonant or by Auger capture into excited states of the neutralized projectile, in the present case the valence excited states 0; rY3 II, and the core excited states 0: 1,3111g,is less well understood at present [g-12,6]. Electron emission as a consequence of these processes can arise in the following way (and can thus serve to identify the process): - The bound core excited states 0: 1,311g will manifest themselves by electron emission via Auger deexcitation to the lower states 0, X, a; b; and - if the lifetime of the repulsive valence excited states 0: 1*3 II, is comparable with the time constant for Auger deexcitation, we can expect to see electron emission via Auger deexcitation of these repulsive states during their dissociation [13,14]. (3) If close to the surface a vibrationally excited 0; 2IQ-like shape resonance is populated, this could lead to the emission of slow electrons (E < 2 eV) by autodetachment of the shape resonance.

(1994)

403-62

2. Experimental details The apparatus was documented previously [15171. Briefly, a monoenergetic and mass analyzed beam of 0: ions (50 eV) impinges grazingly along the [llO] direction of a W(ll0) crystal held at room temperature during the measurements. The spectra are recorded under 90” with respect to the beam direction in the plane formed by the beam direction and the surface normal. The work function of the surface is varied between 5.3 and 1.4 eV by the deposition of various amounts of alkali atoms on the surface. The ion source consists of a Duoplasmatron operating in He gas equipped with an expansion cup 118,191: the He plasma expands into the 0, filled cup and produces the 0: ions in some sort of post-ionisation process. The fraction of excited 0: a 4II, ions in the beam can be varied considerably by operating the ion source at different pressures in the expansion cup: it will become clear from the results discussed in Section 3, that an almost pure 0; X’II, ion beam is obtained when operating the expansion cut at sufficiently high pressure ( > 1 Torr). On the other hand, it will be shown in Section 3.2 that a considerable fraction of excited 0: a 4II, ions is present in the ion beam when working at a factor of three lower pressure in the expansion cup. In Refs. [19,8] we provide convincing evidence that Of beams produced in the manner described above do also contain large amounts of excited species. The exact composition of the beam is, however, of no relevance for the interpretation of the peak structures seen in the spectra. The experimental results (Figs. 2 to 7) display energy spectra of the ejected electrons for grazing incidence (5” with respect to the surface plane) for collisions of almost pure X 211a and mixed O:(XzII,; a4111,) beams with clean and partially alkalated W(110) surfaces. The spectra are recorded as a function of the alkali coverage. The variation of the low energy cutoffs of the spectra displays the variation of the work function with coverage: as a function of the alkali coverage the work function is non-monotonic. Initially a strong decrease of the work function to a minimum (2.1; 1.6; 1.4 eV for Na; KI, Cs, respectively) around about 0.6 ML alkali coverage occurs before the work function rises again

H. Miiller et al./Surface

Science 318 (1994) 403-412

to 2.1 eV (Cs), 2.3 eV (K) and 2.7 eV (Na). Thus, for coverages around 0.6 ML the same work function is found at two different coverages. If the work function would be the only parameter of relevance for the interpretation of the electron spectra at small work functions, identical results should be obtained at two different alkali coverages. Our results show that this is not the case. As compared to our earlier work an improved version for correcting the energy dependent transmission of the electrostatic analyzer (including its deacceleration lens) is employed [8,19]. As a consequence, the overall agreement with simulations of the type described in Refs. [20,21] is better. After post-acceleration of the charged particles to 2 keV, negative ions passing the electrostatic analyzer can also be detected with reasonable efficiency (about 20%); the post-acceleration does not influence the electronic part of the spectra [5,6]. The ion scattering results are shortly summarized in Section 4; more details can be found in Ref. [19].

+ 2

O’( 4s)+o( 3P)

a41T, xa&(

1.3n

3. Discussion The discussion is based on the information available on the 0: and 0, potentials [22-271; those states which appear to be relevant for the following discussion are reproduced in Fig. la, schematically. Fig. lb displays the ionization potentials of several states of 0, (at the 0: equilibrium internuclear distance). They become populated via electron capture by either the 0: X ’ IIs or a 4II u ion as soon as the surface work function is smaller than the respective ionization potentials. Thereby we assume that in any capture process the core state of the parent ion remains unchanged. The excitation energies of the core excited states of atomic oxygen (produced by the dissociation of molecular Rydberg states) and of the oxygen Feshbach resonances are given in Refs. 128,291 (see also Ref. [S]). As was pointed out in Section 1, any core excited atomic or molecular state can decay by electron emission via Auger deexcitation (AD) to lower states (in particular to the respective ground states). If the surface electrons involved in the Auger deexcitation process originate from the Fermi level, the kinetic energy of the emitted electrons corresponds to the

405

+-

3.5 Kl

6

L

0.y

3P)

\

INTERNUCLEAR

(b)

lS)3s+O(

DISTANCE

O,+ X211g O,+ a411u o

16

Fig. 1. (a) Potential curves for the 02 and 0, states involved in the oxygen-surface collision process (schematically). (b) Excitation energies of the 0, states (at the 0: equilibrium internuclear distance) involved in the oxygen-surface collision process (schematically).

406

H. Miller et al. /Surface

difference in energy between the involved initial and final oxygen states (at the projectile-surface distance where the transition occurs). We have indicated in the spectra where this high energy onset due to Auger deexcitation is expected. For molecular transitions it is assumed that this is a vertical transition between the vibrational ground states of the involved electronic molecular states. Electrons with less kinetic energy are emitted when surface electrons from below the Fermi level take part in the Auger process. Sharp structures appear in the spectra when temporary negative ions (Feshbach resonances) decay by autodetachment (AU> without the involvement of surface electrons. Their energetic location is also indicated in the spectra, and is approximately given by the excitation energy of the Feshbach resonance with respect to the final neutral state (here the ground state). Since the formation of the negative ions states requires a work function below about 2 eV [32] the autodetachment structures are normally seen only weakly for Na/W(llO). When the primary oxygen molecular ion is neutralized in an Auger capture (AC) process (involving two surface electrons) to the molecular ground state, we expect to see a relatively structureless feature; its width is approximately given by (IP - 2WF), where IP = 12.2 eV is the ionization energy of the projectile molecule and WF the surface work function. Some conclusions are based on our results for Of collisions with partially alkalated W(110) [8]. Recent summaries discussing the mechanisms for electron emission in collisions of atomic ions with surfaces can be found in Refs. [30-321.

Science 318 (1994) 403-412

co 0 _+ 6.0.. 8

: OML

t

o.or, 0.0

5.0 -1

energy

10.0

15.0

/ eV

Fig. 2. Energy spectra of the electrons emitted in 50 eV collisions of 0: (X’II,) ions (ion source operated under “high-pressure conditions”) with W(110) surfaces partially covered by Na atoms. Coverage increases by * 0.03 ML between each two neighbouring spectra. Bottom curve is for clean W(110). The angle of incidence is 5” with respect to the surface.

3.1. Collisions of 0: X”“g ground state molecular ions with clean and partially alkalated W(ll0)

Such electrons are practically absent in Figs. 2 and 4. We conclude that under “high pressure conditions” the 0: beam is mostly composed of 0: X211, ions. In Ref. [8] we provide evidence that an atomic O+ beam obtained when operating the ion source under “high pressure conditions” is also mostly composed (> 95%) of O+ 4S ground state ions.

The spectra displayed in Figs. 2, 4 and 6 are essentially confined to energies below about 8 eV for all alkali coverages. As will be explained below, this is the maximum energy which can be expected from Auger deexcitation of the core excited states 0: 1,311s formed by the resonant transfer of a surface electron to the ground state 0: X’II, ion into the 3s ~a molecular orbital. Auger deexcitation of core excited states whose parent is the 0: a 4II u ion would produce electrons with up to about 12 eV energy (see Figs. 3, 5 and 7).

3.1.1. Clean W(ll0) and alkali coverages below 0.3 ML It was proposed in Ref. [33] that the repulsive 3aa2 valence excited states 0, * lr311 (configuration 1rru4 1~~~3a,) may become p;pulated by resonant electron transfer even in collisions with clean metals. Following Ref. [13] we estimate that the dissociation time of these states is at least comparable to typical time constants for Auger deexcitation to the states X, a, b. Therefore, the population of the states 0; 1’311u leads to emission of electrons with a broad energy

H. Miiller et al. /Surface

(D

2 *

6.0

0.0

5.0 1

10.0

15.0

pc$q

energy

/ eV

Fi 3. Same as Fig. 2, but for a beam composed of both 0: ! X IIs and a4rI, ions (ion source operated under “low pressure conditions”).

distribution up to about 8 eV. Such a contribution to the spectra is obviously not seen for clean W(110). We conclude that resonant neutralization into the repulsive valence excited states 0; 1Y311U- if it occurs - is very unlikely at large work functions, in particular for clean W(110). This agrees with Fig. lb which suggests that there is no resonance between 0 z 1P311Uand occupied states of the surface for the considered range of work functions. Because we find no indication for dissociative neutralization (into 0; 1T311U)in the spectra, the large probability for dissociation observed for 0: collisions on Ni [34,1,2] cannot be ascribed to dissociative neutralization of $ X’II, into valence excited states, such as 0; ’ III”. For work functions 2 4 eV the 0: projectile can be neutralized by Auger capture into the states X, a, b (configurations 30-s* 19rU4 l?r,2) involving two electrons from the surface. The transition energy between 0: and the neutral oxygen ground state is about 12 eV (see Fig. lb) for a Franck-Condon

Science 318 (1994) 403-412

407

transition at the 0: equilibrium distance. This value is lowered below 11 eV by the image force interaction at projectile-surface distances where Auger capture becomes efficient [20,21]. The remaining energy is not much larger than the energy required to eject an electron in an Auger capture process (twice the work function (5.2 eV for clean W(110)). This explains the weak emission, labeled AC 0: + 0, observed for clean W(110). The absence of a strong contribution of Auger capture to the electron spectra does not imply that the Auger capture process is inefficient: the process may lift the electrons into empty states below the vacuum level instead of ejecting them into the vacuum. Summarizing, for 0: X2111, collisions the Auger capture process leads to the population of bound oxygen molecular states. Therefore it cannot directly lead to molecular dissociation. When scattering from clean metals (and even for moderate alkali coverages) dissociation must be due to some sort of impulsive mechanism [35-371, such as rotational/ vibrational excitation.

-lML

v.u

3.u I

10.0

p&q energy

: OML 15.0

/ eV

Fig. 4. Same as Fig. 2, but for W(110) surfaces by K atoms.

partially

covered

408

H. Miller et al./Surface

1ML

T' 0K

OML 0

pcgg energy

/

eV

Fig. 5. Same as Fig. 3, but for W(110) surfaces by K atoms.

partially

Science 318 (1994) 403-412

Firstly, a peak appears around 9 eV for I3> 0.5 ML. Based on the results of Ref. [8] we attribute it to the autodetachment of the atomic Feshbach resonance O- *2s22p3c4S)3s2 to the final state 03P. Because we cannot exclude that this feature is due to the presence of 0: a 411U excited ions in the beam, its origin will be discussed in Section 3.2. Secondly, an intensity rise occurs for K/W and Cs/W, but not for Na/W, at energies below about 4 eV between 0.5 and 0.7 ML alkali coverage, e.g. at very small work functions. We attribute it tentatively to the Auger deexcitation of the repulsive valence excited states 0; 1,3II, formed by resonant transfer of a surface electron to the 0: ion at sufficiently small work functions (see Fig. lb). The decay of these repulsive states by Auger deexcitation would produce a broad spectrum below about 8 eV. It is likely that for a repulsive initial state the spectrum of the emitted electrons is dominated by low-energy emission (below about 4 eV) [20,21]. For Na/W(llO) the work function obviously never becomes small

covered

3.1.2. Alkali coverage above 0.3 ML Shoulders, denoted by AD 0: c2II,)3s o-s, appear in Figs. 2, 4 and 6 near 8 and 7 eV that are apparently a consequence of the population of core excited states, such as 0; lY311s (configuration 3us2 1rU4 17~~ 3sa,), by resonant transfer of a surface electron to the molecular ion. Fig. lb suggests that this electronic transition process should become allowed for surface work functions below about 3.3 eV (which is the ionization energy of 0; lT311s including its estimated lowering induced by the image force interaction). The population of just these core excited states 0: lY3IIa occurs by near-resonant charge exchange in binary collisions of 0; X 211, with Cs atoms [22,23]. The core excited states are deexcited by Auger deexcitation to the states X and a, respectively (we see no emission which could be attributed to a transition to state b, though). Two new features appear in the coverage range in which the work function of the alkali covered surface becomes smaller than about 2.2 eV:

3.q

1%

1ML

0.0

10.0

energy

OML 15. .O

/ eV

Fig. 6. Same as Fig. 2, but for W(110) surfaces partially by Cs atoms.

covered

H. Miiller et al. /Surface

Science 318 (1994) 403-412

3.2. Collisions of 0; a”‘7g excited molecular with clean and partially alkalated W(ll0)

409

ions

7.0 co 2 *

The spectra in Figs. 3, 5 and 7 extend up to about 12.5 eV. Auger deexcitation of 0, Rydberg states whose parent is the 0: a 4II u ion produce electrons up to this energy (Ref. [27] and Figs. la and lb). We conclude that an appreciable fraction of excited molecular ions 0: a411, is in the beam when operating the ion source under “low pressure conditions”. In Ref. [8] we provide evidence that Of beams obtained in this way are also a mixture (of roughly equal parts) of O+ 4S and *D; *S excited ions. Obviously, the sharp feature around 9 eV, labeled AU 0-*2~*2p~(~S)3s*, seen strongly for K/W and Cs/W, is caused mainly by the presence of 0: a411, ions.

6.0

.O

5.0

1

10.0

15.0

p$Gi

energy

/ eV

Fig. 7. Same as Fig. 3, but for W(110) surfaces by Cs atoms.

enough

to allow for a significant

partially

population

covered

of 0;

L3 nI,.

Other processes, such as the population of repulsive states via mixing of valence and core excited states [24-261, may also contribute to the enhancement of the low energy emission. The population of the repulsive states 0; 1,3111ucould be the reason for the increase observed in the dissociation yield for 0: collisions with alkalated Ni surfaces at small work functions [34,1,2,33]. Vibrationally excited (u > 4) 0; X2111, negative ions (e.g. the oxygen shape resonance) could be formed via the resonant capture of a surface electron by f.i. 0, ground state molecules [12]. Charge transfer to the 0; *IIs shape resonance, followed by its autodetachment with the emission of low energy electrons (E < 2 eV) occurs in binary collisions of H- with 0, [38]. Thus, autodetachment of the 0; 211.a shape resonance, if it is populated in surface collisions, may also contribute to electron emission at low energies.

3.2.1. Clean W(ll0) and alkali coverages below 0.3 ML The emission seen in this coverage range is compatible with Auger capture of 0: a4111, to 0, X, a, and b. Other processes, such as Auger capture to other low lying excited states (as c, C, and A, see Fig. 1) or processes which are a consequence of the population of the 0, states c, C, and A by resonant capture may occur, but are not manifesting themselves in the emission of electrons. If produced, these comparatively weakly bound states may dissociate by some impulsive mechanism. 3.2.2. Alkali coverages above 0.3 ML We attribute the shoulder near 12.5 eV, labeled AD 0; (a 4II “)3s o-s, to the formation of core excited states 0; 3*5II, (configuration l~:lrrt 3s as) by the resonant transfer of a 3s og electron to the 0: a*II, parent ion: it is due to Auger deexcitation of these states to the neutral oxygen ground state. The population of the same core excited states in binary collisions between 0: a 411u and Cs atoms occurs by near-resonant charge exchange [39]. The comparison with Figs. 2, 4 and 6 proves that the additional shoulder appearing near 9 eV, labeled AD 0** 2s22p3c4S)3s, is not due to Auger deexcitation of 0; 1*3IIs (the corresponding shoulder is seen around 8 eV in Figs. 2, 4 and 6). Our data for O+ collisions 181 suggests that the 9 eV shoulder and the

410

H. Miiller et al. /Surface

associated narrow feature have to be attributed to inter- and intra-atomic Auger processes initiated by core excited 0*c4S)3s atoms: we attribute the 9 eV shoulder to Auger deexcitation of excited oxygen atoms in the states 0*(4S)3s. These states are presumably formed by dissociation of the core excited molecular states O$ 3,5II, (see Fig. la) (formed by resonant neutralization of a 4 II ,,) which are comparatively weakly bound only [27]. In agreement with this view the contribution from Auger deexcitation of 0; 3,5II, (below about 12.5 eV) weakens considerably when dissociation of these states becomes efficient with increasing coverage. If the work function becomes sufficiently small (as for K/W and Cs/W in particular), the core excited fragment atoms themselves can resonantly capture a surface electron thus forming the oxygen Feshbach resonance O- * 2~‘2p~(~S)3s’ [28,29,8]. We have explained in Ref. [32] that such resonances can be formed in ion collisions with alkalated surfaces only for a small range of coverages around the work function minimum. Autodetachment of this resonance to the ground state 03P will produce the narrow feature seen around 9 eV, labeled AU O-* 2~‘2p~(~S)3s* [28,29,8]. For coverages beyond 0.7 ML the work function becomes again too large even for Cs/W - to allow for the formation of the resonance by electron capture of the core excited oxygen atom. For these coverages only Auger deexcitation of the core excited atoms is seen. For Na/W the formation of the Feshbach resonance plays a minor role because the work function always stays above 2.1 eV. The features attributed above to core excited oxygen atoms and Feshbach resonances are also seen (weakly) in Figs. 2, 4 and 6. At present we cannot decide whether this comes from the small percentage of excited 0: a 411u ions in the beam or is actually due to small probability for dissociating 0: lp311s molecules (produced by electron capture of 0: X*IIa ions). The dissociation process would also yield core excited 0** 2~*2p~(~S)3s atoms (Ref. [27] and Fig. la>. The emission observed at energies below about 8 eV is a superposition of contributions originating from both 0: X*II, and a411, ions, and we will not attempt to disentangle them. Clearly, the appearance of the weak narrow features around 7 and 5 eV

Science 318 (1994) 403-412

in Figs. 5 and 7 is correlated with the appearance of the peak due to autodetachment of the Feshbach resonance O- * 2~*2p~(~S)3s*. We propose that these weak features are also due to autodetachment of the same resonance, however to the final states ‘D and ‘S, not to 3P as for the dominant feature at 9 eV. These decay channels would be strictly forbidden for free temporary negative ions because of angular momentum conservation [28], but may become open channels under the influence of the perturbation imposed by the surface. Interestingly enough, these decay channels of the Feshbach resonance are not open for 0+ collisions although the resonance is formed [8]. The reason could be that in this case the formation and the decay of the resonance occurs further away from the surface.

4. Additional remarks We have also studied 200 eV 0: collisions under grazing incidence [19] for comparison with time-offlight studies using oxygen ions [34,1,2,33] (see Refs. [1,2] for summaries). The electron spectra are qualitatively similar as the 50 eV results, and therefore do not require any additional discussion. As expected intuitively the features attributed to the autodetachment of O-* 2~*2p~(~S)3s* are less pronounced. We propose that the neutralization mechanisms discussed above which cause the electron emission at 50 eV collision energy do still apply at 200 eV. After the postacceleration of the negatively charged particles to 2 keV behind the electrostatic analyzer [5,6], we have also obtained the energy spectra of the negative oxygen ions scattered into 90”. The measurements were performed for 50 and 200 eV 0; energy. The results can be summarized as follows (for details see Ref. [19]): (1) Only O- fragment ions are found under a 90” scattering angle for both energies. This appears to be reasonable since a violent collision event (such as scattering into 90”) is likely to lead to dissociation. This is also suggested by classical trajectory studies, although they are available for scattering from clean metals only [40,41]. It also appears to be in agreement with the TOF results for 500 eV 0: collisions on clean Ni and K/Ni under grazing incidence and glancing scattering [34,1,2]: these results do also

show that for scattering from both clean and Kcovered Ni the production of atomic negative ions prevails by far. (2) A comparison of the results for 200 eV with the corresponding ones for Of projectiles [19] suggests that the ion scattering spectra are composed from two components, the 90” deflection being mediated by a violent collision with either adsorbed alkali atoms or substrate atoms. At 200 eV the contribution from adsorbate scattering seems to prevail by far, though. (3) For 50 eV beam energy shadowing effects seem to prevent 90” scattering involving adsorbed alkali atoms. A weak (as compared to 200 eV collisions) signal is seen under 90”; the energy loss is governed by the interaction with the substrate. The scattered ion intensity (displaying a m~mum near the work function minimum) decreases strongly towards the alkali saturation coverage (except in the case of es/W) 1191.

5. Summary We have measured and interpreted the spectra of the electrons emitted after grazing incidence of slow 0: X2111, and a 411, ions with parity alkalated W(110) surfaces. On the basis of these results we were able to identify mechanisms leading to the neutralization of the molecular ions and to obtain information on mechanisms leading to the dissociation of the molecules formed in this way: For collisions from clean W(110) and up to alkali coverages of 0.3 ML the observed emission is due to the Auger capture process. At least for 0: X211, projectiles dissociative neutr~~ation seems not to occur, but the dissociation observed in time-offlight-studies must be due to an impulsive mechanism. For coverages beyond about 0.3 ML the formation of core excited molecular oxygen states, 0; 1*311s(configuration 3os2 17r4 lrrs3s o-J for 0: X ‘IIp collisions and 0; Ys II, (configuration 3a,21qQ7r,2 3s as), for 0: a411, collisions, is observed. Auger deexcitation of the core excited states is seen in the electron spectra. Dissociation of is efficient, and leads to the formation of 02 * 37511u core excited oxy en atoms. On the other hand, dissociation of 0 z ‘,QLI, is weak. For alkali coverages

beyond 0.5 ML an oxygen Feshbach resonance is formed by the resonant transfer of a surface electron to the core excited atomic fragments. The decay by autodetachment of the resonance is seen in the spectra. Dissociative neutralization (into the valence excited states 0; tP3II,) is possibly also seen, but only at small surface work functions. Acknowledgement

Financial support of the Deutsche Forschungsgemeinschaft is gratefully acknowledged. References [I] W. Heiland, Surf. Sci. 251/252

(1991) 942. [z] W. Heiland, in: Trends in Physics, EPS - 8, III (1991) 777. f3] S.R. Kasi, C.S. Sass and J.W. Rabalais, Surf. Sci. Rep. 10 (1989) 1. [4] K.J. Snowdon, J. Chem. Sot. Faraday Trans. II 85 (1989) 134. [5] H. Miller, R. Hausmann, H. Brenten and V. Kempter, Surf. Sci. 284 (1993) 129. [6] H. Miiller, R. Hausmann, H. Brenten and V. Kempter, J. Chem. Phys. 179 (1994) 215. [7] H. Miller, R. Hausmann, H. Brenten and V. Kempter, Surf. Sci. 303 (1994) 56. [S] H. Miiller, D. Gador, II. Brenten and V. Kempter, Surf. Sci. 313 (1994) 188. [9] J. Rechtien, U. Imke, K.J. Snowdon, P.H.F. Reijnen, P.J. van den Hoek and A.W. KIeyn, Surf. Sci. 227 (1990) 35. [lo] J.-H. Rechtien, W. Mix, D. Danailov and K.J. Snowdon, Surf. Sci. 271 (1992) 501. [11] J.M. Schins, R.B. Vrijen, W.J. van der Zande and J. Los, Surf. Sci. 280 (1993) 145. [12] P.H.F. Reijnen, P.J. Van den Hoek, A.W. Kleyn, U. Imke and K.J. Snowdon, Surf. Sci. 221 (1989) 427. [13] J.R. Hiskes and A.M. Karo, J. Appf. Phys. 67 (1990) 6621. [14] IS. Bitensky, E.S. Parilis and LA. Wojciechowsky, Nucf. Ins&urn. Methods Phys. Res. B 47 (1996) 243. [U] H. Brenten, H. Miiller and V. Kempter, Z. Phys. D 22 (1992) 563. [16] H. Schall, W. Huber, H. HBrmann, W. Maus-Friedrichs and V. Kempter, Surf. Sci. 210 (1989) 163. [17] H. Brenten, H. Miller, K.H. Knorr, D. Kruse, H. Schall and V. Kempter, Surf. Sci. 243 (1991) 309. [18] L. Vnyi, Atom and Ion Sources (Wiley, London and Budapest, 1977). [19] D. Gador, Diploma Thesis TU Clausthal, 1994, unpublished. [20] P.A. Zeijlmans van Emmichoven, P.A.A.F. Wouters and A. Niehaus, Surf. Sci. 195 (1988) 115. [Zl] P. Eeken, J.M. Fhdt, A. Niehaus and I. Urazgil’din, Surf. Sci. 273 (1992) 160.

412

H. Miller et al. /Surface

[22] W.J. Van der Zande, W. Koot, J.R. Peterson and J. Los, Chem. Phys. L&t. 140 (1987) 175. [23] W.J. Van der Zande, W. Koot, J. Los and J.R. Peterson, Chem. Phys. 89 (1988) 6758. [24] R.P. Saxon and B. Liu, J. Chem. Phys. 67 (1977) 5432. [2.5] R.P. Saxon and B. Liu, J. Chem. Phys. 73 (1980) 870, 876. [26] J. Buenker and SD. Peyerimhoff, J. Chem. Phys. 8 (1975) 324; Chem. Phys. Lett. 34 (1975) 225. [27] P.H. Krupenie, J. Phys. Chem. Ref. Data 1 (1972) 423. [28] J.T. Matese, Phys. Rev. A 10 (1974) 454. [29] G. Schulz, Rev. Mod. Phys. 45 (1973) 423. [30] H. Miller, R. Hausmann, H. Brenten and V. Kempter, Nucl. Jnstrum. Methods Phys. Res. B 78 (1993) 239. [31] H. Brenten, H. Miiller and V. Kempter, in: Ionization of Solids by Heavy Particles, Ed. R.A. Baragiola (Plenum, New York, 1993). [32] H. Brenten, H. Miiller, R. Hausmann, A. Niehaus and V. Kempter, Z. Phys. D 28 (1993) 109.

Science 318 (1994) 403-412 [33] S. Schubert, Doctoral Thesis University Osnabriick, 1989, unpublished. [34] S. Schubert, U. Imke and W. Heiland, Vacuum 41 (1990) 252. [3_5] U. van Slooten, D. Andersson, A.W. Kleyn and E.A. Gislason, Surf. Sci. 274 (1992) 1. [36] AW. KJeyn, Condens. Matter 4 (1992) 8375. [37] A.W. Kleyn, Vol. 31 of Springer Series in Surface Science (Springer, Berlin, 1993) p. 116. [38] V. Esaulov, J.P. Grouard, R.I. Ha& M. Landau, J.L. Montmagnon, F. Pichou and C. Schermann, J. Phys. B: Atom. Mol. Phys. 17 (1984) 1855. [39] W.J. Van der Zande, W. Koot, J.R. Peterson and J. Los, Chem. Phys. 126 (1988) 169. [40] U. van Slooten, E.J.J. Kirchner and A.W. Kleyn, Surf. Sci. 283 (1993) 27. [41] E.J.J. Kirchner, E.J. Baerends, U. van Slooten and A.W. KJeyn, J. Chem. Phys. 97 (1992) 3821.