The statistics of electron emission from clean metal surfaces induced by slow ions: measurement and recent applications

International Journal of Mass Spectrometry and Ion Processes 129 (1993)II-29 0168-I176/93/%06.00 0 1993 - Elsevier Science Publishers B.V. All rights reserved

17

Review

The statistics of electron emission from clean metal surfaces induced by slow ions: measurement and recent applications’ Friedrich Aumayra, Tilmann D. M%rkb9*,Hannspeter Wintera aInstitut ftir Allgemeine Physik, Technische Universitdt Wien, Vienna, Austria bInstitut fiir Ionenphysik, University of Innsbruck, A-6020 Innsbruck, Austria (Received 1 February 1993; accepted 1 April 1993) Abstract We describe investigations on slow ion-induced electron emission from clean gold, based on the related electron emission statistics. Experimental methods for measuring these statistics are demonstrated and typical results from recent studies with various projectile species (singly-, doubly-, and multiply-charged ions as well as cluster ions) are presented and discussed. Key words: Particle induced electron emission; Potential emission; Kinetic emission

1. Introduction

Bombarding the surface of a solid, in particular a clean metal, by slow neutral or ionized atoms, molecules or clusters may cause emission of electrons due to the transfer of potential energy (potential emission PE) and/or kinetic energy (kinetic emission KE) from the projectile to the free and/or bound target electrons. Such processes are of both fundamental interest and considerable practical importance and have therefore been under scrutiny for a long time. The general features of PE have been explained by Hagstrum [I], and a recent review of this field has been given by Varga and Winter [2]. The knowledge about KE has recently been reviewed by Hasselkamp [3]. Both types of processes depend rather critically on the target surface conditions, which is of con* Corresponding

author.

’ Dedicated to the memory of Professor Maximilian Pahl.

siderable relevance for related experimental investigations. As a general distinction, PE results from electronic transitions between the projectile and the surface before an impact has taken place, whereas KE can only be initiated after the projectile has made close contact with the surface. In a more detailed view it can be shown that PE arises from Auger-type processes involving time intervals of the order of lo-l4 s, which is comparable to the flight time of a relatively slow (w < lo5 m s-’ M 60 eV u-l) particle within the distance for its probable electronic interaction with the metal surface (for singly charged ions typically some lo-” m, cf. Hagstrum [I], increasing with the ion charge state q for multicharged ions; see Varga and Winter [2]). Consequently, even rather slow multicharged ions may not be completely neutralized and subsequently de-excited until their impact on the surface, because this would need too many of the relevant electronic transitions. Therefore, PE is expected to become more efficient the lower the impact energy

18

F. Aumayr et aLlInt. .I. Mass Spectrom.

and, moreover, does not involve an impact energy threshold. KE, however, is related to the stopping power of projectiles within the uppermost atomic layers of a solid, and appears as the result of single collisions or collision cascades which usually last not longer than 1O-” s. Only electrons from the vicinity of the surface receiving a minimum kinetic energy of the order of 10 eV may escape across the surface barrier into vacuum. Consequently, KE is commonly believed to appear only above a certain impact energy threshold which, however, is not always very well defined. From the above simple considerations we may already conclude that only for electron emission induced by rather slow particles can the PE and KE processes be regarded as mutually independent. For impact of multicharged ions (MCI - Zq+), electrons are captured resonantly from states near the Fermi edge of a metal surface within a critical distance which is determined by the wavefunctions overlap of the surface density-of-states (S-DOS) with empty projectile states. This critical distance will increase with the projectile charge q and also be larger the lower the surface function W,. Once within this critical region, a slow MCI (impact velocity 2)<< 1 a.u.) will be rapidly further neutralized, according to a characteristic time related to the Fermi velocity uF, which for metals is of the order of 1 a.u. [4]. The multiply excited (hollow) atoms so formed are subject to both resonant ionization (RI) and autoionization (AI), processes which together with the ongoing resonant neutralization (RN) will determine the projectile’s electronic population until it hits the surface [5-81. Finally, for the impact of cluster ions, electrons may be emitted both as the result of collisions between cluster- and surface constituents, and collisions amongst the cluster constituents themselves. Especially for low impact energies, i.e. in the vicinity of the KE threshold, almost no studies have been conducted involving well defined experimental conditions (in particular clean surfaces). The present work describes recent progress in measuring the particle-induced electron emission statistics (cf. section 2), by which means, in contrast to the

Ion Processes

129 (1993) 17-29

common techniques for investigating particleinduced electron emission, we can distinguish amongst different, mutually independent contributions to the experimentally-observable total electron emission. As will be shown later, for particular collision systems this is of special interest in view of the relative importance of PE and KE. We can determine the probabilities for emission of a given number (n = 0, 1,2, etc.) of electrons due to the impact of individual projectiles onto the surface, which results in the so-called statistics of ion-induced electron emission (ES). The latter contains comparably detailed information on the electron emission processes involved, as will be demonstrated for the bombardment of clean polycrystalline gold by slow singly(section 3), doubly- (section 4) and multiply-charged ions (section 5). Finally, in section 6, we will present some results of a first study on ES for cluster ioninduced KE.

2. Determination of the particle-induced emission statistics

electron

Ion-induced electron emission from solid surfaces is usually studied by determining the total electron yield (i.e. the mean number of electrons emitted per projectile particle, conventionally being denoted by 7) and/or the ejected electron energy distribution, dN,/dE,. However, both of these quantities are only of qualitative value insofar as they do not provide a straightforward appreciation of the mechanisms responsible for electron emission. In this situation it is of interest to measure the distribution of the numbers of electrons resulting from the individual emission events. The so-called ES are equivalent to the set of probabilities, W,, for the ejection of a given number of n electrons [9], and are directly related to the total emission yield y

$%+l

n=O

(1)

F. Aumayr et al./lnt. J. Mass Spectrom. Ion Processes 129 (1993)

deceleration

extraction electrode

case

19

17-29

focussing electrode

electron detector (3OkV)

initial ion energy 4.1 keV

fmal ion energy E= 1OOeV Fig. 1. Experimental setup for measuring particle-induced ES from the impact of slow ions on clean polycrystalline gold

Lakits and co-workers [lo-121 and Aumayr et al. [13] have described how such ES can be precisely determined for ionized as well as neutral projectiles. Figure 1 gives a sketch of their apparatus, which has been especially designed for the achievement of low energy ion impact. The primary ions of interest are directed into the statistical detector setup in an UHV environment, and can be decelerated just before hitting the sputter-cleaned polycrystalline gold target. The latter is situated inside a highly transparent, conically shaped mesh 111,121 which backwardly deflects all emitted electrons toward an acceleration lens system and then into a solid state detector biased at a high positive voltage (typically 25 kV)

[l 11.

with respect to the target. The emitted electrons thus hit the detector surface with typically 25 keV energy. Within the time resolution of the detector and subsequent electronics (typically >> 1 ns), the emission of a group of n electrons resulting from the impact of an individual projectile cannot be resolved into n single events but will rather be registered as a single detector pulse equivalent to the n-fold energy of one single 25 keV electron. This gives rise to the raw pulse height distributions as shown in Fig. 2, from which the relative values for W, can be derived after proper corrections for detector resolution and background contributions [13]. Absolute values for W, (including for n = 0) can be derived by means of Eq. (1) if the corresponding

Hi(12 keV) -

Au

s

9

6000

8

E [keV] Fig. 2. Typical uncorrected ES pulse height spectrum for the impact of 12 keV H3’ procedure [12].

IOIIS

on clean gold, and the results of the fitting

F. Aumayr et aLlInt. J. Mass Spectrom. Ion Processes I29 (1993) 17-29

20

For any ionized projectile species, rather low nominal target impact energies down to a few electronvolts times charge state q may be achieved by means of a decelerating four-cylinder lens in front of the target surface (cf. Fig. l), if the initial ion energy spread is sufficiently small.

3. Measurements involving singly-charged ions 6

IO

15

20

25

n

Arg+

0

5

IO

15

1 OOeV

20

25

n Fig. 3. (a) Comparison between the measured electron emission statistics for the impact of 100 eV Argf ions on gold (0) and the ) (141. (b) Individual contriresults of a least-squares fit (-butions to the fitted data, including electron backscattering, (contributions to even numbered peaks have been shaded for clarity).

values for y are known. Otherwise, y has to be determined either in the conventional way by current measurements or with the present setup in the way described by Lakits et al. [11,12]. The latter procedure permits rather precise total electron yield measurements even for y values of less than lop3 electrons per projectile. For highlycharged ions (cf. Fig. 3), the total yield will become so large that the quantity W. can be safely neglected and then y can be evaluated exclusively from the relative ES measurements according to Eq. (1). Figure 3 demonstrates for impact of 100 eV Ar9+ on clean gold, how the experimentally obtained (“raw”) ES have to be evaluated according to the procedures described by Aumayr et al. [13], to obtain the emission probabilities W,.

Experimentally-determined raw ES data for the impact of Ne+ ions on clean gold are shown in Fig. 4, for impact energies ranging from 100 eV up to 16 keV. Note that Ne+ has no long-lived highly excited states [15], so only ground state Ne+ ions are involved. At the lowest impact energy (E 6 15 eV u-l), apparently only one electron is ejected, whereas at higher values of E, the emission of 2,3, etc. electrons gradually occurs [l 11.Consequently, as long as KE is not possible, only PE can contribute to the total electron yield by ejecting one electron, although the available potential energy ( W < 21.6 eV) would in principle suffice to emit up to three electrons. The results presented here are supported by similar measurements for the impact of He+ (see Fig. 5), Ar+ and other singly-charged ions on clean gold t111. Apparently, there is a fundamental difference between PE- and KE related ES. PE involves reasonably well-defined transitions between electronic states of the target surface and the approaching projectiles, and therefore accordingly gives rise to emission of a limited number of electrons. KE, however, involves the dissipation of the projectile kinetic energy amongst a relatively large number of target electrons, which then have comparable but relatively small chances of being ejected. The number of these electrons increases with the transferred kinetic energy and thus with the projectile velocity, whereas the potential energy carried by the approaching ion is transferred via Auger-type electronic transitions to one or a few electrons only. In the following, we demonstrate how such measurements can be extended towards rather small impact

F. Aumayr et aLlInt. J. Mass Spectrom. Ion Processes 129 (1993)

n=

21

17-29

1 tl

Ne'

-

Au

Fig. 4. Raw experimental data of ES from the impact of Nef on clean polycrystalline gold vs. impact energy [18].

energies. As an example, consider the impact of He+ on clean gold. To identify the exact location of the KE threshold, the variation of y vs. impact velocity E is not very significant. In Fig. 5 the measured ES probabilities W, for the emission of n ~3 electrons are plotted. The KE threshold impact energy (or at least an upper limit) can now be determined rather well by the appearance of a second peak (i.e. W, # 0), since the PE causes emission of one electron only (see the above discussion). According to classical dynamics, an upper limit for the onset of KE (i.e. the related KE threshold) will principally be given at that impact velocity where the kinetic energy transfer in headon collisions of projectiles with the quasi-free metal electrons surpasses the metal surface work function, W,. This threshold would be about 300 eV u-’ for a clean gold surface [11,17] and thus considerably higher than the KE threshold

determined here of less than 50eVu-’ for He+ (cf. Fig. 5). The precise origin of KE at such rather low impact velocities is not yet well understood, but it may originate from the autoionisation of highlyexcited quasimolecules which become transiently formed in close encounters of the impinging projectile ions with singular surface atoms. Similar measurements have been carried out with other singly-charged ions [ 11,16,18]. Our observation that PE induced by impact of singly-charged ions on clean gold can eject at most one electron, and the appearance of more electrons with increasing impact energy, which we thus attribute to the onset of KE, can now be used to separate the PE and KE contributions to the total electron emission in the following way. Let P, and K, be the individual probabilities for the emission of n electrons by PE and KE, respec-

22

F. Aumayr et al./Int. J. Mass Spectrom. Ion Processes

I29 (1993) 17-29

4He+ -Au 0

-.-em

---I-

.

0 --.-__

*----

P,

2’ 0

/

c,

/

I

I

,

1

E(keV) Fig. 5. Contributions of PE and KE to the total electron emission statistics for the impact of He+ on clean gold [16].

tively. For the measured probabilities W, the set of equations Wn =

kPiKn_i

(2)

i=O

can be solved for P,, and K,, under the assumptions that (i) P, = 0 for n 22 and (ii) K,, = 0 as long as Wn+l is negligibly small. As an example, in this way our W, data for He+ on gold have been separated into KE and PE contributions, with the results shown in Fig. 5. Furthermore, it is of interest to consider the shapes of the measured ES related solely to KE, since they are often approximated by a Poissonian distribution for the corresponding total KE yield y, i.e. P,(y)

= zeeT

-&=-

pn+ l(Y) pnC-Y)

Equation 3 permits a straightforward comparison with the measured ES ratios, Wn+l/ W,, as shown in Fig. 6 from Ohya et al. [19]. The apparent clear deviations from the Poissonian shape could be related via Monte Carlo simulations to contributions from kinetic electron emission due to backscattered projectiles and/or recoiling target atoms.

4. Measurements involving doubly charged ions PE may become rather important for the impact of highly-charged ions on metal surfaces, due to the enhanced potential energy content of these projectiles [20]. This will now be discussed in somewhat I

0.8 -

I

0

H++Au 14Es

I6keV

I

0

0.6 3” .&

0.4 -

b

0.2 -

,@,

0.0

(

0.2

0.4 *n+l

0.6

1 0.8

’ *n

Fig. 6. Variation of -y/(n + 1) vs. W,, 1(y)/W,(y) for the impact of (1-16 keV) H+ on clean gold ((O), (A) and (0) refer to n = 1, 2 and 3, respectively), and Monte Carlo simulated data; ), a Poissonian ES [19]. (-

F. Aumayr et aLlInt. J. Mass Spectrom.

Ion Processes

129 (1993)

more detail for the impact of slow doubly-charged noble gas ions on clean gold, wherefore ES measurements are very useful to elucidate the interplay of the different electronic transitions governing the PE processes. According to Hagstrum [l], the electronic transitions involved in these PE processes can engage either one or more electrons. The only one-electron transition of interest is RN, cannot directly result in which, however, electron emission, but often acts as a precursor for subsequent two- or more electron-emitting transitions. Auger de-excitation (AD) can cause electron emission if the involved de-excitation energy is larger than the metal surface work function W,. Auger neutralisation (AN) may also give rise to electron emission if the potential energy change in the related neutralisation step surpasses 2 W4, and AI will take place after the formation of a double- (multiply-) excited projectile particle via double RN transitions (DRN). The last process has first been identified by Hagstrum and Becker [21] for the impact of He2+ on Ni single crystal sufaces. Varga et al. [22] have measured electron energy distributions for impact of the doubly-

23

17-29

charged rare gas ions Ne2+ and Xe2+ on clean tungsten. They identified for both Ne2+ and Ar2+ a group of very slow electrons which they related to two successive de-excitation steps, either via two excited X+ states (X+* and X+*‘) in the sequence from X2+ toward the X+’ ground state along the path X2+ 4 RN -_) X+* 4 AD(e-) --t X+*’ ---)AD(e-) --t X +o

(4a)

or, alternatively, along X2+ --$ AN(e-) 4 X+* --) AD(e-) * Xfo

For the occurrence of electron-emission in the steps labelled (e-), sufficiently large de-excitation energies must be made available (see above), as is indeed the case for the collision systems Ne2+, Ar2+ on tungsten. However, in electron energy distributions measured by Wouters et al. [23] for kiloelectronvolt grazing collisions of He2+, Ne2+ and Ar2+ on clean copper, no evidence for transitions as described by

3He2+ -Au 1 t

(4b)

10’1,

I

I

tj

1

E(keV) Fig. 7. Emission probabilities W, for the impact of He’+ on clean gold vs. impact energy [16].

24

F. Aumayr et aLlInt. J. Mass Spectrom. Ion Processes 129 (1993) 17-29

Eq. (4) has been found. These authors assumed that only the transitions described by Eq. (5), which can each give rise to one electron at most, are taking place in the X2+ -+ X+’ neutralization steps X2+ + DRN + X0** -+ AI

+ X+’

(5a)

or X2+ + AN(e-) -+ X+’

(5b)

To clarify this situation, we have measured ES for the impact of He*+, Ne2+ and Ar*+ on clean gold. At 100 eV impact energy and, for all three collision systems, an exclusive PE situation is given (see Figs. 4 and 5 for Ne+ and He+, respectively). Emission of up to three electrons can unambiguously be seen for both He*+ (c.f. Fig. 7) and also Ne*+, but not for Ar*+ ([24]; note that a third electron can be emitted via AN of X+’ to the X0 ground state after the transitions outlined by Eq. (4)). For the impact of He*+, the dependence of W, on ion impact energy is shown in Fig. 7. With decreasing E both W2and W3 go through a minimum and then rise again toward the lowest impact velocities covered by these measurements. This behavior is presumably caused by the increasing probabilities for the transitions described by Eq. (4), in comparison to those covered by Eq. (5). A qualitative explanation for this can be given by comparing the competing neutralization/de-excitation probabilities for the three projectile species He*+, Ne*+ and Ar*+. For the neutralization of He*‘, the first de-excitation step according to Eq. (4) can involve at most 13,6eV, and for both Ne2+ (de-excitation into Nef2s2p6*S) and Ar*+ (de-excitation into Arf3s3p6*S) at most 14.1 eV (cf. Varga et al. [22]). While these first de-excitation steps are almost equal in energy for all three ion species, the recombination energies of the doubly charged ions toward the respective X+’ ground states differ markedly from each other (54.4eV for He, 41,OeV for Ne and 27.6eV for Ar, respectively). Generally, the transition probabilities for the electron emitting processes described by Eq. (5) should be larger the smaller the X*+-X+’ energy steps involved. Keeping this

in mind, we can explain the decreasing probabilities WJ by the competition from processes according to Eq. (5), which should become increasingly more important when changing the projectile species from He*+ via Ne*+ to Ar*+. Whereas Varga et al. [22] discovered a slow electron group for Ar*+tungsten collisions also, we could not find emission of more than two electrons for the impact of Ar*+ on gold. This may be caused by the difference in the S-DOS for tungsten and gold targets. Polycrystalline tungsten has a surface work function W4 = 4.5eV and a large S-DOS just below the Fermi level [25], whereas for gold, W4 = 5.1 eV and the S-DOS is much less dense in the vicinity of the Fermi level. These differences could explain a relative decrease of the probabilities for electron emission due to the first de-excitation processes described by Eq. (4), compared with the electron transition probabilities involved in Eq. (5). Recently, electron energy spectra have been measured by Brenten et al. [26] for the impact of very slow (< 10 eV) He*+ on clean and cesium-covered tungsten, showing a peak at a few electronvolts electron energy, which was ascribed by these authors to emission processes following Eq. (4). 5. Measurements involving multiply-charged ions Emission of slow electrons (E, < 60 eV) induced by the impact of slow (Vi ~2 x 10’ m s-l) MCI on atomically clean, polycrystalline gold has been studied by virtue of the involved ES, both experimentally and by numerical simulation. Various primary ions such as Nq+ (4 = 5-lo), A@ (9 = 5-16;‘; 5 Or 6)’5-lo), Neqf rq+ (q = Xeq+ (q = 6, 8 or 10) and Iq’ (q = 16, 20, 23 or 25) have been extracted from a recoil ion source [27] pumped by the GSI UNILAC heavy ion accelerator. The resulting slow electron emission yields have been the ES All were performed under UHV conditions pressure ,< 3 x lo-’ Pa). As two rather different examples, Fig. 8 shows total electron yields vs. impact velocity for MCI of nitrogen and iodine, respectively. for Nq+

F. Aumayr et al./lnt. J. Mass Spectrom. Ion Processes 129 (1993)

N5’

10 -

0

2s

17-29

a,,,= _u_= w ~~~l~~lT~~~~T~~““” 0 5 IO

N6+ 15

20

v (10 4 m/s) Fig. 8. Measured

total electron

yields vs. impact

velocity for the impact of Nq+ (q = 5 and 6) and Iq+ (q = 16, 20, 23 and 25) on clean polycrystalline gold [ 141.

projectile species practically no variation of the electron yields with impact-velocity can be recognized, rather marked variations can be seen for all four Iq+ projectile species. This already provided some hints as to the probable contribution of several, fundamentally different emission processes to the observed total electron yields, with their relative importance clearly depending not only on the charge state of the primary MCI, but also its structure. Further important information has been obtained from the electron-emission statistics themselves, which in all cases deviated quite markedly from a Poissonian shape, which is to be expected if the observed, relatively large numbers of electrons were emitted as the result of many, statistically independent, single electron emission events. From the systematics of our experimental results we discovered new details on the formation and autoionization of socalled “hollow atoms” (see section l), which are transiently formed during the approach of the MCI towards a metal surface. To support this picture, model calculations based on a recently-developed classical over-barrier approach [6] have been performed, which are in rather good agreement with the experimental data. These calculations

clarified convincingly the above mentioned, different processes contributing to the experimentallyobservable total electron yields [14,28]. Roughly speaking, at least three sources for “above surface” slow electron emission can be distinguished, i.e. autoionization of the multiplyexcited hollow atoms on their approach toward the surface, promotion above the vacuum barrier of electrons previously captured by the projectile due to their self-and image-charge shielding near the surface, and finally “peeling-off” of all electrons still bound in highly-excited projectile states until the very moment of surface impact. As exemplified by Fig. 9, the first mechanism (AI) correlates with a relatively broad emission statistic, whereas the second and third ones together (denoted by “promo+peel”) produce considerably narrower distributions, with the experimentally observed ES being the convolution of all three statistical distributions. 6. Measurements involving cluster-ion induced electron emission Whereas the interactions of clusters with photons, electrons and heavy particles have quite actively been

26

F. Aumayr et aLlInt. J. Mass Spectrom. Ion Processes 129 (1993) 17-29

0

5

10

15

20

25

30

35

40

n

Fig. 9. Modeled contributions to the electron emission statistics for the impact of 188eV A?+ on clean polycrystalline gold, in comparison with the corresponding experimentally-obtained overall ES [14].

studied for over a decade, reliable information on clusters colliding with well-defined solid surfaces are still rather scarce. Recognizing this unsatisfactory situation, we have also used our ES method for studying electron emission induced by bombardment of atomically clean metal surfaces with various slow cluster ions. In a first effort [16], atomically clean polycrystalline gold was bombarded by (N2)i (n
the target surface. All measurements were made with an atomically clean target at background pressures of approximately 10v8 Pa. Again, the total electron emission yields were obtained by measuring the related ES. Whereas neither for Ci, nor for (N& any influence of PE could be identified, the “raw” emission statistics for Nez had to be corrected for a small PE contribution, considering that for the impact of singly-charged ions on clean gold, PE can cause emission of one electron only (cf. section 3). The ES corrected in this way are exclusively due to KE [30] and, given their experimentally-demonstrated Poissonian shape, permitted direct evaluation of the corresponding total KE yields [16]. Grossly different yields were found for different cluster species of about equal velocity and mass, as demonstrated in Fig. 10. This is attributed to the widely different partition of the initial cluster kinetic energy into (a) inelastic binary collision processes among cluster constituents and surface atoms, and (b) vibrational excitation (most and least important for C& and Nez, respectively). An energy of up to several electron-

F. Aumayr et al./lnt. J. Mass Spectrom. Ion Processes 129 (1993)

21

17-29

0.6

0

1

2

3

4

impact velocity Fig. 10. Measured

ratios

5

6

7

v (1 O4 ms-‘)

W,/ W, for the emission of respectively two and one electrons, and related (Nz)& and Cg cluster ions, vs. impact velocity [16].

volts per vibrational bond may, at least temporarily, be stored in the cluster and will not be available for the ejection of “prompt” electrons resulting from KE due to close collisions of the cluster constituents with the target atoms. Similar conclusions could be drawn from the variation of the electron yield vs. cluster size. In contrast to Nez cluster ions, for the (N& cluster ions the electron emission per cluster constituent increases with cluster size, which can be explained by the influence of vibrational excitation, as well. Only for the Ne,f cluster ions, the generally assumed “normal” behaviour [3 l] for cluster impact on solid surfaces could be observed in the present study, i.e. a practically linear increase of the total electron yield with both the cluster size and the impact velocity, at least near the KE threshold, which is probably caused by the almost complete destintegration of Nez cluster ions upon surface impact. However, molecular clusters (in particular C6e) heated up due to collisionallyinduced vibrational excitation might either disintegrate and/or give rise to thermionic electron emission [32-341. With our present experimental setup we cannot yet distinguish such thermionically emitted (and thus somewhat “delayed”) electrons

8

total electron

yields for Ne&,,,,

(see, for example, Yeretzian and Whetten [35]) from the “prompt” (i.e. ejected within typically lo-l3 s-‘) KE-related electrons. It should also be emphasized that the present measurements are of direct interest for the important task of quantitative detection of large cluster species, e.g. biomolecules. Summary and outlook The present survey described recent developments in the field of slow ion-induced electron emission from clean metal surfaces, by making use of ES. By measuring such ES, in many cases both a direct evaluation of the related total electron yield and the separation of the contributions to the latter by respectively PE and KE processes can be achieved. Since these ES measurements are readily performed with rather small primary particle fluxes, they are especially well suited for studies involving “exotic” projectile species as, for example, highly-charged ions or cluster ions. Furthermore, ES measurements permit investigations at rather low impact energies, i.e in the vicinity of the kinetic emission threshold, and

28

F. Aumayr et al./Int. J. Mass Spectrom. Ion Processes 129 (1993) 17-29

also evaluation of the PE contribution above the KE threshold. We have studied in some detail the potential emission induced by doubly- and multiply-charged ions from clean polycrystalline gold. Special emphasis has been laid on the lowest achievable impact energies, where for higher initial projectile charge the projectile self-acceleration due to the related image charge causes non-negligible increase in the final impact energy. For multiply-charged projectiles, three principally different processes adding to the observable electron emission can be identified and their role elucidated by accompanying Monte Carlo model calculations based on the classical over-barrier approach. Finally, the ES method has been applied, for the first time, to the study of cluster-induced electron emission from clean gold. This constitutes a new approach to the dynamics of clusters impinging on a surface. For molecular clusters, in particular those of the fullerene type, some information of the partition of the initial cluster kinetic energy into collisional and vibrational excitation has been obtained. In the latter case, a dissipative heating-up of the reflected clusters possibly leading to thermionic electron emission can be expected, and will be investigated by means of further improved ES measurements. Acknowledgments This work has been supported by Austrian Fonds zur Fijrderung der wissenschaftlichen Forschung (in particular by Projekt Nr. P8315TEC), by Kommission zur Koordination der Kernfusionsforschung at the Austrian Academy of Scienftir ces, and by Austrian Bundesministerium Wissenschaft und Forschung, Wien. Regarding the measurements with multicharged ions at GSI Darmstadt, Germany, the authors express their gratitude to Dr. Rido Mann and other staff members of GSI, for providing their excellent help. Experiments involving cluster ion impact have been performed together with Mrs. K. Toglhofer, Mr. H. Kurz and Dr. P. Scheier. Thanks are also

due to Dr. W. Kratschmer, MPI Heidelberg, Germany, for supplying a purified sample containing x 90% C&, and 10% C7e. References 1 H.D. Hagstrum, Phys. Rev., 96 (1954) 325; 336. 2 P. Varga and H.P. Winter in G. Hohler (Ed.), Particle Induced Electron Emission II, Springer Tracts in Modern Physics, Vol. 123, Berlin, 1992. 3 D. Hasselkamp in G. Hohler (Ed.), Particle Induced Electron Emission II, Springer Tracts in Modern Physics, Vol. 123, Berlin, 1992. 4 N.W. Ashcroft and N.D. Mermin, Solid State Phys., CBS, PA, 1976. 5 J.P. Briand, L. de Billy, P. Charles, S. Essabaa, P. Briand, R. Geller, J.P. Desclaux, S. Bliman and C. Ristori, Phys. Rev. Lett., 65 (1990) 159. 6 J. Burgdiirfer, P. Lerner and F.W. Meyer, Phys. Rev. A, 44 (1991) 5674. 7 H.J. Andrii, A. Simionovici, T. Lamy, A. Brenac, G. Lamboley, A. Pesnelle, S. Andriamonje, A. Fleury, M. Bonnefoy, M. Chassevent and J.J. Bonnet, in W.R. MacGillivray, I.E. McCarty and MC. Standage (Eds.), The Physics of Electronic and Atomic Collisions, IOP, Bristol, 1992, p. 89. 8 H. Kurz, K. Toglhofer, H.P. Winter, F. Aumayr and R. Mann, Phys. Rev. Lett., 69 (1992) 1140. 9 W.O. Hofer and U. Littmark, Nucl. Instrum. Methods, 138 (1976) 67. 10 G. Lakits, F. Aumayr and H.P. Winter, Rev. Sci. Instrum., 60 (1989) 3151. 11 G. Lakits, F. Aumayr, M. Heim and H.P. Winter, Phys. Rev. A, 42 (1990) 5780. 12 G. Lakits and H.P. Winter, Nucl. Instrum. Methods Phys. Res. B, 48 (1990) 597. 13 F. Aumayr, G. Lakits and H.P. Winter, Appl. Surf. Sci., 47 (1991) 139. 14 H. Kurz, F. Aumayr, C. Lemell, K. Tijglhofer and H.P. Winter, Phys. Rev. A, (1993) in press. 15 H.P. Winter, Rev. Sci. Instrum., 53 (1982) 1163. 16 K. Tiiglhofer, F. Aumayr, H. Kurz, H.P. Winter, P. Scheier and T.D. Mark, Europhys. Lett., 22 (1993) 597. K. Toglhofer, F. Aumayr and H.P. Winter, Surf. Sci., 281 (1993) 143. 17 E.V. Alonso, M.A. Alurralde and R.A. Baragiola, Surf. Sci., 166 (1986) L155. 18 H.P. Winter, F. Aumayr and G. Lakits, Nucl. Instrum. Phys. Res. B, 58 (1991) 301. 19 K. Ohya, F. Aumayr and H.P. Winter, Phys. Rev. B, 46 (1992) 3101. 20 M. Delaunay, M. Fehringer, R. Geller, D. Hitz, P. Varga and H.P. Winter, Phys. Rev. B, 35 (1987) 4232. 21 H.D. Hagstrum, and G.E. Becker, Phys. Rev. B, 8 (1973) 107. 22 P. Varga, W. Hofer and H.P. Winter, Surface Sci., 117 (1982) 142.

F. Aumayr et al./Int. J. Mass Spectrom. Ion Processes 129 (1993)

23

24

25 26 27 28

P.A.A.F. Wouters, P.A. Zeijlmans van Emmichoven and A. Niehaus, Surf. Sci., 211/212 (1989) 249. H.P. Winter in W.R. MacGillivray, I.E. McCarty and M.C. Standage (Eds.), The Physics of Electronic and Atomic Collision, IOP, Bristol, 1992, p. 475. B. Feuerbacher and N.E. Christensen, Phys. Rev. B, 10 (1974) 2373. H. Brenten, H. Miiller and V. Kempter, Z. Phys. D, 22 (1992) 563. R. Mann, Z. Phys. D, 3 (1986) 85. F. Aumayr, H. Kurz, K. Tijglhofer and H.P. Winter, Nucl. Instrum. Methods Phys. Res. B, 78 (1993) 99.

17-29

29

29 T.D. Mark, Int. J. Mass Spectrom. Ion Processes, 79 (1992) 1. 30 J.W. Rabalais, H. Bu and C.W. Roux, Phys. Rev. Lett., 69 (1992) 1391. 31 M. Kappes and S. Leutwyler, in G. Stoles, (Ed.), Atomic and Molecular Beam Methods, Vol. I. University Press, New York, 1988, Chapter 15 (and references cited therein). 32 S. Maruyama, M.Y. Lee, R.E. Haufler, Y. Chai and R.E. Smalley, Z. Phys. D, 19 (1991) 409. 33 E.E.B. Campbell, G. Ulmer and I.V. Hertel, Phys. Rev. Lett., 67 (1992) 1986. 34 K.R. Lykke and P. Wurz, J. Phys. Chem., 96 (1992) 3191. 35 C. Yeretzian and R.L. Whetten, Z. Phys. D, 24 (1992) 199.