20 July 2001

Chemical Physics Letters 342 (2001) 536±544

www.elsevier.com/locate/cplett

Unusual temperature dependence in dissociative electron attachment to 1,4-chlorobromobenzene A. Rosa a, W. Barszczewska a,b, D. Nandi a,c, V. Ashok a,c, S.V.K. Kumar a,c, E. Krishnakumar a,c, F. Br uning a,d, E. Illenberger a,* a

Institut f ur Chemie, Physikalische und Theoretische Chemie, Freie Universitat Berlin, Takustrasse 3, D-14195 Berlin, Germany b Department of Chemistry, University of Podlasie, 08-110 Siedlce, Poland c Tata Institute of Fundamental Research, Colaba, Mumbai 4000 005, India d Department of Nuclear Medicine and Radiobiology, Ion Collision Laboratory, University of Sherbrooke, Sherbrooke, Qu e., Canada J1H 5N4 Received 23 February 2001; in ®nal form 22 May 2001

Abstract Electron attachment to 1,4-chlorobromobenzene yields the ionic fragments Cl and Br appearing in the energy range 0±1 eV with Br by about a factor of 70 more intense at room temperature. Increasing the temperature of the target molecule up to 540 K results in an unexpected temperature dependence of the cross-section in that it ®rst increases and after passing a maximum (400 K for Br and 460 K for Cl ) decreases. This behaviour is qualitatively interpreted by the temperature dependent population of the relevant normal modes containing the C±Cl and C±Br stretch vibration. Ó 2001 Elsevier Science B.V. All rights reserved.

1. Introduction Electron attachment processes can exhibit strong temperature e€ects not only with respect to both the value of the cross-section but also its dependence from the primary electron energy [1±4]. Such data are of direct relevance in any electron collision problem where molecules at varying temperatures are present (gas discharges, plasma processing, etc.). In addition, the temperature dependence re¯ects the thermodynamics of the particular dissociative attachment (DA) reaction and also gives some insight into the underlying mechanisms and dynamics. In the

*

Corresponding author. Fax: 49-30-83856612. E-mail address: [email protected] (E. Illenberger).

present contribution we study DA to 1,4-C6 H4 BrCl yielding the competitive DA channels Br and Cl within a low energy resonance between 0 and 1 eV. By increasing the temperature of the target molecule from 300 to 540 K we observe: (i) a characteristic change in the shape of both DA ion yield curves, and (ii) an unusual behaviour of the total DA cross-section for both DA channels in the way that the cross-section ®rst increases with temperature and after passing a maximum decreases with the maximum for the Br channel located at lower temperature. To our knowledge such a temperature dependence has not been observed before. Associative and dissociative electron attachment can take place if a neutral molecule M in the gas phase captures a free electron to form a transient negative ion (TNI) [5,6]

0009-2614/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 9 - 2 6 1 4 ( 0 1 ) 0 0 6 4 3 - 1

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

e ‡M!M

#

…1†

which subsequently evolves along the following pathways: M

#

! M…† ‡ e

…2a†

!R‡X ! M ‡ energy

…2b† …2c†

with channel (2a) representing autodetachment (AA) recovering the neutral molecule, (2b) dissociation into stable fragments and (2c) relaxation into the thermodynamic stable con®guration provided that the target molecule possesses a positive adiabatic electron anity. Electron capture processes are often studied by means of mass spectrometric techniques and from that the term associative electron attachment (AA) simply assigns those attachment processes in which the parent negative ion M is detected within the observation time window of the particular instrument [5,6]. This is the case if: (i) the autodetachment channel (2a) is suciently slow (i.e., in the ls range) so that the thermodynamically unstable ion M # survives the ¯ight time from the interaction region to the detector or (ii) relaxation processes (collisional stabilization, radiative cooling (2c)) are suciently operative to withdraw energy from the initially formed compound state to a level below the detachment threshold. In systems where DA channels are energetically accessible from the attachment resonance (M # ) the parent anion is usually not observed (like in the present system) since DA generally occurs on a time scale much below the ls domain. It has been known since more than three decades that the temperature of the target molecule can have a strong e€ect on DA. For the oxygen molecule it was shown that the cross-section for O formation continuously increases by a factor of 3 when the gas temperature is increased from 300 to 3000 K [7]. At the same time the maximum of the DA cross-section curve is shifted from 6.4 to 4.6 eV. This shift is much more than the thermal energy of the hot molecule (kT (3000 K) ˆ 0.26 eV). For the molecule CF3 Cl the e€ect was even more pronounced [8]: by increasing the tempera-

537

ture from 300 to 800 K the DA cross-section for the Cl channel increases by a factor of 6 with a continuous downshift of the resonance energy and the appearance of a `0 eV' peak (threshold peak) at temperatures above 350 K. In a semiclassical model this behaviour was quantitatively explained by the energy dependent competition between DA and AA in the decay of the TNI along a repulsive potential energy surface [9±11]. So far temperature dependent DA experiments have been carried out by means of electron beam [2,3] and also electron swarm experiments [4]. The temperature dependence for associative attachment is less systematically studied but it appears that the cross-section for AA decreases with temperature [1]. This is also expected as the autodetachment lifetime should decrease with the internal energy of the ion, irrespective whether DA channels are accessible or not. The presently studied system 1,4-bromochlorobenzene possesses a low energy resonance in the energy range 0±1 eV which dissociates into the competitive DA channels C6 H4 ClBr

#

! C6 H4 Cl ‡ Br

…3a†

! C6 H4 Br ‡ Cl

…3b†

with (3a) by about a factor of 70 more intense than (3b) at room temperature. The observed temperature dependence is interpreted in the framework of two-dimensional potential energy curves with the curve crossing between the neutral and relevant anionic state close to the minimum of the neutral con®guration. The cross-section behaviour is then qualitatively explained by the temperature dependent population of the involved vibrationally excited states.

2. Experimental Fig. 1 shows a schematic of the electron attachment spectrometer which has been described earlier [12]. In brief, an electron beam of well de®ned energy (energy resolution 0.1 eV fwhm) is crossed with a molecular beam e€using from a capillary which is connected to an oven. The oven is usually used to evaporate solid samples, in the

538

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

Fig. 1. Schematic of the experimental setup to study the temperature dependence of the cross-section for dissociative electron attachment to 1,4-C6 H4 ClBr.

present case it is connected to the gas inlet system containing the dosing valve. At room temperature 1,4-C6 H4 BrCl is a solid but with sucient vapour pressure that it can be introduced from outside into the vacuum chamber. By passing the heated oven, the molecules undergo many collisions with the walls (>1000) to attain thermal equilibrium. The temperature is controlled by two Pt(1 0 0) resistance thermoelements, one mounted directly at the oven and the other on the base electrode for the ion extraction ®eld which itself carries the capillary for the gas entering the reaction zone (Fig. 1). Irrespective of the heating time there is always some temperature gradient between the oven and the electrode which involves some uncertainty for the actual temperature of the molecules leaving the capillary and entering the collision zone. The assigned temperature refers to the average between the two thermoelement readings which rises continuously up to 28 K for the highest temperature. The electron beam is generated by a trochoidal electron monochromator (TEM) which uses a crossed electric and magnetic ®eld for energy dispersion [13]. The ions are extracted from the interaction area by a small draw-out-®eld (1 V cm 1 ), analyzed by a quadrupole mass ®lter and detected by single pulse counting techniques. All temperature dependences were recorded at a ®xed position of the

dosing valve (located outside the vacuum system at room temperature) which ensures a constant number of molecules per unit time entering the system. The observation is that the reading at the ionization gauge (which measures the number density of molecules at one of the ¯anges outside the heated area) slightly increases with the temperature of the oven. This is due to a decreasing number density in the heated area in favour of an increased density outside of it. If we assume free molecular ¯ow conditions (which is not rigorously the case as the collision region is limited by the two electrodes de®ning the extraction ®eld) the gas density varies as n / T 1=2 in the e€usive molecular beam. We have corrected the ion count rates according to the varying gas density. The electron energy scale is calibrated by a small admixture of SF6 (<1%) yielding the well known SF6 resonance near 0 eV and also the DA signal SF5 which strongly increases with temperature as studied by di€erent laboratories [5,14]. The SF6 system hence serves for energy calibration and also to control the overall stability of the system under varying temperatures. We do here, however, not explicitly report the temperature dependencies of the SF6 and SF5 anion yields since these results are the subject of a forthcoming publication [15]. 3. Results and discussion Figs. 2a,b and 3 show the ion yield curves and their temperature dependences for the two DA channels Br and Cl , respectively, observed from 1,4-C6 H4 BrCl. The spectra were recorded for a ®xed position of the dosing valve and a pressure of 1:3  10 6 mbar (ionization gauge reading at 300 K). The actual pressure in the reaction zone is by 2±3 orders of magnitude higher [16]. The count rates refer to absolute numbers (counts per second) which have been multiplied by the factor 1=2 …T =T0 † …T0 ˆ 300 K† to correct for the varying gas density with temperature in the collision zone. It should be noted that the undissociated ion could not be observed under the present experimental conditions and negative ion formation (studied up to 20 eV) was exclusively restricted to the 0±1 eV

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

539

Fig. 2. (a) Temperature dependence of the Br =C6 H4 ClBr DA cross-section in the range from room temperature to 385 K; (b) temperature dependence of the Br =C6 H4 ClBr DA cross-section in the range from 404 to 541 K.

energy region. This contrasts to the previously studied perhalogenated benzenes C6 F5 X, (X ˆ Cl, Br) [17,18] where intense formation of parent anions near 0 eV in competition to the X DA channel and also DA resonances at higher energies were observed. Returning to Figs. 2 and 3 we see that at room temperature the intensity of Br is by a factor of 70 higher than that of Cl with the Br channel peaking at 0.3 eV and that of Cl at 0.45 eV. Both reactions exhibit pronounced temperature dependences in the way that the peak maximum is shifted towards lower energy and the overall intensity ®rst increases and, after passing a maximum, signi®cantly decreases. For the Br channel the intensity maximum is near 400 K and that for

Cl near 460 K. At the highest temperature the intensity ratio Br =Cl decreases to about ®ve. The shift of the resonance energy and the increasing cross-section by heating the molecule is reminiscent of a DA reaction proceeding along repulsive energy surfaces (Fig. 4) and being subjected to a strong competition between dissociation and autodetachment in the decay of the TNI as, e.g., in the case of the systems O2 and CF3 Cl mentioned in Section 1. While the resonance in O2 is located above 6.5 eV, DA to CF3 Cl yielding Cl has a resonance maximum at 1.4 eV [8] which is closer to the present situation. The temperature dependence in CF3 Cl has successfully been described by means of a semiclassical model [10,11] which may also serve to qualitatively describe the

540

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

Fig. 4. Schematic two-dimensional potential curve to illustrate the temperature dependence of the DA cross-section. Vi is the potential curve of the initial (neutral) state and Vf that of ®nal (anionic) state. On the right-hand side two ion yield curves at room temperature (T1 ) and at elevated temperature (T2 ) are shown. The peak near 0 eV (threshold peak) is due to transitions from vibrational levels close to the crossing point Rc .

Fig. 3. Temperature dependence of Cl =C6 H4 ClBr DA crosssection in the range from room temperature to 541 K.

presently observed temperature dependence and we shall brie¯y recall the essential features in the case of CF3 Cl. The experimental observation is that the DA resonance (located at 1.4 eV at 300 K) is shifted to 1.0 eV at 800 K with a gradual increase of the cross-section. At temperatures above 350 K an additional narrow peak appears near 0 eV (threshold peak) which dominates the Cl yield at 800 K. By comparing with electron scattering data it was shown that the dissociation of the transient anion CF3 Cl # is strongly suppressed by autodetachment in that less than 1% of the initially formed compound ions survive dissociation into Cl ‡ CF3 . The autodetachment rate, on the other

hand, varies with increasing electron energy (decreasing nuclear distance), as indicated by the increasing width of the ionic potential energy curve (dotted area) for distances R < Rc in Fig. 4. For distances R > Rc the system is stable with respect to AA. Since (i) the dissociation asymptote Cl ‡ CF3 is energetically below the level of the curve crossing, (ii) the potential energy curve of the anion possesses only a shallow minimum [10,11] and (iii) dissociation of CF3 Cl # is pronounced impulsive along the C±Cl coordinate (R), the system will exclusively undergo DA once it reaches the crossing point Rc . Impulsive dissociation is mirrored by e€ective kinetic energy release to the fragments which also justi®es description of the process in two-dimensional (diatomic) potential energy curves with the CF3 radical acting as a rigid particle. In the molecular orbital approach, the extra electron is captured into a localized r (C±Cl) MO [6,12]. Heating of the molecule and thus population of vibrational levels extends the Franck±Condon region allowing transitions to the repulsive ionic potential energy curve at lower electron energies and hence higher dissociation probabilities (Fig. 4). This results in a downshift of the DA resonance maximum with increasing temperature, but also in

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

an increase of the cross-section. The e€ect is the more pronounced the stronger autodetachment (AA) dominates the decay of the TNI. The threshold signal (0 eV), ®nally, mirrors transitions from vibrationally excited neutral molecules near the crossing point. From a thermodynamic point of view the energy of the crossing point Rc can be viewed as the activation energy for the attachment of thermal electrons. In the above system the activation energy was determined as 0.26 eV [3]. On the other hand, at 800 K the relative population of the relevant m3 (C±Cl) vibrational modes (¤x…m3 † ˆ 59 meV) close to the level of the crossing point is only a few percent, but the temperature e€ect is nevertheless pronounced due to the strongly varying autodetachment rate in combination with the reciprocal energy dependence of the attachment cross-section. Generally, one expects an increase of the DA cross-section with temperature if the anionic potential energy curve crosses right of the equilibrium distance and a decrease if it crosses left of it (inverted region). In Figs. 2 and 3 we can see that the temperature dependence of the shape of the DA resonance (for both channels) qualitatively shows a comparable behaviour as observed for CF3 Cl in that the peaks are shifted to lower energy merging with a threshold peak (no longer separated from the original DA resonance as in CF3 Cl) resulting in a pronounced asymmetric shape of the ion yield curves at the highest temperatures. The dependence of the DA cross-section, however, di€ers distinctly from that in CF3 Cl as it has a maximum value located near 400 K (Br ) and 460 K (Cl ). On the basis of the short discussion on the CF3 Cl system we may interpret the presently observed temperature dependence in a corresponding way, i.e., by a crossing of the relevant potential energy curves still right of the minimum of the neutral con®guration, however, at a comparatively smaller energy level (a few tenths meV) and with the Br channel below that of Cl . Increasing the temperature initially results in an increasing population of vibrational levels close to the crossing point which in turn: (a) shifts the resonance energy, (b) increases the DA cross-section and (c) results in the appearance of a threshold signal.

541

Due to the low lying crossing point, resonance maximum and threshold peak are no longer separated. Above a certain temperature, however, depopulation of vibrational states close to the crossing point in favour of those above becomes signi®cant resulting in a decreasing DA cross-section. Before considering this qualitative picture in some more detail one has to remember the limitations of such a simple model, in particular when applied to a large molecule as the present target. (a) We can assume that the MOs involved in low energy electron attachment to C6 H4 BrCl possess p character with some more or less signi®cant r (C±X), X ˆ Cl, Br admixture which may also justify description of the process within the framework of two-dimensional potential energy curves along the C±X axis to some extent, but certainly not in a way as in CF3 Cl. In actual practice we have 30 vibrational degrees of freedom which can e€ectively couple after electron capture. (b) The thermodynamic parameters for the present system are not yet established (adiabatic electron anity, energy for the dissociation limits (3a) and (3b). There are no explicit literature values for the C±Cl and C±Br bond dissociation energies necessary to calculate the energy of the DA limits. For the monosubstituted compounds C6 H5 X (X ˆ Cl, Br) we ®nd D(C±Cl) ˆ 4:1  0:1 eV and D(C±Br) ˆ 3:5  0:1 eV [19]. In halogenated hydrocarbons the observation is that substitution of hydrogen by a halogen (except F) appreciably decreases the C±Cl or C±Br dissociation energy. If we assume a comparable trend in the halobenzenes and taking the very well known electron anities EA…Cl† ˆ 3:615  0:004 eV and EA…Br† ˆ 3:364  0:004 eV [19] we expect Cl formation to be approximately thermoneutral and Br formation slightly exothermic. Such a situation is also supported by the nearly symmetric shape of the ion yield curves taken at room temperature indicating that the entire Franck±Condon region for both transitions are located above the respective dissociation limit. In spite of that, dissociation of the anion can be accompanied by energy transfer thereby withdrawing energy from the C±X dissociation coordinate. In the case of an appreciable minimum in the anionic potential

542

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

energy surface (i.e., an appreciable adiabatic electron anity of the target molecule) the system may even return to a con®guration R < Rc , where autodetachment again becomes possible. In other words, in a polyatomic system with an appreciable minimum in the anionic surface, the crossing point Rc can no longer be considered as a point of no return. (c) Our discussion so far is based on the Born± Oppenheimer approximation, i.e., the use of localized potential energy surfaces and the application of the Franck±Condon principle which may be limited in the case of very low electron energies. The low energy domain is still a challenge in electron±molecule collision theory and the complexity of the problem is due to ecient coupling of the electron scattering dynamics with vibrational motion and in the case of polar molecules with rotational motion. A variety of methods such as R-matrix theory [20,21], Feshbach projection operator formalism [22] or the nonlocal resonance model [23,24] have been employed to explain the low energy electron collision phenomena in diatomic molecules. For the present polyatomic system we assume that DA can qualitatively be described in terms of localized potential energy curves. Within that we consider the problem of the changing population of the relevant vibrational states with temperature in some more detail. The C±X stretching vibrations in p-disubstituted benzenes couple to di€erent vibrations of the aromatic ring [25]. For chlorobromobenzene the frequency domains assigned to the C±Cl and C±Br stretch vibrations are located in the middle IR region but also in the far IR (at 270 cm 1 ). Taking the 270 cm 1 C±X stretch vibration as relevant to promote the DA reactions (3a) and (3b) we locate the curve crossing near the v ˆ 1 level for the Br channel and between the v ˆ 1 and v ˆ 2 level for and Cl channel. According to the semiclassical model [10,11] (considering Franck±Condon factors and autodetachment) the DA cross-section as a function of the initial vibrational state increases with the vibrational quantum number for states below the crossing point and decreases for those above. In the present experiment we change the temperature and hence the population of vibra-

tional levels. By increasing the temperature the relative population of the v ˆ 1 level increases and after passing a maximum decreases. By just considering the normal vibrationPat 270 cm 1 and taking the partition function N …v† in the harmonic oscillator approximation we can calculate the relative population of the v ˆ 1 level N …v ˆ 1† P1 V ˆ0 N …v† (N …v† ˆ exp… hx…v ‡ 1=2†=kT † and hx ˆ hc= k…k 1 ˆ 270 cm 1 †) as a function of the temperature. The analysis shows that the population of the ®rst excited level has a maximum slightly below 600 K. Accordingly, we expect a maximum in the cross-section near that temperature, modi®ed by the di€erent Franck±Condon factors of the Cl and Br channel with their slightly di€erent crossing points. At that point we have not carried out explicit calculations of the cross-section taking into account the temperature dependent population of the vibrational states, Franck±Condon factors and autoionization widths, but it is clear, that the Br channel will have the maximum crosssection at lower temperature. We are not aware of a DA cross-section showing a similar temperature dependence. There are many systems with positive temperature dependences [1] including C6 F5 Cl [3] which shows a strongly increasing Cl signal up to a temperature of 700 K. On the other hand, the DA cross-section for the systems Cl =CCl4 [26] and I =CF3 I [2,3] are independent on the temperature. This was interpreted by crossing of the potential energy curves near the minimum of the neutral. In the framework of the present interpretation one would also expect a depopulation of the lowest vibrational levels and the question why these systems do not show a decreasing cross-section at higher temperatures remains so far unanswered. We know of only one DA reaction showing a negative temperature dependence, namely CN =C6 F5 CN [3]. In this case electron attachment proceeds via a strongly localized pCN MO with vibrational energy transfer into the C±CN coordinate. This results in a metastable decay of the transitory negative ion, i.e., the undissociated anion is observable in the mass spectrometric observation window in com-

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

petition to the DA channel CN . The observed e€ect was consequently explained by an increasing autodetachment rate with temperature which results in a decrease of AA and DA [3]. A decrease of the product signal at higher temperature could certainly also be due to a thermal instability of the compound. The problem for the present system is that DA is restricted to the two channels (Cl and Br ) which are both sensitive to temperature. The other halogenated benzenes previously studied by our laboratory [3] exhibited additional DA features at higher energy and additional product ions which were unsensitive to temperature and which served as a monitor for the presence of the undissociated target molecule. On the other hand, those experiments were carried out up to temperatures of 700 K with no indication of thermal instabilities. It is therefore extremely unlikely that the observed decrease of the signal in the present case arises from thermal decomposition of the molecule. In addition at 450 K the average total vibrational energy stored in the molecule is considerably below the C±Br bond dissociation energy so that thermal decomposition (at least in the gas phase) is very unlikely. We note that in a recent elegant crossed beam experiment using Franck±Condon pumping and stimulating Raman scattering [27] dissociative electron attachment to Na2 molecules in selected vibrational (and also rotational) states has been studied. A strong increase of the DA cross-section (more than three orders of magnitude) was observed up to the vibrational level v ˆ 12 while above that level the cross-section slightly decreases. Ab initio calculations in fact showed that the curve crossing for the involved states is located between v ˆ 11 and v ˆ 12 level. In the present experiment we deal with a varying temperature where one expects a smoother dependence than by selecting individual vibrational levels. This may be an indication that a quantitative description of the observed strong temperature dependence within the present onedimensional picture is insucient and additional e€ects associated with intermolecular energy transfer contribute. We also note that in experiments showing a positive dependence of the DA cross-section with

543

temperature, the activation energy (AE) for DA (i.e., the energy level of the curve crossing) is often determined via the Arrhenius equation rth / exp… EA=kT † with rth the cross-section for the threshold peak (thermal electrons). While such a thermodynamic treatment may be adequate for the region kT < AE it is no longer applicable for the temperature range kT  AE or kT > AE. For high temperatures the Arrhenius equation predicts a gradual saturation of the cross-section with temperature while the microscopic treatment predicts a decreasing cross-section above a certain temperature due to a decreasing Franck±Condon overlap. This problem has some analogy to the inverted region in the Marcus theory of electron transfer [28]. Acknowledgements Work supported by the DST (New Delhi) ± DAAD (Bonn) Project Based Personnel Exchange Programm and the Deutsche Forschungsgemeinschaft (DFG). W.B. thanks the State Committee for Scienti®c Research (Poland, grant No 3 T09A 01018 for support). References [1] L.G. Christophorou, J.K. Oltho€, Adv. At. Mol. Opt. Phys. 44 (2000) 155. [2] T. Underwood-Lemmons, T.J. Gergel, J.H. Moore, J. Chem. Phys. 102 (1995) 119. [3] I. Hahndorf, E. Illenberger, Int. J. Mass Spectrom. Ion Proc. 167/168 (1997) 87. [4] P. Spanel, S. Matejcik, D. Smith, J. Phys. B 28 (1995) 2941. [5] L.G. Christorphorou (Ed.), Electron±Molecule Interactions and their Applications, vols. I±II, Academic Press, Orlando, FL, 1984. [6] E. Illenberger, J. Momigny, Gaseous Molecular Ions. An Introduction to Elementary Processes Induced by Ionization, Springer, New York, 1992. [7] W.R. Henderson, W.L. Fite, R.T. Brackmann, Phys. Rev. 183 (1969) 157. [8] I. Hahndorf, E. Illenberger, L. Lehr, J. Manz, Chem. Phys. Lett. 231 (1994) 460. [9] T.F. O'Malley, Phys. Rev. 155 (1967) 59. [10] L. Lehr, W.H. Miller, Chem. Phys. Lett. 250 (1996) 515. [11] L. Lehr, Ph.D. thesis, Fachbereich Chemie, Freie Universitat Berlin, 1996.

544

A. Rosa et al. / Chemical Physics Letters 342 (2001) 536±544

[12] T. Oster, A. K uhn, E. Illenberger, Int. J. Mass Spectrom. Ion Proc. 89 (1989) 1. [13] A. Stamatovic, G.J. Schulz, Rev. Sci. Instr. 41 (1970) 423. [14] C.L. Chen, P.J. Chantry, J. Chem. Phys. 71 (1979) 3897. [15] S. Marienfeld, S. Barsotti, E. Leber, A. Schramm, M.-W. Ruf, H. Hotop, A. Rosa, D. Nandi, S.V.K. Kumar, E. Krishnakumar, E. Illenberger, to be published. [16] Y. Le Coat, N.M. Hedhili, R. Azria, M. Tronc, O. Ing olfsson, E. Illenberger, Chem. Phys. Lett. 296 (1998) 208. [17] P. Tegeder, L. Lehmann, O. Ing olfsson, E. Illenberger, Zeitschr. Phys. Chem. 195 (1996) 21. [18] O. Ing olfsson, F. Weik, E. Illenberger, Int. J. Mass Spectrom. Ion Proc. 155 (1996) 1. [19] D.R. Lide (Ed.), Handbook of Chemistry and Physics, 78 ed., CRC Press, Boca Raton, 1997.

[20] L.A. Morgan, P.G. Burke, C.J. Gillan, J. Phys. B 23 (1990) 90. [21] I.I. Fabrikant, Comm. At. Mol. Phys. 24 (1990) 37. [22] W. Domcke, L.S. Cederbaum, J. Phys. B 14 (1981) 149. [23] W. Domcke, Phys. Rep. 208 (1991) 97. [24] J. Horacek, W. Domcke, H. Nakamura, Z. Phys. D 42 (1997) 181. [25] G. Varsanyi, Vibrational Spectra of Benzene Derivatives, Academic Press, New York, 1969. [26] D. Spence, G.J. Schulz, J. Chem. Phys. 56 (1973) 1800. [27] M. K ulz, M. Keil, A. Kortyna, B. Schellhaaû, J. Hauck, K. Bergmann, W. Meyer, D. Weyh, Phys. Rev. A 53 (1996) 3324. [28] R.A. Marcus, Angew. Chem. 105 (1993) 1161, Nobel lecture.