Detection of C3 in the Circumstellar Shell of IRC+10216

Vibration-rotation lines of C3 have been identified in the circumstellar spectrum of the obscured carbon star IRC+10216. This molecule is of interest in both the chemistry of flames, where it may be involved in the formation of soot, and in astrophysics, where it is a potential building block for carbonaceous grains. This high-resolution infrared detection of the pure carbon chain molecule C3 allows the estimation of the equilibrium C-C bond length, 1.297 angstroms. Possible astrophysical formation and destruction mechanisms for C3 are reviewed, including the relationship between C3 and carbon clusters.

Vibration-rotation lines of C3 have been identified in the circumstellar spectrum of the obscured carbon star IRC+ 10216. This molecule is ofinterest in both the chemistry of flames, where it may be involved in the formation of soot, and in astrophysics, where it is a potential building block for carbonaceous grains. This high-resolution infrared detection of the pure carbon chain molecule C3 allows the estimation of the equilibrium C-C bond length, 1.297 angstroms. Possible astrophysical formation and destruction mechanisms for C3 are reviewed, including the relationship between C3 and carbon clusters. AUNIQUE GROUP OF SPECTRAL lines at 4050 A was first detected in the spectrum of a comet in 1881 (1). It was not until 1951 that A. E. Douglas (2) was able to attribute the 4050 A group to the C3 molecule. The most thorough analysis of the A'HU-AX1' transition of C3 was carried out by Gausset, Herzberg, Lagerquist, and Rosen (3). In addition to comets and electrical discharges, the C3 molecule has been observed in flames (4) and explosions (5) and in the vaporization of carbon (6). There is some low-resolution evidence for the appearance of C3 in the atmospheres of cool stars (7). In flames, C3 may be a critical link in the formation of soot (4). In astrochemistry, pure carbon chains, including C3, have been linked to the formation of circumstellar grains (8) and may be involved in the formation of diffuse interstellar bands (9).
The detection of C3 in carbon stars and in the interstellar medium is inhibited by lack of ultraviolet (UV) flux at 4050 A. Because C3 is linear (3) or quasilinear (10) there is neither a dipole moment nor strong microwave pure rotational transitions. There are three vibrational modes, v1 the symmetric stretch (ug6) at 1224.5 cm-n (11), V2 the bending mode (,,) at 63.1 cm-' (3), and V3 the antisymmetric stretch (ou) at 2040 cm- (12). However, v, is Raman active, not infrared active, and V2 occurs in the far infrared, which remains a difficult spectral region to observe. The antisymmetric stretch, V3, is ideal for monitoring C3 because it has a remarkably strong transition dipole moment of 0.44 D (10)   After appropriately scaling the telluric line strengths to match the air mass of the IRC+10216 spectrum, the lunar spectrum was ratioed to the IRC+ 10216 spectrum to remove the telluric spectrum.
The goal of our observations was to detect the CN vibration-rotation fundamental (15). IRC+ 10216 is frequently used for high sensitivity searches of circumstellar molecules because it is a bright source and lines of circumstellar origin appear against a featureless continuum, not superposed on photospheric lines (16). Although the CN lines were not obviously present (17), a series of strong lines was observed (Fig. 1), which on the basis of combination differences (3) were identified as C3. More R branch lines were identified on an archival spectrum (18 October 1978, resolution 0.026 cm-, S/N = 95) that covered the 2040 to 2170 cm-1 region. Prior to the discovery reported in this paper of the C3 v3 band in the circumstellar envelope of the carbon star IRC+ 10216, no high resolution spectroscopic observations of the infrared spectrum of C3 were available.
At the time this work was completed no laboratory frequencies were available for the 2040 cm-1 C3 lines, thus the frequencies had to be derived from our spectrum. To do this we assumed that, in analogy to the other molecules observed in the infrared with similar line shape and excitation temperature (see below), the lines exist in the circumstellar shell at the terminal expansion velocity, -14 km s-1 relative to the center of mass (16). Assuming this velocity and correcting for a local standard of rest velocity of -26 km s-' for IRC+10216 (18) and for the earth's velocity at the time of the observation, frequencies were measured. Telluric CO2 lines (19) gave the frequency zero point. Subsequent to the submission of this report, we have learned of two research groups reporting laboratory detections of the V3 band of C3 (20). The laboratory observations agree with the work reported here.
The observed line positions (Table 1) were reduced to the molecular constants by means of the customary (21) rotational energv level expression: where B and D are constants, and J is the angular momentum quantum number. The constants obtained from this fit are reported in Table 2.
The high resolution detection of v3 completes the analysis of the C3 fundamental modes ofvibration and allows the derivation of an equilibrium molecular structure. The vibrational dependence of the rotational constants can be represented by (3,11,21) B,,,V,V = 0.41767 -0.0048(v1 + 2 + 0.01345(v2 + 1) -0.00071(v2 + 1)2 + 0.00510(v3 + 9) (2) where t'j is the vibrational quantum number of mode vi. From the value of Be, 0.41767 cm-, the equilibrium C-C bond length (re) of 1.2968 A was calculated. The total error in re is hard to assess because the extremely floppy C3 molecule executes large-amplitude bending motions (10), but ±0.001 A seems a reasonable assumption. The equilibrium bond length of a molecule is an easy quantitv to compute by ab initio quantum chemical calculations. For example, Kraemer, Bunker, and Yoshimine (10)  One unusual feature of the v3 mode is the vibrational dependence of the rotational constant BVIV,V3 For most linear molecules the rotational constants of the antisymmetric stretching mode decrease with increasing vibrational excitation but for C3 the opposite trend is observed. The nonrigid bender calculations of Kraemer, Bunker, and Yoshi-mine (10) predict this effect; at3 (calculated) = -0.003 cm-' compared to a3 (observed, this work) = -0.00510 cm-.
The C3 lines in the spectrum of IRC+ 10216 appear to be saturated and we have used an observer's frame radiative transfer code to model the lines in the circumstellar shell (23). Lacking a detailed knowledge of the spatial distribution, we assume a r-2 density distribution truncated at some inner radius. The assumption of a continuous C3 distribution throughout the envelope, as if the C3 originated in the photosphere similar to CO, C2H2, HCN, and CS (16,(23)(24)(25), results in synthesized line profiles that bear no resemblance to the observations. The line profiles of species with continuous distributions are strongly influenced by higher temperature (T > 100 K) gas in the inner regions (26) (r < 300 R., where 1 R. = 6.7 x 1013 cm) ofthe circumstellar shell (23). Moreover, the (0,1,1) +-(0,1,0) "hot" band, whose lower level lies only 63 cm-l above the ground state, falls within our bandpass. It is not seen. For a continuous distribution of C3, the model's warm gas results in synthetic hot band line strengths that start to rival the cold band line strengths.
The observed rotational distribution of C3 is inconsistent with a photospheric origin and widespread distribution similar to that of the molecules discussed above. Given the small rotational constant of -0.43 cm-', the significant strength in the low J lines (unresolved and -50% deep) and the apparent lack of absorption for lines with J > 20 imply that the C3 absorption arises well away from the photosphere. Using the undetected high excitation (J > 20) lines, we establish an upper limit on X(C3) of 1.5 x 10-7 in the warm, inner envelope (<300 R.). With the assumption of an inner boundary at 350 R., the radiation transport calculations require a range of temperature, from 70 K to -20 K, and a C3 column density of 1(±+ 0.15) x 1015 cm-2.
The strongest C3 lines have optical depths of about unity. The fractional abundance of C3 relative to molecular hydrogen, X(C3), is 1.2 x 10-6 in the region modeled. The strongest synthesized lines in the (0,1,1) <-(0,1,0) hot band are -2% deep, and are undetectable in our present spectrum. Higher signal-to-noise data will be very useful in further constraining the gas temperature. If the actual density distributions were more compact or peaked, then maintaining the same column density would require a larger fractional abundance. However, the range of temperature indicated by the transport calculations points toward an extended distribution. The question arises as to the origin of the C3. The difficulty with a direct photolysis origin for C3 is the paucity of possible parent molecules. Methyl acetylene and its isomer allene, C3H4, are suspected C3 photolysis parents in comets (27). It seems unlikely that such heavy molecules would be abundant enough in IRC+ 10216 (28) to account for the observed C3 and, indeed, this seems to be the case. Upper limits of 3 x 10-7 have been found for the abundance of methyl acetylene (29), 2 x 10-7 for allene (25), and 5 x 10-7 for diacetylene (C4H2) (25). The cyanoacetylene (HC3N) abundance is a few times 10-7 (30), subject to considerable uncertainty, making it difficult to rule out. However, laboratory studies of vacuum UV cyanoacetylene photolysis indicate the photodissociation daughters to be C2H and CN (31).
In addition to direct photolysis, C3 could be the result of photolysis and subsequent neutral or ion gas phase chemistry. Ionmolecule reaction schemes to form C3, initiated bv cosmic ray ionization of helium and (more importantly) acetylene photolysis have been suggested (32), but these manufacture C3 only in the cold outer envelope (r > 1000 R., T < 15 K). The infrared (16) and microwave (33) C2H observations are consistent with an outer envelope mother molecule abundance x(C2H2) -3 X 10-6.
With this assumption, our chemical kinetics and radiation transport calculations indicate that a few percent ofthe required amount of C3 is formed. Even if the outer envelope acetylene abundance were an order of magnitude larger, the cold gas so produced could not account for the absorption in the higher excitation (J > 6) C3 lines.
Carbon condensation products form another possible photolysis parent for C3. Carbon condensation is thought to occur in or near the stellar photosphere (26) and may involve the growth of carbon clusters into macroscopic "grains" (8,34). As nucleation effectively ceases, a remnant distribution of microscopic particles, C, clusters, might survive. In laboratory experiments (8,34), both clusters and macroscopic particles are produced subsequent to the laser vaporization of graphite. Still other experiments (35) suggest that single 3.5-eV photons are effective in fragmenting carbon cluster ions, and even suggest the existence (36) of a "magic" fragment, C3. The intriguing possibility is that C3 is but one photofragment daughter of a much larger carbon cluster, with other daughters (C2, C4, C5, C6, and so on) possible. Although the suggestion that the photolysis of carbon condensation products is a source ofC3 is attractive, the caveat must be made that none of the condensation products, except for grains, have been observed in the circumstellar envelopes of stars. Observational confirmation of these products in circumstellar envelopes will be difficult.
A possible problem with a hypothesized photolysis origin for C3 is that species whose origin is known to be directly photolytic (C2H and CN) occur much further out (16,33) in the circumstellar shell at r -1000 R* and Trot 12 K. However, less energetic photons may be required to photofragment large carbon clusters than are required to fragment the photolysis par-  (31)]. These less energetic photons penetrate the circumstellar envelope more deeply, because the dust is less opaque at the longer wavelengths. Thus C3 could be photolytically produced and be warmer than the C2H. Note that since C3 lacks a permanent dipole moment, the radia-9 SEPTEMBER I988 tive and collisional excitation dynamics controlling the rotational temperature will differ from CN and C2H (as well as SiO, CS, HCN, and SiS) in the sense of C3 being potentially closer to local thermodynamic equilibrium, similar to the situation for other molecules having either no permanent dipole moment (CH4, SiH4, C2H2) or a small one (CO).
Of course, the C3 could be formed on carbon-rich grains either through photolysis or, as is thought to be the case (25) for NH3, CH4, and SiH4, catalytic reactions. With a standard model for the gas and dust density (26) and assuming a grain radius of 0.1 p,m (16), and X(C3) -10-6 at 400 R., we find that 105 carbon atoms must be released per grain. Although this is only 0.1% of the grain mass, this assumes all carbon atoms are released as C3. Note the total relative abundance of the trace species NH3, CH4, and SiH4 is -1 x 10-6 (25), similar to the C3 abundance. However, these presumably catalytically produced species seem to leave the grains in a considerably warmer region (25). This still leaves photolytic interactions with the grains as a possibility.
In addition to the origin of C3, its subsequent chemical behavior is potentially of great importance to the chemistry of the circumstellar shell. Because H2, C2H2, and HCN are highly abundant at r < 100 R. in the circumstellar shell of IRC+ 10216 and because many complex hydrocarbons are observed at r > 1000 R. (28), neutral gas phase reactions between H2, C2H2, and HCN molecules, and C3 are of potential importance. Studies (37) of ground-state C3 reacting with other species indicate that the reaction rates are typically factors of 102 to 104 slower than the gas kinetic collision rate.
Ifthis relatively slow reactivity applies to the molecular collision partners likely to be encountered in the circumstellar envelope then, with the possible exception ofmolecular hydrogen (because of its large relative abundance), neutral gas phase reactions with C3 probably are not important on the time scales involved. The C3 bond energy is 7.3 eV (31) and the ionization potential is 12.1 eV (38). These large values imply a relatively long life for C3 against the eventual photodestruction into the more reactive species C and C2. is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for partial support of this work. Partial support also was provided by the Air Force Astronautics Laboratory grant F 04611-87-K-0020. 29 April 1988; accepted 27 July 1988 Solid-State Ice Volcanism on the Satellites of Uranus DAVID G. JANKOWSKI AND STEVEN W. SQUYRES Voyager images of the uranian satellites Ariel and Miranda show flow features with morphologies indicating that ice has been extruded to the satellites' surfaces in the solid state. These images provide the first observational evidence for solid-state ice volcanism in the solar system. Topographic profiles have been measured across a number of flow features on Ariel. With a simple model of extrusion, spreading, and cooling of a viscous flow, the initial viscosity of the flow material is found to have been no more than about 1016 poise, far lower than expected for H20 ice at the ambient surface temperatures in the uranian system. Sharply reduced viscosities may have resulted from incorporation of ices like NH3 or CG4 in the uranian satellites.

ONE OF THE MOST INTERESTING
results of the Voyager mission to Jupiter, Satum, and Uranus has been the extent to which even small icy satellites of those planets have undergone geologic resurfacing (1)(2)(3)(4)(5). It generally has been uncertain, however, whether the resurfacing has taken place in the liquid or solid state. In the jovian system, buoyancy considerations seem to favor extrusion of warm, mobile ice rather than liquid water (6). However, no landforms clearly indicative of solid-state extrusion have been observed there. On the smaller saturnian satellites, mobility considerations favor extrusion of liquid (7), and the observed morphology is at least consistent with this interpretation. In the uranian system, there is morphologic evidence for extrusion in the solid state (5). In this report we consider this evidence, quantify the morphology of the flows, derive approximate viscosities at the time of extrusion, and consider the implications of these viscosities for the compositions and thermal histories of the uranian satellites.
The morphologic evidence for solid-state resurfacing on the uranian satellites is best developed on Ariel (Figs. 1 and 2). Much of the satellite's surface is transected by a pattern of linear graben-like canyons. On the floors of some of the grabens are deposits that have low crater densities compared to the rest of the satellite (8) and have topographic profiles that appear to be strikingly smooth and convex. They appear to steepen significantly at the margins, in what seem to be flow fronts. Some flows contain medial grooves running parallel to the graben walls. The convex topography and the concentration of the deposits in grabens strongly suggest that these deposits are materials that were extruded to the surface in the solid state, probably in an extensional environment.
In some cases, the flows appear to have been confined by graben walls. In others, however, they clearly have not. Figure 1 shows an instance where the lateral extent of a flow within a canyon remains relatively constant along the canyon's length, whereas the width of the canyon changes from that of the flow to roughly three times it. Figure  2 shows an instance where a flow has spread out across a plains unit, partially burying an impact crater.
It is likely that the sources of most flows were linear fracture systems on the graben floors. This interpretation is supported by (i) the likelihood that extensional fractures associated with grabens would parallel the graben walls; and (ii) the apparent steep flow fronts and limited lateral extent of some flows, indicating that they spread a short distance laterally rather than flowing long distances parallel to the graben walls. In large terrestrial grabens, volcanism is commonly concentrated along vents near the graben axis (9)(10)(11); such may also be the case on Ariel. At least some of the medial grooves may be the juncture of two independent flows that have been extruded from parallel fractures on a graben floor, have spread laterally and met, but have not completely coalesced.
Evidence for solid-state resurfacing is also observed on Miranda (Fig. 3). Miranda's surface consists of two types of materials: an old, heavily cratered terrain and a younger terrain transected by a complex pattem of subparallel bands, scarps, and ridges. Areas of this younger terrain are known as "coronae." In some places where corona materials come into contact with the ancient heavily the National Science Foundation. P.F.B. is an Alfred P. Sloan Fellow and Camille and Henry Dreyfus