JOURNAL OF CHEMICAL PHYSICS

VOLUME 110, NUMBER 14

8 APRIL 1999

Photolysis experiments on SiC mixed clusters: From silicon carbide clusters to silicon-doped fullerenes M. Pellarin,a) C. Ray, J. Lerme´, J. L. Vialle, and M. Broyer Laboratoire de Spectrome´trie Ionique et Mole´culaire (U.M.R. 5579) 43, Bld du 11 Novembre 1918, 69622 Villeurbanne Cedex, France

X. Blase, P. Ke´ghe´lian, P. Me´linon, and A. Perez De´partement de Physique des Mate´riaux (U.M.R. 5586) 43, Bld du 11 Novembre 1918, 69622 Villeurbanne Cedex, France

~Received 28 August 1998; accepted 7 January 1999! Silicon carbon binary clusters are generated in a laser vaporization source from Six C12x mixed targets ~x50 to 50%!. We have first analyzed stoichiometric (SiC) n (n3) cannot be evidenced in abundance mass spectroscopy because of unavoidable mass coincidences. A careful analysis of the photofragmentation behavior of selected sizes relative to the laser fluence nevertheless succeeds in indicating the contribution to the photofragmentation spectra of largely doped heterofullerenes C2n2q Si1 q ~q57 at least! that mainly dissociate by the loss of small even-numbered mixed molecules such as Si2 ,Si3C,... . Both approaches are consistent with the surprising capability of substituting a large number of silicon atoms into fullerenes without destabilizing their cage structure too much. In this respect, a value close to 12 seems to be an upper limit. © 1999 American Institute of Physics. @S0021-9606~99!70514-3#

I. INTRODUCTION

synthesized from the direct growth of clusters from a initially doped carbon-based material, or by starting with preformed C60 ~or C70! precursors interacting with a metallic vapor of the doping element. In the last case, the most probable result is an external coating of the fullerene cage by the foreign atoms.6 Substitution processes can nevertheless be induced by an additional energy supply ~laser, collisions!; the heterofullerene stability is favored by quenching phenomena.7 The fullerene doping through substitution is expected to be made much easier with elements having electronic configurations much closer to carbon than the metals. From this point of view, boron and nitrogen are good candidates. In the bulk phase, boron nitride is isoelectronic to graphite and has the same propensity to form two-dimensional hexagonal sheets. BN nanotubes have then been evidenced,8 even if frustration effects make their endings slightly distorted as compared to pure carbon ones.9 Guo et al. have nicely shown the possibility of substituting up to six boron atoms in fullerenes in a rather easy way.10 Numerous experimental11–13 and theoretical studies13,14 also extended to nitrogen doping have followed. When substituted in a fullerene

The doping of fullerenes with metal atoms is mainly of interest for the processing of fullerene assembled materials having specific physical and chemical properties. These properties are expected to arise from the individual electronic and geometric structures of the doped cages and also from those related to the chemical bonding induced by the dopant atoms. Since the discovery1 and large-scale synthesis of fullerenes,2 numerous studies were devoted to systems among which one can distinguish endohedrally doped ~foreign atom inside the cage!,3 exohedrally doped ~foreign atom outside the cage!,4 and substitutionally doped ~foreign and carbon atom exchanged!5 fullerenes. The two first categories have drained a lot of experimental and theoretical effort from the earliest stages of fullerene science. The third possibility of trapping metal atoms as a part of the fullerene network has been evidenced more recently and has proven to be less easy. Two main experimental approaches can be distinguished concerning the generation of doped fullerenes: they can be a!

Electronic mail: [email protected]

0021-9606/99/110(14)/6927/12/$15.00

6927

© 1999 American Institute of Physics

6928

J. Chem. Phys., Vol. 110, No. 14, 8 April 1999

cage, N and B atoms are responsible for impurity states similar to what is observed with n-type or p-type doped semiconductors.15 The amazing electronic properties of these systems, together with the underlying chemical bonding possibilities they offer, make them ideal candidates for a new class of derived semiconducting materials since they have shown to be synthesized in macroscopic amounts.11 At first glance, silicon should be the more logical candidate for being substituted in the fullerene network since it is the isoelectronic element that is closest to carbon in the periodic table. Unfortunately, it is known for preferring sp 3 -like bonding when carbon can easily generate single, double, or triple bonds. This is quite obvious for pure silicon clusters that are known to adopt compact three-dimensional arrangements.16 For this reason, silicon doping is unlikely, barring strong fullerene network deformations that should rapidly make them unstable as the number of substituted atoms is increased. Experimental studies on silicon-doped heterofullerenes have surprisingly shown the stability of singly and doubly silicon substituted fullerenes. This was previously illustrated through mass abundance spectroscopy17,18 and mobility measurements19 on heterofullerenes produced by laser ablation of a silicon-doped carbon target. In the case of C2n22 Si1 2 clusters, photofragmentation experiments have recently shown the proximity of silicon atoms in the carbon network that was corroborated by ab initio calculations.20 An interesting issue concerning these heterofullerenes is to know to what extent silicon atoms can be substituted without destroying the cage network, and how the doped atoms arrange themselves in the cluster. This paper will try to provide experimental indications about these points. The main characteristics of the experimental setup and the particular fittings for photofragmentation studies will be given in the following section. Section III will deal with the photoevaporation experiments on neutral SiC mixed clusters generated from a silicon carbide target. Their transformation into silicon-doped heterofullerenes will be discussed in relation to the fragmentation processes that can be observed on selected positively charged SiC clusters. In Sec. IV, we will be interested in the photodissociation response of cationic clusters directly produced from a carbon-rich target. Their arrangement as siliconized heterofullerenes will be inferred from the analysis of their photoproducts. II. EXPERIMENT

A schematic view of the experimental setup is given in Fig. 1. Silicon-carbon mixed clusters are generated in a standard laser vaporization source.21 A Nd:YAG laser ~532 nm, 10 Hz! is used to create a high temperature plasma from a Six C12x mixed target. The quenching of the plasma by a pulsed high pressure helium burst insures the cluster growth in the source chamber until the cluster-carrier gas mixture is expanded through a conical nozzle. The supersonic beam is skimmed and directed into an analysis chamber. The only thermodynamically stable Si–C alloy is the conventional crystalline silicon carbide phase with the stoichiometry x 50.5. Targets of this kind and grown as rods ~Morton Inc.!

Pellarin et al.

FIG. 1. Schematic view of the experimental setup.

are, in a first step, used to generate SiC mixed clusters with the same mean composition. They will be referred to as stoichiometric clusters in the following. Mixed clusters with other compositions are obtained from composite rods processed by binding silicon and graphite powders in various proportions.18 The cluster size distribution is studied by a reflection time-of-flight ~RTOF! spectrometer set perpendicular to the incoming cluster beam and described elsewhere.22 The neutral clusters are photoionized by a pulsed ArF excimer laser at a fixed photon energy of 6.4 eV. Clusters directly born as ions in the source are analyzed by a suited triggering of pulsed high voltages applied on the grids of the two-stage accelerating system ~Wiley-McLaren configuration! made of a low electric field extracting zone ~ionization region! followed by a strong field accelerating zone. In addition to mass spectrometric measurements, photofragmentation ~PF! experiments on selected cluster sizes can be realized by taking advantage of the reflectron geometry in a tandem TOF analysis. Accelerated clusters that enter the free flight are mass selected by an electrostatic gate located at the first time focusing point of the spectrometer. The location of this point is defined by the two accelerating fields ratio and chosen as the best compromise between the overall RTOF mass resolution and the mass selection sharpness. Just after this point, the selected clusters are shined on by the focused beam of an XeCl excimer laser ~4.03 eV!, the intensity of which can be discretely attenuated by metallic grids. The charged products ~mass M f ! evaporated from the heated parent cluster ~mass M p ! during the remaining flight time up to the grounded entrance grid of the reflector are dispersed in their returning path to the detector and are then detected at shorter arrival times. The fragmentation mass spectra will not be sensitive to the dissociation processes occurring in the reflection and returning stages. The characteristic time for dissociation is then given by the flight time from the heating zone to the reflector ~50 cm! and is about 0.6AM m s where M is the mass of the parent in a.m.u. The differences between the parent and ionic fragment arrival times are easily converted in mass losses. If E p is the mean initial kinetic energy of accelerated clusters ~only singly charged cations will be considered here!, fragments will carry the kinetic energy E f

J. Chem. Phys., Vol. 110, No. 14, 8 April 1999

Pellarin et al.

6929

tive description, the fragmentation pattern of Fig. 2~a! has been simulated by a sequential bimodal evaporation decay of Si2C and Si3C with respective probabilities 70% and 30% @Fig. 2~b!# deduced from the first-step product intensities. The agreement with experiment is quite good and confirms the sequential character of evaporation from these two products and the constancy of the evaporation branching ratio within a decay chain. This last point is similarly evidenced when different cluster sizes are photofragmented. The photoevaporation of initially stoichiometric clusters is then resulting in a relative impoverishment in silicon.

B. Multiphoton ionization of neutral clusters

FIG. 2. ~a! Photofragmentation mass spectrum of the selected (SiC!40 size ~most likely composition!. The peak at 0 a.m.u. corresponds to the nondissociated precursor clusters. The first-step losses of Si2C and Si3C molecules are indicated. The vertical dotted line separates two size domains recorded with different reflector voltage settings. ~b! Simulation of the sequential decay of Si2C and Si3C as major neutral evaporated molecules.

5Ep3M f /M p in a first approximation, and this gives rise to an experimental difficulty. The voltages applied on the reflector grids (U t ,U k ) are not suited to achieve optimum mass resolution for both parent and fragment clusters. As the kinetic energy ratio M f /M p decreases, the fragment peaks are enlarged compared to the parent. The resolution deterioration becomes definitely awkward when fragments only explore the first repelling field zone (E p 3…

In the case of $ C1 2n % mass clumps, Fig. 7 displays low fluence fragmentation patterns for three different selected sizes. They are rather similar, and apart from the small peaks for a 24 a.m.u. (C2) or a 48 a.m.u. loss (C4) that are reminiscences of pure carbon fullerene fragmentation, and for a 68 a.m.u. (Si2C) loss that is an indication of a small amount of stoichiometric isomers selected at the same time ~Sec. III!, they should reflect the dissociation sequences of siliconized 1 isomass species such as C2n-7 Si1 3 ,C2n214Si6 ,..., in descending amounts. The strong resemblance between these spectra shows that they are ruled by a generic sequence of evaporation common to any selected size. This should help us in trying to deconvoluate the photoinduced dissociation response of silicon containing clusters that are simultaneously selected. Even if one can get rid of the pure C1 2n cluster contribution, obtaining quantitative information from such complicated spectra is far from easy. Different decay chains are superimposed and the relative intensities among photoproducts may depend on the laser fluence through the activation energies for single or multistep evaporative processes and the photoabsorption cross section of the different selected parents. Nevertheless, it is clear that when an even-numbered cluster is selected, all the isomass compositions, if present, also do correspond to even-numbered species as seen before. Moreover, all the photoproducts detected in Fig. 7 can be assigned to even-numbered clusters, as the mass separation between them is nothing else than a multiple of 8 a.m.u. ~see Sec. III!. In an equivalent way, all the evaporative decay chains from which they emerge consist of an even total num-

Pellarin et al.

ber of silicon and carbon atoms. This feature is an unquestionable signature of a fullerenelike structure for these photodissociated clusters. Two important questions remain to be answered: what is the most siliconized isomass composition C2n2q Si1 q ~q maximum! that contributes to such a dissociation pattern, and what is the exact nature of the evaporated neutral molecules? Getting reliable information about these points is not straightforward and requires a careful but somewhat difficult analysis of the PF mass spectra. To guide the reader, we can first summarize the process of successive elimination exposed in the following section. In a first stage, we have shown that recording low laser fluence PF spectra of $ C1 2n % mass bunches is a means to simplify them by eliminating the contribution of the most abundant and most carbon-rich isomass clusters ~C1 2n in this case!. In a second stage, we will infer the contribution of the next isomass clusters containing silicon (C2n27 Si1 3 ). Finally, this information will be used to deduce the contribution, and therefore the presence, of more siliconized species (C2n214Si1 6 ) necessary to account for extra features in the spectra. The same procedure applied to a $ C2n21 Si1% mass bunch will also give evidence for the fragmentation signature of C2n28 Si1 4 and C2n215Si1 7 species. The problem of mass coincidences again prevents unambiguous assignments for the compositions of both the charged photoproducts and the corresponding neutral lost molecules. We will make the assumption that the amount of intermediate charged products containing more than three silicon atoms is weak enough to be neglected with regard to their fragility toward photoexcitation. They will be labeled as 1 1 belonging to C1 2p , C2p21 Si , or C2p22 Si2 series. To proceed further, it must be kept in mind that the photoproduct sizes and relative intensities result from the convolution of both their intrinsic stability and the probability of the decay chain by which they are connected to their parent, or, in other words, the nature of the neutral preferentially evaporated constituents. In Fig. 7, the absence of detectable Si loss ~28 a.m.u.! is in favor of a direct evaporation of the Si2 molecule ~peak A!, which is consistent with previous observations on the dissociation of C2n22 Si1 2 clusters. The most probable product is C2n27 Si1 rather than C2n214Si1 4 , as discussed above. For similar reasons, peak B ~96 a.m.u. loss! may originate from a direct Si3C loss from C2n27 Si1 3 or from a SiC loss from the intermediate product C2n27 Si1. In both cases the remaining composition is C1 2n28 . Although the two-step process cannot be completely ruled out since the C2n27 Si1 intermediate product is expected to evaporate a SiC molecule,20 the former hypothesis is certainly the most likely owing to the apparently high activation energy for a direct SiC evaporation. This one is not detected as a first-step process ~40 a.m.u.!, the C2n28 Si1 2 product being not apparent. The A and B largest peaks can then be considered as the major products of the C2n27 Si1 3 dissociation. In the left part of the spectra, peaks detected for pure carbon sizes are partly the result of the next C2 sequential decay from the C1 2n28 intermediate fullerene product. Two small peaks are also detected for 64 and 80 a.m.u. losses. The most probable explanation for these features is a direct loss of SiC3 and Si2C2 molecules rather than sequential ~SiC1C2 or Si21C2! pro-

J. Chem. Phys., Vol. 110, No. 14, 8 April 1999

FIG. 8. Comparison between the low fluence fragmentation spectra of $ C1 66% ~a! and $ C69Si1% ~b! selected clumps. The neutral evaporated molecules corresponding to the most abundant products are specified. Residual peaks from the C69Si1 sequential evaporation (SiC1C21C2...) are indicated by 1 crosses ~x! in ~b!. The photoproducts assigned to the C1 or 2n, C2n21 Si C2n22 Si1 most probable compositions are indicated by full circles ~d!, 2 empty circles ~s!, and crosses ~1!, respectively.

cesses. In a sequential scheme, the intermediate products 1 C2 p22 Si1 would contain at least one silicon 2 or C2p21 Si atom prior to the C2 evaporation. This is unlikely since the photofragmentation of C2n21 Si1 or C2n22 Si1 hetero2 fullerenes has shown that the C2 ejection is always observed after the complete elimination of silicon via SiC or Si2 former evaporation steps.20 At this point, the fragmentation pattern of C2n27 Si1 3 is at least well defined. Additional information can be obtained in the same way by studying the fragmentation of $ C2n21 Si1% rather than $ C1 2n % selected clumps. For a priori singly siliconized heterofullerenes, low fluence fragmentation patterns are also very similar, whatever the selected size. An example of such spectrum is given in Fig. 8~b! for $ C69Si1% and compared to the one of $ C1 66% in Fig. 8~a!. Exponentially decreasing peaks at 40, 64, and 88 a.m.u. are the relics of the most abundant C69Si1 clusters. They correspond to the initial SiC and the later sequential C2 evaporations detected in previous experiments on clusters grown from more weakly Si-doped targets.20 The other observed products then belong to the evaporation sequences from more siliconized isomass clus1 ters such as C62Si1 4 ,C55Si7 ,... and so on. As discussed for 1 1 C2n27 Si3 @C59Si3 in Fig. 8~a!#, the Si2 and Si3C losses are quite obvious, but in this case, one can observe a large amount of photoproducts that can be assigned to Si4 and Si4C2 global losses. These decays are consistent with a favorable loss of four silicon atoms from the second most abundant C62Si1 4 clusters in the selected bunch to provide the stable C1 and C1 62 60 intermediate products. In the left part of the spectra, peaks detected for pure carbon sizes ~full circles! are partly the result of the next C2 sequential decay from these intermediate products. Concerning the Si3C loss, the possible Si21SiC sequential decay invoked for C2n27 Si1 3 clusters is unlikely here because an initial Si2 loss ~C62Si1 2

Pellarin et al.

6935

FIG. 9. Comparison between the low fluence fragmentation spectra of $ C1 66% ~top! and $ C69Si1% ~bottom! selected clumps restricted to the left side of spectra displayed in Fig. 8. The horizontal scales of both spectra are relatively shifted by 16 ~a! and 40 a.m.u. ~b!. In the upper part of Fig. 9~a!, the mass loss at 152 a.m.u. (C51Si1) is indicated by dashed line ~1!.

intermediate product! would proceed further by another Si2 rather than SiC loss. This is in favor of a direct Si3C loss in both cases. Weak Si2C2 and SiC3 losses in Fig. 8~b! can be explained as for $ C1 66% . At this point, it is not possible to decide if the Si4 and Si4C2 evaporations are direct or sequential ~Si21Si2 and Si21Si2C2 or Si21Si21C2, for instance! 1 As for C2n27 Si1 3 , the photodissociation from C2n28 Si4 is clear. The remaining question is the origin of smaller peaks in the left side of fragmentation spectra ~below 120 and 138 a.m.u., respectively! in Fig. 8. Some of them ~pure fullerene products! can in part be explained by the sequential C2 decay following the total elimination of silicon from these sizes. If these light products are assumed to arise from the 1 dissociation of C2n27 Si1 3 or C2n28 Si4 , they will therefore result from a direct evaporation of large Sin Cm molecules ~with n50, 1, 2 and m@n!. A sequential decay with the Si3C and Si2 molecule evaporations as the earliest steps would not be suited to explain the birth of products containing up to two silicon atoms. But, on the other hand, if large and stable Sin Cm molecules were evaporated as a whole, this should lead to similar photoproduct size distributions for 1 both C2n27 Si1 3 and C2n28 Si4 , which is not the case in Fig. 8. The hypothesis that these sizes are the precursors of the lightest photoproducts must be eliminated. Figures 9~a! and 9~b! show a great similarity between the interesting regions 1 of the $ C1 66% and $ C69Si % fragmentation mass spectra in Fig. 8 if relative shifts of 16 or 40 a.m.u. are, respectively, performed in the horizontal axis. This is an indication of a sequential decay scheme from the more siliconized C52Si1 6 or 1 1 C55Si1 clusters selected in the respective C and C Si $ % $ % 69 7 66 clumps. These shifts indeed amount to comparing the decay 1 chains from $ C1 66% and $ C69Si % as if they originate from intermediate products related to Si3C and Si4 losses, respectively, in the case of Fig. 9~a! or to Si2 and Si3C losses in the case of Fig. 9~b! ~see Fig. 8 for a complete display!. At this point, and in a sequential decay scheme, these intermediate products are assigned to clusters having the same number of 1 silicon atoms: C51Si1 3 or C55Si3 @case of Fig. 9~a!# and 1 1 C52Si4 or C54Si4 @case of Fig. 9~b!#, if they are respectively 1 assumed to originate from C52Si1 6 and C55Si7 . This explains

6936

J. Chem. Phys., Vol. 110, No. 14, 8 April 1999

the similarity observed in the photoproduct distributions. This similarity is not perfect because, although they are important, the Si2, Si3C, or even Si4 losses are not the only ones that may be involved in the first stages of sequential decays, and their order of occurrence may vary. Specific size stabilities for the photoproducts may also occur. Nevertheless, the most likely explanation is that small photoproducts 1 1 in Fig. 9 arise from C52Si1 6 or C55Si7 ~C2n214Si6 or 1 C2n215Si7 in general!. The corresponding decays bring into play the evaporation of neutral molecules that can be inferred from the fragmentation patterns of less siliconized C51Si1 3 or 1 1 C54Si1 clusters ~C Si or C Si in general!. As an 2n27 3 2n28 4 4 illustration, the peak labeled ~1! in Fig. 9~a! is the first onesilicon containing cluster (C51Si1) in this size region. It can be explained by the concurrent losses of Si2 and Si3C mol1 ecules from C52Si1 6 . The fact that the C53Si product is not detected is well understood since it contains more carbon atoms than its supposed precursor (C52Si1 6 ) does and cannot be connected to it by any decay sequence. Owing to their superimposition, the richness in the competing dissociation channels and the dependence of activation energies with the internal degree of excitation of the fragments, these features are difficult to describe more quantitatively in a similar way as stoichiometric clusters in Sec. III. However, there is little doubt about the heterofullerene nature of clusters containing up to seven silicon atoms at least. Finally, a too low signal has prevented us from observing the fragmentation signature 1 of C2n216Si1 8 clusters in the $ C2n22 Si2 % corresponding clumps. V. SUMMARY AND CONCLUSION

The multiphoton ionization mass spectroscopy of initially stoichiometric SiC clusters has shown that doped heterofullerenes appear as stable products as soon as a sufficient amount of silicon atoms is evaporated. This is confirmed by the laser fragmentation study on selected sizes. The cooling of photoexcited SiC clusters is ruled by the evaporation of Si2C and Si3C molecules. Sublimation experiments performed on silicon carbide indicate that Si2C, SiC2, SiC, and Si3C are the main mixed molecules present in the vapor above the bulk.39 A quantitative comparison with clusters is not possible, just like in the case of pure bulk carbon40 and bulk silicon,41 because finite size effects are of main importance with regard to the surface arrangement in small particles. For instance, SiC2 is found to be a major product for smaller sizes than those presented in this work. Nevertheless, a qualitative agreement enlightens the certainly compact nature of these clusters in which the chemical ordering of the silicon carbide structure shall be retained at least on the average.25,42 It is interesting to notice the great similarity observed between the silicon evaporation that leads to the appearance of fullerene networks in clusters and the separation among silicon carbide and graphite phases when annealing bulk silicon carbide.43 At this point, it is possible to only speculate about the structural transformation from SiC clusters to heterofullerenes. Since photoexcited clusters evacuate Si rich neutral molecules and since this process is effective during quite long decay chains, the remaining silicon atoms

Pellarin et al.

must migrate to the cluster surface that must rearrange between two consecutive evaporation steps. This silicon surface segregation will induce the growth of a carbon-rich inner core. Does that core remain compact ~diamondlike bonding! or soon develop as a cage network that will be the skeleton of the future heterofullerene? Molecular dynamics studies should check these speculations. Anyway, the detection of heterofullerenes from the annealing of silicon carbide clusters is not the misleading result of the possible simultaneous selection of isomer clusters already having an approaching structure. A study is in progress in order to analyze the sizes at which they appear in the mass spectra as a function of both the size of their precursors and the mean composition of the latter by changing the composition of the targets. Preliminary results show that heterofullerene products can be formed as soon as an upper limit for the number of remaining silicon atoms is reached. This limit is close to what is inferred in Sec. III and is little dependent on the size of the precursors. This one must be large enough so that heterofullerenes appear for an assessed total number of atoms which is above the lower limit (n532) for pure carbon fullerene stability. Small SiC annealed clusters are indeed unable to produce heterofullerenes owing to the too poor amount of carbon atoms they initially contain. On the other hand, direct photofragmentation experiments on mixed cold cations have definitely shown the possibility of at least doping up to seven silicon atoms into carbon fullerenes. This is quite surprising with regard to the large sp 2 character of carbon hybridization within the graphitelike network of fullerenes, which is hardly compatible with the single bonding tendency of silicon ~preferred sp 3 hybridization!. The signature of such a misfit is enlightened in the results of previous computational studies on the stability and structure of C2n21 Si1 and C2n22 Si1 2 heterofullerenes where silicon atoms induce slight deformation of the cages. They indeed tend to pop out from the spheres and partially accommodate their dangling bonds by pairing as close neighbors.20 It will then be interesting to extend such studies to a larger number of silicon doping atoms. Fragmentation pathways may give indications about their atomic arrangement within the fullerenes. Concerning pure carbon fullerenes, the nature and the energetics of the fragmentation mechanisms have been the subject of controversy from both experimental and theoretical point of views, but the statistical behavior of the unimolecular decomposition of fullerenes ruled by the C2 unit loss is now commonly acknowledged. The cohesive energy of molecules that are candidates to be evaporated from a cluster is often a good criterion for describing dissociation processes, as in the case of silicon clusters where Si6 and Si10 are preferentially highly bounded fragmentation products.44 But just as for fullerenes, the preferential evaporation of SiC and Si2 from C2n21 Si1 and C2n22 Si1 2 , respectively, is better explained by the stability of the remaining pure carbon fullerene products, although SiC and Si2 have lower cohesive energies than C2 in any case.45 The geometric arrangement of silicon in heterofullerenes is certainly responsible for a low activation energy of these processes. This is particularly true for C2n22 Si1 2, where the silicon atoms, even if they are not predicted to be

Pellarin et al.

J. Chem. Phys., Vol. 110, No. 14, 8 April 1999

FIG. 10. Symbolic ball and stick representation of a hypothetical C55Si5 heterofullerene with five doped silicon atoms arranged as a pair and a triplet of second neighbors ~Ref. 20!. The bond distances and angles are not the result of an optimized calculated geometry and are just distorted for a convenient display.

first neighbors, are close enough to be considered as part of a nearly preformed molecule bridged above the fullerene structure ~Fig. 10!. The observation of the Si3C evaporation from more siliconized heterofullerenes could be explained by the almost preformation of this molecule: a central carbon atom being connected to three silicon atoms at the endings of a roughly equilateral triangle ~Fig. 10!. In pure carbon fullerenes, the direct evaporation of even-numbered molecules larger than C2, such as C4 ,C6..., has proven to be negligible.33,38 The situation is different for silicon heterofullerenes, where tetramers such as Si3C, and to a less extent Si2C2 and SiC3, are detected. A further analysis should give more precise indications on the even possible evaporation of larger molecules like Si4C2 or Si5C. The unzipping mechanism invoked to account for the evaporation of evennumbered carbon chains in pure fullerenes37 is unlikely in the case of heterofullerenes since the atoms are not equivalent. The cluster reorganization that will follow the evaporation of the Sin Cm molecules as a whole must be favored here by the residual internal energy provided by the laser excitation. Molecular dynamics studies on the annealing and fragmentation of fullerenes have indeed shown the high degree of surface reorganization they can undergo without breaking.46 It is also interesting to note the completely different behavior upon fragmentation between silicon- and boron-doped heterofullerenes where the main products of evaporation are found to be C2 rather than BC or B2 units.10 Reactivity experiments indicate that boron atoms are dispersed at the surface and are not likely to stay close together as confirmed theoretically and contrary to silicon. An easier substitution of boron is also evidenced in mass spectroscopy. On the contrary, theoretical studies show the tendency for silicon-doped atoms to gather close to each other instead of being dispersed at the cluster surface. For instance, the C2n22 Si1 2 configura-

6937

tions where the silicon atoms belong to the same cycle at the surface are more stable than those with two silicon atoms placed at opposite sides in the fullerene cage. By extension, in more doped heterofullerenes with up to seven silicon atoms, the concentration of the doping element in a restricted area should strongly destabilize the cage structure, which is not observed. An alternative explanation could be that strongly doped heterofullerenes do not adopt the basic arrangement of carbon fullerenes ~hexagons 112 pentagons!47 any longer. Some isomeric structures could allow a more homogeneous distribution of silicon atoms. For instance, among the various families of polyhedrons that can be generated from the Euler’s theorem, the one consisting of replacing the 12 pentagons of the fullerenes by six squares is particularly interesting.48 It has mainly been used as the basis of theoretical studies on the boron–nitrogen analogs of the fullerenes since it allows us to construct cage molecules with a full alternation of B–N bonds thanks to the even parity of their cycles.49 The magic size sequence that can be inferred from these structures is very similar to the fullerene one. By analogy with these studies, one can speculate that the square faces in such cages could be interesting hosts for silicon doping atoms. This hypothesis will be checked in light of planned experiments on the transformation of SiC clusters into heterofullerenes, as discussed above. By showing the possibility of a rather large doping of silicon into carbon fullerenes, the next stage will be to study their properties when assembled on a substrate. The synthesis of large amounts of such heterofullerenes is certainly far from being easy, as shown for boron- or nitrogen-doped analogs.11 It is less ambitious, but nevertheless interesting, to process thin films by means of their ballistic deposition after being produced in the gas phase. Previous experiments of this kind performed on silicon carbide clusters50 are now enlarged to carbon-rich clusters that are expected to retain the fullerene network. 1

H. W. Kroto, J. R. Heath, S. C. O’Brien, R. F. Curl, and R. E. Smalley, Nature ~London! 318, 162 ~1985!. 2 W. Kra¨tshmer, L. D. Lamb, K. Fostiropoulos, and D. R. Huffman, Nature ~London! 347, 354 ~1990!. 3 J. R. Heath, S. C. O’Brien, Q. Zhang, Y. Liu, R. F. Curl, H. W. Kroto, F. K. Tittel, and R. E. Smalley, J. Am. Chem. Soc. 107, 7779 ~1985!; Y. Chai, T. Guo, C. Jin, R. E. Haufler, L. P. Felipe Chibante, J. Fure, L. Wang, J. M. Alford, and R. E. Smalley, J. Chem. Phys. 95, 7564 ~1991!; K. Shelimov and M. F. Jarrold, J. Am. Chem. Soc. 118, 1139 ~1996!. 4 L. M. Roth, Y. Huang, J. T. Schwelder, C. J. Cassady, D. Ben-Amotz, B. Khar, and B. S. Freiser, J. Am. Chem. Soc. 113, 6298 ~1991!; Y. Huang and B. S. Freiser, ibid. 113, 8187 ~1991!; 113, 9418 ~1991!. 5 D. E. Clemmer, J. M. Hunter, K. B. Shelimov, and M. F. Jarrold, Nature ~London! 372, 248 ~1994!; W. Branz, I. M. L. Billas, N. Malinowski, F. Tast, M. Heinebrodt, and T. P. Martin, J. Chem. Phys. 109, 3425 ~1998!. 6 T. P. Martin, N. Malinowski, U. Zimmermann, U. Na¨her, and H. Shaber, J. Chem. Phys. 99, 4210 ~1993!; U. Zimmermann, N. Malinowski, U. Na¨her, S. Frank, and T. P. Martin, Phys. Rev. Lett. 72, 3542 ~1994!. 7 F. Tast, N. Malinowski, S. Frank, M. Heinebrodt, I. M. L. Billas, and T. P. Martin, Phys. Rev. Lett. 77, 3529 ~1996!. 8 N. G. Chopra, R. J. Luyken, K. Cherrey, V. H. Crespi, M. L. Cohen, S. G. Louie, and A. Zettl, Science 269, 966 ~1995!. 9 X. Blase, A. De Vita, J. C. Charlier, and R. Car, Phys. Rev. Lett. 80, 1666 ~1998!. 10 T. Guo, C. Jin, and R. E. Smalley, J. Chem. Phys. 95, 4948 ~1991!. 11 H.-J. Muhr, R. Nesper, B. Schnyder, and R. Ko¨tz, Chem. Phys. Lett. 249, 399 ~1996!; J. C. Hummelen, B. Knight, J. Pavlovich, R. Gonza´lez, and F. Wudi, Science 269, 1554 ~1995!; Z. C. Ying, R. L. Hettich, R. N. Comp-

6938

J. Chem. Phys., Vol. 110, No. 14, 8 April 1999

ton, and R. E. Haufler, J. Phys. B 29, 4935 ~1996!; R. Yu, M. Mengxiong, D. Cheng, S. Yang, Z. Liu, and L. Zheng, J. Phys. Chem. 99, 1818 ~1995!. 12 S. J. La Placa, P. A. Roland, and J. J. Wynne, Chem. Phys. Lett. 190, 163 ~1992!; J. F. Christian, Z. Wan, and S. Anderson, J. Phys. Chem. 96, 10,597 ~1992!; T. Pradeep, V. Vijayakrishnan, A. K. Santra, and C. N. R. Rao, ibid. 95, 10,564 ~1991!. 13 W. Andreoni, A. Curioni, K. Holezer, K. Prassides, M. Keshavarz-K, J. C. Hummelen, and F. Wudl, J. Am. Chem. Soc. 118, 11,335 ~1996!; T. Pichler, M. Knupfer, M. S. Golden, S. Haffner, R. Friedlein, J. Fink, W. Andreoni, A. Curioni, M. Keshavarz-K, C. Bellavia-Lund, A. Sastre, J. C. Hummelen, and F. Wudl, Phys. Rev. Lett. 78, 4249 ~1996!; S. Haffner, T. Pichler, M. Knupfer, B. Umlauf, R. Friedlein, M. S. Golden, J. Fink, M. Keshavarz-K, C. Bellavia-Lund, A. Sastre, J. C. Hummelen, and F. Wudl, Eur. Phys. J. B 1, 11 ~1998!. 14 F. Chen, D. Singh, and S. A. Jansen, J. Phys. Chem. 97, 10,958 ~1993!; K. Esfarjani, K. Ohno, and Y. Kawazoe, Phys. Rev. B 50, 17,830 ~1994!; S.-H. Wang, F. Chen, Y. C. Fann, M. Kashani, M. Malaty, and S. A. Jansen, J. Phys. Chem. 99, 6801 ~1995!. 15 W. Andreoni, F. Gygi, and M. Parinello, Chem. Phys. Lett. 190, 159 ~1992!; N. Kurita, K. Kobayashi, H. Kumahora, K. Tago, and K. Osawa, ibid. 198, 95 ~1992!; K. Kobayashi and N. Kurita, Phys. Rev. Lett. 70, 3542 ~1993!; N. Kurita, K. Kobayashi, H. Kumahora, and K. Tago, Phys. Rev. B 48, 4850 ~1993!. 16 ~a! Theory: K. Raghavachari and C. McMichael Rohlfing, J. Chem. Phys. 89, 2219 ~1988!; U. Ro¨thlisberger, W. Andreoni, and M. Parinello, Phys. Rev. Lett. 72, 665 ~1994!; K.-M. Ho, A. A. Shvartsburg, B. Pan, Z.-Y. Lu, C.-Z. Wang, J. G. Wacker, J. L. Fye, and M. F. Jarrold, Nature ~London! 392, 582 ~1998!; ~b! Experiment: M. F. Jarrold and J. E. Bower, J. Chem. Phys. 96, 9190 ~1990!; J. M. Alford, R. T. Laaksonen, and R. E. Smalley, ibid. 94, 2618 ~1990!. 17 T. Kimura, T. Sugai, and H. Shinohara, Chem. Phys. Lett. 256, 269 ~1996!. 18 M. Pellarin, C. Ray, P. Me´linon, J. Lerme´, J. L. Vialle, P. Ke´ghe´lian, A. Perez, and M. Broyer, Chem. Phys. Lett. 277, 96 ~1997!. 19 J. L. Fye and M. F. Jarrold, J. Phys. Chem. 101, 1836 ~1997!. 20 C. Ray, M. Pellarin, J. Lerme´, J. L. Vialle, M. Broyer, X. Balse, P. Me´linon, P. Ke´ghe´lian, and A. Perez, Phys. Rev. Lett. 80, 5365 ~1998!. 21 M. Pellarin, B. Baguenard, M. Broyer, J. Lerme´, J. L. Vialle, and A. Perez, J. Chem. Phys. 98, 944 ~1993!. 22 J. L. Vialle, B. Baguenard, A. Bourgey, E. Cottancin, J. Lerme´, B. Palpant, M. Pellarin, F. Valadier, and M. Broyer, Rev. Sci. Instrum. 68, 2312 ~1997!. 23 H. Haberland, H. Kommeier, C. Ludewigt, and A. Risch, Rev. Sci. Instrum. 62, 2368 ~1991!; S. Wei and A. W. Castelman, Jr., Int. J. Mass Spectrom. Ion Processes 131, 233 ~1994!. 24 C.-R. Wang, R.-B. Huang, Z.-Y. Lui, and Z.-W. Zheng, Chem. Phys. Lett. 201, 23 ~1995!. 25 C. Kittel, Introduction to Solid State Physics ~Wiley, New York, 1976!. 26 C. Bre´chignac, Ph. Cahuzac, N. Kebaı¨li, J. Leygnier, and A. Sarfati, Phys. Rev. Lett. 68, 3916 ~1991!. 27 K. Fuke, K. Tsukamoto, F. Misaizu, and M. Sanekata, J. Chem. Phys. 99, 7807 ~1993!. 28 S. B. H. Bach, J. E. Bruce, M. A. Cheeseman, R. Ramanathan, C. H. Watson, J. A. Zimmerman, and J. R. Eyler, Advances in Metal and Semiconductor Clusters, edited by M. A. Duncan ~JAI, London, 1994!, Vol. 2, p. 47. 29 T. P. Martin, U. Na¨her, T. Bergmann, H. Go¨hlich, and T. Lange, Chem.

Pellarin et al. Phys. Lett. 183, 119 ~1991!; C. Bre´chignac, Ph. Cahuzac, M. de Frutos, P. Garnier, and N. Kebiaı¨li, J. Chem. Phys. 103, 6631 ~1995!; R. Antoine, Ph. Dugourd, D. Rayane, and M. Broyer, ibid. 104, 110 ~1996!. 30 E. A. Rohlfing, D. M. Cox, and A. Kaldor, J. Chem. Phys. 81, 3322 ~1984!. 31 C. Bre´chignac, H. Busch, Ph. Cahuzac, and J. Leynier, J. Chem. Phys. 101, 6992 ~1994!; U. Na¨her and K. Hansen, ibid. 101, 5367 ~1994!. 32 C. Klots, J. Chem. Phys. 83, 5854 ~1985!; Z. Phys. D 5, 83 ~1987!. 33 P. Wurz and K. R. Lykke, J. Phys. Chem. 96, 10129 ~1992!; H. Hohmann, C. Callegari, S. Furrer, D. Grosenick, E. E. B. Campbell, and I. V. Hertel, Phys. Rev. Lett. 73, 1919 ~1994!. 34 ~a! Theory: K. Raghavachari and J. S. Binkley, J. Chem. Phys. 87, 2191 ~1987!; ~b! Experiment: G. von Helden, P. R. Kemper, N. G. Gotts, and M. T. Bowers, Science 259, 1300 ~1993!; J. M. Hunter, J. L. Fye, and M. F. Jarrold, J. Chem. Phys. 99, 1785 ~1993!; H. Handschuh, G. Gantefo¨r, B. Kessler, P. S. Bechthold, and W. Eberhardt, Phys. Rev. Lett. 74, 1095 ~1994!. 35 K. S. Pitzer and E. Clementi, J. Am. Chem. Soc. 81, 4477 ~1959!; R. Hoffmann, Tetrahedron 22, 521 ~1966!. 36 ~a! Experiment: A. Nakajima, T. Takuwa, K. Nakao, M. Gomei, R. Kishi, S. Iwata, and K. Kaya, J. Chem. Phys. 103, 2050 ~1995!; D. Parent, Int. J. Mass Spectrom. Ion Processes 116, 257 ~1992!; ~b! Theory: G. Foudrakis, A. Zdetsis, M. Mu¨hlha¨user, B. Engels, and S. Peyerhoff, J. Chem. Phys. 101, 6790 ~1994!; M. Bertolus, V. Brenner, and P. Millie´, Eur. Phys. J. D 1, 197 ~1998!. 37 S. C. O’Brien, J. R. Heath, R. F. Curl, and R. E. Smalley, J. Chem. Phys. 88, 220 ~1988!. 38 P. P. Radi, M. T. Hsu, J. Brodbelt-Lustig, M. Rincon, and M. T. Bowers, J. Chem. Phys. 88, 2809 ~1987!; P. Sheier, B. Du¨nser, R. Wo¨rgo¨tter, D. Muigg, S. Matt, O. Echt, M. Foltin, and T. D. Ma¨rk, Phys. Rev. Lett. 77, 2654 ~1996!. 39 J. Drowart, G. de Maria, and M. G. Inghram, J. Chem. Phys. 29, 1015 ~1958!; R. W. Schmude, Jr. and K. A. Gingerich, J. Phys. Chem. A 101, 2610 ~1997!. 40 R. E. Honig, J. Chem. Phys. 22, 126 ~1954!. 41 R. E. Honig, J. Chem. Phys. 22, 1610 ~1954!. 42 F. Finocchi, G. Galli, M. Parinello, and C. M. Bertoni, Physica B 185, 379 ~1993!. 43 H. Koinuma, M-S. Kim, and M. Yoshimoto, Jpn. J. Appl. Phys., Part 1 34, 3720 ~1995!; M. A. Capano, S. D. Walck, P. T. Murray, D. Dempsey, and J. T. Grant, Appl. Phys. Lett. 64, 3413 ~1994!. 44 ~a! Theory: K. Raghavachari and C. McMichael Rohlfing, Chem. Phys. Lett. 143, 428 ~1988!; ~b! Experiment: L. A. Bloomfield, R. R. Freeman, and W. L. Brown, Phys. Rev. Lett. 54, 2246 ~1985!; M. F. Jarrold and E. Honea, J. Phys. Chem. 95, 9185 ~1991!. 45 Handbook of Chemistry and Physics, 73rd edited by D. R. Lide ~CRC, Cleveland, 1992!. 46 C. Xu and G. E. Scuseria, Phys. Rev. Lett. 72, 669 ~1994!. 47 P. W. Fowler and D. E. Manolopoulos, An Atlas of Fullerenes ~Clarendon, Oxford, 1995!. 48 D. Babic and N. Trinajstic, Chem. Phys. Lett. 237, 239 ~1995!. 49 G. Seifert, P. W. Fowler, D. Mitchell, D. Porezag, and Th. Frauenheim, Chem. Phys. Lett. 268, 352 ~1997!. 50 P. Me´linon, P. Ke´ghe´lian, A. Perez, C. Ray, J. Lerme´, M. Pellarin, M. Broyer, M. Boudeulle, B. Champagnon, and J. L. Rousset, Phys. Rev. B 58, 16481 ~1998!.