8182 J. Phys. Chem. zyxwvut 1993,97, 8182-8192 Carbon Cluster Cations with up to 84 Atoms: Structures, Formation Mechanism, and Reactivity Gert von Helden, Ming-Teb Hsu, Nigel Gotts, and Michael T. Bowers' Department of Chemistry, University of California, Santa Barbara, California 931 06 Received: March 30, 1993; zyxwvuts In Final Form: May 25, 1993 Carbon clusters are generated by laser desorption. Mass-selected beams are then pulse injected into an ion chromatography (IC) device. This devicetemporally and spatially separates thebeam into its isomericcomponents. Arrival time distributions (ATDs) are then measured at the detector. From these distributions, accurate mobilities are obtained for each isomeric component, along with the fractional abundance of each isomer. Different isomer structures are calculated using quantum chemical methods. A Monte Carlo technique uses these structures to obtain accurate theoretical mobilities. Comparison of theory with experiment allows unambiguous structural assignment of the various families of isomers present in the cluster beam. The results indicate that, for carbon cluster cations, linear structures exist up to Clo+. Several families of planar ring systems begin with monocyclic rings (ring I), which first appear at C7+ and persist beyond Ca+. Bicyclic rings (ring 11) are first observed at Czl+ and persist beyond Ca+, followed by tricyclic rings (ring 111, initiated at C30+) and tetracyclic rings (ring IV, initiated at Ca+). A 3-dimensional family we label as zyx 3-D rings begins at CZ9+, whose structure is not yet unambiguously assigned. This family never exceeds 5% of the ions at any cluster size. Finally, the first fullereneis observed at C30+, with this family dominating above C50+. The growth pattern of carbon, beginning with C atoms, is shown to be linear zyxwv - monocyclic rings - polycyclic rings - fullerenes. No graphitic or "cup" shaped isomers are observed, eliminating these species as building blocks for fullerenes. Our structural data, when coupled with recently published annealing studies, indicate that fullerenes are formed from isomerization of hot planar ring systems and that monocyclic rings and fullerenes are close in energy between C~O+ and C36+ with fullerenes dominating above c36+. Reactions of C7+ to CIS+ with zy 02 and NO are reported and indicate that linear chains are generally much more reactive than rings. Finally, Cm and C70 are made with up to four positive charges but retain the fullerene cage structure. I. Introduction Even before the synthesis of macroscopic quantities of fullerenes,' carbonclusters were the subjectof an intenseresearch effort. Thevast amount of research up to 1988 has been reviewed by Welter and van &e.* Early studies on small carbon cluster ions with up to about 30 atoms used high-frequency spark evaporation: laser evaporation followed by direct detection of positive or negative ions: thermal evaporation of carbon,5 and secondary ion mass spectrometry.6 All thesestudies showed strong intensity variations with carbon cluster size in mass spectra. For example, higher intensities were observed- for clusters with an odd number of atoms up to Cg+. For larger clusters, a periodicity in intensity for every four atoms was observed.- These periodicities were interpreted as consistent with carbon chains for n I 9 and monocyclic rings for larger A breakthrough in the field occurred with the design of pulsed laser vaporization sources coupled with a supersonic expansion using an inert carrier gas.* These sources are capable of generating intense cluster beams having up to several hundreds of atoms in a cluster.gJ0 Remarkable size distributions were observed for carbon cluster cation^.^ For example, essentially no clusters were observed from about 26 to about 32 at0ms;9 clusters with an odd number of atoms for n > 30 were either absent or very weak? and enhanced intensities were observed for "magic" clusters containing 50,60, and 70 at0ms.~J0 It was then proposed that Cm was composed of a closed quasi-spherical carbon network containing 12 pentagons and 20 hexagons, and the name buckminsterfullerene was These authors argued that Cm and C70 were also closed-shell cagecompounds, and eventually it became common to describe all even carbon clusters with greater than 30 atoms as "fullerenes". Experimental determination of structures for any of these clusters remained elusive, however, until Cm was first isolated1 and its structure investigated by traditional means. 0022-3654/93/2097-8 182$04.00/0 For clusters other than the few fullerenes that could be isolated in bulk, unambiguous structural evidence was lacking, with the exceptionof somevery small clusters where spectroscopic methods were applied and at least some molecular parameters deter- mined.2J1 Reactivity studies12suggested that C7+, CS+, and Cg+ exist in two different isomeric forms, a reactiveform (presumably linear) and a less reactive form (presumably cyclic). For the negativeions, interpretation of photoelectron spectral) suggested that clusters with up to nine atoms are linear, Clo- exists in both linear and cyclic forms, and above Clo- the ions and/or neutrals are monocyclic rings. Coulomb explosion methods" have shown the possible presence of different isomers for C4-, C r , and c6-. Structural information also has been deduced from metastable mass ~pectr0metry.l~ These studies indicatestructural transitions occurring near Cl0+ and C30+. For clusters with n < 24, the main metastable neutral loss was found to be Cp, consistent with the possible existence of linear and ring like clusters. Above C,O+, only C2 loss was observed for clusters with an even number of atoms. Between CN+and C30+ primary neutral fragment losses included Clo and C14, presumably intact monocyclic rings. Numerous computationalstudieson carbon clusters have been undertaken.16 High-level ab initio calculations have mainly focused on clusters containing up to 11 atomslh,b*dJ.i and Cm.16j* Recently, however, calculations have been done on clusters of intermediate ~ize.lC.~.~fJ~ Results indicate that, for clusters with up to 11 atoms, the cluster ground state is either monocyclic or linear with the possibility of both structures coexisting for some cluster sizes. Recent MP2 and LDA calculations on C20, C14, and other ~lusters~~.~LJJ~ indicate a cagelike fullerene structure may be lowest in energy followed by cup or graphitic like "open fullerene" structures and then monocyclic rings. One of these studies16f proposed the C20 "cup" as a possible precursor in Cm formation. This proposal is in contrast to experimentalresults17 that indicate monocyclic rings are the dominant clusters for both 0 1993 American Chemical Society