6. Conclusions and future work
We have made high spatial resolution and sensitivity CO(J= and 21) observations of the semiregular carbon star TT Cyg with the IRAM Plateau de Bure interferometer. They show that this star is surrounded by a large, geometrically thin shell of gas that expands with a velocity of 12.6. The age of the shell is estimated to be 7000 yr. It has an almost perfect overall spherical symmetry, e.g., its radius varies by less than 3% over the sphere. The width varies by a factor of three, but the emitting gas is rather evenly distributed when averaged over a solid angle of about 0.2 steradians. Assuming a homogeneous medium and =10-3, we estimate a molecular hydrogen density of 250 cm-3, a gas kinetic temperature of 100 K, and a mass of 0.007 for the shell (adopting the Hipparcos distance of 510 pc). These estimates are rather uncertain since neither the density nor the temperature are well constrained by the observational data, and in addition there is good evidence for a highly clumped medium. There is no evidence for matter immediately inside or outside the shell. The brightness there is 5% of the shell peak brightness when averaged azimuthally. We find no evidence for any particular structure (i.e., deviations from a Gaussian distribution) in the radial direction of the shell brightness distribution. The shell centre is clearly displaced by 17 (position angle -20o) with respect to the star. We tentatively interpret this in terms of TT Cyg being a member of a binary system. We put forward several arguments for a shell medium that consists almost entirely of a large number of small (1") clumps: e.g., the patchy brightness distributions, the interferometer flux being close to the single-dish flux even in areas of very extended emission, the constant total emission as a function of line-of-sight velocity, and the protection of the CO molecules against photodissociation. In this case the density required to fit the observational data is much higher, 104 cm-3, and the kinetic temperature is considerably lower, 20 K.
TT Cyg is presently losing mass at a quite modest rate, 310, and with a low expansion velocity, 3.8. Both values are low even when compared with the average values obtained for a sample of visually bright carbon stars, 1.510 and 12.5 The systemic velocities estimated from the centre and the shell emissions agree perfectly, and the final value is estimated to be very accurate, -27.30.1 in the LSR system.
We argue that TT Cyg has gone through a phase of increased, coordinated mass ejection, that may be related to a recent He-shell flash. At this point, we do not know how the effects of this internal thermo-nuclear runaway is transferred to the mass ejection mechanism. It seems very likely that the star had a low-velocity wind with a moderate mass loss rate before the ejection. One would therefore expect some effects of interacting winds. Primarily, this may affect the morphology of the shell, as well as provide swept-up gas. To what extent these effects are present is still uncertain, and consequently also the details of the mass ejection.
It is clear that more work is required before the shell formation process is identified and understood in detail. The proposed scenarios (a brief high mass loss rate epoch and interacting winds with swept-up gas, in both cases possibly connected with a He-shell flash) can be tested in various ways. One possibility is to observe in detail detached CO shell sources with different shell ages. Statistically, the younger shells should contain less mass, and perhaps also have considerably larger widths if interacting winds play a rôle. The first interferometer results on the very young (103 yr) shell around U Cam indicate that the linear width of the shell is very similar to that of the TT Cyg shell, but the shell expansion velocity is higher by about a factor of two, the mass is lower by almost a factor of ten, and the kinetic temperature is lower by a factor of two (Lindqvist et al. 1999). Taken at face values, and provided that the mass ejections of U Cam and TT Cyg were very similar, this would support an interacting wind scenario, but we caution here that these results are obtained assuming a homogeneous medium. One cannot exclude the possibility that the CO emission does not trace the density distribution, e.g., due to photodissociation or lack of excitation. Unfortunately, other molecular radio line emissions have been possible to detect only towards the youngest CO shells, U Cam (Bujarrabal & Cernicharo 1994; Lindqvist et al. 1996) and R Scl (Olofsson et al. 1996). Therefore, the development of alternative observational methods are important. An obvious possibility is to observe the dust emission with high spatial resolution. The relation between the dust and the gas is bound to be different in the two scenarios. Unfortunately, the angular resolutions of the IRAS and ISO observations have not been high enough for exploring the detailed structure of the CO shells. The detected dust shells, in which no CO emission has been found (Olofsson et al. 1993), seem to be much broader than the CO shells (Waters et al. 1994; Izumiura et al. 1996, 1997). Thus, the connection between these dust shells and the CO shells is not clear. We note that in the hydrodynamical simulations of Steffen et al. (1998) geometrically thick, detached dust shells are produced as a consequence of the long period of low mass loss rate following a He-shell flash. The first results using stellar light scattered (by atoms and/or dust) in the shell of U Ant suggest that at least the outer edge of the CO emission also defines a sharp decrease in the density distribution (González Delgado et al., in prep.).
It is also important to numerically simulate, including hydrodynamical effects and using a two-fluid medium (dust and gas), the evolution of expanding thin shells and interacting winds [see e.g., Mastrodemos et al. (1996), Steffen et al. (1998)]. Recently, Steffen & Schönberner (1999) have simulated the result of a He-shell flash induced mass ejection and subsequent wind interaction and obtain results that are very similar to those observed for the CO shell of TT Cyg. Considering the clumpy structure of the shell gas, the evolution of primordial structures and the development of new ones due to instabilities are interesting. Myasnikov et al. (in preparation) have shown that clumpy structures, of the type found in the TT Cyg shell, may be ascribed to the development of Rayleigh-Taylor instabilities in the interaction zone (comparable in size with the width of the TT Cyg shell) between a massive stellar wind and a previous slower and less massive one. If, on the other hand, the winds are clumpy already from the start, the interaction of the winds becomes a more complicated phenomenon to study. Furthermore, it is possible that the CO shells are related to the multiple spherical shells and halos seen around many planetary nebulae (Frank et al. 1994; Stanghellini & Pasquali 1995). Finally, the shell clumps may appear in compressed form in planetary nebulae. A beautiful illustration of this is provided by the cometary knots in the Helix nebula (O'Dell & Handron 1996). O'Dell & Handron estimate that there are about 3500 `detectable' cometary knots in the Helix nebula. Meaburn et al. (1998) found that the morphological properties of the knots are remarkably similar. Extinction and CO radio line emission both give knot masses and densities of 10-5 and 105-106 cm- 3, respectively (Huggins et al. 1992; Meaburn et al. 1998). The diameters of the largest knots are 31015 cm, and the CO line widths are narrow, 1-2. One may conclude that the knot properties are relatively similar to those of our clumps, except for their higher densities and masses. On the other hand, we do not believe that the TT Cyg shell is related to the shell structures seen in the HST images of CRL2688 (Sahai et al. 1998). These correspond to a much shorter time scale, and may possibly be related to mass loss variations during the (long) pulsational cycles of the central star at the tip of the AGB and subsequent interactions.
© European Southern Observatory (ESO) 2000
Online publication: December 17, 1999