Astron. Astrophys. 363, 1115-1122 (2000)

4. Discussion

4.1. Synthesized samples

Our synthesized samples are pure and large high-quality single crystals. Compared to our samples, the synthesized enstatite sample by Jäger et al. (1998) is micro crystalline; they formed melt at 1700 , and cooled it to room temperature at  . At such high cooling rate, small and highly imperfect crystals of clinoenstatite, which were transformed from the high-temperature phase (protoenstatite), should be formed. Moreover, also forsterite () and silica () were formed as a result of incomplete peripheral reaction between forsterite and melt. In contrast, we synthesized single crystals of orthoenstatite, which is the stable phase of at room temperature and 1 atm., with the flux method (Ozima 1982). If enstatite is synthesized at temperature higher than 985 , protoenstatite is formed, and this transforms to clinoenstatite having polysynthetic twinning as a metastable phase by rapid cooling (e.g., Heubner 1980), and orthoenstatite cannot be obtained. This is the reason why we used the flux method at low temperature below 985  in stead of using the CZ or FZ (floating-zone) method, which can grow larger and more perfect single crystals from their own melts without incorporating impurities (of flux). We applied very slow cooling rates (as slow as 0.2 K/hr) to obtain large and clear single crystals (up to 2 cm in length).

4.2. Spectra of samples

In general, there is a shift of the peak wavelengths of about 0.2-0.3 , when spherical particles are in a medium instead of vacuum (Schmidt 1981). The shape of the crystalline silicates such as pyroxenes and olivines were observed under an SEM. They all show irregularities and clearly differ from spheres. The particles of our pyroxenes and forsterites are dispersed on bulk materials (KBr, PE, NaCl, Ge, Si, etc) after which the spectra are measured in mid and far infrared region. The peak positions are nearly the same in a medium as in the air although the peak strengths are a little changed due to irregularity in shapes. In the case of an irregular shape, the medium effects on the peak positions are minor. The same result was found by Colangeli et al. (1995).

As a next step, calculated spectra of spherical particles are compared to the calculated spectra of particles with a continuous distribution of ellipsoids (CDE; Bohren & Huffman 1983) using our derived optical constants of bulk samples (En, Fo100, Fo90, and Fa) (Sogawa & Suto et al. in preparation). Clearly these spectra are different. The calculated spectra of CDE in a medium are very similar to those in vacuum, but the strength of the peaks are higher by (medium index) times the vacuum values. The calculated spectra of CDE in medium are similar to the measured spectra of fine particles of En, Fo100 and Fo90 (Sogawa et al. 1999; Jäger et al. 1998). Therefore, it is unnecessary to consider the shift of the peak positions measured in medium and in vacuum.

The results of the enstatite samples are similar to those of Jäger et al. (1998) except for some peaks in the far-infrared region. In spite of the low resolution, our measured spectra of the orthoenstatite clearly show the two strong peaks at about 70 . The existence of these sharp and clear features are confirmed for the first time in the far infrared region. The two peaks also clearly appear in the calculated CDE spectrum of natural enstatite reported by Jäger et al. (1998). The discrepancy may be due to the difference in the dispersion of the particles in PE sheets and the difference in quality of crystals.

In contrast to the spectra of the amorphous pyroxenes in this paper with two broad bands at about 10-10.4  and 19.2-21.4 , the spectra of the glassy bronzite, pyroxene glass, and amorphous enstatite showed two broad bands at 9.5 and 18.5  (Dorschner et al. 1988), 9.5 and 18.8  (Jäger et al. 1994), and 9.89 and 19.21  (Brucato et al. 1999), respectively. The difference in the spectra between our amorphous pyroxenes and their glass might be due to the procedure for preparation, that is, the peak positions are strongly influenced by the quenching method (Jäger et al. 1994), or difference of the chemical compositions (the glasses contain Fe elements in the previous study, but not in this study).

4.3. Comparion with observed spectra of the 7  band

In oxygen-rich circumstellar shells, the 7  band was observed to be correlated with dust, and this carrier remains to be identified (Goebel et al. 1994). In this work, the 7  band is detected in the spectra of the crystalline pyroxenes (the clinoenstatite, diopside, natural orthopyroxene from Bambel and Ichinome-gata, and augite), the diopside glass and the enstatite gel, but only marginally in the enstatite glass. It is not clear if this band is due to some impurity (possible contamination during preparation for the measurements of spectra). The 7  band appears only as a hump in the clinoenstatite (crystalline silicates) which does not contain Ca. On the other hand, as for the diopside glass, the 7  band becomes stronger after hydration. In this case, this 7  band may be due to Ca contained materials, such as carbonates glass in the diopside glass (Knacke & Krätschmer 1980).

4.4. Comparion with observed spectra NGC 6302

In connection with Ca-containing mineral, the spectra of diopside are compared with the ISO observation of the planetary nebula NGC 6302 (Lim et al. 2000). The spectrum of NGC 6302 is characterized by very strong and narrow forbidden emission lines and many relatively sharp features due to crystalline silicates (see Fig. 3). The 43  feature and the strong and very broad emission feature peaking at about 65  are proposed to be identified with crystalline ice of 45 K (Barlow 1998; Lim et al. 2000). The observed peak position of 65  is however at a little longer wavelength than those of laboratory measurements of ice (Smith et al. 1994).

Fig. 3. Comparison of the continuum subtracted spectrum of NGC 6302 with the spectrum of diopside multiplied by a Planck curve of 50 K (arbitrary scale). The very narrow and strong features are forbidden emission lines.

The strong and broad peak of 65  of NGC 6302, is similar to the spectrum of crystalline diopside multiplied by a 50 K Planck function. However, the peak position is not exactly the same and has a difference of about 1  from the laboratory spectrum of diopside. Our recent measurements of spectra at liquid He temperature show that the peak position shifts about 1  to the shorter wavelengths compared to the peak position of 65.7 , measured at room temperature (Chihara et al. 2000). Further, the spike at about 68.93  of NGC 6302 might be due to forsterite (Barlow 1998; Lim et al. 2000). This is strongly supported by our recent measurements that forsterite has a sharper and stronger peak at liquid He temperature and that the peak positions in far infrared region shift to shorter wavelengths by about 1-0.3  compared the spectra at room temperature, i.e. the feature at 69.6  shifts to 68.8 . When crystalline pyroxenes and forsterite are cooled down to liquid He temperature, the spectra show sharper and stronger peaks, which also shift to shorter wavelengths compared to those at room temperature (Chihara et al. 2000). This is similar to the result of Mennella et al. (1998). Looking into the another peaks in Fig. 3, the observed peaks at 30, 34, 40, and 45  are also very similar to the peaks of the diopside at room temperature.

The chemical abundances of NGC 6302 show that Mg, Al, Ca, and Fe are depleted in gas. In particularly, Ca is a factor of 100 less abundant with respect to the Sun and B stars. These elements may be in the form of dust (Pottasch & Beintema 1999). Considering the depletion of the elements in the gas, the presence of Ca-rich pyroxene might be possible. Most peaks of NGC 6302, can be identified with diopside, crystalline water ice and forsterite. The broadness of the observed band at 65  indicates a blend of diopside and crystalline water ice.

Many new emission peaks at wavelengths between 20 and 45  have been detected in dust shells around evolved oxygen-rich stars and young stars (ISO results) (Molster 2000). Crystalline silicates such as olivines and pyroxenes are attributed to most of these emission peaks. These identifications are reasonable based on the present measurements. For example, the detected peak at 40.5  in HD 100546 (Malfait et al. 1998), He 2-113 and BD+30 3639 (Waters et al. 1998) is due to crystalline pyroxene, and this band commonly appears in the spectra of synthetic enstatite and Ca-rich pyroxene (diopside and augite). Crystalline water ice is tentatively identified with the 43  hump (Waters et al. 1996; Molster et al. 1999a). Furthermore, by adding these identifications, we indicate pyroxenes as another promising candidate for the carriers of humps at about 43-44 . Pyroxenes (ortho-enstatite, orthopyroxene from Bambel, and orthopyroxene from Ichinome-gata) have a strong band at 43-44 . Still more, the detection of the double peaks at about 50 and about 70  may confirm the identification of pyroxene group same as forsterite detected at about 69  in the young star HD 100546 (Malfait et al. 1998) and NGC6302 (Lim et al. 2000).

The spectra of diopside and enstatite glasses have a very broad band at around 10.0-10.3  with a half-width of about 3 . These spectra are similar to the spectrum of Elias 1, but show a peak at slightly shorter wavelength than the spectrum of Elias 1; the spectrum shows peak at about 10.6  with half-width of about 4.6 , which might be due to large particles (Hanner et al. 1994a).

In this paper, we investigate the spectra of only Fe-poor Mg-Fe-Ca pyroxenes. We will further investigate the spectra of Fe-rich pyroxene, and make clear the correlation between the band and chemical composition in the near future.

© European Southern Observatory (ESO) 2000

Online publication: December 5, 2000
helpdesk@link.springer.de