Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 350, L27-L30 (1999)

Previous Section Next Section Title Page Table of Contents

3. The formation of hot dust

The ISOCAM data clearly confirmed the existence of a strong - and steadily increasing - IR excess, see Fig. 1. We conclude that this is the signature of very hot dust that has recently formed in an extended, optically thin shell around the star. Furthermore there is indication of change in the spectral energy distribution (SED), getting redder over time. In addition, the lack of an excess in the ISO filter centered on the 7.7 µm PAH feature indicates that the dust is formed in an H-poor environment. One conclusion is that the formation of dust is continuing but at variable rates or in an episodic manner.

[FIGURE] Fig. 1. ISOCAM photometry of Sakurai's object at different epochs. Diamonds are from Feb 1997, squares from Sept and Oct 1997 and triangles from Feb 1998. UBVRi and JHK data are by us and from the literature (see text).

For two reasons the dust observed in Sakurai's object must have formed most recently: First Sakurai's object has not been detected as an IRAS source in the 80s (compare with Table 1), second the location of the dust excludes an age older than the He-flash because dust from the AGB is much further out and does not remain intact so close to a PN central star.

Moreover we recorded an unprecedented increase in the IR flux by a factor of ten over one year. This of course is an indication of continuous or episodic mass loss.

Fig. 2 shows a fit to ISO and UBVRi data for February 1997 and February 1998 using a spherically symmetric dust radiative transfer model (Groenewegen 1993). The observational data are the ISO fluxes discussed above, and UBVRi photometry from the literature. For Feb. 1997 they are interpolated from values listed in Duerbeck et al. (1997), for Feb. 1998 they are from Liller et al. (1998). The near-IR data shown in Fig. 1 is our data taken with the 1.5m Carlos Sanchez telescope (Tenerife). We adopt a visual foreground extinction of 2.2 mag (Pollacco 1999).

[FIGURE] Fig. 2. ISOCAM photometry of Sakurai's object with a fit to the SED, February 1997 (solid line) and February 1998 (dashed line).

For the central star we use a state of the art H-deficient model atmosphere of a temperature appropriate for the respective epoch (see Asplund et al. 1997, 1999 for details), specifically we have used [FORMULA] = 6000 K for Feb. 1997, [FORMULA] = 5500 K for Feb. 1998. For the dust we used amorphous carbon grains (optical constants from Rouleau & Martin (1991) for the "AC" species). Since the laboratory species may not be fully representative of astronomical dust, the opacity used may not be appropriate.

The fit to the Feb. 1997 data assumes a dust condensation temperature of 1500 K. A good fit could only be achieved using a finite value for the outer radius of the dust shell. This is not surprising as the dust shell has been expanding for a very limited time. Eyres et al. (1998) report a mean temperature of 680 K (Apr. 97) for the dust using a Planck fit to SWS data. For a distance of 1.5 kpc (Kimeswenger & Kerber 1998; Kerber et al. 1999a), we derive a luminosity of 650 L[FORMULA] and a mass loss rate of 2.3 [FORMULA] M[FORMULA]/yr. For a distance of 5.5 kpc, these numbers are 8750 L[FORMULA] and 8.4 [FORMULA] M[FORMULA]/yr, respectively. These numbers assume a gas-to-dust ratio of 200 and an expansion velocity typical for AGB stars of 10 km s-1. If the expansion velocity is larger (see below) the mass loss rate is larger by the same factor. The inner shell radius is 17.2 stellar radii (which follows from the effective temperature of the star and the dust condensation temperature), and the outer radius is 25 times the inner dust radius. The latter number is accurate to within a factor of 1.5. For a distance of 1.5 (5.5) kpc this implies that the dust has traveled about 7.2 (26.2) [FORMULA] cm from the star since the Helium flash occurred, assuming radial expansion. Normalized to a time-scale since the flash of two years, this implies a typical expansion velocity of 110 (420) km s-1 to within a factor of 1.5. This indicates an expansion velocity much higher than in AGB stars, but is quite similar to the velocity found for the ejecta in A 58 (Pollacco et al. 1992).

The fit to the Feb. 1998 data is less good, and required some fine-tuning. In the model shown the dust at the inner radius has a temperature of 875 K, and the (assumed constant) mass loss rate is 9.0 [FORMULA] M[FORMULA]/yr (3.3 [FORMULA] M[FORMULA]/yr for a distance of 5.5 kpc). There was no need to change the luminosity. Since we actually show that the mass loss rate has increased by a factor of 40 over a time-scale of a year, more sophisticated modeling should include a time-dependent mass loss rate. This is beyond the scope of the present paper. The fit to the ISO data can be improved by adopting a higher condensation temperature. The fit in the optical becomes much worse then, but one could then invoke non-spherical mass-loss, as seen in some of the other late He-flash objects (see next section). Mid-IR observations by Käufl & Stecklum (1998) from June 98 using TIMMI show more change indicating that the dust formation is time dependent.

Combining our near-IR observations and those reported by Kamath & Ashok (1999) a period of relative stability, lasting from late March to late May 97 seems to exist; that is exactly between two ISO epochs (Feb 97 and Sept 97) which show a strong increase in the mid-IR. It would be tempting to speculate that the rate of mass loss varies on rather short time scales of weeks or few months, but comparison of the two fits in Fig. 2 clearly show that the near-IR is the region of least change. Apparently the effects of dust formation and cooling counteract, making conclusions based on the near-IR alone very difficult.

The observed dimmings in the visual, - as the dust became optically thick - are therefore the direct consequence of this mass loss, see Fig. 3 for the light curve provided by the AAVSO (Mattei 1999). The observed light curve and physical mechanism resemble those of R CrB stars (Clayton 1996), which produce clouds of dust at irregular intervals, which only become (visually) apparent when blocking our line of sight. The lack of change in the visual between epochs 1 and 2, could then be interpreted as an indication of non-spherical mass loss as seen in R CrB stars.

[FIGURE] Fig. 3. Visual light curve of Sakurai's object (provided by the AAVSO). The arrows denote limits, while the solid diamonds represent the epochs of the four ISOCAM observations.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: October 4, 1999