3 October 1997
Chemical Physics Letters 277 Ž1997. 96–104
Silicon–carbon mixed clusters b b M. Pellarin a , C. Ray a , P. Melinon , J. Lerme´ a , J.L. Vialle a , P. Keghelian , ´ ´ ´ b a A. Perez , M. Broyer a
(UMR C.N.R.S. 5579), UniÕersite´ Claude Bernard-Lyon 1, Bat. Laboratoire de Spectrometrie ´ Ionique et Moleculaire ´ ˆ 205, 43 Bd. du 11 NoÕembre 1918, 69622 Villeurbanne Cedex, France b (UMR C.N.R.S. 5586), UniÕersite´ Claude Bernard-Lyon 1, Bat. Departement de Physique des Materiaux ´ ´ ˆ 203, 43 Bd. du 11 NoÕembre 1918, 69622 Villeurbanne Cedex, France Received 25 June 1997; in final form 18 July 1997
Abstract Binary clusters Si nC m are produced in a laser vaporization source from Si xC 1yx mixed targets. Different composition regimes are investigated by abundance and photoinduced dissociation mass spectroscopy. In the case of stoichiometric silicon–carbide clusters Ž x s 0.5., no clear magic size is detected but rather the marked effects of a lower binding energy of silicon as compared to carbon atoms. The structural changes induced by the presence of foreign atoms are discussed in the case of carbon-rich clusters Ž x < 1. for which the formation of heterofullerenes is observed. q 1997 Elsevier Science B.V.
1. Introduction Although silicon and carbon are contiguous within the same column of the periodic table, their bonding and chemical properties are rather different. The high flexibility of carbon bonds allows the emergence of a wide range of cluster geometries. As the cluster size increases, the most favorable forms evolve from linear Žchains. to planar and cage-like arrangements Žrings and fullerenes., from one to three-dimensional structures w1x. On the contrary, silicon is known to prefer multidirectional single bonds that merely generate three-dimensional structures Žstuffed fullerenes. even from the smallest sizes w2x. This behavior is also apparent in the bulk where the absence of a silicon graphitic phase is noticeable. Extensive theoretical and experimental studies have focused on pure clusters that were nicely illustrated, in the case of carbon, by the finding of
fullerenes and other three-dimensional related graphitic structures w3,4x. More recently, silicon– carbon ŽSi–C. mixed clusters have stimulated interest. Silicon carbide ŽSiC. is an attractive technological material because of possible applications in the field of heat and radiation-resistant devices for electronics. Cluster-assembled thin films from such systems may also be important for material science w5x. If early studies on Si–C clusters were mainly devoted to small molecules of astrophysical interest w6x, they now deal more precisely with the following questions: in what way can the pure cluster structures and properties be affected by the presence of foreign atoms and how a transition from C-like to Si-like structures occurs as a function of the Si:C molar concentration Žhomogeneousness, shape, dimensionality.? To our knowledge, except for the investigations of silicon doped heterofullerenes by Kimura et al. w7x and the recent results of Fye and
0009-2614r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. PII S 0 0 0 9 - 2 6 1 4 Ž 9 7 . 0 0 8 6 9 - 5
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
Jarrold w8x on the mobility of C n Siq clusters, most of these studies concern small systems Ž n q m - 10. w9–11x. In this Letter, we report some experimental observations on Si–C binary clusters in a wide mass range Ž n q m - 100.. These results are intended to explore the structure of Si nC m silicon–carbide clusters Ž n ; m. and to address the influence of silicon foreign atoms on the structure of carbon clusters.
2. Experiment Silicon–carbon binary clusters are generated in a laser vaporization source described in a previous paper w12x. This technique has proven to be very efficient for producing mono as well as multi-elemental clusters of refractory materials w13x. The non-equilibrium growth processes involved in such a source allow the free clusters to be frozen in the form of exotic geometric structures or homogeneous atomic arrangements from elements that are not miscible under standard themodynamical conditions. Mixed clusters with any mean composition can then be obtained from a laser-induced plasma having the same stoichiometry ŽSi x C 1yx .. Even if thermodynamical constraints prevent the target being processed in the form of a homogenenous alloy, the same plasma composition can be obtained from inhomogeneous alloys or composites provided that their average stoichiometry is at least unchanged on the scale of the vaporization laser spot Ža few 100 mm in diameter.. The stable structures obtained by mixing carbon and silicon are the cubic, hexagonal or rhombohedral stoichiometric Ž x s 0.5. crystalline phases and all their related polytypic forms.The most direct experiment was to produce and to study clusters obtained from such a stoichiometric target built as a 3 mm diameter SiC rod ŽMorton.. To achieve different compositions, specific targets have been processed in the form of composite rods prepared from pressed and baked mixtures of carbon-graphite and silicon powders ŽGoodFellow. initially bound with a graphite cement ŽAremco.. The RBS ŽRutherford Back Scattering. analysis of these samples has shown that the oxygen contamination is almost negligible. The size distribution of binary clusters born as
97
cations or anions is analyzed by high resolution mass spectroscopy using a time-of-flight mass spectrometer ŽTOFMS. in a reflectron configuration. Clusters of a given size can also be selected by an electrostatic gate and then photodissociated by an XeCl excimer laser in their free flight region. The dissociation products are spatially dispersed during their reflection towards the detector which allows their size and relative abundances to be measured. Finally, we have studied the neutral size distribution by ionizing the clusters with the focused beam of an XeCl excimer laser.
3. Results 3.1. Silicon–carbide clusters Fig. 1 shows a mass spectrum of Si nCq m clusters directly produced from a silicon–carbide rod Ž x s 0.5.. The apparent complexity of this spectrum simply originates from the isotopic distributions of silicon Ž 28 Si Ž92.2%., 29 Si Ž4.7%., 30 Si Ž3.1%.. and carbon Ž 12 C Ž98.9%., 13 C Ž1.1%.. atoms. The inset displays a comparison between an enlarged part of the experimental spectrum Ža. and the result of a simulation Žb. in the same region. If one assumes that clusters of a given nuclearity Žsame total number of atoms N s n q m. are formed by the random stacking of atoms from a vapor of fixed composition and that cluster abundances do not depend on this nuclearity, the mass peak intensities can be calculated from a generalized binomial distribution involving all the possible isotopes weighted by their natural abundances and the Si:C molar concentration ratio. A peak width close to the experimental one is introduced to allow an easier comparison between Ža. and Žb.. The simulation roughly fits the experimental abundances. A pattern made of four peak clumps with decreasing intensities is periodically reproduced. The most intense peaks correspond to mixed clusters formed with the most abundant isotopes of carbon and silicon Ž 12 C and 28 Si.. The periodicity of four mass units originates from the reinforced abundance of near stoichiometric Ž n ; m. binary clusters such as 2 8 Si n 1 2 C q and m 28 Si n" 112 Cq m . 2 . The smaller peaks correspond to combinations involving less abundant isotopes. As
98
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
Fig. 1. TOF mass spectrum of Si nCq m clusters directly obtained from a silicon carbide rod. The inset shows a comparison between an enlarged part of the experimental spectrum Ža. and the corresponding simulation Žb..
q Si nCq m and Si n " 3 C m . 7 clusters have the same mass, a clear analysis of the spectra is made all the more difficult as their abundances are close to each other. This is particularly true when the binomial distribution width on both sides of the mean cluster composition Ž n ; m. is large Ž x ( 0.5 and n 4 1.. Nevertheless, from a comparison between experiment and different simulations, it can be stated that the cluster mean composition is close to the one of the target and that the stacking of carbon and silicon atoms on cationic precursors during the growth process is roughly statistical. Taking into account the signalto-noise ratio and the experimental reproducibility, the irregularities that may occur in the mass spectra are too weak to be reasonably assigned to particularly stable structures. Clear ‘‘magic sizes’’ are therefore not detected contrary to pure carbon w14,15x and, to a lesser extent, to pure silicon clusters w16,17x. The same behavior has been observed on negatively charged clusters under the same experimental conditions and throughout the same mass range. To obtain further information on the composition and stability of Si nCq m clusters, we have recorded photodissociation mass spectra on some selected sizes. Fig. 2a shows the fragmentation pattern in the vicinity of the most abundant predicted Si 25 Cq 21 com-
bination. This composition is the most likely, considering the arrival time and the requirement of an Si:C atomic ratio as close as possible to the target stoichiometry Ž x ( 0.5 or n ( m.. The fragmentation products are undoubtedly the result of sequential and combined evaporations of mainly Si 2 C and secondarily Si 3 C neutral molecules. A similar behavior is observed for most of the sizes except for the smallest Ž n q m - 10. where the evaporation of Si 2 C and SiC 2 prevails ŽFig. 2b.. We can make a comparison with the sublimation of bulk silicon carbide which is ruled by the Si monomer evaporation w18x. The difference between both situations may be explained by energetic balances. It is similar to what is encountered in the case of pure silicon clusters that are known to dissociate through the ejection of large neutral fragments ŽSi 6 , Si 7 , Si 10 . w19,20x when Si atoms are evaporated from the bulk w21x. Possible activation barriers and surface segregation of silicon could also play a significant role. A more precise interpretation of these results would require the knowledge of various Si nCq m cluster binding energies which are not available for the moment. To get an idea of the size distribution of neutral SiC clusters, they are analyzed in the TOFMS after ionization by the beam of an XeCl excimer laser
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
99
q Fig. 2. Photofragmentation mass spectra of the size selected Si 25 Cq 21 and Si 4 C 5 positive clusters. The mass peaks at shorter arrival times correspond to charged products having evaporated neutral molecules through single or multistep processes. These molecules are only indicated for the largest products.
Ž308 nm.. The ionization potentials of such clusters have not been measured up to now but are certainly intermediate between those of pure carbon and pure silicon clusters. The work function of bulk silicon Ž4.85 eV w22x. is a lower limit and the direct singlephoton ionization of these clusters is impossible whatever their size Ž hn ( 4 eV.. Multi-photon ionization can only be achieved by focusing the laser beam which unavoidably results in cluster heating and fragmentation. Fast sequential dissociation processes occur in the stage of cluster acceleration which leads to a size distribution shifted towards the small sizes as compared to the one emerging from the source. As clusters carry on evaporating during their free flight, their fragments will be dispersed by the reflector and will consequently congest the mass spectrum. One can get rid of this awkward effect by simply removing the reflector and by operating the spectrometer in a conventional direct configuration. The most striking result is that almost at the detected clusters are carbon fullerenes as shown in Fig. 3. As
in similar experiments performed in the same conditions but with a pure carbon target, the high stability of Cq 60 and the absence of odd-numbered species are noticeable. The evaporation processes during the cooling of the accelerated clusters are certainly similar to those detected in fragmentation experiments and must lead to a silicon impoverishment within clusters through the successive loss of fragments always richer in silicon than in carbon. The observation of a large number of such molecules Žinset in Fig. 3., either directly ejected as ions or photoionized after their ejection, is consistent with this hypothesis. One can then believe that the acceleration stage duration Ža few ms. and the initial excitation energy were sufficient for near stoichiometric SiC clusters to be ionized and to loose, at least, all their silicon atoms before entering the free flight region of the spectrometer. For example, if Si 2 C is supposed to be the main evaporated fragment in such a sequence, the Cq 60 cluster should originate from a neutral parent cluster close to Si 120 C 120 . In the last
100
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
Fig. 3. TOF mass spectrum of Si nC m clusters obtained by high fluence photoionization of initially neutral species Žsee text.. The transmitted clusters are almost all carbon fullerenes Ž n G 32.. The bell shape of the spectrum is merely a consequence of the deflecting voltage applied to direct the clusters onto the TOF apparatus axis. Silicon just appears in the smallest products Ženlarged portion..
stages of their formation, the ‘‘pure’’ carbon clusters are hot enough to assume structural rearrangements to the most stable form of fullerenes. An additional study, not reported here, has revealed that such clusters undergo a unimolecular evaporation of C 2 entities throughout their flight in the spectrometer. This is consistent with previous experiments w23,24x and shows that these clusters are far to be completely cooled when they enter the free flight zone following the ionization and acceleration stages. The weak Ž . amount of small Cq n carbon clusters n - 30 and the absence of odd-numbered larger sizes Ž n ) 30. is consistent with what we have observed in a similar experiment performed in the same conditions but from a pure carbon target. The high cohesive energy of fullerenes explains why these clusters act as bottlenecks in the size distribution shift during fragmentation. 3.2. Carbon rich and silicon rich clusters The most simple investigation on the structure of such mixed systems from TOF mass spectra first
consists in identifying the different composition sequences that involve an increasing number of foreign q q . atoms ŽXq n , AX ny1, A 2 X ny2, . . . , and then in comparing the relative size abundances between these series. Carbon rich clusters are produced from a silicon–carbon composite rod with an Si:C molar composition of about 5:100. The mass spectrum of such positively charged clusters is shown in Fig. 4. The solid grey line connects the most prominent peaks of the fullerenes series Cq 2 n that appears clearly in this size domain. This size distribution is very similar to previous observations on the abundance studies on pure carbon clusters w14x. In particular, a transition to a ‘‘fullerene regime’’ is clearly apparent through both the gradual disappearance of odd-numbered clusters as the size is increasing and the enq q hanced abundances of Cq 44, C 50 and C 60. The possiq ble weak amount of C ny 7 Si 3 mixed clusters with the same mass may somewhat blur and modify this sequence by comparison with pure carbon clusters. This will be omitted in the following for simplicity. The dashed and solid black lines connect the se-
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
101
q Fig. 4. TOF mass spectrum of carbon rich clusters Žfullerene size domain.. The mass peaks of Cq and C ny2 Siq n , C ny1 Si 2 series are connected by grey solid, black dashed and black solid lines respectively. The region close to Cq is enlarged in the inset and the enhanced 60 abundances detected for n s 44, 50 and 60 are indicated.
quences built from C 2 ny1 Siq and C 2 ny2 Siq 2 clusters. The same marked features are observed as in the case of pure species with the same nuclearity. This is an indication that at least one or two carbon atoms can be replaced by silicon without important structural changes of the fullerenes. The evidence for the stability of nitrogen or boron-doped fullerenes w25,26x has been already given by theoretical w27x as well as experimental studies, particularly in the case of boron substitution w28x. The present results confirm and extend previous experiments on silicondoped hetero-fullerenes w7,8x. Going further is rather difficult because the C 2 ny7 Siq 3 cluster mass peaks are merged with those of Cq 2 n and because the more strongly doped the fullerenes produced Žfrom silicon richer targets., the more awkward will be the mass peak overlap of an increasing number of clusters with different compositions. Preliminary fragmentation studies on these heterofullerenes show that they sequentially loose neutral dimers exactly like pure fullerenes but that the most easily evaporated molecules are those containing the maximum number of silicon atoms ŽSiC or Si 2 . considering the
. initial cluster composition ŽC 2 ny1 Siq or C 2 ny2 Siq 2 . This highlights the special nature and the weakness of the silicon bonding in the fullerene network. In the ring or planar structure domain Ž n - 30., q q q the enhanced abundances of Cq 11, C 15, C 19 and C 23 q q clusters w14x are mirrored in the C 11 Si , C 15 Si , C 18 Siq, C 22 Siq w7x and to a lesser extent, C 11 Siq 2, q. q q Ž C 15 Siq 2 C 14 Si 2 , C 17 Si 2 , C 21 Si 2 prominent peak sequences ŽFig. 5.. This seems to indicate that below n ( 5 silicon prefers to be attached to the surface of almost unperturbed carbon structures and that above this limit it accepts being inserted in the carbon cluster network without appreciably destabilizing the host structure just like with fullerenes. The same investigation on silicon-rich clusters generated from a composite rod with a 100:5 molar concentration is made more difficult by the lack of strong prominent features in the mass spectra of pure silicon clusters w16x. Only the enhanced stabilities of q q q Siq 6 , Si 11 or even Si 15 and Si 19 can be used as a q guide. Up to n ( 11, the Si n , Si ny1Cq and Si ny2 Cq 2 sequences are very similar which is in favor of a simple substitution of carbon atoms within the sili-
102
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
q Fig. 5. TOF mass spectrum of carbon rich clusters. The mass peaks of Cq and C n Siq n , C n Si 2 series are connected by grey solid, black dashed and black solid lines respectively. The sizes of the more intense peaks of pure carbon clusters are also indicated.
con clusters without any strong perturbation. For larger sizes, the situation is less clear but more information should be obtained from the photodissociation studies. These experiments are in progress and will be reported in a forthcoming paper.
4. Discussion The silicon atom is expected to fit into a three-dimensional network where its propensity to form multidirectional single bonds as in a tetrahedral environment can be satisfied. Nevertheless, at least up to two atoms can visibly be inserted in the two-dimensional network of the fullerenes. This is not so surprising in the case of small fullerenes where the surface curvature and the number of pentagons compared to hexagons are relatively important. The s–p hybridization is then different from the case of a pure planar graphitic sheet and allows a sufficient amount of sp 3 character for the silicon atom to be more easily incorporated, even at the cost of slight geometric deformations. In the case of larger fullerenes Žtubes. w4x, the situation should be different and the silicon atom should be preferentially located in the regions of larger curvature. This remains to be investigated.
Previous studies on small SiC binary clusters give evidence of structural changes among clusters having the same nuclearity as a function of their composition. The three-dimensional character of these structures particularly tends to increase with the number of silicon atoms w9,10x. The present study is not suited to illustrate such behavior but the affinity of silicon for a larger atomic coordination appears in the case of small carbon-rich clusters where some indications are in favor of a transition from an external to an internal position of the silicon atom in the carbon network. In the reverse case of silicon-rich clusters, which adopt cage-like arrangements early, the ability of carbon atoms to form any kind of bond makes their direct insertion very easy. However, more important structural changes seem to be induced in this case as a consequence of a stronger relative carbon bonding. The absence of particular stable sizes of large silicon–carbide clusters is not surprising. The formation of exotic structures such as those of carbon buckminsterfullerenes is unlikely in the absence of a graphitic phase of the bulk SiC. For instance the predicted high stability of a ŽSiC. 60 molecule w29x made of a Si 60 cage surrounding and covalently bound to a C 60 fullerene was not confirmed in this experiment. The same conclusion can be drawn con-
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
cerning the Si 30 C 30 similar form of C 60 . As discussed in the case of the boron–nitrogen analogues of the fullerenes w26x, such a structure should not present any particular stability owing the occurrence of fivefold rings that forbid a regular alternation of Si and C atoms over the whole cluster. The stabilities of modified structures ŽŽSiC.12 , ŽSiC.16 , ŽSiC. 28 , . . . . involving squares instead of pentagons so as to avoid frustration effects w30x were not detected either. The remaining question is to establish the geometry of these systems and more particularly the relative arrangement of silicon and carbon atoms as a function of size. Large SiC clusters should be described as cut-out pieces of a diamond-like network with a suitably reconstructed surface or as amorphous and homogeneous arrangements w31x, when smaller clusters could eventually present strong surface effects with a possible segregation of one element. If they exist, such effects are expected to strongly appear in the case where the Si:C molar ratio deviates from unity. Is it possible to establish a parallel between the properties of such clusters and those of the corresponding macroscopic materials? It is well known that, in the crystalline phase, thermodynamical constraints lead to a separation between a stoichiometric Si 0.5 C 0.5 phase and a phase containing the element in excess. This should correspond to a segregation phenomenon within clusters. The analysis of the size abundances of silicon and carbon rich cluster ions discloses no mark of any difference between the sticking probabilities of Si and C atoms, depending on the phase Žpure or mixed. they would be part of. Moreover, whatever the size and composition of mixed clusters, the more easily evaporated fragments contain both elements even if silicon is more abundant on average. One could expect, in the case of silicon rich segregated systems for instance, the appearence of photoproducts that would be the signature of pure silicon cluster dissociation ŽSi 6 or Si 10 . since silicon clusters are certainly much less bound than silicon–carbide clusters. The results of a preliminary study on such systems are inconsistent with this hypothesis and are rather in favor of homogeneous cluster structures in which the carbon and silicon networks are strongly interdependent. Finally, this study has shown that homogeneous clusters of any relative composition of carbon and
103
silicon can be easily produced. This leaves the door open for processing amorphous and homogeneous nanostructured thin films with the same composition. They can be obtained by means of the low energy cluster beam deposition technique ŽLECBD. w5x that takes advantage of the cluster ‘‘memory effect’’ w32x so as to get rid of phase segregation phenomena inherent in conventional atomic vapor or plasma deposition experiments w33,34x. For instance, such method could then be interesting to investigate the partial destruction of the chemical order in Si–C films as the Si:C molar ratio departs from unity. In continuous amorphous samples, this chemical order is known to strongly vary in the vicinity of the equimolar composition Ž x s 0.5.. The study of thin films processed by depositing clusters of well controlled stoichiometry near this critical value should permit a better understanding of this phenomenon.
Acknowledgements The authors thank Fernand Valadier for his technical assistance in processing the silicon–carbon targets.
References w1x K. Raghavachari, J.S. Binkley, J. Chem. Phys. 87 Ž1987.. w2x K. Raghavachari, C.M. Rohlfing, J. Chem. Phys. 89 Ž1988. 2219. w3x H.W. Kroto, J.R. Heath, S.C. O’Brien, R.F. Curl, R.E. Smalley, Nature 318 Ž1985. 162. w4x H.W. Kroto, J.E. Fisher, D.E. Cox ŽEds.., The Fullerenes, Pergamon Press, Oxford, 1993. w5x M. Broyer, M. Pellarin, B. Baguenard, J. Lerme, ´ J.L. Vialle, P. Melinon, J. Tuaillon, V. Dupuis, B. Prevel, A. Perez, Mat. ´ ´ ´ Sci. Forum 232 Ž1996. 27. w6x J.D. Priscilla-Marquez, S.C. Gay, C.M.L. Rittby, W.R.M. Graham, J. Chem. Phys 102 Ž1995. 6354, and references therein. w7x T. Kimura, T. Sugai, H. Shinohara, Chem. Phys. Lett. 265 Ž1996. 269. w8x J.L. Fye, M.F. Jarrold, J. Chem. Phys. A 101 Ž1997. 1836. w9x G. Foudakis, A. Zdetsis, M. Muhlhauser, B. Engels, S.D. ¨ ¨ Peyerimhoff, J. Chem. Phys. 101 Ž1994. 6790. w10x S. Hunsicker, R.O. Jones, J. Chem. Phys. 105 Ž1996. 5048, and references therein. w11x A. Nakajima, T. Takuwa, K. Nakao, M. Gomei, R. Kishi, S. Iwata, K. Kaya, J. Chem. Phys. 103 Ž1995. 2050.
104
M. Pellarin et al.r Chemical Physics Letters 277 (1997) 96–104
w12x M. Pellarin, B. Baguenard, M. Broyer, J. Lerme, ´ J.L. Vialle, A. Perez, J. Chem. Phys. 98 Ž1993. 944. w13x J.L. Rousset, A.M. Cadrot, F.J. Cadete Santos Aires, A. Renouprez, P. Melinon, A. Perez, M. Pellarin, J.L. Vialle, M. ´ Broyer, J. Chem. Phys. 102 Ž1995. 8574. w14x E.A. Rohlfing, D.M. Cox, A. Kaldor, J. Chem. Phys. 81 Ž1984. 3322. w15x H. Handschuh, G. Gantefor, ¨ B. Kessler, P.S. Bechthold, W. Eberhardt, Phys. Rev. Lett. 74 Ž1994. 1095. w16x Q.-L. Zhang, Y. Liu, R.F. Curl, F.K. Tittel, R.E. Smalley, J. Chem. Phys. 88 Ž1988. 1670. w17x M.F. Jarrold, J.E. Bower, J. Chem. Phys. 96 Ž1992. 9180. w18x R.W. Schmude Jr., K.A. Gingerich, J. Phys. Chem. A 101 Ž1997. 2610. w19x K. Raghavachari, C.M. Rohlfing, Chem. Phys. Lett. 143 Ž1988. 428. w20x M.F. Jarrold, E.C. Honea, J. Phys. Chem. 95 Ž1991. 9181, and references therein. w21x R.E. Honig, J. Chem. Phys. 22 Ž1954. 1610. w22x Handbook of Chemistry and Physics, 72nd ed., CRC, Cleveland, 1991r1992. w23x S.C. O’Brien, J.R. Heath, R.F. Curl, R.E. Smalley, J. Chem. Phys. 88 Ž1987. 220.
w24x M. Foltin, M. Lezius, P. Scheier, T.D. Mark, ¨ J. Chem. Phys. 98 Ž1993. 9624. w25x J.C. Hummelen, B. Knight, J. Pavlovich, R. Gonzalez, F. Wudl, Science 269 Ž1995. 1555. w26x H.-J. Muhr, R. Nesper, B. Schnyder, R. Kotz, ¨ Chem. Phys. Lett. 249 Ž1996. 399. w27x S. Wang, F. Chen, Y.-C. Fann, M. Kashani, M. Malaty, S.A. Jansen, J. Phys. Chem 99 Ž1995. 6801. w28x T. Guo, C. Jin, R.E. Smalley, J. Phys. Chem. 95 Ž1991. 4948. w29x S. Osawa, M. Harada, E. Osawa, B. Kiran, E.D. Jemmis, Fullerene Sci. Technol. 3 Ž1995. 225. w30x G. Seifert, P.W. Fowler, D. Mitchell, D. Porezag, Th. Frauenheim, Chem. Phys. Lett. 268 Ž1997. 352. w31x F. Finocchi, G. Galli, M. Parinello, C.M. Bertoni, Physica B 185 Ž1993. 379. w32x V. Paillard, P. Melinon, J.P. Perez, V. Dupuis, A. Perez, J.L. ´ Loubet, H. Pascal, A. Tonck, M. Fallavier, Nanostruct. Mat. 4 Ž1994. 759. w33x H. Koinuma, M.-S. Kim, M. Yoshimoto, Jpn. J. Appl. Phys. 34 Ž1995. 3720. w34x M.A. Capano, S.D. Walck, P.T. Murray, D. Dempsey, J.T. Grant, Appl. Phys. Lett. 64 Ž1994. 3413.
