Astron. Astrophys. 323, 449-460 (1997)

1. Introduction

After the termination of He-core burning, low- and intermediate-mass stars begin to ascend along the asymptotic giant branch (AGB). Initially, during the early AGB evolution (E-AGB), the stars have a burning shell of He while the H-shell is inactive. Later they proceed to the stage of a double burning shell, at which the He-shell predominantly burns during shell flashes or thermal pulses (TP-AGB) and develop a degenerate carbon-oxygen core (Schwarzschild and Härm 1965). The AGB evolution is dominated by mass loss which determines the final mass of the stars (cf. Vassiliadis & Wood 1993 and references therein) as well as by the thermal pulses which may affect the mass loss. Thermal pulses, on the other hand, cause the third dredge-up of processed material in the burning shells, in particular ¹² C, into the surface layer, making the star evolve into a carbon star under certain conditions (Sugimoto & Nomoto 1975, Iben 1975). The interplay between the mass loss and the thermal pulses is crucial for understanding the AGB evolution.

Recently several studies have shown that considerable changes in mass loss occur on a short time scale during TP-AGB evolution. Willems & de Jong (1988) explained the excess far-infrared emission among a significant fraction of optical carbon stars as due to a cold detached dust shell caused by a very recent termination of high mass loss. They suggest that the thermal pulse which had turned the star into a carbon star caused a temporal cessation of mass loss activity. Based on their discovery of detached, geometrically thin CO gas envelopes around a few carbon stars with excess far-infrared emission Olofsson, Eriksson, & Gustafsson (1988) and Olofsson et al. (1990) suggested that the CO envelopes were produced in a high mass loss phase with very short duration, which necessarily implies a corresponding detached dust shell which is also responsible for the excess far-infrared emission. Zijlstra et al. (1992) and Hashimoto (1994) analysed the IRAS photometric data to find that dozens of oxygen-rich AGB stars show similar excess infrared emission, presumably also produced by a detached dust shell. These results show that a temporal decrease or increase of mass loss rate occurs among oxygen-rich AGB stars as well as carbon stars.

Details are not very clear, however, about the actual relationship between the thermal pulses and the mass loss. The study of the mass loss history on the AGB on time scales of the order of 1,000 to 10,000 years is necessary to detect remnants of different thermal pulses that must have occurred during AGB evolution. Such long term history of mass loss can be traced only by studying the spatial distribution of dust grains through their far-infrared emission, since observations of molecular gas are less suitable for this purpose due to photodissociation by the interstellar UV radiation field. Hacking et al. (1985) first pointed out that dozens of AGB stars were extended in the IRAS survey data of the 60µm or 100µm band, showing that the IRAS database could be used for such studies. Gillett et al. (1988) examined the structure of a dust shell extending more than 1pc around R CrB with the IRAS survey data. Stencel, Pesce, & Bauer (1988a, b) further demonstrated the potential of the IRAS database for the investigation of extended dust shells through their studies of supergiant stars based on the IRAS survey scan data. Hawkins (1990) found that W Hya, an AGB star with the variable type SRb, possesses a very extended dust shell in the IRAS data. Moreover, a detailed analysis was made by Young et al. (1993a) of the structure of dust shells around many AGB stars through fitting a ellipsoidal dust shell model to the 60µm IRAS survey data. Their method was very sensitive to detect such an extended emission component. However, their assumptions that the dust shell is isothermal and single may be too simple, which might result in overlooking important features of some dust shells.

Using Pyramid Maximum Entropy (PME) image reconstruction techniques (Bontekoe et al. 1994), it has become possible to explore the geometry of dust shells around AGB stars in reconstructed High resolution IRAS (HIRAS) images, which are suitable for investigating the influence of thermal pulses on mass loss. HIRAS images allow us to examine these dust shells without introducing any assumptions about their geometry. Using the HIRAS processor we initiated a program to study the structure of extended dust shells around optically bright carbon stars. We selected carbon stars which (1) have a 60µm flux density greater than 5 Jy in the IRAS Point Source Catalog (PSC, Joint IRAS Science Working Group 1988), (2) are listed in the general catalogue of cool galactic carbon stars (Stephenson 1989), and (3) show an excess emission at 60µm and/or 100µm. We follow the definition of 60µm and 100µm excesses by Zijlstra et al. (1992) and Loup (1991), respectively. Fifty-one stars were selected and fourty-two have been processed for the 60µm images. Background confusion is too severe for the remaining 9 stars to be properly processed. Well resolved sources were further examined in the 100µm images. Waters et al. (1994) reported the first HIRAS results of a very extended, probably detached, dust shell around the N-type carbon star U Hya. They could determine directly the position of the inner shell edge to find that the age of the dust shell is compatible with the interpulse period for an intermediate mass star.

Here we report the discovery of a double detached dust shell surrounding the N-type carbon star U Ant (IRAS10329-3918, see Table 1), which has a very large excess at 60µm and is the brightest among the stars with a detached CO gas envelope (Olofsson et al. 1990, 1993). This star has two entries at 60 and 100µm in the IRAS Small Scale Structure Catalog (Joint IRAS Science Working Group 1988). The HIRAS images are analysed on the basis of a double detached dust shell model with temperature and density gradients. Relating the two shells to two consecutive thermal pulses gives a selfconsistent determination of the distance, luminosity, interpulse period, and core mass, which leads to a new method of distance determination of AGB stars. Further development can be expected in the distance determination of the Galactic carbon stars, which is one of the major obstacles for understanding their evolutionary status.

Table 1. Observational properties of U Ant

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de