Coalescence reactions in laser-induced fullerene desorption: the role of fragments

Transcription

1 Z Phys D 37, (1996) Coalescence reactions in laser-induced fullerene desorption: the role of fragments R Mitzner, B Winter, Ch Kusch, EEB Campbell, IV Hertel* Max-Born-Institut fu r Nichtlineare Optik und Kurzzeitspektroskopie, Postfach 1107, D Berlin, Germany Received: 18 September 1995/Final version; 16 November 1995 Abstract Coalescence reactions of molecules during the laser desorption from fullerene thin films have been investigated using nanosecond and picosecond laser pulses in order to directly identify the primary reaction steps Reflectron time-of-flight (RETOF) mass spectroscopy was used to obtain the mass and velocity distributions of the desorbed ions The coalescence products are seen to be formed predominantly in collisions between fragment ions and PACS: 3690#f; 6810Jy; 8265Yh 1 Introduction Pulsed laser desorption of fullerenes has been used extensively as an ion source for time-of-flight mass spectroscopy (LDMS) [1] and to produce beams of gas phase neutral and charged fullerenes used in molecular beam experiments [2] Under certain desorption conditions coalescence of the desorbed species can occur to produce larger fullerenes [1a, 3] In spite of the importance these reactions may have for understanding and optimising the production of large fullerene molecules very little is known about the details of the formation mechanism Experiments in which desorption was carried out under the influence of a cooling gas [3a] were the first to show that the coalescence products show maxima in the mass distributions at approximately integer multiples of The growth mechanism was suggested to involve the coalescence of intact molecules accompanied by C emission and recapture of C molecules to produce the mass peak distribution around ( ) [3b, 4] In another study reactive fragments produced in LID from O thin films have been interpreted to be the favourable reagents in gas phase product formation [5] * Also at: Fachbereich Physik, Freie Universität Berlin, Berlin, Germany It is still unclear as to whether these coalescence products are formed entirely in the gas phase or whether there is a contribution due to aggregation or oligomerisation on the surface Studies of UV (laser or lamp) irradiated films have shown evidence of photochemically induced covalently bound molecules on the surface [6] Fullerene-surface collision experiments with the coalescence products [3, 6] have provided evidence that these products are stable, closed cage fullerenes On the other hand, in a recent ion drift tube study the coexistence of structural isomers having a compact closed-cage fullerene structure and a dimer structure, shaped more like a peanut or dumbbell, have been identified under fullerene LID conditions [7] In this paper we continue our recent investigations on the LID of [8, 9] The results reported here provide more direct evidence that the coalescence products formed upon UV LID from fullerene thin films are largely due to collisions between fragment ions and neutral molecules 2 Experimental set-up A description of the experimental set-up has already been given in detail elsewhere [9] and will only be briefly described here films (300 nm thick) were deposited on single crystal Si(111) by evaporation of pure from an oven at a temperature of ca 560 K The lasers used in the experiment were a XeCl excimer laser (308 nm, repetition rate 10 Hz, pulse length 20 ns) and a Nd:YAG laser (266 nm, 10 Hz, 30 ps) focused by a 200 mm quartz lens onto the film surface The irradiated spot size was typically 1 mm Variation of the laser pulse energy was accomplished with optical filters leaving the laser pulse shape unchanged The TOF measurements were performed at a base pressure of 10 Torr Desorbed ions were allowed to drift for a distance of 55 mm before being extracted by a pulsed electric field and mass selected in a refelctron time-of-flight mass spectrometer The velocity distributions were determined by changing the time delay between the laser pulse and the pulsing of the extraction

2 90 field and appropriately transforming flight time into velocities 3 Results and discussion Mass spectra of ions produced by desorption with ns laser pulses are shown in Fig 1 as a function of laser fluence for a fixed delay time between laser pulse and extraction field At low fluences (Fig 1a: 19 mj/cm ) only and a very small contribution from and fragment ions can be observed For an intermediate fluence (Fig 1b: 25 mj/cm ) fragment species become more intense and can be observed down to At the same time coalescence products appear in the mass spectrum The highest dimer [10] mass product under these conditions is (not, although there may just be an indication of a mass peak emerging from the noise) The mass distribution of the coalescence products is highly asymmetrical with the ion yield falling off more or less uniformly towards smaller species down to Note that the coalescence species observed,, correlate directly with the fragment ions that can be seen in the mass spectrum At higher fluences (Fig 1c: 39 mj/cm ) the coalescence mass distribution becomes increasingly symmetric, centred around Under these conditions the relative amount of fragments is clearly increased compared to the 25 mj/cm fluence case In Fig 1c one can also observe the formation of trimer species The threshold fluence for the trimers lies slightly higher than that for the dimers as a consequence of the higher density of desorbed species required for their formation Their mass distribution at fluences close to the threshold for their formation is very similar to the dimer mass distribution as can be seen in Fig 1c This provides a strong indication that they are predominantly formed by collisions between dimer ions and neutral (or large neutral fullerene fragments) At still higher fluences, where the dimer distribution remains symmetric about, the trimer mass distribution shifts towards lower masses For LID with 30 ps pulses from the fourth harmonic of the Nd:YAG laser we obtain mass spectra which show a qualitatively similar dependence on desorption laser fluence In particular, the highest dimer mass which appears in the threshold region in The main difference is that the intensities of the fragments and coalesced species are smaller relative to the signal (and much smaller relative to the total amount of material desorbed [9]) for desorption with ps pulses From this first inspection of raw mass spectra the following conclusions concerning product formation may be drawn The fact that experimental conditions may be chosen such that the mass spectral distribution abruptly cuts off at and that the smaller coalescence products can be observed down to a mass corresponding to the sum of with the smallest observable fragment ion is very suggestive that at least one of the reagents is a fragment This is a reasonable assumption since such a fragment would be likely to be more reactive towards coalescence than a stable Furthermore, accounting for the fact that the rate constant for ionization of a molecule Fig 1a c TOF mass spectra of direct fullerene ions including, fragments and coalescence products obtained for 308 nm laser desorption (pulse width 20 ns) from 300 nm thick /Si(111) film for the three fluences as indicated in the figure Delay time was 60 μs [11] is larger than the rate constant for C evaporation [12], ie that ionisation is likely to occur before fragmentation, we would expect the fragment molecules to carry the charge in the collisions leading to coalescence products C #C P C, 254n(30 C #C P C etc (1) The products thus formed may cool during their passage through the mass spectrometer by C evaporation or photon emission For higher laser fluences where on the order of 50 or more monolayers per laser shot are desorbed [9] many collisions between fragments and already coalesced species are likely to occur This will lead to the addition of C (or higher) units to the primary coalescence products giving rise to C with n'118 Such a mechanism was already proposed in the earlier literature to account for formation of large fullerene species during laser desorption [1a, 3] An alternative mechanism could involve the exchange of carbon atoms in collisions between dimer

3 91 Fig 2a c Fluence dependence of the desorbed ion yield using 308 nm, 20 ns pulses for desorption The delay time was 60 μs The error bars represent the standard deviation of the measured ion yields a (full circles and full line) and integrated fragment ion intensities (open circles and dashed line) b Integrated dimer intensity: and C 120 display the ion intensities of the direct dimer products (see text) and the total product ion intensities (direct plus secondary products: see text for details), respectively c Integrated trimer intensity ions and neutral fullerenes which, at lower fluences and hence lower internal energies, would have led to the formation of stable trimers For a more quantitative presentation of the evolution of ions, their fragments and their coalescence products the signal dependence of each of these species on the laser fluence for a fixed delay time of 60 μs is shown in Fig 2 Figure 2a displays the integrated fragment ion signal summed over through (open circles and dashed line) along with the integrated ion signal (full circles and solid line) Figure 2b and c shows the corresponding data for products with masses centered near and The dimer product signal has been separated into two contributions as motivated by the discussion above: (i) the integrated signal summed over coalescence products with mass4 (open circles and dashed line) and (ii) the integrated signal summed over all dimer products (full circles and solid line) The curves in the figure are drawn to guide the eye The appearence of coalescence products coincides approximately with the maximum yield at a laser fluence of approximately 25 mj/cm, with the maximum in the dimer signal coinciding with the maximum in the fragment ion intensity For increasing laser fluences the and fragment ion signals (top panel) decrease while the total coalescence product signal increases as would be expected if the products grow at the expense of the monomers until reaching a maximum at around 60 mj/cm The onset of the trimer signal corresponds reasonably well with the fluence value at which dimer species with n'118 appear It should be stressed here that Fig 2 refers to a delay between desorption and extraction of 60 μs (the same delay used to obtain the mass spectra in Fig 1) For other values of the delay the various maxima and thresholds are shifted but the overall conclusions remain the same Unfortunately, the data presented in Fig 2 does not provide direct support for the formation mechanism discussed above since the and fragment ion signals show a very similar dependence on laser fluence Obviously, the problem with unequivocally identifying the underlying coalescence mechanism is that the fragmentation rate and particle density cannot easily be varied independently The information which is needed in order to provide more conclusive experimental evidence for the above suggested formation scheme (1) is the dependence of the velocities of the different ion signals as a function of laser fluence Such velocity distributions are shown in Figs 3 and 4 for two nanosecond laser fluences 25 mj/cm and 39 mj/cm, respectively For both laser fluences the and fragment ion velocity distributions are bimodal (panels (a) and (b), respectively) For lower laser fluences (using nanosecond pulses) we can only observe the slow peak The fragment ion yield (centre panels) is the sum of through The curves are modified Maxwell-Boltzmann fits to the experimental data The velocity distributions for the parent and fragment ions have been discussed in detail elsewhere [9] where we have shown that the slow peak component is due to purely thermal desorption processes with the desorbed molecules being in thermal equilibrium on desorption (ns pulses) The fast component observed in desorption with ns pulses is attributed to a thermal explosion due to a high pressure build-up in the top few monolayers, leading to the desorption of fast molecular ions [9] As is evident from Figs 3 and 4, the velocities of the coalesced species correlate predominantly with the velocities of the slow, thermal components of the and fragment ion distributions The velocity distribution of the trimer species (not shown) is very similar to that of the dimers One can also see that the apparently very low intensity of fragment ions in the slow component is due to the efficient coalescence of these species and not to a very much larger fragmentation rate of the fast ions compared to the slow ions Note that the maximum intensity of the coalesced species is more than a factor of two higher than the remaining fragments as indicated by the normalisation factor given in the figures Summing up the fragment ion yield and the dimer ion yield in Fig 3 approximately reproduces the shape of the velocity distribution The

4 92 Fig 3 a, b fragment and c dimer ion velocity distributions obtained from laser desorbed thin film Laser parameters: 25 mj/cm, 20 ns pulse width, 308 nm wavelength The bimodal distribution is fitted by two modified Maxwellians (slow and fast component), see [9] for details The translational energies given on the upper x-axis refer to the ions neutral molecules are expected to have a very similar velocity distribution to the slow component of the ions [9] (the proportion of ions, produced by thermionic emission, is expected to be much larger for the fast peak than for the slow thermal peak) so that, if our proposed formation mechanism is correct, we would expect the coalesced species to behave as shown in Figs 3 and 4 The lack of a fast velocity component for the coalescence products is thus presumably due to the much lower proportion of neutral species expected for the fast desorbed particles, as mentioned above, which greatly reduces the number of collisions which could lead to coalescence The high velocities of these fast fragment ions also means that they will very rapidly leave the high density interaction region where the collisions occur thus further decreasing the probability of coalescence Figure 5 shows similar velocity distributions for desorption with 30 ps pulses We have shown previously [9] that the slow velocity component for desorption with Fig 4a c As Fig 3 for a fluence of 39 mj/cm ps pulses also shows thermal characteristics although the internal temperature of the desorbed molecules was seen to be less than the translational temperature and the ions were attributed mainly to a direct two-photon ionisation rather than to thermionic emission As a consequence of the lower internal energy, the proportion of fragment ions observed is much lower than for desorption with ns pulses and the velocity distribution of the fragment ions is correlated with the high energy tail of the slow component of the velocity distribution The dimer ion intensity is also considerably lower than for desorption with ns pulses and is, as also seen in the ns case, about a factor of two larger than the fragment intensity The velocity distribution of the dimers is very similar to that of the fragment ions providing very convincing evidence for our proposed formation mechanism The results discussed here can be compared with the results of gas phase collision experiments carried out under single collision conditions (One has to bear in mind here that the conditions in the laser desorption experiment are not as straightforward as in the collision experiments and multiple collisions are certainly occuring in the desorption plume) The energetic barrier for fusion of

5 93 Fig 5a c As Fig 3 with laser parameters: 9 mj/cm, 30 ps pulse width, 266 nm wavelength with neutral to produce a ion which is stable on a μs timescale has been shown to be ca 66 ev on the centre of mass energy scale [13] This is a much larger kinetic energy than those which are available in the laser desorption experiments However, if we simply consider the total energy available to the colliding system (translational plus internal energy) and ignore possible kinematic effects we may expect to be able to start to see coalescence of with when the total energy is approximately ev (accounting for ca ev internal energy in the projectile ion used in the collision experiments) ie ev internal energy per colliding This energy corresponds to a temperature of about K which is certainly accessible for ns fluences beyond 25 mj/cm [9] We can thus not completely rule out a small contribution to the coalescene signal from this mechanism at high laser fluences The barrier for fusion of with has not yet been determined (experiments are in progress) but is expected to be significantly lower than that for # The discussion above has been concerned with laser desorption from films of which are thick enough to rule out any influence of the underlying substrate Care was always taken when averaging the mass spectra to ensure that only that maximum number of pulses was allowed for which the next following shot would quantitatively reproduce the former mass spectrum and that the underlying substrate was never exposed The results obtained under these conditions lead us to conclude that the coalescence products are produced by gas phase reactions We have, however, observed a remarkable effect when investigating the dependence of the mass spectra on the number of laser shots up to and beyond the number of shots which should completely remove the fullerene film in the centre of the laser focus Figure 6a shows the integrated intensities for and the integrated fragment intensities for to as a function of laser shot for desorption with 25 mj/cm ns pulses Under these desorption conditions the substrate is exposed in the centre of the laser focus after approximately 15 laser shots This corresponds to a minimum in the ion intensity, however, the intensity of the desorbed monomer ions strongly increases as the number of laser shots is increased further, reaching a maximum at about 40 shots The fragment ions show a similar but less pronounced effect with the maximum occuring at smaller shot numbers as the fragment size decreases This may be an indication that the temperature of the desorbed fullerenes is decreasing as the number of shots is increasing In Fig 6b we have plotted the integrated intensity of the dimer which reaches a minimum at 15 and then increases again to give a second maximum for laser shots, following the trend in the fragment ion intensities in Fig 6a In Fig 6c and d we compare the results obtained with a film of the same thickness (300 nm) on a stainless steel substrate The and fragment ion intensities show a similar dependence on laser shot as was observed for the Si substrate The coalescence products, on the other hand, show a more or less constant intensity up to the 15th laser shot and then fall rapidly in intensity and are not observed beyond ca laser shots We attribute the similar dependence of the and fragment ion intensities in both cases to a reflection of the laser light at the substrate surface (Si and stainless steel have similar reflectivities for the wavelength used in these experiments) thus effectively giving a higher laser fluence when the fullerene film has been almost completely removed in the centre of the beam focus As the number of shots is increased further the material being removed comes from an area further and further away from the focus centre and thus the temperature of the desorbed species will decrease as will the overall density of desorbed species This density effect could explain the rapid drop in intensity of coalescence products after 15 shots on the stainless steel substrate even although the fragment ion and intensities reach their maximum at 30 and 40 laser shots, respectively The reason for the different results obtained for the coalescence products formed on desorption from a Si substrate is not yet understood The effect is not only observed for the dimers but a similar effect can also be seen for trimers and even tetramers at slightly higher laser fluences Figure 7 shows the results obtained for desorption with ca 30 mj/cm The integrated trimer and tetramer intensities remain fairly constant with increasing number of laser shots until about n"15 and then rise to a maximum with the position of the maximum occuring after a smaller number of laser

6 94 Fig 6a d Dependence of ion signal on number of laser shots Laser fluence 25 mj/cm, 308 nm, 20 ns pulses a and fragment ion signal from 300 nm film on Si, b coalescence products from 300 nm film on Si, c and fragment ion signal from 300 nm film on stainless steel, d coalescence products from 300 nm film on stainless steel In all cases shown the underlying substrate could be seen with the naked eye after 15 laser shots is observed when colliding with SiO compared with targets such as graphite [15] It may be that a strong distortion or even break-up of the fullerene cage is occuring in the first few monolayers close to the Si surface which is leading to an enhanced probability of coalescence reactions A similar dependence of coalescence products on laser shot number can also be observed when silicates (mica) are used as the substrate providing colloborative evidence for the role of Si in these reactions A full explanation of this effect will, however, have to await the results of more detailed experiments 4 Conclusions Fig 7 Coalescence products observed on desorption from 300 nm thick film on Si (308 nm, 20 ns, 30 mj/cm ) as a function of number of laser shots shots for the higher coalescence products The reason for the initial increase in the dimer intensity with number of laser shots which can be seen in Figs 6 and 7 is also unclear and maybe indicative of the occurence of photoinduced surface processes The effect decreases with increasing substrate temperature and cannot be observed for high laser fluences ('50 mj/cm ) It is known that reactions can occur between fullerenes and Si to form Si-C bonds between the first monolayer and the Si surface [14] There is also some evidence from collision experiments that an increased fragmentation of Investigations of the mass spectra and velocity distributions of the ions produced during laser desorption of thin films have provided information on the mechanisms leading to coalescence of fullerenes For films which are thick enough to rule out any substrate influence the main contribution is seen to be due to gas phase reactions between fullerene fragment ions ( to ) and neutral An increase in coalescence reactions can be observed, using Si as substrate, as the number of laser shots is increased sufficiently to expose the underlying substrate in the centre of the laser focus This was not observed on desorption of from a stainless steel substrate The effect is attributed to reactions occuring between the fullerenes and the Si surface leading to distortion or break-up of the fullerene cage thus enhancing the probability for coalescence

7 95 References 1 eg (a) Ulmer, G, Campbell, EEB, Kühnle, R, Busmann, H-G, Hertel, IV: Chem Phys Lett 182, 114 (1991); (b) Bethune, DS, Meijer, G, Tang, WC, Rosen, HJ: Chem Phys Lett 174, 219 (1990); (c) Yeretzian, Ch, Hansen, K, Alvarez, MM, Min, KS, Gillan, EG, Holzer, K, Kaner, RB, Whetten, RL: Chem Phys Lett 196, 337 (1992); (d) Yeretzian, Ch, Alvarez, MM, DiCamillo, B, Whetten, RL: SPIE Proc 1857, 60 (1993); (e) Campbell, EEB, Tellgmann, R, Ru chardt, C, Gerst, M, Ebenhoch, J: NATO ASI 443, 27 Landon New York; (Kluwer, 1994) 2 (a) Campbell, EEB, Ehlich, R, Westerburg, M, Hertel, IV: in Electronic and atomic collisions T Andersen, B Fastrup, F Folkmann, H Knudsen, N Andersen, (eds) Invited Paper of XVIII ICPEAC Aarhus (1993) AIP Conference Proc 295 (AIP 1993) ; (b) Sprang, H, Mahlkow, A, Campbell, EEB: Chem Phys Lett 227, 91 (1994) 3 (a) Yeretzian, Ch, Hansen, K, Diederich, FD, Whetten, RL: Nature 359, 44 (1992); (b) Yeretzian, Ch, Hansen, K, Diederich, F, Whetten, RL: Z Phys D 26, 300 (1993); (c) Beck, RD, Weis, P, Bräuchle, G, Kappes, MM: J Chem Phys 100, 262 (1994) 4 Hansen, K, Yeretzian, Ch, Whetten, RL: Chem Phys Lett 218, 462 (1994) 5 (a) Beck, RD, Stoermer, C, Schulz, Ch, Michel, R, Weis, P, Bräuchle, G, Kappes, M: J Chem Phys 101, 3243 (1994); (b) Beck, RD, Bra uchle, G, Stoermer, C, Kappes, M: J Chem Phys 102, 540 (1995) 6 (a) Rao, AM, Ping Zhou, Kai-An Wang, Hager, GT, Holden, JM, Ying Wang, Lee, W-T, Xiang-Xin Bi, Eklund, PC, Cornett, DS, Duncan, MA, Amster, IJ: Science 259, 955 (1993); (b) Cornett, DS, Amster, IJ, Duncan, MA, Rao, AM, Eklund, PC: J Phys Chem 97, 5036 (1993); (c) Akselrod, L, Byrne, HJ, Thomsen, C, Roth, S: Chem Phys Lett 215, 131 (1993) 7 Hunter, JM, Fye, JL, Boivin, NM, Jarrold, MF: J Phys Chem 98, 7440 (1994) 8 Campbell, EEB, Hertel, IV, Kusch, Ch, Mitzner, R, Winter, B: Synth Met (in press) 9 Winter, B, Mitzner, R, Kusch, Ch, Campbell, EEB: J Chem Phys (submitted for publication) 10 Here the term dimer refers to a particle having the approximate mass of ( ) but with non-specified geometry 11 Klots, CE: Chem Phys Lett 186, 73 (1991) 12 Klots, CE: Z Phys D 20, 105 (1991) 13 Rohmund, F, Campbell, EEB, Knospe, O, Seifert, G, Schmidt, R: Phys Rev Lett (submitted for publication) 14 Hu, C-W, Kasuya, A, Suto, S, Wawro, A, Nishina, Y: Proc of ISSPIC 7, Book of Abstracts, p 232 (1994) 15 Hamza, AV: Balooch, M, Moalem, M, Olander, DR: Chem Phys Lett 228, 117 (1994)