Astron. Astrophys. 349, 605-618 (1999)

3. Results

3.1. Structure of the nebula

Images of DR18 in H and the J, H, and K bands are presented in Fig. 1. At J, the overall appearance of the nebulosity matches well that observed in H, with the emission peaking near the position of the central star. The morphology changes in the H band, where the nebula is bounded to the East (left) by a bright rim. This is most clearly seen in the K band image, where the bright rim becomes the most prominent feature in the nebula. Faint filamentary extensions of the nebula cab also be seen towards the North and the East.

Fig. 1. Images of DR 18 in H and the J, H, and K filters. Each image is in size. North is at the top and East on the left.

Narrow band images of the nebula in the 2 µm window, shown in Fig. 2, can provide additional insights on the nature of the emission in different areas of the nebula. The bright rim dominates the emission in both of the images centered on the H₂ transitions, as well as in the broader band image centered at 2.26 µm. Fainter emission is seen in these filters in the parts of the nebula closer to the central star, but it is the Br emission which dominates in this region. The overall appearance of the nebula in Br is similar to that seen in the J band. The main morphological difference between the images centered on the two H₂ transitions is the fact that the outer edge of the arc-shaped nebula is outlined by a maximum in the (1,0) H₂ image, which does not appear in any of the two other narrow filter images. In both the 2.248 µm and 2.260 µm filters, the intensity tends to increase towards the inner edge of the arc. Concerning the comparison between the 2.248 µm and 2.260 µm images one may expect that, if the nebula emitted most of the luminosity seen in those bands in the (2,1) H₂ line, the threefold increase in filter width would imply a decrease of the brightness ratio of the nebula with respect to any star by a factor of 3 when passing from the 2.248 µm to the 2.260 µm filter. This is not what is seen in our images, where the brightness ratio of the nebula is seen to decrease slightly with respect to those of the stars, but by much less than a factor of 3. We take this as an indication that the dominant process responsible for the intensity observed in these bands is continuum emission, rather than emission in the (2,1) H₂ line.

Fig. 2. Images of DR 18 in four narrow band filters in the K band. The 2.122 µm filter is centered on the (1,0) H2 line; 2.166 µm, on the Br line; 2.248 µm, on the (2,1) H2 line; and 2.260 µm, on a region used to measure the CO continuum in cool stars, which in this case includes also the (2,1) H2 line. The size and orientation are the same as in Fig. 1.

To better appreciate the structure of the nebulosity, especially in the proximities of the central star, we have artificially removed the stars in all the images. This was done using some of the utilities available in DAOPHOT: first, the task PSTSELECT was used to select a set of suitable reference stars for the determination of the point spread function, which was subsequently calculated using the task PSF. Stellar images were then located in each frame using the task DAOFIND, and the scaled point spread function at the position of each one was subtracted using the task ALLSTAR. The results are shown in Fig. 3. The most dramatic effect of the subtraction of the stellar images takes place in the Br filter, in which removal of the central star reveals a bright crescent of emission peaking about east of it. Such a peak is absent or much less prominent in the other narrow band images, but can also be identified in the H, J, and H band images. Given that intense hydrogen recombination lines fall in the latter two bands, it seems clear that this maximum is mostly due to recombination line emission.

Fig. 3. A selection of broad and narrow band images in which the stellar images have been artificially subtracted using the method described in the text. Some residuals are still seen at the positions of the brightest stars. The size and orientation are the same as in Fig. 1.

Approximate flux calibration of the narrow band images has been performed by assuming that the flux emitted by the field stars is the same in all the narrow band filters, and the same in turn as the flux derived from the K band photometry. By using different box sizes around the nebula in the star-subtracted images, we estimate that the nebular integrated fluxes measured in this way have accuracies of % or worse. The integrated flux values obtained for the different filters are 0.38 Jy at 2.122 µm, 0.45 Jy at 2.166 µm, 0.40 Jy at 2.248 µm, and 0.29 Jy at 2.26 µm. While the quoted accuracy is too low for significative line ratios to be derived from these numbers, we take these results as a confirmation that the flux recorded in the different images is indeed dominated by the continuum, although with a significant contribution from H₂ and HI lines.

3.2. Spectra of the central star

Fig. 4 shows the spectrum of the central star in the K window. For the sake of comparison, we show as well the spectrum of the brightest star appearing in the southeastern quadrant of the K band image. The spectrum of the central star appears featureless at the resolution and signal-to-noise ratio used here. The absence of even Br in the spectrum shown here is however an artifact of the reduction: as mentioned in Sect. 2.2, the bright star used for the correction of telluric absorption is also expected to show Br absorption, and therefore this feature cancels out when ratioing both spectra. The reduced spectrum of the central star thus implies that its Br absorption has a similar equivalent width to that of a G0 star. We can in fact measure this absorption in the spectrum shown in the lower panel of Fig. 4, which clearly corresponds to that of a reddened early M-type giant (Kleinmann & Hall 1986). Since its Br absorption should be negligible, the false emission appearing in its reduced spectrum at 2.166 µm is actually the reversed B absorption of the star used for telluric correction. The measured equivalent width of this feature in the spectrum of the M star, , thus gives indirectly the equivalent width of Br in the spectrum of the central star. In the spectral classification scheme in the 2 µm region developed by Hanson et al. 1996, this value is typical of a main sequence star with a spectral type between O8 and B2. The absence of any other obvious lines in the spectrum is consistent with this classification. The slope of the spectrum over the K window may be due to a combination of foreground reddening and intrinsic excess emission in this spectral region. As discussed in Sect. 3.3, both effects are likely to be present in this object.

Fig. 4. Spectra of the DR 18 central star and the brightest star in K in the same field (here called the "nearby star"). Both stars are marked in the K band image under the two spectra. The principal spectral features visible in the spectrum of the nearby star, that we classify as an early M type supergiant, are indicated. The Br feature was not removed from the spectrum of the star used to correct for telluric absorption; this artificially suppresses the Br feature in the spectrum of the central star, and makes a false emission feature appear in the spectrum of the nearby star.

The spectral type can be better constrained using the visible spectrum presented in Fig. 5. Because of the large foreground extinction, the blue part of the spectrum, where most of the features used for classification are found, has a poor signal-to-noise ratio. Nevertheless, the spectrum can be shown to be later than O-type due to the absence of the HeII line at 5411 Å. On the other hand, the HeI line at 4428 Å is clearly seen, with a depth similar to that of the H line at 4340 Å, indicating a very early B type. We therefore classify the central star as B0-B1 (see Jacoby et al. 1984 for an atlas of stellar spectra at a similar resolution and covering a similar wavelength interval). Note also the appearance of strong interstellar lines, due to the high extinction towards the star, which include the diffuse interstellar band (DIB) features at 5780 Å and 6284 Å, as well as the Na I D doublet at 5890 Å, whose strength is usually observed to be similar to that of the 5780 Å feature (Dorschner et al. 1977). Using band strength to reddening ratios observed towards other lines of sight (Jenniskens et al. 1994) we find that the equivalent widths of these features, Å, are consistent with an extinction of mag toward the star, as discussed in Sect. 4.2.

Fig. 5. Spectrum of the DR 18 central star at visible wavelengths, with the main stellar and interstellar features identified. The bands marked as O2 and H2O are telluric. The steep slope in the blue is a consequence of heavy extinction. Note that the vertical scale is logarithmic.

3.3. Photometry

The near-infrared color-color diagram is a useful tool to discern young stellar objects from the unrelated background and foreground population, thanks to the excess emission displayed by stars still surrounded by circumstellar disks. By modeling the disk surrounding an intermediate mass star, such as a Herbig Ae/Be star, Lada & Adams 1992 showed that these objects occupy a region of the , that is inaccessible to normal photospheres reddened by foreground extinction. Objects occupying that region of the diagram are thus likely to be in the early stages of their evolution. Since this is most likely to be the case of the central star illuminating the nebulosity of DR 18, the , diagram has the potential of revealing fainter, roughly coeval companions that may have formed together with the central star, thus allowing a better understanding of the star formation process in the DR 18 region. The census of young stars obtained in this way may be incomplete for several reasons. Some of them may have already dissipated their disks, or may have an insufficient amount of circumstellar material close to the star for it to emit significantly at 2 µm. Moreover, stars heavily obscured by foreground dust would be missed at J, and it would not be possible to place them in the color-color diagram. Therefore, it must be kept in mind that the sample of young stars identified in this way is likely to be biased towards objects with large amounts of circumstellar material lying in regions of relatively low extinction.

The definition of the region of infrared excesses requires the adoption of a reddening vector. Although the near-infrared extinction law has been observed to be more universal than at visible wavelengths (Mathis 1990), departures are sometimes observed in star forming regions. The slope of the redening vector in the , diagram, , tends to decrease in such regions as the total-to-selective absorption ratio increases, what is commonly interpreted as an effect of increased grain size. Thus, in the Ophiuchi cloud, where (Vrba et al. 1993), is found to be 1.57 (Kenyon et al. 1998), as compared to the standard values of 3.1 and 1.7 (Rieke & Lebofsky 1985). Similarly, Wilking et al. 1997 find in the star forming region around R Coronae Australis, where (Vrba & Rydgren 1984). However, this behavior may not be general: in Chamaeleon I, another star forming region with a larger than average (; Steenman & Thé 1989), Comerón et al. 1999 find a , diagram where the distribution of objects, almost entirely composed of background stars, traces a reddening vector with and is incompatible with a slope as low as 1.5. In the Cygnus region, Terranegra et al. 1994 find , but Torres-Dodgen et al. 1991 find in the direction of Cygnus OB2, which is only 30´ away from DR 18.

Fig. 6 shows the color-color diagram of the stars in the imaged area. Only stars with , , were retained to ensure color indexes accurate to 0.1 mag. The solid curves are the loci populated by unreddened main sequence stars and giants (Bessell & Brett 1988). The straight lines are limiting reddening vectors with and , according to the discussion in the previous paragraph. Their origin is placed at the intrinsic colors of an A0 star, as those would be the stars for which extinction would cause the reddest at a given ; for any reasonable slope of the reddening vector, M giants should always have a smaller than reddened A stars with the same . This argument is not strictly true if the slope of 1.7 is adopted for the reddening vector, as unreddened late M dwarfs would be even redder in . However, M dwarfs should be far too faint to be detected at the distance of DR 18.

Fig. 6. Color-color diagram of the stars detected in the J, H, and K bands. The solid curves are the loci occupied by normal main sequence (lower branch) and giant stars (upper branch) (Bessell & Brett 1988). The dotted line is the reddening vector of the normal reddening law (Rieke & Lebofsky 1985), and the dashed line is a reddening vector with a slope of 1.5, as observed in some star forming regions.

The distribution of stars in the color-color diagram shows that the field is dominated at low extinctions by sources with K-band excess, regardless of the adopted extinction law. The interpretation at higher extinctions critically depends on the slope of the reddening vector: a slope of 1.7 places nearly all highly obscured objects in the zone of circumstellar excesses, a situation that we judge very unlikely. A slope of 1.5 is however consistent with the vast majority of objects with being reddened background stars. The upper envelope of the distribution of points in Fig. 6 is in agreement with the 1.5 slope, as no stars lie in the region above it, which should be inaccessible to stars both with and without circumstellar material. Therefore, in view of the distribution of stars in the , diagram, we believe that a value of is favoured.

The spatial distribution of the sources with infrared excess is presented in Fig. 7. In it, sources having an observed color more than 0.2 mag redder than the limiting reddening vector at their position are marked on the K-band image. The inhomogeneity in their distribution is apparent, and a weak trend for them to lie along a band running from the southeast to the northwest may be appreciated. The scarcity of sources with infrared excess in the northeastern quadrant of the images is partly an obvious consequence of the lower observed overall density of sources in that area. This may in turn be caused by a higher extinction, confirmed by the fact that the objects with the reddest colors are generally found in that quadrant. However, a closer analysis of the data shows that the inhomogeneity of the overall distribution is accentuated when only infrared excess sources are taken into account, what we take as an evidence for a real clustering of such sources in the western and southern halves of the field. The central star of DR 18 thus appears to be the brightest member of a loose aggregate of young stars, with the less luminous members possibly illuminating the filaments extending to the North and to the East of the arc nebula. A wider area JHK survey reaching to similar magnitude limits as the one presented here should easily confirm or disprove the spatial distribution of sources hinted here.

Fig. 7. Spatial distribution of the stars in the field of DR 18 appearing in the infrared excess region of Fig. 6. The stars lying more than 0.2 mag to the right of the limiting reddening vector with slope 1.5 in that figure are surrounded by a circle in this K-band image.

© European Southern Observatory (ESO) 1999

Online publication: September 2, 1999
helpdesk.link@springer.de