distilled water vapor from a gas reservoir at 14 torr and room temperature into the microscope until the partial H20 pressure rises to 4 x 1 0-6 torr. We grow films of ice by opening for 10 s a shutter that covers the sample at all times when the ice is not viewed (rate of growth, 18 gm/hour). The resulting film is smooth and homogeneous. Bright-field imaging reveals no sign of crystallinity even at high magnification. There is no indication of nee- dle or stalactite-like surface features, and the diffraction pattern directly after deposition does not show crystalline diffraction peaks for depo- sition temperatures below about 120 K. The form transitions are observed to occur simultaneously across the entire substrate. Low-electron dose imaging techniques minimize sample degrada- tion during viewing. Electron beam damage does not change the diffraction patterns on time scales of each measurement, and we move to a different part of the ice layer in subsequent measurements. 12. A. H. Narten, C. G. Venkatesh, S. A. Rice, J. Chem. Phys. 64,1106 (1976). 13. T. C. Sivakumar, S. A. Rice, M. C. Sceats, ibid. 69, 3468 (1978). 14. L. G. Dowell and A. P. Rinfret, Nature 188, 1144 (1960). 15. A. Hallbrucker and E. Mayer, Icarus 90, 176 (1991). 16. B. Schmitt, S. Espinasse, R. J. A. Grim, J. M. Greenberg, J. Klinger, in Proceedings of the In- ternational Workshop on Physics and Mechanics of Cometary Materials, T. D. G. J. Hunts, Ed. (European Space Agency, Paris, 1989), vol. ESA SP-302, pp. 65-69. 17. N. J. Sack and R. A. Baragiola, Phys. Rev. 8 48, 9973 (1993). 18. L. Pauling, The Nature of the Chemical Bond (Cornell Univ. Press, Ithaca, NY, 1960), chap. 12, pp. 449-504. 19. Our choice of nomenclature for the amorphous forms is given by the success of describing amorphous forms by random network models and follows previous suggestions to name the amor- phous form l, [J. Dubochet, J. Chang, R. Free- man, J. Lepault, A. W. McDowall, Ultramicroscopy 10, 55 (1982)], 1as (21), or la (17). We prefer this notation to that used in studies related to liquid water, that is, H20(as), H2O(as,l), or (H2O),,. Although the Roman numeral polymorph designa- tions are generally applied only to the crystalline phases, the structure of the low-density amor- phous form is well represented by a disordered ice network and the high-density form by such a network including interstitial water. We believe that alternative models for high-density ice based on one of the 11 high-pressure phases of water (4, 12) are unlikely because they are either outside their stability regime or demand a form of proton ordering [B. Kamb, in Water and Aqueous Solu- tions, R. A. Home, Ed. (Wiley-lnterscience, New York, 1973), pp. 9-24]. Of course, we realize that an amorphous form may not have a unique struc- ture. For example, the high-density form might have a varying content of interstitial water if dep- osition conditions vary. The justification for giving names to these forms is that there are well- defined transition zones between the different forms. Alternative amorphous forms have been made in the laboratory (and similarly bear some relation to ice 1) by pressure-induced amorphos- ing hexagonal ice 1h at 77 K and 10 kbar [0. Mishima, L. D. Calvert, E. Whalley, Nature 310, 393 (1984)]. This synthetic ice shows a similar high- to low-density transition, but at a higher temperature of 105 to 128 K [Y. P. Handa, 0. Mishima, E. Whalley, J. Chem. Phys. 84, 2766 (1986); A. Bizid, L. Bosio, A. Defrain, M. Oumez- zine, ibid. 87, 2225 (1987); J. Tse, ibid. 96, 5482 (1992)]. 20. L. C. Allen, in Physics and Chemistry of Ice, E. Whalley, S. J. Jones, L. W. Gold, Eds. (Royal Society of Canada, Ottawa, 1974), pp. 13-18. 21. A. Kouchi, Nature 330, 550 (1987); J. Cryst. Growth 99, 1220 (1990). 22. R. J. Speedy, J. Phys. Chem. 96, 2322 (1992). 756 23. J. A. Ghormley, J. Chem. Phys. 48, 503 (1968); A. Hallbrucker, E. Mayer, G. P. Johari, J. Phys. Chem. 93, 7751 (1989). 24. A. G. G. M. Tielens, W. Hagen, J. M. Greenberg, J. Phys. Chem. 87, 4220 (1983); G. A. Baratta and G. Stazzulla, Astron. Astrophys. 240, 429 (1990). 25. J. M. Greenberg, A. J. Yencha, J. W. Corbett, H. L. Frisch, Mem. Soc. R. Sci. Liege 6e serie, tome IlIl, 425 (1972); W. Hagen, L. J. Allamandola, J. M. Greenberg, Astrophys. Space Sci. 65, 215 (1979). 26. P. Jenniskens et al., Astron. Astrophys. 273, 583 (1993). 27. W. A. Schutte, thesis, University of Leiden, Lei- den, Netherlands (1988). 28. A. Kouchi and T. Kuroda, Nature 344, 134 (1990). 29. M. Duncan, T. Quinn, S. Tremaine, Astron. J. 94, 1330 (1987). 30. J. A. Ghormley, J. Chem. Phys. 48,1321 (1967). 31. We acknowledge the work of G. Palmer, who is responsible for a number of important modifica- tions to the electron microscope, and A. Breon, who automated the reduction process of large batches of diffraction patterns. M. A. Wilson as- sisted in the analysis of the diffraction patterns. We thank W. A. Schutte for permission to repro- duce figure 111.1.4 from his thesis in Fig. 4. This report benefited from discussions with A. Po- horille, A. G. G. M. Tielens, L. J. Allamandola, F. Freund, and S. Chang. This work was supported by grants from the Exobiology and Planetary Materials and Geochemistry Programs of the Na- tional Aeronautics and Space Administration and was performed while P.J. held a National Re- search Council-Ames Research Center Research Associateship. 3 February 1994; accepted 20 June 1994 Infrared Laser Spectroscopy of the Linear C13 Carbon Cluster T. F. Giesen,* A. Van Orden, H. J. Hwangt R. S. Fellers, R. A. Provenqal, R. J. Saykally The infrared absorption spectrum of a linear, 13-atom carbon cluster (C13) has been observed by the use of a supersonic cluster beam-diode laser spectrometer. Seventy-six rovibrational transitions were measured near 1809 wave numbers and assigned to an antisymmetric stretching fundamental in the 15 + ground state of C13. This definitive structural characterization of a carbon cluster in the intermediate size range between C1O and C20 is in apparent conflict with theoretical calculations, which predict that clusters of this size should exist as planar monocyclic rings. The structure and bonding in pure carbon molecules have been of interest for many years because of the importance of these species in many contexts, ranging from dust- grain formation in the interstellar medium to soot formation in combustion systems. Re- cently, this interest in carbon clusters has intensified because of the dramatic emer- gence of fullerene science, centered primar- ily about the discovery and characterization of the C60 molecule and other members of this "third form of carbon." A review article by Weltner and Van Zee describes research conducted before 1989 (1). Experimental (2) and theoretical (3) evidence suggests that the formation of C6( and other fullerenes proceeds by a mechanism in which small carbon clusters undergo condensation from linear chains through monocyclic rings and finally to large three-dimensional, cage- like structures. A detailed characterization of this mechanism, as well as those for related processes like soot formation, re- quires a thorough understanding of how the structure and bonding evolve in smaller carbon clusters as the cluster size increases. Department of Chemistry, University of California, Berkeley, CA 94720, USA. *Present address: I. Physikalisches Institut, der Uni- versitat zu Koin, 50937 Koin, Germany. tPresent address: Department of Chemistry, Kyung Hee University, Seoul, 130-701, Korea. SCIENCE * VOL. 265 * 5 AUGUST 1994 Extensive theoretical and experimental efforts have been under way for several years to elucidate these mechanisms. From theo- retical considerations, the picture that has emerged is that the small, odd-numbered clusters of C3 to C9 exist exclusively as cumulenic linear chains with 'Ig+ ground electronic states, whereas the even-num- bered clusters of C4, C6, and C8 have two nearly isoenergetic structural isomers, a 3 - linear chain and a nearly planar singfet cyclic ring (4-6). Above C9 a transition occurs, and the ground-state structures of both even and odd clusters become planar monocyclic rings, while the corresponding linear structures become relatively high in energy (5, 7, 8). This trend is thought to continue for C10 to C20. As the cluster size increases toward C20, high-level ab initio calculations become unfeasible, although several calculations at lower levels of theory have been reported (9, 10). Ion mobility measurements indicate that a rich variety of structures begins to form above C20, includ- ing planar monocyclic and polycyclic rings (1 1). Some calculations suggest that clusters as small as C18 may exist as fullerenes (10), although there is no experimental evidence to support this suggestion. However, it is clear that fullerenes become the most stable structures at sizes larger than C30. Experimental verification of this picture iffi ...... RWAYAWRIffi~ on March 26, 2016 Downloaded from on March 26, 2016 Downloaded from on March 26, 2016 Downloaded from on March 26, 2016 Downloaded from