METASTABLE DECAY AND COLLISIONS OF SPUTTERED METAL AND SILICON CLUSTER IONS

(1)

HAL Id: jpa-00229421

https://hal.archives-ouvertes.fr/jpa-00229421

Submitted on 1 Jan 1989

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are published 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.

METASTABLE DECAY AND COLLISIONS OF SPUTTERED METAL AND SILICON CLUSTER

IONS

W. Begemann, K. Meiwes-Broer, H. Lutz

To cite this version:

W. Begemann, K. Meiwes-Broer, H. Lutz. METASTABLE DECAY AND COLLISIONS OF SPUT- TERED METAL AND SILICON CLUSTER IONS. Journal de Physique Colloques, 1989, 50 (C2), pp.C2-133-C2-139. �10.1051/jphyscol:1989223�. �jpa-00229421�

(2)

METASTABLE DECAY AND COLLISIONS OF SPUTTERED METAL AND SILICON CLUSTER IONS

W. BEGEMANN, K.H. MEIWES-BROER and ^H.O. ^LUTZ

Fakultdt fiir Physik, Universitdt Bielefeld, 0-4800 Bielefeld 1, F.R.G.

RQsumB-Les agrkgats ioniques, positifs ou nhgatifs, sont obtenus par bombardement d'une surface par des ions de gaz rares. Une corrBlation grossibre entre le sputtering yield Y et le nombre maximum d'atomes n dans l'agrBgat produit est mise en Bvidence. La largeur en Anergie cinBtique des ions formBs d6croit avec n, tout en restant en gBneral plus importante que celle des agr6gats neutres ionisks. La dissociation unimolkculaire qui est dominante conduit B des anomalies en intensite dans les spectres de masse des agrkgats. Les taux de dissociation dkpendent fortement du temps et mettent en 6vidence des diffkrences entre les 6nergies de liaison des agrAgats. La distribution en masse des fragments produits par collision d'agrkgats sur une cible de gaz rare est trbs lide aux taux de dissociation.

Abstract-Cluster ions and anions are produced by rare gas ion bombardment of surfaces.

A rough correlation between sputtering yield Y and maximum number of atoms in the cluster n is found. Kinetic energy spreads decrease with n, generally being larger than those of ionized neutrals. Strong unimolecular decay leads to intensity anomalies in cluster mass spectra. Decay rate constants are strongly time dependent and reveal differences in cluster binding energies. Fragment patterns resulting from keV cluster collisions in a rare gas target turn out to be strongly related to the decay rates.

Sputtering by ion beam bombardment of solid surfaces is one means for the production of neutral, positively and negatively charged clusters even from materials with high melting temperature. Ion-induced erosion is ltnown since 1853 / I / and is now explained by collision cascades originating from the impacts of single primary ions 12-41. Theoretical concepts are available to calculate the average number of emitted particles per projectile atom (sputtering yield Y) for single elemental amorphous and polycristalline targets /3,5/; Y depends mainly on the projectile-target atom mass ratio, projectile energy, angle of incidence and surface binding energy. In addition, the effect of target cristallinity upon Y is understood as well /4,6/. The emission of charged particles and clusters, on the other hand, still needs extended experimental and theoretical effort. All au- thors agree that a cluster stems from atoms of a collision cascade excited by a single projectile.

Disagreement exists, however, whether entire clusters leave the surface with the atoms originating from neighbouring lattice sites /7,8/, or whether cluster formation is a statistical recombination process of independently sputtered atoms from one collision cascade /7,9-121. None of the models could so far have been definitely supported by experiments.

Common features of clustering models are a decrease with n of widths in translational energy distributions and in the intensities of sputtered clusters. This behaviour has been verified in experiments and led to the conclusion that mainly the low energy parts of the collision cascades contribute to cluster emission 18,131. For a further understanding of the emission process detailed cluster properties like binding energies or ionization potentials have to be regarded. In addition, special care has to be talten on the fate of the clusters between emission from the surface and detection, i.e, the decay behaviour of excited clusters may influence the interpretation of, e.g.,

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1989223

(3)

C2-134 JOURNAL DE PHYSIQUE

energy distribution measurements. Only a few experimental investigations of cluster decomposition after sputtering are published /14-17/. In most cases monomer evaporation is dominant, although fission is also possible /16/'

In this communication we will inslinly concentrate on the decay behaviour of charged clusters after production by sputtering, or after collisions in a rare gas target.

The experimental set-up has been described in detail elsewhere 1171. In short, cluster ions are sputtered by Xe+ bombardment (10-25 lteV,

-

¹⁰⁰^,uA/cm2)of bulk target material, see Fig. 1.

Ion Sources 10 or 25 keV X e '

Wien Filter Gas Cell

\

L e n s System Target

TOF-Reflectron

e-

Det. l

Det. 2

Grid System

Fig, 1-Experimental set-up for the investigation of energy distributions, the metastable decay and collisions of sputtered cluster ions

After accelerating the charged particles over a preselected distance (1.7 or 18.5 mm) a Wien filter (COLUTRON 600 B) is used for velocity and mass selection. Cluster fragmentation during acceleration leads to a misfit of the velocity and to the fragment's deflection. Those clusters, however, which originate from a fragmeritation process after the acceleration is completed, possess, though lower energy, still their parent's velocity, and are also transmitted through the Wien filter.

The experiment is, therefore, sensitive to fragmentation of clusters at times greater than the acceleration time t o . By means of energy analysis in front of detector 1, or by employing the time- of-flight reflectron (detector 2), the fragment masses and intensities are determined. The energy analyzer in front of detector 1 is also used for kinetic energy determination of sputtered cluster ions.

3-Results a n d Discussion A. Sputtering of cl~uster ions

Positive and negative cluster ions of numerous materials have been sputtered and analyzed by mass spectrometry. In some cases the clusters contain more than 100 atoms (e-g., Cu, /15,18/), with other materials only small clusters are found (e.g., Si,, n

5

11 1191. Table 1 compiles for some selected materials the maximum atom numbers nmaz of cluster ions detected with some significance. The low n,,,,, of

Si;

corresponds to a low value of Y, while the high nmaz of Cuz corresponds to a high Y; values for Al;,

Mo;,

and

W$

are in between. nmaz of Sin and

Cu,

is independent of the ion's polarity. This comparison indicates a rough correlation between Y and n,,, but shows also the limit of the significance of Y on the cluster ion emission. Evidently, Y must show strong statistical fluctuations as, e.g., in the case of copper, clusters with more than 100 atoms are found while the corresponding average sputtering yield is only 15. We note that

(4)

Table 1-Maximum number of atoms n,,, in sputtered positively charged cluster ions (20 keV Xe+, 100 ,uA/cm2). In addition, theoretical and experimental sputtering yields Y are given for 20 lceV Xe+ bombardment /4/.

the employed primary ion density is too low to allow a correlated action of two different projectile ions.

3 11-15

7 7 Aln

Si, Gun

Man

Wn

n,,,, has been measured for different Xe+ fluxes and energies. The values given in table 1 are

obtained from mass spectra under ^cleansurface conditions at a target chamber pressure of lo-' mbar. Target contamination usually yields a lower ^n,,,. Higher detection efficiency, on the other hand, could possibly slightly enhance n,,,.

To determine the translational energy of mass selected clusters a repeller grid energy analyzer is employed in front of detector 1 (see Fig. 1). Fig. 2 gives the normalized ion intensities of selected silicon clusters as function of the repelling voltage.

11

>

100 x 30

.% 20

ENERGY ANALYSIS OF

I

REPELLER VOLTAGE (V)

Fig. 2-Energy analysis of Wien filter-selected sputtered

Si;

(n=1-9). The acceleration (target) voltage is marlced by an arrow. Intensity above 1.8 lceV originates from ions with excess energy due to the sputtering process.

Cluster ions of all sizes have significant intensity at energies well beyond the acceleration energy of 1.8 keV. As expected, the monomer and dimer excess energies E,, are high and extend to about 100 eV. The trimer and the larger cluster ions are repelled at about 1840 eV, i.e., E,,240 eV.

For

Si,

to

Si,

we find identical curves. Their tails extend to energies below the target potential which is a sign of fragmentation during acceleration, close to the surface. The energy spreads are thoughout larger than those which have been measured for positive silicon clusters, sputtered by 4 le\J ^~ ^r /8/. ⁺ Other systems like W$ show similar broad energy distributions shifting to lower energies with increasing n. As a matter of fact, however, such energy distributions suffer from contributions from strong decay processes directly at the surface. Consequently the measured spectra do not only originate from clusters which are formed as such in the sputteri~lg process.

(5)

C2-136 JOURNAL DE PHYSIQUE

B. Metastable decay of sputtered clusters

The fragment ion content in the beam is determined by energy analysis either with the repeller grid or the reflectron, see Fig. 1. In each case fragmentation is observed in a well defined time window which opens after the acceleration time to. By variation of the acceleration distance, target potential and the cluster mass, to ranges from 50 ns to 3.5 ps; overall drift times from 0.5 ps to 250 ^/ISaIe employed. Resulting decay rates for sputtered A l i , Si: and C u i have been published recently 116,171. Sputtered cluster ions for n

<

3 turn out to be stable within the expriment's time window. Larger clusters decompose with rates which are strongly dependent on n. A close correlation between decay rates and intensity anomalies in mass spectra is found thus proving that the shapes of sputtered metal cluster intensity distributions largely originate from decay processes.

The dominant decay process is the evaporation of a (neutral) atom.

A theoretical description of unimolecular cluster decay is given by the evaporative ensemble model of Klots 1201. It is based on the stabilization of excited clusters by evaporation of neutral monomers. An evaporative ensemble consists of the n-clusters formed directly in a production process, as well as those of the same size created as fragment ions from an n f l parent cluster immediately after leaving the surface. Therefore, clusters are characterized by a temperature range or internal energy band width roughly as large as the energy release due to evaporation.

Consequently, clusters of a defined size have different lifetimes which depend on the time between production and analysis. The resulting decay rates are calculated to increase-for a given time window-linearly with n. Such a behaviour is in rough agreement with the experiment 116,171.

To investigate cluster decomposition as close as possible to the production event we studied the diEerence in decay processes between the acceleration over distances of 1.7 mm and 18.5 mm. In order to get a rough estimate of the cluster-specific decay rate constants

K,

(to, At) an exponential decay is assumed. Such values will depend on to and the time window length (until detection) At.

Although the evaporative ensemble is not characterized by one temperature, K , (to, At) as an average over the evaporative ensemble will reveal differences in the behaviour of, e.g., sputtered A l i and Cuk (Fig. 3). Note that a direct comparison to other experiments on cluster decay is only possible when the same to and At is employed.

The Cu: decay rates are governed by a pronounced even-odd oscillation. Maxima in Kn(to, At) are related to minima in mass spectra, and vice versa. For Al: pronounced steps show at n=8 and n=15. According to the jelliurn model 1211 such steps are in agreement with the enhanced binding energies of Al:, n=7,14, and of C u z , n=9,21. These findings clearly prove that sputtered metal cluster ions undergo substantial fragmentation. As a consequence cluster properties which are measured some time after production are not necessarily related to the sputtering process.

E.g., tllc energy distributions of Si; (Fig. 2) contain contributions from larger clusters which decomposed shortly after production.

C. Collision-induced processes

Cluster formation by sputtering is determined by the aggregation as well as charge exchange processes near the surface. Those combined processes malce a theoretical description very difficult.

Therefore, in order to study cluster excitation and subsequent decay under defined conditions we investigate 1.8 lteV collisions of mass-defined metal and silicon clusters in a rare gas target. The geometry of the experiment selects mainly low angle scattering processes where cluster excitation energies remain typically below 10 eV. Resulting fragment patterns show the dominant appearance of distinct fragment masses 1191 and opens a way to determine relative ionization potentials 122,231 and electron affinities 1231. Integrated fragment intensities reveal the collision-induced fragmentation behaviour of the parent clusters. Fig. 4 gives such normalized fragment intensities

I(C

Fray)/Io of 1.8 lceV Cu; in an argon gas target.

The lowest trace represents fragment intensities under single collision conditions. A pronounced even-odd alternation is seen where the higher fragment intensities appear for even n.

(6)

depending on n), and of the time window length At (here: l p s

<

At

<

2 . 5 ~ ~ ) . The strong variations in I<, (to, At) originate from binding energy differences.

CIF of 1.8keV Cu; in Ar T

v 1.10-'rnbar v\/

o 5 - 1 0 - ~ ,"---v

Fig. 4-Integral fragment intensities I(C Frag) of 1.8 keV Cu$ collisions in argon, normalized to the sum of all ions transmitted. Different target gas pressures are employed.

Note that the unimolecular decay has the same n-dependence, c.f. Fig. 3. Therefore, collision- induced fragmentation at these energies appears to be initiated by cluster excitation and followed by energy randomization. Subsequent decay is then governed by rate constants which show the same n-dependence like those measured for the sputtering process. Consequently, after energy randomization the cluster forgets its history and decomposes according to the rules of unimolecular decay. Non-resonant photofragmentation (PF) would probably give a similar result; e.g., in the

(7)

C2-138 JOURNAL DE PHYSIQUE

case of potassium cluster ions n

>

5, P F decay rates are qualitatively similar to those shown in Fig. 3 1241.

Multiple collisions at higher argon densities result in fragment cascading. As a result the relative amplitudes of the even-odd alternation are partly washed out, see Fig. 4. For a determination of absolute fragmentation cross sections their pressure dependence has to be regarded and extrapolated to zero pressure. In this manner collision-induced fragmentation cross sections of 1.8 lceV Cu, (2

5

n

5

9) in argon have been measured to range from 5 to 80

A2

1191.

Sputtering of bulk surfaces is a general means for the production of neutral and charged clusters. The maximum number of atoms n,,, of metal and semiconductor cluster ions depends strongly on the' material; a rough correlation between sputtering yield Y and n,,, is found with n,,, being thoughout larger than Y. Cluster ion trandational energy distributions extend to high energies. For a detailed understanding of the emission process, however, such measurements are of limited value since cluster fragmentation might prohibit unique energy determinations. Decay rates of sputtered cluster ions are strongly size and material dependent. Therefore, a thorough theoretical picture of cluster emission has to include the decomposition behaviour.

We thanlc J. Tiggesbiiumker for helpful discussions. This work has been supported by the Deutsche Forschungsgemeinschaft (DFG).

References

/ I / Grove, W.R., Phil. Mag. 5 (1853) 203.

/2/ Thompson, M.W., Phil. Mag. 18 (1968) 377.

/3/ Sigmund, P., Phys. Rev. 184 (1969) 383 and 187 (1969) 786.

/4/ Sputtering by Particle Bombardment I, Ed. R. Behrisch, Topics in Applied Physics, Vol. 47, Springer, Berlin 1981.

/5/ Sigmund, P., in Ref. 4.

/6/ Hofer, W.O., Gnaser, H., Nucl. Instr. Meth. Phys. Res. B18 (1987) 605.

/7/ Staudenmaier, G., Rad. Effects 18 (1973) 181.

/8/ Wittrnaaclc, K., Phys. Lett. 69A (1979) 322.

/9/ Oechsner, H., Gerhard, W., Surf. Sci. 44 (1974) 480.

/lo/

Gerhard, W., 2. Phys. B22 (1975) 31 and

Gerhard, W., Oechsner, H., 2. Phys. B22 (1975) 41.

/11/ Konnen, G.P., Tip, A., de Vries, A.E., Rad. Effects 2 1 (1974) 269 and 26 (1975) 23.

/12/ Winograd, N., Harrison, D.E., Garrison, B.J., Surf. Sci. 78 (1978) 467.

1131 Herzog, R.F.K., Poschenrieder, W.P., Satkiewicz, F.G., Rad. Effects 18 (1973) 199.

/14/ Ens, W., Beavis, R., Standing, K.G., Phys. Rev. Lett. 50 (1983) 27.

1151 Katalcuse, I., Ichihara, T., Fujita, Y., Matsuo, T., Sakurai, T., Matsuda, H., Int. J. Mass Spectrom. Ion Processes 67 (1985) 229.

/16/ Begemann, W., Meiwes-Broer, K.H., Lutz, H.O., Phys. Rev. Lett. 56 (1986) 2248.

/17/ Begemann, W., Dreihofer, S., Meiwes-Broer, K.H., Lutz, H.O., 2. Phys. D 3 (1986) 183.

/18/ Begemann,

W.,

Dreihofer, S., Meiwes-Broer, K.H., Lutz, H.O., in Physics and Chemistry of Small Clusters, NATO AS1 Ser. B 158 (1987) 269.

/19/ Begemann, W., Dreihofer, S., Gantefor, G., Sielcmann, H.R., Meiwes-Broer, K.H., Lutz, H.O.

in Elemental and Molecular Clusters, Eds. G. Benedelc, T.P. Martin, G. Pacchioni, Springer Series in Materials Science 6, Springer Verlag Berlin (1988) 230.

1201 Klots, C.E., J. Chem. Phys. 83 (1985) 5854;

Klots, C.E., 2. Phys.

D

5 (1987) 83.

/21/ a) Martins, J.L.,

Car,

R., Buttet, J., Surf. Sci. 106 (1981) 265.

b) Elcardt, W., Ber. Bunsenges. Phys. Chem. 88 (1984) 289.

(8)

/22/ Jarrolcl, M.F., Bower, J.E., ICrauss, J.S., J. Chem. Phys. 86 (1987) 3876.

1231 Begernann, W., Hector, R., Liu, Y.Y., Tiggesb&uiumker, J., Meiwes-Broer, K.H., Lute, H.O., 2.

Phys. D, to be published (1988).

/24/ Brdcllignac, C., Cahuzac, Ph., Roux, J.-Ph., Pavolini, D., and Spiegelmann, F., J. Chem.

Phys. 87 (1987) 5694