In a study of the chromospheric Mg II h & k emission of the bright N-type carbon star TX Piscium with the IUE we found that the Mg II lines were severely blended by resonance lines of Mn I and Fe I in absorption (Eriksson et al. 1986). Simple estimates of the column densities needed to produce this absorption made us search for a circumstellar envelope around the star at CO mm wavelengths. This search led to a positive identification, and later we found that the vast majority of the apparently brightest N-type stars in the near infrared show circumstellar CO emission (Olofsson et al. 1988, 1993). We also found that about 5% of these CO line profiles show signatures that indicate envelopes that are markedly detached from the stellar surface, and we later verified this interpretation by detailed maps of four stars, which showed that the CO line emitting gas was distributed in the form of large, and geometrically thin, shells centred on the stars (Olofsson et al. 1992, 1996). One interpretation of this fact that agreed with physical and statistical data was that the detached shells are produced in connection with helium burning that occurs in short pulses, flashes, on the later part of the evolution along the asymptotic giant branch (cf. Olofsson et al. 1990). The maps of the sources with detached shells indicated that the shells are almost entirely empty, at least of CO, so that the mass loss was suggested to be switched off, after a probable rapid increase, as a result of the shell flash. We also found, from the detailed maps, that the CO emission indicated a clumpy circumstellar medium, with each shell consisting of apparently several tens of CO clumps (Olofsson et al. 1992, 1996; Bergman et al. 1993). Since the poor spatial resolution of single radio telescopes limits the map resolution to typically we decided to search for the shells also at visual and near infrared wavelengths, i.e. as scattered photospheric emission in the Na D and K I 769.9 nm resonance lines. If detections could be made, these data could be expected to be complementary to the CO data since they sample the interior parts of the shell, in view of the decline of the illuminating radiative flux with the distance from the stellar surface.
The envelopes around several red giants ( Ori, Bernat & Lambert, 1975, 1976, Bernat et al. 1978, Honeycutt et al. 1980, Mauron & Caux 1992, and references therein; µ Cep, Mauron et al. 1986, Mauron & Querci 1990; R Leo, Lambert & van den Bout 1978; Her and o Cet, Mauron & Caux 1992, g Her and the miras R Aql, R Leo and V Hya, Plez & Lambert 1994; Per, Peg and CE Tau, Mauron & Guilain 1995, and the miras W Hya and R Hya, Guilain & Mauron 1996) have been traced and studied in these lines; for some stars in considerable detail with important scientific results. V Hya is the only carbon star envelope that has been investigated with this method, except the carbon-rich planetary nebula BD , from which Dinerstein & Sneden 1988 reported Na I emission. Guilain & Mauron (1996) report lower limits on the K I emission from the carbon stars TX Psc, Y CVn and U Hya.
We made the first discovery for R Scl (reported in Edvardsson 1990) with the CAT-CES at ESO, both in the Na D and the K I 769.9 nm line. These studies continued for this star and for others with the same equipment at several subsequent observing runs. With the Nordic Optical Telescope, and with the ESO 2.2 m, we also tried direct imaging with coronographic methods, starting in 1991, however, with little success. The problems clearly are due to the faintness of the envelope which makes the scattered light from the star in the atmosphere or the telescope very problematic. However, in 1994 we succeeded to image the envelopes of several stars with the ESO 3.6 m telescope.
Subsequently we present the spectroscopic observations of the K I line in Sect. 2, and discuss the results in Sect. 3. The results of the Na I and imaging observations are discussed in later papers.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998