4. Discussion and results
4.1. The Red Rectangle
The Red Rectangle nebula (AFCRL 618-1343, AFGL 915) is a remarkable carbon-rich symmetrical nebulosity embedding the ninth V magnitude post-AGB star HD 44179. The photographic images in red light by give the object a rectangular shape of nearly 1 arcmin on the sky and show spikes marking an X-shape (Cohen et al. 1975). Many observations have been carried out at high angular resolution in the infrared. In 1993 December, adaptive optics images were obtained through narrow band filters centered on 1.647µm (bandwith: 18 nm) and 0.850µm (bandwidth: 75 nm) at the 3.60-m CFHT (Roddier et al. 1995). In 1991 May, J, H, K, L and M maps were obtained using the COME-ON+ experiment at the ESO 3.60-m (Rouan 1993). Speckle images were reconstructed from CFHT data (Hawaii) on 1990 December in L and M and from KPNO 3.80-m data (Kitt Peak) on 1987 November in K (Cruzalèbes et al. 1996). L images were also produced from the same CFHT speckle data set (Tessier et al. 1990). A tomographic reconstructed map was also produced at K (Leinert & Haas 1989). Earlier, the sizes of both the extended and the compact emissions were estimated from K, L and M observations (Dainty et al. 1985) and from H, K, L and M measurements (Dyck et al. 1984).
Fig. 2 shows our Lucy-deconvolved map of the Red Rectangle nebula, obtained at K (2.2µm) with a total integration time of 2000 s. The envelope is clearly non spherical, roughly rectangular. The general orientation of the nebula is of (from North to East). At a level of 0.1% (related to the dynamic range of 1000:1), its dimensions are 2.2 . As previously seen in speckle maps (Cruzalèbes et al. 1996), the overall shape of the nebula is bipolar. The X shape is clearly visible, interpreted as light scattered by material distributed on the surface of a bicone (Roddier et al. 1995). The total angle of the cone is about and its axis is aligned on the axis of the rectangular nebula ( from North to East). The North-Eastern lobe is more extended than the South-Western one, as seen in the previous IR speckle maps (Cruzalebes et al. 1996). A secondary maximum is also present at a level of intensity of 20 %, at a distance of from the central maximum and at a position angle of . Recent radiative transfer modelling of the dust shell has explained the shape of the nebula and led to the extraction of some dust parameters as well as to an estimation of the mass of the dust shell of (Lopez et al. 1997b).
4.2. VY Canis Majoris
The object VY CMa (M3I-II) is one of the strongest emitters in the infrared (Hyland et al. 1969). It has a reflection nebulous structure. Although it has been intensively observed with photometry, polarimetry, imagery and interferometry, it is still not clear today whether it is a very young star still in formation, in a pre-main-sequence stage of its evolution (Herbig 1970), or an evolved one, in a post-main-sequence star (Hyland et al. 1969). Observations carried out on 1981 January at the ESO 3.60-m telescope at La Silla from 1 to 5µm have shown that VY CMa is resolved at all wavelengths and that its diameter decreases from at 1µm to at 2.2µm and then increases to at 4.8µm (Bensammar et al. 1985).
Fig. 3 shows that the dust shell surrounding the central star is clearly resolved at K (with a total integration time of 260 s) and quasi spherical with a full width at 0.1% of the maximum of about . It also shows a structure at a distance of about in the SW direction with a position angle of . This quasi-circular clump, never seen before, has a relative intensity level of 0.4 % and an angular diameter of about . It may be an accumulation of dust possibly accreted by a non visible companion star. This hypothesis needs further observations at other wavelenghts to be confirmed.
4.3. TX Piscium
The semi-regular variable of spectral type between C5,2 and C7,2 TX Piscium is one of the carbon stars best observed at HRA. Lunar Occultation events were recorded between 1992 and 1994 at wavelengths ranging from 0.55 to 3.6µm (Richichi et al. 1995). The authors interpreted substantial departure from the simple photospheric model of circular disk as due to the presence of warm dust in an optically thin shell very close to the central star, possibly clumpy, and/or the presence of large star spots, areas of lower temperature or results of obscuration by intervening dust clumps very close to the star. Recent determination of the angular diameter has also been determined at 0.712µm (12 nm of bandwidth) and at 0.8µm (22 nm of bandwidth) at the Mark III Optical Interferometer from 1989 august to 1992 september, finding a diameter varying between 9.4 mas (millisecond of arc) and 11.1 mas (Quirrenbach et al. 1994). Clumps were also detected by CO mapping with the IRAM 30-m telescope in the (2-1) and (1-0) transitions, interpreted as evidence for erratic and non spherically symmetric outflow of matter in the envelope (Heske et al. 1989).
Fig. 4 shows the Lucy-reconstructed map of TX Piscium. The dust shell surrounding the central star is well resolved in K (840 s of total integration time), roughly circular with a diameter of about at a level of 0.1% (related to the dynamic range of 1000:1). A clump, with a size smaller than and a level of intensity of about 2%, is detected in the SW direction (position angle of ) close to the central star at about of distance. This tends to confirm the clumpy interpretation of the Lunar Occultation data (Richichi et al. 1995), even if the clumps suggested by LO were supposed to be much closer to the central star (some milliseconds of arc).
The M1.5 supergiant and visual binary Antares has a well known dust shell extending up to many stellar radii from the star. No significant amount of dust closer to was found from the visibility measurements at 11 µm with the ISI interferometer (Bester et al. 1996). At this wavelength, the authors have measured an angular diameter of 44.4 2 mas (millisecond of arc). The angular diameter of the supergiant was determined with the ESO 3.60-m telescope by means of lunar occultation (Richichi & Lisi 1990). The authors claimed a good agreement between their photospheric angular diameter of 41.3 mas at 2.43µm, and the later interferometric measurements at 0.6µm (bandwidth of 55 nm) (Bedding et al. 1994). They suspected possible deviations from the simple model of a circular, homogeneous photospheric disk. The presence of non-centro-symmetric features was established from interferometric data, leading to a "disk + 1 spot" model (Tuthill 1994). In addition, observations at various wavelengths and with various techniques revealed that Antares and its blue companion star are surrounded by a shell composed of ionized gas and dust extending about in diameter (Bloemhof et al. 1984). Ionization calculations based on ultraviolet high resolution spectroscopic measurements led to the estimation of the inner radius of the expanding circumstellar envelope around the M-star of about 8.5 (stellar radii) (van der Hucht et al. 1980).
Fig. 5 shows our diffraction-limited map of Antares in the K-band (570 s of total integration time). The circumstellar dust shell is roughly ovalized, elongated to the East (position angle of ) and lies on about (at 0.1% level). Many clumps are detected close to the star: one at a distance of with a position angle of (barely separated from the central star); another at a distance of with a position angle of ; and the last at a distance of (well resolved) with a position angle of and a relative intensity of 2%. This West-East extension is in apparent contradiction with the 10µm images of IRTF, which found a North-South extended emission (Danchi et al. 1992). The origin of that misagreement is not clear, maybe instrumental (residual smearing of the 10µm images from the chopper which had a North-South throw). Further high angular observations at higher dynamic range are needed to understand the various orientations of the extended emission.
4.5. o Ceti
Although o Ceti belongs to the first targets of early speckle interferometry (Bonneau et al. 1982), departures from circular rotational symmetry were noted only recently. Many high angular resolution observations at optical wavelengths were interpreted in terms of an ellipticity of the photosphere (Karovska et al. 1991; Haniff et al. 1992; Wilson et al. 1992; Quirrenbach et al. 1992). A very recent modelling work, based on visibility observations at 11µm with the ISI interferometer during the time period 1988-1995, proposed different non-spherical models to fit the interferometric data, one of which emphasizing inhomogeneities or clumps within the dust shell (Lopez et al. 1997a).
For this star only of our sample, we carried out multiple infrared wavelength imaging with COME-ON+. We obtained images in J at 1.25µm (Fig. 6), H at 1.65µm (Fig. 7) and K at 2.2µm (Fig. 8) with respectively 600 s, 1400 s and 600 s of total integration time. In each infrared band, the dust shell is clearly resolved. Our observations confirm that the distribution of the circumstellar matter around o Ceti is asymmetrical.
The circumstellar dust shell in J measures about (at 0.1% level). Two elongated and parallel low-level structures appear in J along an axis oriented at north to east. The brightest one (that to the south) lies at a distance of about from the central star, has an extension of about 1.0 with a level of intensity of 0.7%. The faintest one (that to the north) lies at a distance of about from the central star, is separated from the southern structure by , has an extension at 0.1% of about 0.8 , and a maximum intensity of 0.3%. These 2 elongated structures, never seen before, suggest that clumps of dust may be present there. These clumps are not in concordance with the position of the companion (a white dwarf surrounded by an accretion disk): speckle observations of 1990 showed the companion at a distance of and a position angle of (Karovska et al. 1993).
The H map shows a quasi square dust shell, at a level of 0.1%, North-South oriented. A secondary maximum is seen very close to the central star, at a distance of (the angular resolution is ), at a position angle of , with an relative intensity of about 50%. The shape of the shell can be seen as a small vertical X. The opening angle of its arms is about . One can notice that the upper-right arm (to the North-West) roughly follows the axis of the low level structures appearing in J (position angle of for this arm).
The K map confirms the X shape of the shell, its orientation and its opening angle. In this spectral band, the shell is more extended, roughly at 0.1% level. The upper-right arm is still in the direction of the J elongated structures (P.A. of ). This X shape of the shell was never suspected before. Even the best maps previously published (Haniff et al. 1992) taken at 700.7, 709.9 and 650.0 nm using non-redundant aperture masking experiment at Palomar do not show this kind of morphology. One must notice that these maps concerned the photosphere and the molecular atmosphere of Mira, although the COME-ON+ maps of Figs. 6 to 8 concern the dust shell.
As for the Red Rectangle, we are tempted to interpret this small X shape as due to the light scattered by the dust material distributed at the surface of a bicone. The presence of a close companion star tends to make this hypothesis more convincing (Morris 1987). Further high angular observations at a lower level (with a higher dynamic range) are really needed to confirm the X shape of the shell and the presence of the elongated structures appearing in J, possibly aligned on the X.
© European Southern Observatory (ESO) 1998
Online publication: September 8, 1998