Astron. Astrophys. 332, L25-L28 (1998)

3. Identification and modelling

The spectrum is characterized by a large variety of narrow and broad emission features superposed on a continuum. From earlier photometric observations (e.g. Hu et al. 1989; Malfait et al. in press), it is known that the circumstellar continuum consists of a warm and a cool component which intersect at 6 µm. Immediately apparent circumstellar emission features are the 3.29, 6.24, 7.9, 8.6 and 11.3 µm PAH bands, a series of mid-IR bands that are discussed below, and a broad shoulder around 60 µm due to crystalline water ice, besides atomic lines from , at 63 µm, and the 158 µm line. The latter may indicate the presence of a photon dominated region around the star, but may also be due to background contamination. A detailed analysis of the PAH features will be carried out when the data for a broader sample are available. An interesting remark that can already been made is that since a band at 3.45 and one at 3.52 µm is present, some hydrogenation has occured (Schutte et al. 1990).

The wealth of solid-state emission features precludes a clear definition of the circumstellar continuum. Therefore, in interpreting the spectrum, we have adopted a strategy in which the emission lines are iteratively subtracted, as is illustrated on Fig. 1 and Fig. 2. The strong emission bands at 10.2, 11.4, 16.5, 19.8, 23.8, 27.9, and 33.7 µm, and weaker ones at 10.5, 12.0, 21.7, 31.3, 36.3, and 69 µm closely match in position and strength those of crystalline forsterite () as determined from laboratory spectra (Koike et al. 1993). These labspectra are obtained for forsterite particles with a radius of 0.5 µm. Such small grains should, however, already have been removed from the HD 100546 disk due to radiation pressure and also to Poynting-Robertson drag. It is, however, also likely that collisional replenishment with small particles has occured.

Fig. 2. Short wavelength part of the spectrum

Other crystalline olivines with a different Mg/Fe ratio produce similar spectra as forsterite, but agree less well with the observations. The dust of HD 100546 clearly contains much more olivines than pyroxenes, though the 40.5 µm emission feature points to the presence of some crystalline clino-pyroxene. The best discriminant for the Mg/Fe ratio probably is the longest-wavelength emission feature, which occurs near 69 µm for pure forsterite and at 73 µm or more for mixtures with more than 10% iron (Koike et al. 1993); in the LWS spectrum of HD 100546 a weak but distinct emission feature is observed at 69 µm (Fig. 3). The strengths of the features between 11 and 30 µm are correctly described with a unique temperature of 210 5 K, but cooler particles ( 40-55 K) have to be invoked in order to account for the longer-wavelength features (Fig. 4). These temperatures have been derived by interpreting the different strengths of the forsterite features in the spectrum compared to lab-measurements (Koike et al. 1993) as due to temperature effects, using the model by Waters et al. (1988) (method similar to Bouwman et al., 1997).

Fig. 3. Part of the LWS spectrum displaying the 69 µm forsterite emission. The location in wavelength of this feature is critically dependent on the Mg/Fe ratio of the crystalline olivine dust. Moreover, the presence of this weak feature in the spectrum indicates that the crystalline forsterite is not only confined to the inner part of the dust disk. In addition to this emission peak, we also see the [OI]-line at 63 µm.

Fig. 4. Ratio between the strengths of the forsterite features in the spectrum and in the laboratory measurement (Koike et al. 1993) (expressed in arbitrary units), modelled with two components with temperatures of 210 K and 40-55 K.

After subtraction of the crystalline forsterite features from the SWS spectrum, a broad 10 µm band persists, that is partly affected by PAH emission, but mostly due to amorphous silicate. As was anticipated from a study of the IRAS LRS spectrum (Grady et al. 1997), this feature is best matched with olivines rather than pyroxenes, with a large Mg/Fe ratio. After accounting for this amorphous silicate component and the accompanying 18 µm feature, broad emission bands persist around 23 and 60 µm. The former can be reproduced successfully with the optical constants of (Henning et al. 1995), the latter by those of crystalline water ice. The water ice band at 43 µm is much weaker than the one at 60 µm, from which it follows that the water ice is located in the outer parts of the disk, with a temperature below 50 K, arguably as a coating on a crystalline silicate core (Omont et al. 1990). Barlow (1997) noted that the 60 µm -ice feature in NGC 6302 is accompanied by another broad feature between 87 and 98 µm; as can be seen on Fig. 1, this feature is also apparent in the LWS spectrum of HD 100546.

Once all these solid-state features are removed (but not before) it is possible to represent the underlying hot and cold continua with a model. Applying the optically-thin model developed by Waters et al. (1988), while adopting the usual emissivity law , a best fit for the cold continuum is found for a dust surface density law dropping as and for inner and outer disk temperatures of 210 10 and 43 2 K, respectively. Fitting the hot continuum requires an emissivity law and a steeper density drop, proportional to , yielding inner and outer temperatures of 1550 50 and 350 20 K, respectively.

More radiative modelling will be needed in order to place firm constraints on the amount of different dust components, though we can already say that the fraction of crystalline over amorphous silicates will at most be 0.1, since a small degree of crystallisation changes the optical properties drastically (see eg. laboratory measurements by Hallenbeck et al. (1997))

© European Southern Observatory (ESO) 1998

Online publication: March 23, 1998
helpdesk@link.springer.de