Astron. Astrophys. 328, 290-310 (1997)

2. Observations and results

We obtained images of AFGL2688 in a number of observing sessions (see Table 1 for summary). Near-IR images were all obtained with the 3.8m United Kingdom Infrared Telescope ¹ using the facility near-IR camera IRCAM, which employed a 62 58 pixel InSb array (McLean et al. 1986). Images were obtained at a pixel scale of 0.62" in the J, H and K bands in June 1990, and in October 1990 in the K band, in the H₂ S(1) 1-0 line at 2.122 m with a 1% spectral bandwidth filter, and at 3.4 m with a 4% bandwidth filter (nbL). During June 1993 images were obtained at 0.62" pixel scale in the J and K bands, at nbL, in the H₂ S(1) 1-0 line and in the continuum at 2.104 m with narrow bandwidth filters, as well as at 0.31" pixel scale in the H₂ S(1) 1-0 line. All observations were taken in stare mode, with stares of various integration times being alternated between source and sky, and a sky frame subtracted from the source frame to remove the sky background emission. The sky frame was used in each case to flat-field the source image. In all cases an integration identical to the source observation was made looking at a cold stop in order to approximately subtract the dark current. Observations of the standard stars BS8143 and HD203856 were used to flux calibrate the images, and to determine the point spread function (psf), which was approximately 1.2" (FWHM) for the images with 0.6" pixels and 0.6" (FWHM) for the image with 0.3" pixels (since we had fewer than 2 pixels per resolution element, our spatial resolution is limited by the pixel size rather than by the seeing or the telescope performance).

Table 1. Observation details

Mid-IR images were made at the NASA Infrared Telescope Facility ² during May 1991. We used the Berkeley/Livermore Mid-IR Array Camera, which employs a Hughes 10 64 pixel Si:Ga array detector (Arens et al. 1987a, 1987b; Keto et al. 1992). Images at 8.8 m, 10 m, and 11.5 m were made using a 10% spectral bandwidth circular variable filter (CVF), and images at 8.2 m and 9.7 m were taken using a 1.3% spectral bandwidth CVF. In each case a mosaic was assembled out of 3 to 5 10 64 pixel panels in order to cover the extent of the IR nebula. Sky subtraction was achieved by nodding and chopping as described by Ball et al. (1992), and flux calibration was done by comparing observations of the standard star Her. After each frame was sky subtracted and flat-fielded, any bad pixels were removed by interpolation using IRAF. The frames were then assembled automatically into a mosaic using a program which minimises the variance between frames in the overlap regions (Meixner 1993). The point source calibrator Her was used to determine the psf, which was found to be approximately circular, with FWHM 1.5": this is considerably worse than the diffraction limit for this telescope, and rather worse than we have experienced on other observing runs at the IRTF.

An image was obtained at a wavelength of 19.2 m at the 60-inch IR telescope on Mt. Lemmon, using the UCSD/Minnesota mid-IR camera (Piña 1994). These data were sky subtracted, flat fielded and flux calibrated in a similar fashion to the IRTF images. Careful comparison of this image with a variety of point source calibrators taken through the night indicated that AFGL2688 was basically unresolved at this wavelength, and that the source size is less than 2" (FWHM).

Image reduction for the near-IR and mid-IR images was done partly in Starlink's IRCAM package and partly using IRAF ³. The 19.2 m image was reduced using tasks developed in the IDL software package. Further details of those images presented here are listed in Table 1.

The 1993 J-band image is shown in Fig. 1. Both of our broadband K images show the same structure and details as Fig. 4c in Latter et al. (1993), namely N-S lobes similar to the optical nebula, plus a pair of E-W "ears." We do not present these, nor the H-band image, which is identical to Fig. 4b of Latter et al. (1993), and very similar to our J-band image, because they add little to the discussion in this paper, and because the broadband K images contain a complex mix of continuum and H₂ emission, whereas our 2.104 m narrowband image contains only continuum flux, with no contribution from any lines (Hora & Latter 1994), and so is more informative. For broadband H or K images, the interested reader is encouraged to consult Latter et al. (1993). The narrowband L image (3.4 m, nbL) in Fig. 2 shows the N-S lobes almost joined together, giving the appearance that the Egg contains a tadpole (unusual offspring for Cygnus, the Swan). The three mid-IR images at wavelengths of 8.8 m, 10.0 m and 11.5 m, appear more or less identical (only one is shown in Fig. 3). The other two images (8.2 and 9.7 m) have significantly lower S/N, due to the smaller bandwidth of the observations, and do not show any further detail. Despite the rather worse than usual spatial resolution, our mid-IR images are consistent with those presented by Hora et al. (1995), which were taken at UKIRT in 1993.

Fig. 1. UKIRT J-band image of AFGL2688, presented in grey-scale with contour overlay. The grey scale and the contours are logarithmic, pixel scale is 0.62". The upper contour is 2.1mJy/arcsec2 and the lower contour 2.1 Jy/arcsec2.

Fig. 2. Narrowband L (nbL) image of AFGL2688, presented as Fig. 1. Upper contour 117mJy/arcsec2, lower 0.29mJy/arcsec2.

Fig. 3. IRTF mid-IR image of AFGL2688, taken with a 10% bandwidth CVF, centred on a wavelength of 8.8 m again presented as Fig. 1. Pixel scale is 0.39". Upper contour 39Jy/arcsec2, lower 0.39Jy/arcsec2.

While there are clear differences between the images at different wavelengths, there are a number of morphological features in common. At the lowest flux levels, there is a hint of a rectangular structure in all the images where the vertices are the prominent N-S lobes and the fainter E-W lobes seen more prominently at longer wavelengths. The rectangular structure dominates the mid-IR images. The rectangle is largest in the J-band image, and smallest in the mid-IR and the horns which are so striking at I (Latter et al., 1993) and at J become progressively less apparent with longer wavelengths, consistent with the decrease in scattering efficiency of the dust and increase in thermal emission as the wavelength is increased. In all the near-IR images, there is a significant E-W asymmetry in the nebula, the emission appearing more extended to the western edge of the center than to the eastern edge. At the higher flux levels, the N lobe appears to orient almost due N, whilst the S lobe appears skewed significantly W of S. As we move to longer wavelengths, the angular separation of the N and S lobes decreases with increasing wavelength and the western skew of the S lobe with respect to the N lobe becomes more apparent. At mid-IR wavelengths the N-S lobes are joined together and are clearly skewed in angle.

The 1993 H₂ S(1) 1-0 images are presented in Figs. 4a (0.62"/pixel) and 4c (0.31"/pixel). The "ears" seen in broadband K images are seen clearly in Fig. 4c. The 2.104 m continuum image is presented in Fig. 4b, and we now see that, with reasonable S/N, part of this E and W blob structure is visible in scattered continuum as well as in the line emission. With a larger field of view, in Fig. 4a, we see two additional features - two very faint blobs on an axis through the center of the nebula at position angle 75 , and at about 11" from the center of the nebula. There is a finite chance that these two blobs are ghosts because we did collect some ghost images in some of the narrowband observations during these observing runs; however, we made checks using single standard stars and all these checks failed to produce any pairs of ghosts consistent with these two features, at this or any other wavelength.

Fig. 4. Narrowband images in the 2 m region of AFGL2688 from UKIRT, presented as Fig. 1. From top to bottom: a 2.122 m image, upper contour 4.4mJy/arcsec2, lower 34 Jy/arcsec2. b 2.104 m image, upper contour 20mJy/arcsec2, lower 44 Jy/arcsec2. c 2.122 m image at 0.31"/pixel, upper contour 4.1mJy/arcsec2, lower 41 Jy/arcsec2.

The 2.122 m image presented in Fig. 4a contains a large contribution from scattered continuum emission. We have subtracted the 2.104 m continuum image from the 2.122 m image to yield a nearly pure H₂ image which we present in Fig. 5. Uncertainties in the flux calibration ⁴ of the original images presented here will impart some errors into Fig. 5, and so the precise details of the structure revealed in Fig. 5 must be treated with some caution. However, the gross morphology should be correct. Given the S/N of the various observations, Fig. 5 herein agrees with the H₂ observations of AFGL2688 in Beckwith, Beck & Gatley (1984) and Smith et al. (1990), but is of considerably higher S/N and so reveals fainter structure than any previous images at this wavelength.

Fig. 5. Difference between Figs. 4a and 4b, which represents the pure molecular hydrogen emission, presented as Fig. 1 again. Upper contour 390 Jy/arcsec2, lower 12 Jy/arcsec2.

In molecular hydrogen the nebula is spectacular. The molecular hydrogen emission is more than just photogenic; it is very revealing. The E-W structure ascribed by Latter et al. to a torus is highly clumped, indicating that the nebula as a whole probably has a very inhomogeneous structure. The filaments linking the N and E blobs and their radial velocities (Smith et al. 1990) very strongly suggest that there is a physical link between these two blobs. A fainter but significant filament also connects the S and W H₂ blobs suggesting a physical link for this pair as well. The two new H₂ blobs at PA 70 and the elongation of the mid-IR images delineate yet another axis through the middle of the nebula. The two detached H₂ blobs or ansae are highly reminiscent of the ansae and FLIERs (Fast Low-Ionization Emission Regions; Balick et al. 1993a,b) that appear in about 20% of the sample of bright and moderately compact planetary nebulae imaged by Balick (1987). These features possess high velocities relative to the surrounding nebular gas and usually have smaller dynamical ages. The possibility that bow-shock excitation is responsible for many of the observed characteristics of PN FLIERS was considered by Balick et al. It would be of some interest to determine the velocities of the detached H2 ansae in AFGL2688 to see if they too are moving at high velocities.

In Fig. 6 we present a dust temperature map generated from the ratio of 11.5 m to 8.8 m emission. Before making far-reaching conclusions from this map, we point out that the interpretation, and even the validity, of such a map must be questionable in such a nebula. Temperature and optical depth maps can be derived for observations of optically thin nebulae, under the assumption that the optical depths at the two wavelengths are very similar at each position, and that the nebular temperature is very similar at the two wavelengths. In the case of an extended, optically thick nebula, the radiation from below the unit optical depth surface is largely lost, and the temperature at a given position is largely determined by the depth to which we can see. Deutsch (1990) assumed an optically thin nebula and derived, for the center of the AFGL2688 nebula, an optical depth at 9.8 m (0.037) larger than that at 8.3 m (0.013), which appears inconsistent with properties of the amorphous carbon grains that are believed to comprise the majority of the grains in the Egg nebula and that have a larger opacity at 8.3 m than at 9.8 m. In the model which we present in Sect. 5, the dust nebula is optically thick and the temperature in the inner region changes very rapidly with depth. Because we see to somewhat different depths at 8.8 and 11.5 m, the intensity ratio maps will no longer fulfil the assumptions used in deriving temperatures and optical depths. Thus one must be extremely cautious in interpreting Fig. 6. Nevertheless, we find that the derived temperatures range from 120K to 170K, which is a reasonable range, consistent with the range of temperatures predicted by the model in Sect. 5. The morphology of the temperature map is ambiguous. One could interpret the E and W minima as being consistent with the presence of a strongly absorbing torus, whose plane lies on the line connecting the E and W minima (Sect. 3), as predicted by the standard model. However, our alternative model (Sect. 4) also predicts temperature minima behind the material that is shadowing part of the lobes. Hence, this temperature map could be taken to be consistent with either model.

Fig. 6. Dust temperature map made using the 8.8 and 11.5 m images. Temperatures are contoured linearly from 120K to 172K. Note that these temperatures may be misleading, as explained in the text.

© European Southern Observatory (ESO) 1997

Online publication: March 24, 1998

helpdesk.link@springer.de