Astron. Astrophys. 339, 904-916 (1998)

5. Discussion

We compared our spectroscopic results with the astronomical data of the source AFGL 4106. AFGL 4106 is an evolved binary (Molster et al. 1998), which experienced a huge mass loss. Former IRAS-LRS measurements have shown that a large amount of amorphous silicates is present in the circumstellar shell. On 22 July 1996, ISO-SWS observed this object in an AOT01 speed 3 observing mode. The spectrum revealed the existence of a significant amount of crystalline silicates. For comparison with our laboratory data we subtracted a kind of continuum from the ISO-SWS spectrum. The result of this procedure is plotted in Fig. 8. In the same figure we show normalized MAC multiplied by a 100 K Planck function. The dust temperature of 100 K is a very rough estimate of the mean dust temperature of the object. The relative strength of the bands in the spectrum is sensitive to the temperature of the dust, which can be easily seen by a comparison with the spectra in Figs. 2 and 3.

Fig. 8. Comparison of the continuum subtracted spectrum of AFGL 4106 with the MAC of forsterite and clinoenstatite multiplied by a Planck curve of 100 K and normalized to one.

Although there is a good agreement between the continuum-subtracted spectrum of the source AFGL 4106 and the laboratory data of forsterite, not every peak in the spectrum can be matched. Quite a number of the remaining peaks can be fitted with synthetic enstatite (see Fig. 8). We found that only the very magnesium-rich crystalline silicates can explain the spectrum of AFGL 4106. The 69.7 µm feature, which is present in some other sources is unfortunately too weak to be detected in this source. This band is a suitable indicator for the presence of very Mg-rich crystalline olivines (see Sect. 3).

Also in our own planetary system there is ample evidence for the existence of crystalline silicates like in the spectrum of comet Hale-Bopp (Crovisier et al. 1997). A detailed comparison of the emission features between 6 and 45 µm with the silicate spectra presented in this paper supports the presence of a magnesium-rich olivine (forsterite) additional to an amorphous silicate component.

Laboratory IR transmission spectra of individual dust particles (Sandford & Walker 1985) collected in the stratosphere also show the occurrence of a wide variability of crystalline olivines and pyroxenes. However there is no indication for a larger abundance of the pure Mg silicate materials like forsterite or enstatite.

Still there are some differences between the ISO-SWS spectrum and the laboratory measurements, in the width, strength, and position of some of the features. Differences in width can be caused by differences in size and shapes of the grains. Despite our efforts to produce small grains, coagulation could have taken place during the production of the pellets used for the transmission measurements. It is quite unrealistic to assume that the grinding process produces the same grain shapes as are present in circumstellar outflows. Differences in strength can be partially explained by the fact that there is a temperature distribution instead of one single temperature. In addition, we have to keep in mind that the ISO-SWS spectrum shown here consists of five independent spectra of different wavelengths regions. Although we took pains to attach the different spectra correctly to each other, it still remains a source of uncertainty in the occurrence and strength of some features, especially those situated near the edges of the spectral regions.

Discrepancies in the peak positions are not easy to explain by matrix effects of KBr or PE. Calculations of the absorption behaviour of spheres in the Rayleigh limit, with the optical data of olivine determined by Steyer (1974), in a KBr matrix and vacuum, have shown that there are shifts of different band centres ranging from 0 to 0.6 µm to longer wavelengths for the sample embedded in KBr/PE. Additionally, we found differences in shapes and peak strengths. However, the formation of exactly spherical grains in the circumstellar outflow is not expected. An approach more realistic than the spheres would be the calculation of the absorption efficiency of a continuous distribution of ellipsoids (CDE) (Bohren & Huffman 1983). We show the result of these calculations in Fig. 9. It demonstrates that there is no change in the peak positions for the different matrices, but the intensity of the peaks in KBr is about 1.4 times higher than in vacuum over the whole wavelength range. Since only very limited information about shape and size distributions of interstellar silicate grains is available it is very difficult to quantify matrix and shape effects in order to compare laboratory and astronomical spectra in an appropriate manner.

Fig. 9. Comparison of the calculated absorption efficiency of a CDE in KBr and in vacuum from the optical constants of olivine (Steyer 1974).

Mindful of all considerations mentioned before, we want to point out that there are some bands that are unexplainably strong, like those at 20.6, 21.5, 26.1 and the plateau from 30 to 45 µm. It is also likely that other materials are present in the circumstellar shell, for example Fe-containing oxides, Ca-containing silicates, sulfates and H₂O ice.

The details of silicate dust formation in oxygen-rich shells are still unknown. First ideas were based on thermodynamical equilibrium considerations of non-homogenous nucleation postulating condensation sequences (Tielens et al. 1990). More recent investigations found a lot of indications for incomplete condensation and dust formation as a non-equilibrium process. One indication is for example the detection of SiO molecules in the outer regions of circumstellar shells. Based on laboratory experiments, Nuth (1996) described grain formation as a complex process starting with the nucleation of SiO and formation of SiO_x clusters. Other underoxidized species like AlO_x, FeO_x and MgO and metallic components are also formed, which can react in internal oxidation/reduction reactions forming an underoxidized low-density silicate material, a so-called "chaotic" silicate. Three phases, forsterite, enstatite and SiO₂ were also found in experimentally produced Mg-SiO smokes (Rietmeijer et al. 1986) as one form of chaotic silicate material. This demonstrates that the same phase separations could take place in the chaotic silicates described by Nuth & Hecht (1990) and in inhomogeneous samples produced by cooling of molten material (discussed in this paper). It is interesting to see that different starting materials provide the same products and phase separations.

One remaining question is how the crystalline material is formed in circumstellar shells and what we can say about chemical processing and aging of silicate dust during the transition from the circumstellar environment to the interstellar medium and the star-forming regions. Crystalline dust features are prominent in the ISO SWS spectra of stars on the Asymptotic Giant Branch (AGB) and those of Red Supergiants, if the color temperature of the dust is low (typically less than 200-300 K). A low color temperature for stars with dusty outflows corresponds to high dust optical depth, and thus with high mass loss rates. There are indications that the crystalline dust features are more prominent in post-AGB stars and planetary nebulae (PNe), suggesting perhaps an increase in the abundance of crystalline silicates. The dust shell in post-AGB stars and PNe, which is the remnant from the high mass loss phase on the AGB, expands and cools to temperatures below 100 K.

Hallenbeck et al. (1998) have observed that annealing of amorphous magnesium silicate smoke at temperatures slightly above 1000 K for more than 10 h causes remarkable changes of the profile of the 10 µm band, which indicate internal ordering processes of the arrangement of atoms leading to crystalline structures. According to this results the requirement for the crystallization process is a sufficient high temperature ( 900 K) and/or a corresponding annealing time. Recently, Gail and Sedlmayer (1998) have shown that corresponding conditions for annealing processes are given in M giants. Clearly, many questions concerning the origin and evolution of crystalline dust in the outflows of evolved stars remain open.

A related issue is the apparent absence of crystalline silicates in the interstellar medium (ISM). This suggests that crystalline silicates are not produced with sufficient abundance in the outflows of evolved stars to be detectable, or they are destroyed in the ISM. The high abundance of Mg-rich crystalline silicates in some young stars (Malfait et al. 1998) suggests that such grains are produced during the star formation and planet formation process, from annealing of amorphous grains.

In this paper we have demonstrated that crystalline Mg-rich silicates are responsible for the observed sharp peaks at longer wavelengths in the oxygen-rich circumstellar environments seen in the ISO spectra. We consider this as one of the most important findings of the ISO mission.

© European Southern Observatory (ESO) 1998

Online publication: October 22, 1998
helpdesk.link@springer.de