Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 346, 441-452 (1999)

Previous Section Next Section Title Page Table of Contents

3. The central source

In Paper I, we determined the spectral type of the central star to be roughly early K on the basis of a high resolution spectrum in the range 6100Å to 6800Å. We have now used our low resolution spectrum to make a better determination of the spectral type as well as the amount of extinction and veiling present in the system. Fig. 1 shows the measured spectrum with our calibrated B, V, R observations shown as squares. The excellent correspondence between the broadband fluxes and the spectrum shows our spectrophotometry is well calibrated.

[FIGURE] Fig. 1. Low resolution spectrum (thick line) with K2 III template spectrum (dotted line; Silva & Cornell 1992). The template is also shown offset for clarity (thin line). The boxes are wide-band B, V, and R photometry (upper limit for B).

We have labeled and identified the significant emission and absorption lines visible in this spectrum. Metal absorption lines typical of late-type stars and emission lines typical of T Tauri stars (Balmer and various forbidden lines) are clearly present, while strong molecular bands are absent.

3.1. Spectral type and veiling

We compared our spectrum to the template spectra of Silva & Cornell (1992) available in electronic form. The strength of the Fe I absorption lines in the region 4900Å to 6500Å implies the spectral type cannot be earlier than [FORMULA]G6. Conversely, the absence of strong TiO bands at 5170Å and 6160Å constrains the spectral type to be earlier than [FORMULA]K5. The TiO band at 5170Å may be weakly present, suggesting a best spectral type of [FORMULA]K1. The high-luminosity models are also more favored because of the weakness of the bands given the strength of the absorption lines. The K2 III and G8-K0 III spectra give the best comparison, though the K1-2 I and G9-K0 V spectra compare almost as well. Table 2 lists our assessments of a range of models including these four and several other, less likely spectral types.


Table 2. Spectral class comparisons

In our comparisons with the template spectra, we principally used the strength of the absorption lines and bands to constrain the spectral type. It is also possible to use the general shape of the spectrum to constrain the extinction, as well as other effects. We applied the standard extinction curve, tabulated by Whittet (1992), to the template spectra, and determined the best value of [FORMULA] for each spectral type. These extinction values are also given in Table 2. Fig. 1 includes the best-fit spectral template (K2 III), adjusted for 2.3 magnitudes of extinction.

There are several other effects in addition to the extinction which can change the continuum shape. First, the extinction may not follow the standard Galactic extinction curve, and may instead be anomalous. Anomalous extinction has been seen for stars in dense clouds. Cardelli et al. (1989) have presented a parameterized form of the extinction law as a function of [FORMULA]. Values of [FORMULA] up to 5.6 have been seen for stars in the star forming regions in Orion (see e.g. Cardelli et al. 1989). Another possible effect is the scattering of the light by dust grains. If the stellar source we observe is not actually the star, but instead a reflection of the star off dust clouds, then the emission will be altered by a scattering law, [FORMULA]. Finally, the presence of a veiling continuum may contribute a blue component to the stellar flux. A veiling continuum could, for example, be due to hot gas at the accretion disk boundary layer, and is frequently used as a measure of the accretion activity in T Tauri stars (e.g., Hartmann & Kenyon 1990; Batalha & Basri 1993). We compared our observed spectrum with models including all of these effects. We started with the best model with normal extinction, no veiling, and no scattering, and slowly added additional effects to judge the importance. We hoped to constrain the possible range of [FORMULA] values, to determine roughly the amount of veiling, and to test if a reflected model may be discriminated from a direct image model on the basis of just the continuum shape.

We first attempted to add a significant veiling continuum, in an attempt to make the model and observations agree for [FORMULA]Å, where there is a clear excess for the simplest model. One constraint on the amount of veiling is independent of the continuum shape: it is not possible to add a large veiling continuum without reducing the strength of the absorption lines too much. The G-band is particularly sensitive to this effect, as it is a significant, deep line in the blue, but the Fe I lines at [FORMULA] also show this effect. For comparison with Hartmann & Kenyon (1990), we determined the strength of the veiling continuum by measuring the flux ratio [FORMULA] at 5200Å. When [FORMULA], the G-band is somewhat reduced, for [FORMULA], the G-band is no longer deep enough, but the other redder lines are still acceptable, and at [FORMULA] all of the Fe I lines are too weak compared with the observations. With an optically thick continuum ([FORMULA]), it is not possible to add enough veiling emission to explain the blue emission at [FORMULA]Å without completely washing out the absorption lines. As the veiling emission becomes more optically thin, the ratio of the Balmer emission lines to the continuum increases. In this case, a large fraction of the blue emission may be explained by blended Balmer lines. A model with [FORMULA] and [FORMULA] agrees with the Balmer line flux in the blue region, as well as with the observed H[FORMULA] and H[FORMULA] line fluxes, but it does not explain all of the observed flux in the blue region. If the optical depth of this veiling is decreased much below 0.2, too much flux is produced in H[FORMULA] to allow the veiling to explain the blue emission. We conclude that the level of veiling is [FORMULA], due to a marginally optically thin gas, and that, while some of the excess emission seen in the range [FORMULA]Å may be due to Balmer line emission, much of it must be due to a variety of other emission lines, to photometric uncertainty or to calibration errors at the edges of the spectrum.

The value of [FORMULA] is not strongly constrained by the shape of the spectrum in this relatively narrow spectral region. It is possible to increase the [FORMULA] value and still achieve acceptable fits by simultaneously increasing the extinction. There are small changes in the shape of the continuum, but for the range [FORMULA] the models are acceptable. When [FORMULA] is too low, the region 4800Å [FORMULA]Å has too much flux in the model, and when [FORMULA] is too high, this region has too little flux. A difficulty in determining the value of [FORMULA] is the clear presence of a blue excess and the possible presence of a near-IR excess from circumstellar material. We are unable to use the photometry beyond R to improve the constraint of [FORMULA] because these bands may be contaminated by emission from a circumstellar disk. While the best value of [FORMULA] is roughly 3.1, we cannot rule out values in the range [FORMULA].

All of the above conclusions assume we see the star directly. If the flux we see is the result of scattering, we must adjust the spectral shape by [FORMULA] (see e.g., Whittet 1992). It is not possible to find an acceptable fit with normal extinction and scattering. However, by including anomalous extinction, it is possible to find an acceptable fit. The best scattering model has [FORMULA] and [FORMULA], using the K2 III spectral template. As before, we can decrease [FORMULA] if we decrease the value of [FORMULA]. Acceptable fits are found for the range [FORMULA]. The best fit with scattered emission and anomalous extinction is essentially indistinguishable from the best model without scattered emission. We conclude that it is not possible to distinguish the scattering model and direct model on the basis of the spectral continuum shape alone, though Occam's razor would suggest the simplest model is favorable. Below, we will discuss the arguments in favor and against the scattered light model.

3.2. Spectral energy distribution

In Fig. 2, we show the broadband spectral energy distribution for the central source from 0.45 µm to 1100 µm, based on the IRAS data and our broadband photometry. The fluxes for the central source at wavelengths longer than L include the JCMT observations reported above and the IRAS observations. These points, especially the IRAS points, should be viewed with some care because of the large beam size. As noted above, there is evidence in our 800 µm data of some extended emission at these wavelengths. However, the fact that the IRAS source does not appear extended in the deconvolved image suggests that the level of contamination is small. Also, there is evidence at other wavelengths of only a limited number of other young stars in the area, all substantially fainter than Holoea at 3.6 µm.

[FIGURE] Fig. 2. Spectral Energy Distribution for IRAS 05327+3404 (Holoea). The thick solid line is a stellar spectrum with temperature 4750 K and extinction [FORMULA] for comparison with the observations. The thin solid line is the best Calvet et al. (1994) model. The dashed curve is the best model from Adams & Shu (1986). See Table fluxes for photometry values.

The spectral energy distribution (SED) shows a peak around 60 - 100 µm, a non-thermal distribution on the blue side of the peak, and a submillimeter spectral shape of [FORMULA] where [FORMULA] is in the range 3.5 - 4. The submillimeter spectral index of 3.5 - 4 may be understood in terms of a thermal spectrum from dust grains with an emissivity proportional to [FORMULA], where [FORMULA]. The range 0.5 - 1.0 for [FORMULA] is typical for dust in the submillimeter (e.g., Whittet 1992; Beckwith & Sargent 1991). The peak of the spectrum shows that the coolest emission components have a temperature around 25K. The SED between 100 µm and 3 µm has a slope which is typical of flat spectrum sources (see e.g. Calvet et al. 1994). The flux in the optical bands principally represents emission from the photosphere of the central star, with some moderate absorption ([FORMULA]; see discussion above). For comparison, we have included in the figure a model spectrum (Kurucz 1991) for a star with a temperature of 4750 K and an extinction of [FORMULA]. The flux in the near IR (J, H, K) represents emission from a warm component in excess of the stellar photosphere. The measured nbL flux is surprisingly high, especially when compared with K. It is unlikely that this is due to variability, as the nbL image was taken within minutes of the JHK images. We have no explanation for the extreme [FORMULA] color.

Several groups have made models of the expected SED of YSOs at various stages in their evolution. Models by Adams & Shu (1986) include an extended dust envelope, a central protostar, and an active accretion disk, which means that the energy generated through viscosity in the disk raises the temperature of the disk. The thermal structure of the accretion disk and the temperature of the central source are fixed by the accretion rate and the processing efficiencies, while the thermal structure of the envelope is calculated based on the resulting star + disk luminosity. In models by Calvet et al. (1994), the central disk is passive, which means it does not generate its own heat. In this case, the disk thermal structure is determined by the reprocessing of the stellar radiation field. Both of these groups present models with various choices for the free parameters.

We have compared the observed SED of IRAS 05327+3404 (Holoea) with the models of these two groups. We find that none of the models are able to describe the entire SED particularly well. Of the Calvet et al. models, those which are most successful have viewing angles nearly pole-on, or large values of the centrifugal radius (implying high rotation rates). However, if the viewing angle and opening angle of the central hole are chosen to give sufficient optical emission, then the near or mid IR regions have much too little flux. The best of the Calvet et al. models is shown in Fig. 2 as a solid curve. This has parameters [FORMULA] g cm-2, [FORMULA] AU, [FORMULA]o, and [FORMULA]o, and is shown in their Fig. 6. Of the models by Adams & Shu (1986), the best comparison is a model with the highest rotation rate, shown as a dashed line in the figure. This model has parameters of [FORMULA], [FORMULA] km s-1, [FORMULA], [FORMULA], and [FORMULA] rad s-1 and is shown in their Fig. 6. Adams & Shu point out that higher rotation rates make the general shape of the SED wider by allowing a better view of the central source while maintaining large amounts of cool material at large radii. Thus, both sets of models suggest that a high cloud rotation rate is necessary to explain the width of the observed SED.

There is a possible alternative explanation for the combination of the large far-IR flux and the optically visible central star: the system may be a binary star, in which a still-enshrouded star is the source of the far-IR flux. There are a small number of systems consisting of a T Tauri star and an embedded infrared companion. Konesko et al. (1997) discuss the evolutionary status of this class of systems. If this is the case for Holoea, there is a natural dividing line in the SED: the flux from L to the submillimeter would be due to the embedded star while the flux from the optical to K would be due to the optically visible star. However, the L-band image shows a point source which is coincident with the optical and near-IR stellar source to within 0:001, the accuracy of the alignment of the images. Second, there is other evidence for the necessary circumstellar material with a wide central hole, as implied by the SED models. This evidence comes from the morphology of the reflection nebula, and is discussed below.

Even if Holoea is not a binary, we are hesitant to draw very strong conclusions about the properties of the circumstellar material present in Holoea on the basis of the model comparisons. The models produce SEDs which agree with observed SEDs in a general sense, but are probably still too simplified to predict details. Current models all make simplifying assumptions to reduce the computational difficulty of the modelling problem. For example, they invoke spherical symmetry to avoid a 3-D radiative transfer problem. Certain of the models by Calvet et al. (1994) break the spherical symmetry by including a central hole, but these are calculated under the assumption of spherical symmetry, with a correction for the additional heating on the conical surface of the hole due to the central star. The authors offer warnings about the approximations involved, which we will heed. Before strong conclusions can be made about the geometry of the system, more detailed, 3-D radiative transfer models are necessary. However, we can draw some conclusions: First, the central source suffers only moderate extinction, implying that we are looking along a central hole of some kind, though as we discussed above, it is possible that we are seeing the central star via a reflection down the central hole. Second, the large amount of emission in the far IR suggests that large amounts of dust surround the source at a large distance, corresponding to the outer extremities of the circumstellar disk or to a large circumstellar envelope. Finally, the high level of mid-IR emission suggests the presence of warm circumstellar material on a smaller scale, presumably an accretion disk or the remnant material of an accretion disk. We are hesitant to use the term accretion disk since the small amount of veiling seen in the optical spectrum (see above) suggests that the current accretion activity is fairly small.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: May 21, 1999