Z. Phys. D 40, 323–326 (1997)

ZEITSCHRIFT ¨ PHYSIK D FUR

c Springer-Verlag 1997

Collision induced fragmentation of small sodium cluster ions M. Barat, J.C. Brenot, H. Dunet, J.A. Fayeton Laboratoire des Collisions Atomiques et Mol´eculaires (Unit´e de Recherche Associ´ee au CNRS No. 281), Universit´e Paris-Sud XI, F-91405 Orsay Cedex, France Received: 4 July 1996

Abstract. The basic mechanisms of collision induced fragmentation of small sodium cluster ions (Na+n n < 9) at keV collision energy are investigated by measuring the velocity vectors of the two fragments employing a new type of coincidence experiment. The results suggest that in most of the cases the lost of one Na atom can be interpreted as relevant of an impulsive mechanism. On the other hand, the lost of one Na+ ion seems to require transition to an electronically excited state of the cluster. PACS: 36.40.Qv

Fig. 1. Experimental set-up

1 Introduction Collision induced fragmentation of metallic cluster ions was essentially investigated by measuring total cross sections and branching ratios for the various fragmentation channels as a function of collision energy [1]. In the present experiment, a detailed study of the fragmentation process is carried out owing to a simultaneous measurement of the velocity vectors of both fragments giving, for example, information on the sharing of the energy transferred by the collision between kinetic and internal energies of the fragments or on their spatial orientation with respect to the collision plane. This technique was first successfully applied to the dissociation of Na+2 ions in collision with an helium target at keV energy [2] revealing the competition between two dissociation mechanisms: – (i) Impulsive Mechanism: (IM) in which a close encounter between the He atom and one of the Na+ cores transfers some momentum to the Na+2 ion resulting in a vibrational excitation and a deflection of the molecule. If the transferred energy is large enough, the Na+2 ion dissociates. – (ii) Electronic Mechanism: (EM) which dominates for collisions at large impact parameters, in which the molecular ion is borne in a dissociative or weakly bound electronically excited state, a mechanism somewhat similar to photodissociation.

The aim of this paper is to present an extension of such an analysis to small ionic sodium clusters. 2 Experimental procedure Only a brief description of the experiment is given here. Sodium clusters are formed in an adiabatic expansion issued from an oven heated at about 800 degrees Celsius. The beam is ionised by 40 eV electron impact and accelerated at keV energies, focused and mass selected in a Wien filter. The cluster beam crosses at 90 deg. a cold target beam issued from a supersonic expansion. As shown in Fig. 1, the neutral and ionic fragments are separated in a parallel plate electrostatic selector. The neutral fragments fly in straight line through the analyser and reach a position sensitive detector (PSD) from which the location Yn Zn and the arrival time Tn are determined. The deflected ions are received on a second PSD from which the same kind of information (Yi Zi Ti ) are obtained. Two types of experiments have been performed depending of the beam intensity. The ‘complete’ experiment requires the beam to be chopped at a frequency of typically 1 MHz allowing the full determination of the velocity vectors of the fragments. Unfortunately 99% of the beam is lost in the chopping procedure, therefore for low intensity beams an unchopped beam is used allowing to mea-

324

sure only the difference of arrival times of the two fragments in addition to their spatial localisation. This insures that the two fragments come from the same collision. In this type of experiment that we call the ‘ZZ correlation’, the significant information comes from the spatial correlation of the fragments as explained below. 3 Vectorial analysis of Na+3 and Na+4 fragmentation In such multiparametric experiment, it is important to select the most interesting types of data to be correlated. A significant choice is provided by the representation (Fig. 2), as a contour map, of the fragmentation intensity as a function of the relative kinetic energy of the fragments Erel and of the scattering angle χ, the deflection of the cluster centre of mass (CM). For Na+2 , two groups of structures show up [2]: structures (I) and (II) at very small χ values are attributed to dissociation of excited states of Na+2 via an EM mechanism. The third structure (III) is very different appearing at large scattering angle with Erel values increasing with χ. This third process is attributed to the IM mechanism. Notice that the (Erel , χ) dependence of structure (III) is well reproduced by a simple model which just consider the energy transferred in the collision of the He target with a single Na atom [2]. The same type of correlation maps are shown at similar values of the relative collision energy for the Na+3 + He ⇒ Na+2 + Na + He Na+4 + He ⇒ Na+3 + Na + He dominant fragmentation channels which are unambiguously assigned. As compared to the Na+2 results, the first striking feature is the absence of EM mechanism at small χ angles. For both systems, the only significant structure appears as a peak stretching towards the large χ values with Erel slightly increasing with χ. This result showing a large deviation of the CM of the dissociating fragments suggests that an IM mechanism mediates the population of these channels. The curves Eimp (χ) corresponding to the energy transferred during a binary He Na elastic collision are drawn on the same graphs. As for the Na+2 case, the contours follow the Eimp (χ) curve specially for the largest energy transfers (Erel > 0.4 eV) consistently with the model. Notice however that the major contribution, specially for Na+4 , comes from the smallest Erel values for which the corresponding contours do not follow the Eimp (χ) curves. This behaviour indicates that a significant part of the transferred energy is converted into internal energy of the ionic fragment.

Fig. 2. Complete analysis of the fragmentation of Na cluster ions at 80 eV CM collision energy. Left hand side: correlation contour maps in the frame of the CM scattering angle χ and the relative kinetic energy of the fragments Erel (χ) coordinates. All contour levels show logarithmic intensity scale with a factor 1.6 between each contour. From top to bottom: Na+2 + He ⇒ Na+ + Na + He , Na+3 + He ⇒ Na+2 + Na + He , Na+4 + He ⇒ Na+3 + Na + He. The solid curves represent Eimp (χ) given by the impulsive model for the two limiting values of the internal energy of the cluster: Eint = 0 and Eint equals to the dissociation energy Ed . The dashed curve is for Eint = E∗ (see text). Right hand side: Φ angular distribution for the same reactions. Notice that for Na+2 the distribution corresponds exclusively to structure III

4 The ZZ correlation Na+2

dissociation, it was shown that fragmentaFor tion dominantly occurs around the angles Φ = 0 and 180 deg. Such dissociation in the collision plane was an additional test of the validity of the IM mechanism. By convention, Φ = 0 and 180 deg. correspond to the smallest and largest deviation of the ionic fragment respectively. Notice that the symmetry of the Na+2 system is responsible for the double distribution in Φ as observed with the present accuracy. For Na+3 and Na+4 , the same conclusion is reached but the major contribution comes from Φ = 0 deg, indicating that the neutral Na fragment is side scattered.

The ‘ZZ correlation’ technique consists in displaying the correlation between the Z components (vertical in Fig. 1) of the positions of both fragments on the detectors. The Z components have been chosen because that of the ionic fragments is not affected by the electric field inside the analyser. This representation which gives, with a good approximation, the correlation between the VZ components of the fragment velocities allows to estimate the deviation of the CM of the fragments, as sketched in Fig. 3. Such a type of correlation is shown for the Na+3 + He ⇒ Na+2 + Na + He channel in order to

325

Fig. 4. Cross sections for the production of Na+n−1 and Na+n−2 fragments at 2 keV collision energy. Full squares: experimental results. Open squares: model calculations with an initial internal energy E∗ . Open circles: model calculations with E∗ = 0. Experimental data have been arbitrarily normalised to the calculations for Na+6 ⇒ Na+5

Fig. 3. ‘ZZ correlation’ contour maps between the vertical (Z) positions of the ionic and neutral fragments for two fragmentation channels. The symbols (+) and (-) indicate a deflection above and below the incident beam direction respectively. The dashed lines, with a slope equal to the ionic to neutral mass ratio, correspond to a zero deflection of the CM. The relationship between the ‘ZZ correlation’ and the CM angular deflection is schematised for the two types of fragmentation mechanism

check the consistency of the two experimental methods. Indeed it is readily seen that the two velocity vectors are found on the same side of the incident beam, indicating a net deflection of the Na+3 CM, a signature of a close encounter, consistently with the results of the ‘complete’ analysis discussed above. A fully different pattern is found for the other fragmentation channel Na+3 + He ⇒ Na+ + Na2 + He. As seen on Fig. 3, the two fragments are found on each side of the axis of the incident beam indicating a negligible deflection of the CM as compared to the previous channel and despite a largest energy defect. Such result speaks in favour of an EM mechanism as found for the Na+2 dissociation via the population of electronically excited states. Similar ZZ patterns are observed for the Na+4 fragmentation. From these examples, one can infer that fragmentation giving Na+ + Nan−1 fragments involves dissociation of excited states of the cluster ion. We assumed that these channels should primarily correspond to the lost of one Na2 dimer in the Na+3 dissociation for example, a channel less endothermic than that corresponding to the production of two Na atoms. In fact, we searched

for channels giving 3 and more fragments by looking at two successive hits on the neutral detector and corresponding to the same event. In these preliminary attempts only the localisation of one hit can be achieved within the 5 µs dead time of the position encoding. However, double hit can be identified by the time channel of our device provided that the second atom hits the detector at least 20 ns later. This time has to be compared with a typical 200 ns time of flight distribution of the fragments. Preliminary experiments did not show such triple fragmentation. However works are in progress to improve these measurements.

5 Fragmentation cross sections The relative fragmentation cross sections have been determined as the ratio between the intensity of the ionic fragments Na+n−p and that of the incident beam Na+n . Since the ions are post-accelerated at 3 keV and detected on the same PSD, the difference in the detection efficiency between the incident ions and the fragment ions should not significantly influence these relative measurements. The results are presented on Fig. 4 for dissociation of 2 keV Na+n (n < 9) cluster ions into Na+n−1 and Na+n−2 fragments. These data are compared with a simple model based on the IM mechanism as suggested by the results of the ‘complete’ experiment for the smallest clusters and by the ‘ZZ correlation’ for the largest ones. In this calculation, for each θ scattering angle in the He Na CM frame, the energy transferred Eimp (θ) is determined

326

by the elastic scattering of the Na core in a binary Na+ He collision. Eimp (θ) is transformed into internal energy of the cluster. The internal energy is then given by the sum of Eimp (θ) and E∗ , the initial internal energy of the cluster ion as estimated from energy loss measurements of unfragmented cluster ions [3]. Assuming a statistical redistribution of the internal energy, the relative fragmentation probability can be estimated in the framework of the RRK theory as modified by Engelking [4, 5]. For each θ angle, this probability is weighted by the Na+ He elastic differential cross section computed with the exponential potential derived by Kita et al [6] and integrated from θmin , the angle corresponding to the minimum energy transfer needed for this fragmentation channel. The endothermicities are taken from [5] for the monomer and dimer fragmentation and are derived from the total cluster energy [7] for the other channels. The total cross section for the production of a fragment Na+n−p then obtained is multiplied by n to account for the number of Na atoms in the cluster: Z π Σ(θ)Kn,p (θ)sinθdθ σT (n, p) = 2πn Pn−1 θmin i=1 Kn,i (θ) These calculations compared to the experimental data in Fig. 4 call for the following remarks. The odd-even oscillations in the n dependence of the cross sections, well reproduced by the model, just reflect the fragmentation endothermicities. More interestingly, the Na+n−2 channel is also reasonably well reproduced speaking in favour of a mechanism in which the initial energy transferred to one atom is then redistributed among the others Na atoms. Obviously the dimer formation in Na+3 fragmentation via the EM mechanism as suggested by the ‘ZZ correlation’ cannot be accounted for by a calculation based on the IM model. Finally the huge cross section for the monomer formation from Na+8 is not reproduced by the model calculations.

6 Conclusion The present dynamic study based on the multicoincidence detection of the fragments gives strong indications on the mechanisms of collision induced fragmentation of metallic clusters. An impulsive mechanism should be responsible for the dominant channels, in particular for the monomer and dimer production. On the other hand, the same type of analysis suggests that the most endothermic channel leading to Na+ production is induced by an electronic transition. A deeper investigation of the internal energy redistribution between kinetic and internal energy of the fragments is still needed. In the future, our efforts will also be directed towards the research of multiple fragmentation. This work was supported by the E.U.Human Capital and Mobility Program through the Collision Induced Cluster Dynamics Network under contract number CHRX-CT-940643

References 1. see e.g. D.A. Hales, C.X. Su, Li Lian and P.B. Armentrout, J. Chem. Phys. 100, 1049 (1994) 2. J.C. Brenot, H. Dunet, J.A. Fayeton, M. Barat and M. Winter, Phys. Rev. Lett. 77, 1246 (1996) 3. J.A. Fayeton et al., to be published 4. P.C. Engelking, J. Chem. Phys. 87, 936 (1987) 5. C. Br´echignac, Ph. Cahuzac, J. Leygnier, and J. Weiner, J. Chem. Phys. 90, 1492 (1989) 6. S. Kita, K. Noda and H. Inouye, J. Chem. Phys. 63, 4930 (1975) 7. D. M. Lindsay private communication quoted in J.Ph. Roux, Th`ese, Universit´e Paris Sud (1988)