Mass loss is a fundamental process during stellar evolution on the asymptotic giant branch (AGB). It limits the lifetime and the maximum luminosity reached on the AGB and puts a high lower mass limit for supernova precursors (Blöcker 1995), it affects the elemental and isotopic abundances at the stellar surface (Forestini & Charbonnel 1997) and the stellar contribution to nucleosynthesis, and it provides one of the most important replenishment processes for the interstellar medium. Even though its existence is well established, the details and the mechanism behind it remain uncertain (Olofsson 1996). It appears that the average mass loss rate increases during the AGB evolution, but there is probably also a dependence on main sequence mass, so that the more massive stars reach higher mass loss rates (Habing 1996). On shorter time scales (a few years up to 104 yr) there is ample evidence for variations. For instance, the marked 60µm-excesses of a number of carbon stars were interpreted as arising from detached dust shells by Willems & de Jong (1988), and for a sample of M stars by Zijlstra et al. (1992). Possibly the best examples of episodic mass loss are the detached CO and dust shells that have been detected towards seven AGB-stars, all carbon stars (Olofsson et al. 1990, 1996; Lindqvist et al. 1996, 1999; Waters et al. 1994; Izumiura et al. 1996, 1997). A similar dust shell may have been detected towards one M star (Hashimoto et al. 1998). The most spectacular result is probably the large (radius35") and remarkably thin CO shell found around the carbon star TT Cyg (Olofsson et al. 1998; width/radius 0.04 in the region covered by their maps). The CO radio line emission appears to probe the mass loss history for 104 yr, after which the CO molecules become rapidly photodissociated. The dust emission is maintained for a longer time, but the interpretation of these data are hampered by the poor spatial resolution.
The central stars of the `detached shell'-systems are irregular or semiregular variables, which presently have low mass loss rate winds (10-8-10) that expand at low velocities (5, as opposed to the shells which expand at 15-25). The estimates of the shell masses are uncertain, but, in general the result is about 0.01 (Groenewegen & de Jong 1994; Olofsson et al. 1996; Izumiura et al. 1997). The corresponding mass loss rates depend on whether or not swept-up material plays a rôle and the time scale of ejection, which may be as low as a few hundred years as judged from the TT Cyg results (Olofsson et al. 1998). In any case, these stars have certainly gone through relatively drastic changes in their mass loss behaviour. If the process responsible for these changes can be identified, some light may also be shed on the mechanism behind the mass loss itself. So far, a connection with a He-shell flash seems the most probable explanation (Olofsson et al. 1990; Izumiura et al. 1997; Schröder et al. 1998, 1999; Steffen & Schönberner 1999). If so, these shells may be one of very few ways to investigate this astrophysically very important process.
The presence of a detached CO shell around TT Cyg was first suggested by Olofsson et al. (1990), and Olofsson et al. (1996) provided direct evidence for such a shell of diameter 70". A high spatial resolution investigation of the CO shell around TT Cyg, using the IRAM interferometer on Plateau de Bure, was started by Olofsson et al. (1998). Here, we present interferometer observations that now cover the entire shell. At the end we discuss the constraints that these put on the possible shell formation scenarios. TT Cyg is a short-period (120d) semi-regularly variable (SRb) carbon star. Its Hipparcos position is (2000) = 19h40m5702, (2000) = 32o37´057, and its galactic latitude is 4.9o with the galactic plane to the southeast along the position angle 35o. The Hipparcos distance is 510 pc, but the parallax is uncertain, 1.960.8 mas.
© European Southern Observatory (ESO) 2000
Online publication: December 17, 1999