4.1. An IR companion to HD 97300?
As noted in Sect. 3, we detect a secondary peak about (240 AU) north of the optical star HD 97300. This peak of emission is visible in the lw8 (10.7-12.0 µm) and lw9 (14.0-16.0 µm) filters, but not at shorter wavelengths. On the contrary, the peak of emission at the position of HD 97300 is clearly seen in the lw4, lw5 and lw8 filters, but not in lw9. Both components are clearly seen in the lw8 image. Contamination from extended emission does not allow us to determine reliable fluxes for the two components. In the CVF images, which have poorer spatial resolution than the filter ones, the two components are not separated.
The situation is reminiscent of the case of LkH198, where an embedded companion was found near the optical star by Lagage et al. (1993) in their ground-based 10 µm images of the region. Note that in LkH198 most of the far-infrared emission is due to optically thin dust heated by the visible star (Butner & Natta 1995). This analogy, and the dominance of the IEBs in the region, suggest to us that also in the case of HD 97300 the embedded companion does not play a major role in the energetics of the system.
4.2. The IEBs carriers
To acquire a better understanding of the nature of the IEBs, we have fitted the spectra of Fig. 2 using the dust model of Siebenmorgen & Krügel (1992). This model computes the emission per unit mass of dust heated by radiation of known intensity and spectral distribution. The dust consists of a mixture of large grains, very small grains and organic molecules. We assume here that the molecules are polycyclic aromatic hydrocarbons (PAHs), though other band carriers have been suggested (e.g. Sakata et al. 1984, Duley 1989, Papoular et al. 1989).
Only large grains are in thermal equilibrium with the incident radiation field. PAHs and small grains show strong temperature fluctuations which are treated following Siebenmorgen et al. (1992). The radiation field is that of a star of effective temperature =10700 K and luminosity =35 . We assume that the projected separation between the star and the position at which the spectrum has been measured equals the physical distance, and neglect any intervening absorption. The computed spectra are smoothed to the CVF spectral resolution.
In our model, the IEBs are due to PAHs of fixed size and structure. In reality, we expect a mixture of PAHs of different size and structure. Their average composition will depend on the environment and is thus likely to change from place to place. Nevertheless, we consider for simplicity only two types of PAHs, one formed by 30 carbon atoms and one by 300. Both kinds contain 3% of the amount of carbon in large grains. In our dust model we assume a carbon abundance in solids relative to hydrogen atoms of 3 10-4. The absorption cross sections of PAHs have been calculated following Schutte et al. (1993). For each IR resonance, we specify the center wavelength , the integrated cross section , and the shape of the line. For the latter we assume a Lorentzian profile with a damping constant . These numbers are listed in Table 1 and kept fixed in the calculations. For comparison, we also give the resonance cross sections of Schutte at al. (1993).
Table 1. PAH Properties
For features due to C-H bonds, we additionally need the hydrogenation parameter =[H]/[C], i.e. the ratio of hydrogen to carbon atoms in the PAH. This parameter is allowed to vary in the nebula, but is assumed to be equal for both small and large PAHs. The resonances at 11.3 and 12.5 µm are due to out-of-plane C-H bending; that at 11.3 µm is caused by isolated H atoms, while the 12.5 µm feature is caused by two adjacent H atoms. We assume that a fraction of all out-of-plane C-H bending modes are due to isolated H atoms, and the rest (1-f) to two adjacent ones. Also may vary within the nebula.
For each position, we adjust , and the gas column density until a satisfactory fit is obtained. The results are shown as solid lines in Fig. 2. Table 2 gives the position in Column 1, the corresponding values of the projected distance from HD 97300 in Column 2, the gas column density () in Column 3, in Column 4 and in Column 5. The hydrogenation parameter is higher at larger distance from the star, while decreases. This is expected (Schutte et al., 1990) since the radiation field intensity decreases as the distance from the star increases.
Table 2. Model Parameters
The agreement in Fig. 2 between model and data is generally good, considering the observational uncertainties. There are a few discrepancies, as in position 1, where the predicted intensity at 8.6µm is too strong, or in position 4, where it is too weak. Also the flux between the features is sometimes underestimated, but the peak intensities and band shapes of the resonances themselves are fitted quite well. At some places, one has a definite impression that there is a feature which is not included in the model, for example, at position 6 at 7µm.
One important characteristic of these models is the fact that, in order to obtain a good fit of the observations, we find it necessary to postulate that the features have extended wings. We model the absorption profile of each feature in a purely descriptive manner. The simplest picture we can construct is to consider that the bands can be described by classical oscillators. The absorption coefficient () of such a driven damped oscillator is a Lorentzian profile
where . The integrated cross section and the damping constant are given in Table 1.
We are not in a position to give a physical explanation for the damping constant and restrict, therefore, their discussion to a comment. Cum granu salis , when in an atom an excited level has an average lifetime , where A is the Einstein coefficient for spontaneous transition, the probability to find the atom there decays like . Because of Heisenberg's uncertainty principle , the energy of the upper level is then only defined to an accuracy . This leads naturally to a Lorentzian emission profile, and A may be identified with the damping constant, .
In this picture, the line width is determined by the timescale . For PAH resonances with a width of m, the characteristic time would be 10-12 s. This would imply immense values for A (1012 s-1) as well as for the associated dipole moment Debye because . A way out of the dilemma would be to assume that the IEBs arise from a superposition of many narrow lines.
Allamandola et al. (1989) have interpreted s as follows: After the absorption of a UV photon and once the PAH has arrived at the final electronic state from which it will emit the IR features, it takes about 10-12 s to statistically distribute the energy of the UV photon (or what is left of it) among the various vibrational levels (mostly of low vibrational quantum number, ). This shuffling among the population of the levels continues while the IR photons are being emitted, so that the mean lifetime of a vibrational level is also only 10-12 s, which explains the observed line width.
4.3. Energetics and continuum emission
Our models include not only PAHs but also large grains and very small particles of carbon and silicates. We assume that the grains have a power-law size distribution () with Å, for the large grains and 10Å, for the small particles. Small grains are 10% in mass relatively to large grains. In Fig. 3 we show the contribution of the individual dust population to the total IR emission. The spectrum is calculated for the ring (position 7). The relative contribution of the different dust species to the emitted spectrum does not depend significantly on the distance from HD 97300.
Large grains absorb 75% of the stellar radiation and emit it predominantly in the far IR. They are responsible for the IR excess observed at longer wavelengths. Given the moderate luminosity of the star the large grains are too cold to contribute to the emission observed in the spectral range of our CVF scan.
Very small grains absorb 8% of the total energy. As for PAHs, the temperature of the very small grains fluctuates violently, but the absence of any resonance smoothes out their emission over a large wavelength interval. The small grain emission peaks at about 50-60µm, and is unimportant in the range of wavelengths we have observed. However, it should be noted that the properties of very small grains are very uncertain. Their absorption cross section, computed assuming that the grains are spheres and applying the Mie theory for continuous media, is very likely underestimated. There is an enormous difference in the absorption properties between a PAH made of 300 C atoms and a very small carbon particle of the same size.
PAHs absorb about 17% of the stellar luminosity (4% is absorbed by PAH molecules and 13% by PAH clusters). They re-radiate it in the IEBs, which account for most of the emission in the CVF wavelength range.
The CVF spectra show evidence of continuum emission at wavelengths around 7µm and in the interval 9-11µm. Continuum emission is also seen in the interval 14-16 µm. Fig. 4 shows for the four narrow band filters the observed intensity as function of the distance from the star along the direction indicated in Fig. 1. The morphology of the region is very similar at all wavelengths.
The morphological similarity between the emission in the IEBs and in the continuum suggests that both are due to the same carriers, and that UV photon excitation must be the main excitation mechanism. Similar conclusions have been reached by Cesarsky et al. (1996b,c), Boulanger et al. (1996) and Uchida et al. (1997) in their analysis of the CAM spectra of a number of reflection nebulae. Our models include continuum emission from PAHs only at µm (see Fig. 3). Together with emission by large grains, they can roughly account for the continumm emission observed in the lw9 filter, but not for the continuum at shorter wavelengths, even when we allowed the size range of the very small grains to vary (see Natta et al. 1993 for a discussion of the dependence of very small grain emission on grain parameters). We suspect that the adopted continuum cross sections of transiently heated species are incorrect in the infrared. In fact, these and other ISO results could be used to better constrain the continuum cross section of these species.
4.4. Mass of circumstellar material
An interesting possibility opened up by our observations is to derive the mass of circumstellar material from the observed intensity of the PAH features.
We know the geometry of the emitting region, i.e., the approximate distance of the grains from the star. For a given distance and stellar luminosity, the integrated 6 to 14µm flux is directly proportional to the number of C atoms in PAHs. This is because the PAHs account for the total emission in this spectral region (Fig. 3). The fraction of C atoms in PAHs is known to within a factor of three, so we can directly convert the carbon column density derived by fitting the feature intensities into a hydrogen column density . The uncertainty on is probably comparable to the uncertainty that affects its determination from sub-millimeter continuum observations (see, for example, Krügel & Siebenmorgen, 1994). The mass of gas and dust can then be computed from the values of averaged over the region of interest.
We derive a total mass of the circumstellar material in a region of about 0.03 pc radius of about 0.07 of which 0.03 is in the elongated ring structure and only 0.003 in the central region of 1600 AU radius. Our estimates agree with the upper limit in a region of size (2000 AU radius) derived from the 1.3mm flux ( mJy; Henning et al. 1993).
4.5. The origin of the ring
The ring structure is clearly seen in the enhancement of the column density in position 7 (see Table 2). It has a projected size of about 0.0450.03 pc , a thickness of about 0.01 pc and a mass of 0.03 . Its density is of the order of cm-3.
The presence of PAHs in the ring indicates that it is made of interstellar matter, rather than of matter ejected from the star. Its morphology suggests that the ring structure has been created by the interaction of the star with the surrounding matter.
It is possible that the ring results from the interaction of a stellar wind with the environment. In this hypothesis, the material in the ring is swept-up gas. The ring coincides with the inner wall of a three-dimensional cavity created by the wind. If we take a ring radius 0.02 pc and an average ambient density cm-3 (as derived from the column densities in position 8; see Table 2), the swept-up mass is 0.015, comparable to the estimated ring mass. Since there is no evidence of a stellar wind at the present time, the ring formation must be due to an episode of mass-loss, which has died out very recently. Assuming that the ring is expanding at 5 km s-1 (there is no dynamical information available), we estimate the momentum in the ring to be gr cm s-1. If we take as a typical timescale the dynamical age of the ring expansion (4000 yr), we find that the ring momentum could have been provided by a wind episode characterized by velocity 350 km s-1 (Finkenzeller and Mundt 1984) and mass-loss rate yr-1 typical of winds in Herbig Ae/Be stars (Nisini et al. 1995). Note that these estimates depend on our assumption of an expansion velocity for the ring, that would be very interesting to measure directly.
A second possibility is that the ring is due to the action of the radiation pressure from the star. The radiation pressure at the ring (/( c R2)) is erg cm-3, comparable with the thermal pressure of the ambient medium , with cm-3 and K, the typical temperature of the large grains at the distance of the ring.
The ring is not centered on the star, which is closer to the NW edge than to the SE. This asymmetry may be related to a density gradient in the ambient medium, if the density is higher to the NW than to the SE. Since HD 97300 is very likely located near the front edge of the Chamalion I cloud, the best tracer of its immediate environment is an optically thick molecule such as 12CO. The 12CO map of Mattila et al. (1989) shows an enhancement of the radiation temperature just to the NW of HD 97300.
Finally, it is interesting that HD 97300 is not the only Herbig AeBe star with a ring. A similar structure (although about 3.5 times larger) is seen in scattered light in the younger and more deeply embedded star LkH198 (Leinert et al. 1991). Also in that case the ring has an elliptical shape and the exciting star is shifted from its center. Both LkH198 and HD 97300 have an embedded companion with projected separation much smaller than the size of the ring. It is possible that binarity plays a role in shaping the ring. In LkH198, there is a CO outflow (Levreault 1988), roughly aligned with the reflection nebulosity, whose association to LkH198 itself is, however, uncertain. We do not know if there is PAH emission associated with the LkH198 ring.
© European Southern Observatory (ESO) 1998
Online publication: September 30, 1998