C₅₈ on HOPG: Soft-landing adsorption and thermal desorption

Abstract

Electron-impact induced dissociation/ionization of C₆₀ molecules was used to produce an intense beam of C₅₈⁺ ions. This was directed towards a HOPG (highly oriented pyrolytic graphite) substrate under nominally perpendicular impact conditions in order to generate deposits by soft-landing (kinetic energy < 0.1 eV atom⁻¹). Deposited C₅₈ molecules could subsequently be thermally desorbed intactly. Thermal desorption mass spectra of the deposits exhibit only C₅₈. Surface deposited C₅₈ can react with background gases to generate hydride derivatives C₅₈H_n which are also desorbable. The apparent desorption energy of C₅₈ and C₅₈H_n molecules from the HOPG surface varies with increasing adsorbate coverage from 2 ± 0.1 to 2.2 ± 0.1 eV as determined by a Redhead analysis. These values are 0.7 ± 0.1 eV larger than found for C₆₀ desorbed from the same substrate.

---

1. Introduction

The fabrication and properties of fullerene-based solids have been of interest since the first mass spectrometric observation¹ and the first bulk production of C₆₀.² Most such investigations have focused on C₆₀ and C₇₀ solids and solid films³ because the constituting molecules can be most easily prepared (generally by solvent extraction from the carbon soot created by arc-discharge of graphite electrodes).² However, several other (higher) fullerenes such as C₇₆ have also been solvent extracted, isolated and studied in macroscopic (generally milligram) quantities.⁴ A common feature of all such readily isolable fullerenes is that their structure conforms to the isolated pentagon rule (IPR),⁵i.e. each pentagon is surrounded by carbon hexagons. C₆₀ is the smallest and most prominent example of these IPR-fullerenes. Below this atom count, all possible “classical” fullerene cages (i.e. closed shell structures comprising only carbon pentagons and hexagons) contain at least two adjoining pentagons. Such adjoining pentagons and the associated strain are thought to confer much higher reactivity to these small non-IPR fullerenes,⁶ thus rationalizing the apparent absence of fullerenes smaller than C₆₀ in early mass spectral analyses of raw soots, soot sublimates and soot solvent extracts.

More recently, Piskoti et al.⁷ have inferred the bulk scale preparation by arc discharge of a much smaller fullerene that drastically violates the IPR rule: C₃₆. So far, it has only proven possible to extract this species from arc discharge (and laser ablation) soots in the form of hydride (C₃₆H₄, C₃₆H₆) and oxyhydride (C₃₆H₄O) derivatives^8,9 indicating high reactivity of the bare cluster precursors-probably towards moisture. An analogously passivated and solvent extractable C₅₀Cl₁₀ cage has recently been observed in the soot generated by graphite arc discharge in a helium/CCl₄ atmosphere.¹⁰

As yet, however, there have been no direct probes of pure solids or of thin films derived from non-IPR fullerenes. According to theoretical predictions such solids should exhibit interesting electronic and chemical properties.^11,12 Recent developments in “preparative mass spectrometry”, i.e. of intense ion beam sources, more efficient mass selection methods and optimal low kinetic energy deposition schemes have made possible a novel approach to this question. The desired molecular species are generated as ions in gas-phase, mass selected and then soft-landed onto a clean surface. This strategy has already been used extensively to probe the surface reactivity of transition metal clusters.^13,14 More recently it has been used to prepare low coverages (0.04 ML) of endohedral metallofullerenes for STM/STS studies.¹⁵ In this, the first of a series of papers on non-IPR fullerenes, we have used preparative mass spectrometry to generate and characterize (intact) deposits of C₅₈ on HOPG at coverages up to monolayers (ML).

The choice of C₅₈ was motivated in part by the particularly large contrast in physical properties expected on the basis of theoretical predictions and experimental findings versus solid C₆₀. Based on photodetachment probes of C₅₈⁻ generated by laser vaporization of graphite, C₅₈ is thought to have a significantly higher electron affinity of 3.5 eV ^16,17versus 2.63 eV for C₆₀.¹⁷ Furthermore, DFT calculations indicate that the fragmentation of isolated C₅₈ (by C₂ loss) requires ca. 2.2 eV less energy than the lowest energy dissociation channel of C₆₀.^12,18 According to these calculations,^12,18 there are two plausible quasi-equienergetic ground state structures: one is a “conventional” non-IPR fullerene characterized by three pairs of adjoining pentagons the other is a “non-conventional” cage containing a heptagonal ring. Finally, an earlier study has inferred a particularly small HOMO–LUMO separation of ca. 0.1 eV, for the preferred ground state of C₅₈.¹⁹ This is in contrast to C₆₀ , which has a much larger molecular HOMO–LUMO separation and which forms a wide band gap semiconducting solid under ambient conditions (E_gap ∼ 1.6 eV ¹⁹). Conceivably, solid C₅₈ may have a much narrower band gap.

2. Experimental

Fig. 1 shows a schematic drawing of the apparatus. C₅₈⁺ ions were produced by electron-impact induced ionization/dissociation of C₆₀ molecules issuing from a Knudsen cell (500 °C; C₆₀ powder (Aldrich), electron energy 50 eV). The positive ions created in this way were guided by a series of electrostatic lenses through three stages of differential pumping, bent by 90° in an ion mirror, mass filtered in a quadrupole mass filter (Extrel) and then soft-landed. The mass filter removes all undesired ions, particularly the abundant C₆₀⁺, but also smaller fragments. The ion mirror avoids a direct line of sight between source and target surface and prevents neutral fullerene molecules from reaching the surface. In the final stage of the electrostatic system the mass-selected ion beam was focused onto the target surface (highly oriented pyrolytic graphite (HOPG), size 6 × 8 mm, ZYI quality, SPI Supplies) and decelerated to a kinetic energy of 6 eV, i.e. about 0.1 eV/carbon atom under conditions of nominally perpendicular impact (unless otherwise noted). The impact energy is determined by a variable retarding field, the full width of the kinetic energy distribution is ca. 2 eV. The base pressure in the deposition chamber was 5 × 10⁻¹⁰ mbar. Such soft-landing conditions are crucial in order to avoid impact-mediated fragmentation as well as to maximize the effective sticking probability. The corresponding C₅₈⁺ flux on the target was measured via monitoring the deposition (=neutralization) current by means of a picoamperometer (Keithley). Usually the C₅₈⁺ flux during the deposition was kept constant at a level between 0.2 and 1 nA. After the deposition, the C₅₈⁺ flux was switched off by closing the valve between ion beam line and UHV-deposition chamber. Subsequently, the sample was moved from the deposition chamber into the analysis chamber by means of a long-stroke manipulator and placed in front of (ca. 2.5 cm from the ion source opening) a second quadrupole mass filter (Extrel) equipped with an electron impact ionizer (electron energy 70 eV). The pressure during transfer and during desorption was in the range between 5 × 10⁻¹⁰ and 1 × 10⁻⁹ mbar. The desorption spectra were taken by resistively heating the sample from room temperature up to 1000 K. All spectra (TDS and MS, see below) were taken at a heating rate of 9 K s⁻¹. The desorbed neutral fullerene molecules were again ionized, mass filtered, detected by means of an electron multiplier and counted in a multichannel scaler (EG&G, Turbo-MCS). We operated in two modes: (1) thermal desorption spectroscopy (TDS) – here the mass filter was set at a constant mass and the ion signal was measured as function of target temperature and (2) mass scanning (MS) – for this we quickly (25–50 ms u⁻¹) scanned the mass spectrometer over the mass range of interest while heating the target and accumulated the ion signal as function of mass. The HOPG sample was heated repeatedly up to 1000 K as a routine procedure prior to C₅₈⁺ or C₆₀⁺ deposition.²⁰ In addition to removing other residual surface contaminants, this procedure results in a considerable depletion of the initial hydrogen content in the outermost layers of the substrate. Sample temperature was monitored with a K-type thermocouple attached to the front surface of the sample. Reference experiments were performed with C₆₀⁺ ions at higher ion flux (up to 20 nA). In order to achieve comparable coverages this higher intensity was compensated by shorter deposition times. All deposition experiments were conducted with HOPG samples kept at room temperature. Effective coverages were determined under the assumption that neutralization currents correspond to one electron per deposited monocation.

Fig. 1 Experimental setup: the fullerene ion beam is created in an electron impact ionization source, bent by 90° in an ion mirror (electrostatic quadrupole), mass selected in a quadrupole mass filter (QMS1), decelerated and soft landed on a target surface (HOPG) which is in turn mounted on a temperature variable UHV-manipulator. After deposition the target is transferred in vacuo to bring it into line-of-sight of a second mass filter (QMS2) equipped with an electron impact ionizer. For desorption spectra the target is resistively heated to 1000 K at a rate of 9 K s⁻¹.

3. Results and discussion

In order to study the solid deposits created we applied thermal desorption measurements coupled with electron impact mass spectrometry of desorbed species. This is a well established method to determine the binding energy of molecules on surfaces – and more generally of surface bound molecules interacting with surrounding species.²⁰ In the present case, such measurements can provide information on: (1) ion impact/fragmentation and sticking on the surface, (2) molecule–surface interactions and (3) molecule–molecule interactions (at higher coverages). We compare the results obtained for C₅₈-based deposits with those of analogous C₆₀ films.

3.1. Soft-landing adsorption

Two separate samples were prepared by exposing HOPG to nearly the same doses of slow (6 eV) C₆₀ and C₅₈ cations (dose 1.9 × 10¹³ and 2.6 × 10¹³ ions respectively, corresponding to a coverage of 9% and 13% of a monolayer, assuming an effective area of 1 nm² per fullerene molecule and a unit sticking coefficient). In each case, we subsequently recorded the mass spectrum of molecules desorbed as a result of raising the substrate temperature up to 1000 K (at a rate of 16 K s⁻¹). Fig. 2a shows such a spectrum over the mass range 650–750 u as taken during thermal desorption of the C₆₀ deposit. Only one dominant mass peak at 720 u is observed, clearly indicating that the deposition-desorption cycle does not lead to excessive fragmentation of C₆₀ molecules – insofar as this is evidenced by desorbable species. This is not surprising since the impact energy of less than 6eV is well below the fragmentation threshold of isolated C₆₀ (10–11 eV).¹⁸ The small peak at 696 u reflects ionization induced fragmentation of desorbed C₆₀ molecules within the TDS mass spectrometer ion source. (When ionizing pure C₆₀ vapor in the primary electron impact ion source under comparable ionization conditions (i.e. an electron energy of 70 eV), we observe a C₅₈⁺/C₆₀⁺ ion ratio of 1 : 20, very similar to the ratio we observe after desorption). We also looked for signs of pick-up processes (i.e. adduct formation with surface contaminants), C₆₀ derived dimers and/or other coalescence products.^21–23 Mass spectra recorded over a broad range including scans centered around 1400 u showed no traces of species other than C₆₀.

Fig. 2 Mass spectra of thermally desorbed fullerenes as obtained using QMS2 by heating up the substrate to 1000 K (see Fig. 1 caption). (a) Ion signal after exposure of the target surface to a dose of 50 nA min = 1.9 × 10¹³ C₆₀⁺ ions at a base pressure of 5 × 10⁻¹⁰ mbar. (b) Ion signal after exposure of the target to a dose of 120 nA min = 4.5 × 10¹³ C₅₈⁺ ions at a base pressure of 5 × 10⁻¹⁰ mbar. (c) Ion signal after exposure of the target to a dose of 70 nA min = 2.6 × 10¹³ C₅₈⁺ ions under rather poor vacuum conditions (base pressure 2 × 10⁻⁸ mbar). Note the tail to higher masses which is attributed to hydrogen uptake.

Fig. 2b shows the corresponding measurement for a C₅₈-based deposit rapidly heated up to 1000 K. The peak at 696 u as the dominating spectral feature proves that the film deposited contains C₅₈ molecules as the only desorbable species. As a corollary, the mass spectrum implies that C₅₈ can be deposited without fragmentation under soft-landing conditions. The small signal at the mass of C₅₆ is again most likely due to fragmentation upon ionization (cf. C₆₀ desorption). Similarly, no indication of C₅₈–C₅₈ dimers could be identified in a mass spectrum taken in the 900–1800 u range.

3.2. Dependence on incident kinetic energy

Fig. 3c documents the nearly linear dependence of the integrated TDS signal on C₅₈⁺ dose (conditions as described in 3.4). This proportionality could only be observed for impinging ions having kinetic energies lower than about 10 eV. Higher kinetic energies lead first to non-linearity and then to complete loss of the thermal desorption signal. These observations are consistent with increasing importance of competing channels such as back-scattering and “destructive sticking” to yield non-desorbable sample modifications (e.g. via fragmentation and/or covalent surface attachment). In order to avoid these processes we reduced the kinetic energy of impinging ions as far as possible, while still allowing sufficient flux to the surface (E_kin <6 eV) to enable desorption measurements.

Fig. 3 (a) Thermal desorption spectra (TDS) taken after exposing the HOPG to four different C₅₈⁺-doses: 0.15, 0.37, 0.78 and 2.36 × 10¹³ ions. The resolution of the mass spectrometer QMS2 was set to allow simultaneous transmission of 695–705 u which assures that both C₅₈ and hydrogen adducts are detected. With increasing exposure, the peak in the TDS shifts from 790 K to 820 K. (b) Corresponding ion signal after exposure to 8.6 × 10¹² C₆₀⁺ ions. Note that the desorption maximum is 220 K lower than the corresponding C₅₈ signal. (c) Integral signal intensity of desorbed C₅₈versus the dose of incoming ions.

Determination of the absolute yield of desorbed particles was beyond the scope of this study, requiring calibration against other techniques sensitive to surface coverage. In contrast, relative desorption yields could be readily measured. At 6 eV impact, the integrated desorbed C₅₈ ion signal (normalized against the flux of the impinging ions) was found to be roughly 2/3 that of the corresponding C₆₀ signal. There are several possible explanations: (i) different final charge states of adsorbed molecules (neutral C₆₀ ↔ C₅₈⁻), (ii) smaller ionization cross section of C₅₈, (iii) smaller sticking coefficient of impinging C₅₈⁺ at the same impact energy or (iv) larger “destructive sticking” cross section in the case of C₅₈ molecules. Note that for comparable coverages we have seen no significant modifications of the C₆₀ or C₅₈ TDS traces resulting from repeatedly performed adsorption–desorption cycles.

3.3. Hydride formation

For long C₅₈⁺ exposures at pressures above 5 × 10⁻⁹ mbar, the C₅₈-derived mass spectral peak obtained upon subsequent desorption was observed to be measurably broadened. In particular, at the low mass resolution settings used in order to obtain adequate signal-to-noise, a tailing to higher mass was observed which can extend for up to ca. ten mass units from the primary C₅₈⁺ isotopomer at m/z 696 (Fig. 2c). This indicates the formation of C₅₈H_n (n < 8) on the surface. By comparison, a high resolution mass spectrum of C₅₈⁺ gas-phase molecular ions impinging on the substrate reveals the natural isotope distribution of C₅₈ covering a narrow mass range of essentially three atomic mass units, i.e. there are no measurable hydrides in the primary ion beam. Note that deposited C60 is unreactive under comparable conditions.

At this stage we can only speculate about the origin of the hydrogen containing species. We find that the tail of the mass spectrum slowly shifts towards heavier molecules C₅₈H_n, from n = 0 to n = 8, when keeping the freshly created C₅₈ film in vacuum at base pressure for extended periods of time. Measurements of the outcomes of controlled exposures to candidate reagents such as H, H₂ and H₂O are ongoing.

3.4. Activation energies for desorption

Various C₅₈ films were created by exposing the HOPG surface to C₅₈⁺ flux for different time periods. In each case, the intensity of the subsequently desorbed mass-selected C₅₈ flux was monitored while raising the sample temperature with a constant heating rate β of 9 K s⁻¹ from room temperature up to 1000 K. In order to reach an acceptable signal to noise ratio, all such thermal desorption spectra were recorded with a low mass resolution (filter window 695–705 u) which allows for simultaneous detection of C₅₈H_n species also desorbed from surface. Fig. 3a shows a series of corresponding C₅₈ + C₅₈H_n – TD spectra. For small coverages, the desorption bands are peaked at T_m ∼ 790 K. The desorption maximum shifts slightly towards higher temperatures with increasing coverage and levels off at 830 K.

Redhead’s analysis^24,25 of first order desorption kinetics was applied to the data: dΘ/dt = νΘexp(−E/RT), where Θ is the coverage, ν the pre-exponential factor and E the activation energy for desorption. Both ν and E are assumed to be coverage independent and the activation energy can be directly related to the location of desorption peak, T_m, by a simple equation, E = RT_m[ln(νT_m/β) − 3.64].²⁴ Thus, for coverages lower than 10¹³ molecule cm⁻² we obtained a value of the apparent activation energy for desorption of 2.0 ± 0.06 eV. For considerably higher coverages, the latter energy increases up to 2.2 eV. Note that under presently achievable experimental conditions, increased coverage may also be associated with increased hydride content. An estimation of the net C₅₈–HOPG binding energy requires a determination of the lateral C₅₈–C₅₈ interaction. This is presently an open problem because of the still unknown charge state of the deposited C₅₈.

By comparison, Fig. 3b shows a C₆₀-TD spectrum taken under the same conditions. For coverages on the order of 10¹³ molecule cm⁻² the C₆₀-TD spectra exhibit maxima at T_m ∼ 570 K. Redhead’s analysis yields 1.38 ± 0.06 eV as the activation energy E for C₆₀ desorption, in reasonable agreement with previous TDS measurements by Ulbricht et al. for neutral C₆₀ monolayers sublimed onto HOPG.²⁶ These authors obtained an activation energy of 1.69 eV using a different heating rate and analysis procedure. Recent calculations of the binding energy for a C₆₀ molecule (physi)sorbed onto HOPG yield a binding energy of 0.968 eV.²⁷ The corresponding interactions are essentially of van der Waals type. This value is considerably lower than the E(C₆₀) values derived from TDS. This may be rationalized in terms of lateral C₆₀–C₆₀ interactions which serves to raise the effective desorption energy in experiment. The corresponding van der Waals dimer binding energy has been calculated to 0.28 eV.²⁸

3.5. Comparison between C₅₈ and C₆₀

Two primary inferences may be drawn from Fig. 3. First, the large difference ΔT_m of about 220 K between C₅₈ and C₆₀ desorprtion maxima implies a surprisingly large difference in desorption activation energy of about 0.7 ± 0.1 eV. There are two possible explanations for this: (i) the lateral C₅₈–C₅₈ interactions may be significantly stronger than for C₆₀ (e.g. due to the adjacent pentagons in the non-IPR cages) and/or (ii) C₅₈ may bind more strongly to HOPG. Given the electron affinities (ε_A(C₅₈) = 3.5 eV, and ε_A(C₆₀) = 2.63 eV) and HOPG work function (ϕ = 4.4 eV,²⁹ 4.475 eV ³⁰), spontaneous surface-to-molecule electron transfer at a van der Waals separation is energetically possible for C₅₈ but not for C₆₀. Such additional image charge stabilization would lead to significantly stronger binding of C₅₈.³¹

The second inference is that TDS traces obtained for C₅₈ are much wider than observed for C₆₀. This feature may be attributable to multiple isomers in the beam of impinging C₅₈⁺ ions (see below). Additionally, the low symmetry cage structures likely pertaining have a much wider spectrum of possible (binding) orientations with respect to the HOPG substrate than is the case for C₆₀. The prevalence of adjacent pentagon units or of other unusual structural features may also significantly change the interaction of the mobile C₅₈^−δ cages with step edges as mediators in a desorption scenario (step edge as a trap for C₅₈^−δ, longer residence times, stronger bonds, etc.).

We have used the electron affinity determined for an isomer distribution of C₅₈⁻ as generated by laser vaporization of graphite in order to rationalize the observation of enhanced surface binding vs. C₆₀. Such an isomer distribution could also include “non-classical” closed-shell isomers having heptagonal rings. According to recent theoretical predictions (Hartree–Fock and density functional), the stabilities of the optimal classical (three pairs of adjacent pentagons, C_3v symmetry) and non-classical (seven-ring) C₅₈ isomers are nearly equal.¹⁸ Note that significant amounts of open-shell isomers such as “cups”, polycyclic rings or “sticks” are most likely not present in our experiment: ion mobility measurements of C₅₈⁺ ions produced by laser vaporization of graphite show that almost 90% of the detected C₅₈⁺ ions are closed shell.³² Furthermore, these measurements show that C₅₈⁺ ions produced by fragmentation of C₆₀⁺ ions (itself generated by graphite laser vaporization) consist of closed shell isomers only.³³ A more recent high resolution ion mobility study of C₅₈⁺ generated by laser ablation/ionization of fullerene soot likely probed an isomer distribution close to that used here: only closed shell structures were observed.³⁴

In future work we intend to probe the surface deposits by means of scanning tunneling/force microscopies (accessing absolute sticking coefficients, information on aggregate formation and surface diffusion as well as a measure of absolute desorption yield) as well as X-ray photoelectron spectroscopy (electronic structure). Variation of the substrate work function should facilitate more detailed understanding of substrate-to-molecule spontaneous electron transfer. Variation of the heating rate coupled with a more detailed analysis of TDS measurements should provide information on the spread of binding energies induced by system heterogeneities. Note that the sensitivity of analytical HPLC with spectroscopic detection of eluting fractions is sufficient to probe the amounts deposited here provided they are readily soluable. In addition, mass selected fullerene ion fluxes at least an order of magnitude higher than those used in the present study are within easy reach by improving the ion source configuration. This should facilitate a determination of the isomer content of the desorbable material by wet chemical means.

Summary

We describe a new route towards the formation of solid materials based on small fullerenes (C_x, 50 < x < 60). Electron impact ionization/fragmentation of C₆₀ was used to generate an intense beam of slow C₅₈⁺ ions. This allowed for the deposition of films by perpendicular impact onto HOPG within reasonable time periods. The C₅₈ deposit created in this way exhibits significantly higher thermal stability than a comparable C₆₀ deposit. Both molecules survive the adsorption-desorption cycle. A desorption activation energy in the range from 2 eV was determined by means of thermal desorption spectroscopy. In contrast to C₆₀, C₅₈ thin films react with residual gases at room temperature to generate hydride derivatives.

Acknowledgements

This research was supported by the Deutsche Forschungsgemeinschaft with a grant administered by the Center for Functional Nanostructures (CFN).

References

1. H. W. Kroto, J. R. Heath, S. C. O’Brien, R. F. Curl and R. E. Smalley, Nature, 1985, 318, 162.
2. W. Krätschmer, L. D. Lamb, K. Fostiropoulos and D. R. Huffman, Nature, 1990, 347, 354.
3. (a) See, for example: J. E. Fischer and D. E. Cox, J. Phys. Chem. Solids, 1992, 53, v , (Special Issue: Physics and chemistry of fullerene-based solids); (b) V. P. Dravid, S. Liu and M. M. Kappes, Chem. Phys. Lett., 1991, 185, 75; (c) L. Akselrod, H. J. Byrne, C. Thomsen, A. Mittelbach and S. Roth, Chem. Phys. Lett., 1993, 212, 384; (d) S. Liu, Y. Lu, M. M. Kappes and J. A. Ibers, Science, 1991, 254, 408.
4. F. Diederich, R. Ettl, Y. Rubin, R. L. Whetten, R. D. Beck, M. Alvarez, S. Anz, D. Sensharma, F. Wudl, K. C. Khemani and A. Koch, Science, 1991, 252, 548.
5. H. W. Kroto, Nature, 1987, 329, 529.
6. S. Petrie and D. K. Böhme, Nature, 1993, 365, 426.
7. C. Piskoti, J. Yarger and A. Zettl, Nature, 1998, 393, 771.
8. A. Koshio, M. Inakuma, T. Sugai and H. Shinohara, J. Am. Chem. Soc., 2000, 122, 398.
9. A. Koshio, M. Inakuma, Z. W. Wang, T. Sugai and H. Shinohara, J. Phys. Chem. B, 2000, 104, 7908.
10. S.-Y. Xie, F. Gao, X. Lu, R. Huang, C. Wang, X. Zhang, M. Liu, S. Deng and L. Zheng, Science, 2004, 304, 699.
11. E. Hernandez, P. Ordejon and H. Terrones, Phys. Rev. B, 2001, 63, 193403.
12. S. Diaz-Tendero, M. Alcami and F. Martin, J. Chem. Phys., 2003, 119, 5545.
13. A. Sanchez, S. Abbet, U. Heiz, W.-D. Schneider, H. Häkkinen, R. N. Barnett and U. Landman, J. Phys. Chem. A, 1999, 103, 9573.
14. S. Abbet, K. Judai, L. Klinger and U. Heiz, Pure Appl. Chem., 2002, 74, 1527.
15. R. Klingeler, C. Breuer, I. Wirth, A. Blanchard, P. S. Bechthold, M. Neeb and W. Eberhardt, Surf. Sci., 2004, 553, 95.
16. O. V. Boltalina, I. N. Joffe, L. N. Sidorov, G. Seifert and K. Vietze, J. Am. Chem. Soc., 2000, 122, 9745.
17. S. H. Yang, C. L. Pettiette, J. Conceicao, O. Chesnovsky and R. E. Smalley, Chem. Phys. Lett., 1987, 139, 233.
18. D. Boese and G. E. Scuseira, Chem. Phys. Lett., 1998, 294, 233.
19. B. L. Zhang, C. H. Xu, C. Z. Wang, C. T. Chan and K. M. Ho, Phys. Rev. B, 1992, 46, 7333.
20. T. Zecho, A. Güttler, X. Sha, B. Jackson and J. Küppers, J. Chem. Phys., 2002, 117, 8486.
21. G. Ulmer, E. E. B. Campbell, R. Kühnle, H.-G. Busmann and I. V. Hertel, Chem. Phys. Lett., 1991, 182, 114.
22. C. Yeretzian, K. Hansen, F. Diederich and R. L. Whetten, Nature, 1992, 359, 44.
23. K. R. Lykke, D. H. Parker and P. Wurz, Int. J. Mass Spectrom. Ion Processes, 1994, 138, 149.
24. P. A. Redhead, Vacuum, 1962, 12, 203.
25. A. M. de Jong and J. W. Niemantsverdriet, Surf. Sci., 1990, 233, 355.
26. H. Ulbricht, G. Moos and T. Hertel, Phys. Rev. Lett., 2003, 90, 095501.
27. P. A. Gravil, M. Devel, Ph. Lambin, X. Bouju, Ch. Girard and A. A. Lucas, Phys. Rev. B, 1996, 53, 1622.
28. (a) L. A. Girifalco, J. Phys. Chem., 1992, 96, 858; (b) L. A. Girifalco, M. Hodak and R. S. Lee, Phys. Rev. B, 2000, 62, 013104.
29. H. Ago, Th. Kugler, F. Cacialli, K. Petritsch, R. H. Friend, W. R. Salaneck, Y. Ono, T. Yamabe and K. Tanaka, Synth. Met., 1999, 103, 2494.
30. W. N. Hansen and G. J. Hansen, Surf. Sci., 2001, 481, 172.
31. K. Hermann, K. Freihube, T. Greber, A. Böttcher, R. Grobecker, G. Ertl and D. Fick, Surf. Sci., 1994, 313, L806.
32. G. v. Helden, M.-T. Hsu, N. Gotts and M. T. Bowers, J. Phys. Chem., 1993, 97, 8182.
33. J. Hunter, J. Fye and M. F. Jarrold, Science, 1993, 260, 784.
34. A. Shvartsburg, R. Hudgins, P. Dugourd, R. Gutierrez, T. Frauenheim and M. Jarrold, Phys. Rev. Lett., 2000, 84, 2421.

---