New, sub-o.l-arcsec radio maps of two young planetary nebulae

Transcription

1 Mon. Not. R. Astron. Soc. 284, (1997) New, sub-o.l-arcsec radio maps of two young planetary nebulae M. Bryce/ A. Pedlar,2 T. Muxlow,2 P. Thomasson2 and G. Mellema 3 'Department of Physics and Astronomy, The University of Manchester, Oxford Road, Manchester M13 9PL 2University of Manchester, Nuffield Radio Astronomy Laboratories, lodrell Bank, Cheshire SKll 9DL 3Stockholm Observatory, S Saltsjobaden, Sweden Accepted 1996 August 12. Received 1996 July 24; in original form 1996 May 20 ABSTRACT Sub-O.l-arcsec maps of the 5-GHz continuum emission from two young planetary nebulae (NGC 7027 and BD ) have been obtained by combining MERLIN and VLA observations, resulting in the highest spatial resolution radio maps of these (or any) planetary nebulae to date. Small clumps of emission are resolved within the essentially ring-like structure of each nebula. These radio maps are at a very similar spatial resolution to HST WFPC2 images, and thus for the first time it is possible to compare radio and optical images of young planetary nebulae at very high spatial resolution. Key words: planetary nebulae: individual: NGC individual: BD radio continuum: ISM. planetary nebulae: 1 INTRODUCTION As part of a larger, ongoing project to investigate young planetary nebulae (PNe), high-resolution 5-GHz radio continuum maps have been obtained of two young PNe, NGC7027 and BD These maps combine both Very Large Array (VLA) and Multi Element Radio Linked Interferometer Network (MERLIN) data, and are believed to be the first maps of PNe to use data from both interferometer arrays. The MERLIN array at 5 GHz can provide spatial resolution of up to 40 mas for sources of up to arcsec in linear dimension under the most favourable circumstances: however, since the two targets here have dimensions -12 x 8 arcsec 2 (NGC 7027) and -7 x 6 arcsec 2 (BD ), it was necessary to combine the very highresolution MERLIN data with VLA data which provides the short baselines required for complete maps of these targets. The resulting maps have a resolution of better than 0.1 arcsec, which is sufficient to resolve many of the structural details. (At a typical PN distance of 1 kpc, 0.1 arcsec=4.85 x 10-4 pc.) These high-resolution radio maps are ideally suited to comparison with HST images, which have a similar spatial resolution. The two targets in this paper have been well studied in a range of wavelengths and are often taken as prototypical examples of young PNe. Previous VLA maps of these objects, such as those of Basart & Daub (1987, hereafter BD87), Masson (1989), Hajian, Terzian & Bignell (1993) and Kawamura & Masson (1996) show that both appear as elliptical rings of emission. These 6-cm VLA maps have resolutions -1 arcsec, about an order of magnitude larger than the new observations presented here. Masson (1989) used an ellipsoidal shell model and three epochs of observation to measure proper motions in NGC 7027 and hence to derive a distance of 880 ± 150 pc, consistent with the distance of 700 ± 100 pc derived from two epochs of observation by Hajian et al. (1993). A similar, but more sophisticated, investigation by Kawamura & Masson (1996) led to a distance of 1500±400 pc to BD , substantially lower than Hajian et al.'s (1993) distance of 2700 ± 800 pc. Kinematical ages for the two nebulae were derived to be -600 yr (NGC7027: Masson 1989) and 900±200 yr (BD : Kawamura & Masson 1996). Ground-based optical images (e.g. Balick 1987) show NGC7027 as a very bright knot, offset to the north-west of the central star, with a fainter extension to the south-east. This PN is unusual in that the optical and radio images are very different. Optical images of BD (Balick 1987) show an elliptical shell, similar to the one seen in the radio maps. Both nebulae show evidence of very faint, extended optical emission surrounding the main nebular shells. Seaton (1979) investigated the extinction of radiation from NGC7027 by considering radio continuum emission and hydrogen and helium emission lines from this nebula. He concluded that for A. ~ 6363 A an extinction model that includes a contribution from local nebular extinction as well as from interstellar extinction was necessary to explain the 1997 RAS

2 816 M Bryce et al. data. Walton et al. (1988) compared a 2-cm VLA continuum map of NGC7027 to narrow-band continuum and HfJ images and produced an extinction map which showed that the local extinction was markedly less in the area of the bright optical knot. Walsh & Clegg (1994) investigated the polarized emission-line profiles from NGC7027 and discovered that up to 44 per cent of the emission from the fainter parts of the nebula was polarized, although the bright knot was only 0.2±0.1 per cent polarized. They concluded that these observations confirm the existence of a dusty halo around the ionized region, which has been breached at the position of the bright knot. BD also shows evidence of localized nebular extinction. [Sm]9532-A and [On] 3727-A narrow-band Wide Field and Planetary Camera 2 (WFPC2) images presented by Arnaud, Borkowski & Harrington (1996) show several dusty clumps, particularly in the north and north-eastern regions of this PN. 2 THE OBSERVATIONS The initial observations of both targets were made using the MERLIN array at 4994 MHz. BD was observed for almost 12 h on 1995 April 8-9 and NGC 7027 was observed for approximately 14 h on 1996 April Because of the large field of view the data were taken in spectral-line mode (15 x I-MHz channels in left and right circular polarization) with short (4s) integration times. The data were flux calibrated against 3C286 assuming a flux density of Jy. Phase and gain variations were calibrated by rapid cycling to nearby phase calibrators ( and ). Image deconvolution was performed using the AlPS maximum-entropy package VTESS. As expected, NGC 7027 was rather large to be mapped using just MERLIN at this frequency and only the sharp edges of the bright limbs were resolved. The map of BD showed the expected ring-like structure but surprisingly the ring of emission was seen to have a rectangular appearance as opposed to the elliptical shape observed in previous, lower resolution VLA maps. Here again, however, the lack of short baselines meant that the larger scale structure of the nebula was resolved out. VLA A-array data was obtained for both targets on 1995 June 30. Each data set contained about 1 h of on-target observation spread across 3 h of observing time. These data were processed using standard VLA procedures and concatenated to the MERLIN data after conversion to pseudospectral-line mode. It was felt that the map of NGC 7027 would benefit from further short spacings and therefore a B-array data set of about 1 h integration, obtained by Terzian et al. in 1989 (Hajian et al. 1993), was also added to the data. The final map is shown in Fig. l(a) opposite. The raw image was made with the IMAGR software, with a robustness factor of 0, corresponding to a weighting scheme midway between uniform and natural, and the final map was obtained with VTESS, with a resultant beamsize of 62 x 48 mas 2 The resulting map of BD is shown in Fig. l(b). The dirty map was also made using the IMAGR software in AlPS, this time using a robustness factor of 4 which corresponds to approximately natural weighting. The final deconvolution was again made using VTESS which gave a resultant beamsize of 82 x 78 mas 2 3 DISCUSSION The new observations presented in this paper include data at only one radio wavelength and hence it is not possible to solve for both electron temperature and optical depth across the objects. Observations are underway on MERLIN at 1.5 GHz, which (together with proposed 1.5-GHz VLA observations) will allow the derivation of electron temperature, optical depth and emission measure on scales of order 0.15 arcsec (limited by the lower resolution of the 1.5-GHz observations). In the meantime, the new data are compared to the results of BD87 in order to derive some preliminary results. 3.1 The compact radio structure The peak flux density observed in NGC 7027 is 6.8 x 10-4 Jy beam - \ which corresponds to a brightness temperature Tb = K, occurring at a point in the south-west limb. BD87 measured their peak brightness temperature at 6 cm (7500 K) in the north-east limb; corresponding values of optical depth r = 0.9 and electron temperature Te = K were derived at this position. The value of Tb measured from the new map at this point in the north-east limb is 10250K. Incidentally, when the new map is convolved to BD87's beamsize, Tb derived from the convolved map is within 3 per cent of BD87's value, which shows that the flux calibration of the new, high-resolution map is consistent with that of the old map. The much higher value for Tb measured from the high-resolution map implies that the emission is coming from compact structure, unresolved by the relatively large beamsize used by BD87. A simple calculation shows that BD87's beam would contain just over 500 of the new beams but their maximum detection contains just less than 400 times the flux density measured in the new beam at the same position. Comparison of the map shown in Fig. l(a) with the radio map produced by BD87 shows that the structure of the nebula is much more clearly resolved in the new map. Improved values of optical depth can be calculated, using the electron temperatures Te derived by BD87 (which are unlikely to vary over small regions of the nebula), combined with the brightness temperatures measured at the same spatial position from the new maps. Using the familiar relationship r = -In (1 - Tb/Te) the optical depths at the positions of peak flux density in both the north-east and the south-west limbs were calculated to be r = 1.66 and 2.05, respectively (Te = and K). Following BD87's analysis, the optical depth can now be written E r,,= 21T135 ' v. where v is measured in GHz, E = JNe~ds is the emission measure in cm- 6 pc along the line of sight through the nebula and Ne and ~ are the electron and ion densities. Then at the peak flux density position on the new map, where r=2.05, E=2.67 x 10 8 cm- 6 pc. A tentative estimate of electron density can now be made using the definition of E. If it is assumed that Ne = ~ and the line of sight through the nebula intersects a single, constant density clump, then (1)

3 Opposite p. 816, MNRAS, 284 (a) Figure 1. The 6-cm radio maps of (a) NGC 7027 and (b) BD derived from combined MERLIN and VLA data. The elliptical beam sizes are 62 x 48 mas 2 and 82 x 78 mas 2 respectively. The scalebar is in units of!!1y per beam in each case. The contour levels in map (a) start at 5 x 10-5 Jy beam - I with linear increments of 1.0 x 10-4 Jy beam- I and those in map (b) start at 8 x 10-5 Jy beam- I with linear increments of 1.2 x 10-4 Jy beam- I RAS, MNRAS 284,

4 (b) Figure 1 - continued

5 Figure 2. HST images of (a) NGC 7027 through the 400A wide F547N filter and (b) BD through the narrow-band HfJ F487N filter, obtained using WFPC2. The resolutions are 0.11 arcsec and 64 mas, respectively. The scales are expressed in count pixel - I where one count is equivalent to (a) 3.83 x 10 - III Wm - 2 sr- I HZ - I and (b) 3.06 x 10-4 Wm -2 sr - I. The contour levels in (a) start at 20 count pixel - I with a linear 50 count pixel- I increment and in (b) start at 100 count pixel - I with a linear 100 count pixel - I increment.

6 (b) Figure 2 - continued 1997 RAS, MNRAS 284,

7 where As is the path-length through the clump. Taking As = X 10-4 pc (== 0.1 arcsec at a distance to NGC 7027 of 880 pc) then for the peak brightness position in the southwest limb, Ne=7.9 x 10 5 cm- 3 In practice, the true value of Ne may well be less than this if there are more emitting clumps down the line of sight. Middlemass (1990) used a number of diagnostic optical line ratios to derive a very consistent value of Ne cm -3 for NGC A similar analysis can be performed for BD In this case, the peak flux density measured from the new map is 7.07 x 10-4 Jy beam-i, measured in the northern limb. BD87 also recorded their peak flux density measurement at this position and derived an electron temperature of 5690 K. (When the new map was convolved to BD87's resolution, the peak flux densities agreed to within 10 per cent.) A value of r = 3.17 was derived for the optical depth at this point and hence the emission measure was calculated to be 1.62 x 10 8 CI?-6 pc. Again, taking As ==0.1 arcsec (7.3 x 10-4 pc at a distance of 1500 pc), the electron density can be estimated as Ne=4.7 x 10 5 cm- 3 In this case, BD87's beam would contain some 250 of the higher resolution beams and they detected only about 90 times as much flux. Here again, it is clear from Fig. 1 (b) that the structure of the bright shell of emission is much more clearly defined than in the corresponding map by BD87. It is difficult to estimate the depth along the line of sight of this shell, however it is apparent that if ~he bright ring is elongated in a uniform way, this elongatlon must be almost exactly along the line of sight. 3.2 Comparison to HST images The high spatial resolution achieved in the radio maps means that these maps are ideally suited to direct comparison with HST WFPC2 optical images, which have very similar spatial resolution. WFPC2 images of these two nebulae have been obtained from the HST archive! and are shown in Figs 2(a) and (b) opposite p The image of NGC 7027 was obtained through the 400 A wide filter centred on 547 nm and may include [NI], [01] and [Fen] emission lines as well as the stellar and nebular continuum. Most of the im~g~ fell o~ one ofthe three WF chips (pixel size 99 mas). ThiS image is undersampled since the WF CCD pixels are rather larger than the width of the nominal HST pointspread function (PSF) (45 mas). For the purposes of this ~aper the image resolution was derived (somewhat simplistically) to be 0.11 arcsec FWHM from the convolution of two Gaussians having FWHM equal to the pixel dimension and the nominal HST PSF width, respectively. Note that the diagonal line in the north-west comer of the image is a processing artefact from the main chip boundary. The image presented in Fig. 2(a) was flux calibrated in the usual way, assuming a continuum spectrum of constant flux density through the filter. The image of BD was taken using the narrow H{3 filter and the PC chip (pixel size 45 mas 2 ). This image is also somewhat undersampled; here, IBased on obs~rvations made with the NASAJESA Hubble Space Te~escope, o~tamed from ~he data archive at the Space Telescope SCience Institute. STScI IS operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS (2) Sub-O.l-arcsec radio maps o/young PNe 817 the image resolution was derived to be 64 mas FWHM. In this case, since the image is dominated by line emission, the flux calibration is in terms of the total flux observed through the filter rather than per unit frequency. The absolute pointing of HST is not as well defined as that of the radio maps and therefore the absolute alignment of the HST images with the radio maps was achieved by a process of trial and error. In each case, the position of the central star in the optical nebula was aligned with an estimate of the centre of the radio nebula, calculated from the mid-points between the brightest parts of the limbs of emission along the major and minor axes of each nebula. The trial alignments were assessed by eye and small adjustments were made to the position of the central star until radio and optical contours gave the best match, particularly across features that appeared in both wavelengths. The usual procedure to derive extinction maps is to convert the radio flux density into a theoretically equivalent optical emission-line flux density and compare the resultant map with a direct, flux-calibrated image in the same optical emission line. This is the method used by Walton et al. (1988). However, there are reasons why such an approach is not appro~riate here. As detailed in the previous section, a full analysis of the new data at high spatial resolution must await follow-up observations at other radio frequencies and therefore neither the electron temperatures nor the emission measures are determined at high spatial resolution across these targets. Moreover, in both targets it is concluded that the very high spatial resolution permits detection of small-scale, optically thick clumps. Here, as the total flux is not being collected, an accurate conversion to optical emission-line flux density cannot be made. It is also possible that. these radio maps may still be missing flux from large spatial features that are resolved out at this high resolution, although comparison of the total flux observed in each nebula here to the total flux measured in each using much lower resolution array configurations indicates that such missing flux is of the order of a few per cent at most. The conversion between radio flux density and emission-line flux density is dependent on r;..53. The derived Te maps of BD87 show that Te varies considerably across each of the nebulae and so a fully spatially resolved calculation would be required, otherwise further errors of 10 per cent or more would ensue. Finally, this method of producing an extinction map is strictly only applicable to the measurement of interstellar extinction. In both of these nebulae, it is clear that much of the observed extinction is due to local dust and therefore a three-dimensional model of the dust distribu ~ion throug~ ea~h nebula would be required to properly mterpret extmction maps. Although, for the reasons outlined above, it would not be particularly instructive to create new extinction maps from the high-resolution data presented in this paper, nevertheless it is still interesting to investigate in more detail the comparison between optical and radio images of these two PNe. The similarities between the two images of BD shown in Figs l(b) and 2(b) are striking. To aid the comparison, both images were first smoothed to the same resolution (83 mas) by convolving each with the appropriate Gaussian beam. The HST image was scaled arbitrarily to the radio map scale by matching the peak surface brightnesses in the small, bright, diagonal feature

8 818 M Bryce et al. apparent in both images just north of the centre of the western limb. Finally, the difference map shown in Fig. 3 was constructed by dividing the convolved radio map by the convolved and scaled optical map. Light tones ( < 1) in the difference map indicate an excess of optical brightness, or possibly optically thick radio emission and dark tones ( > 1) indicate an excess of radio brightness. Bearing in mind the arbitrary scaling, two features immediately become apparent. First, there is a gradient from east to west of increasing relative radio brightness, and secondly there is a pronounced patch of high radio brightness in the north-east corner of the nebula. These observations confirm the findings of Lame et al. (1995) who have compared Hoc and HfJ HST images with VLA radio images. The radio emission in this second feature is more than 10 times brighter than the (scaled) optical emission, and is almost certainly due to deep optical obscuration by a local region of dust. Close comparison of the difference map with the optical map shows that the entire northern limb contains irregular, small-scale dust obscuration, whereas the eastern limb is apparently free from localized dusty clumps at this resolution. The southern and western limbs show some localized variation of up to a factor of 2 but not nearly as extreme as in the northern limb. Arnaud et al. (1996) compare the [Sm]9532-A and [0 n]3727-a HST images and find that, particularly in the north-east corner of the shell, dusty clumps and filaments on scales of -0.1 arcsec cause enhanced obscuration of the shorter wavelength image. The overall brightness gradient may be an indicator of the inclination of this nebula, with the eastern, optically brighter side closer to the observer and the slightly obscured western side further away. The optical and radio images of NGC 7027 are superficially quite different and so it was not thought worthwhile to construct a difference map. Instead, the two maps were again convolved to the same resolution (0.11 arcsec) and the radio map contours were plotted against the grey-scale optical image (Fig. 4). It is clear from this plot that despite the ~~~~ '.. ~., ~.~. t "'f.;f '... ~t, '.,.. 01.~. - -,, ; "'- 0.., C'I." ~ -Z ~ z :::i 0 w c ~ ,, RIGHT ASCENSION (J2000) Figure 3. A difference map of BD made by dividing the radio map (Fig. 1b) by the HST image (Fig. 2b). Both maps were first convolved to a resolution of 83 mas and scaled to the same flux density using as an arbitrary reference point the diagonal feature on the west limb,

9 obvious differences between the two images, they are very similar in size and the radio contours frequently follow the optical features both at large scales, such as in tracing the southern corner of the bright shell, and at small scales, such as in the protruding finger of emission west of the bright knot. This would suggest that the obscuring dust in the nebula is very closely associated with the radio-bright shell. The bipolar optical structure seen in the HST image is likely to be mostly emission scattered from a dusty bipolar shell, whereas the bright knot may well indicate the point at which the dusty shell has been breached. The radio shell shows an extremely steep fall-off in emission along the outer edges of the north-east and (particularly) south-west limbs. It is likely that the material that has constrained the dusty shell illuminated in the optical image to a bipolar shape is also constraining the ionized, radio-emitting gas into a bright ring of emission, viewed almost sideways on, and hence the Sub-O.l-arcsec radio maps of young PNe 819 bright north-east and south-west limbs are in fact the limbbrightened sides of this ring. 4 CONCLUSIONS The radio maps presented in Fig. 1 are the first such images of PNe at sub-o.l-arcsec resolution to be produced. Using MERLIN data combined with VLA data it is possible to produce high spatial resolution 5-GHz radio maps of PNe that are very similar in spatial scale to HST WFPC2 optical images, thus permitting a direct comparison at high spatial resolution between optical and radio emission from these objects. The radio map of NGC7027 presented here, when compared to the WFPC2 image, shows the well-known pattern of extinction across the nebular ring, except in the region of the bright optical knot in the north-west corner. o z o ~ z o w c 12 :J ~ ~ ~~ ~ ~ ~ ~ L-~ RIGHT ASCENSION Figure 4. The HSTimage of NGC 7027 is shown as a grey-scale, with the same conversion factor to units of surface brightness as in Fig. 2(a). The contours are derived from the radio map; the first contour is at 4 x 10 ~4 Jy beam - J and the linear increment is also 4 x 10-4 Jy beam - J. Both images were first convolved to a resolution of 0.11 arcsec RAS, MNRAS 284,

10 820 M Bryce et al. What is interesting, however, is that the two images show remarkable similarities in many of the small-scale features that are resolved at high resolution. The radio and optical images of BD are superficially very similar, but the relative difference map presented here confirms the findings of Lame et al. (1995) and Arnaud et al. (1996) who showed that this nebula contains marked, small-scale, dusty clumps and filaments which are particularly prevalent in the northern limb of the main ring of emission. A multifrequency study using MERLIN will provide accurate electron temperatures and emission measures on scales of 0.15 arcsec: this is now underway. However, preliminary analysis indicates that both of the PNe discussed here show some optically thick emission at 5 GHz, with peak values of 't" = 2.05 and 3.17 derived for NGC 7027 and BD , respectively. This implies that in each case material is concentrated into small, dense clumps; tentative upper limits for electron density were derived to be Ne = 7.9 X 10 5 and 4.7 x 10 5 cm- 3, respectively. The radio maps of both of the PNe show small-scale structures that will be suitable for proper-motion studies on time-scales of a few years. ACKNOWLEDGMENTS We thank Nial Tanvir at the UK HST coordinating facility for his help and advice in processing the HST images shown in this paper, and also staff at the VLA centre in Socorro and at Jodrell Bank for their help in reducing the radio data. We are grateful to Yervaint Terzian and co-workers for allowing us to use their B-array data on NGC 7027 and to Pat Harrington for sending us the abstract of Lame et al. and the preprint by Arnaud et al. detailing their work on BD MB is in receipt of a University of Manchester research fellowship. This work relied largely on computing facilities provided by the Starlink project, and was based on radio observations made with MERLIN, a national facility operated by the University of Manchester on behalf of PPARC, and the VLA, a facility of the US National Radio Astronomy Observatory, which is operated by Associated Universities, Inc., under cooperative agreement with the National Science Foundation. REFERENCES Arnaud K, Borkowski K J., Harrington J. P., 1996, ApJ, 462, L75 Balick B., 1987, AJ, 94, 671 Basart J. P., Daub C. T., 1987, ApJ, 317, 412 (BD87) Hajian A H., Terzian Y., Bignell c., 1993, AJ, 106, 1965 Kawamura J., Masson c., 1996, ApJ, 461, 282 Lame N. J., Harrington P., Borkowski K, White S., 1995, BAAS, 187,8106L Masson C. R, 1989, ApJ, 336, 294 Middlemass D., 1990, MNRAS, 244, 294 Seaton M. J., 1979, MNRAS, 187,785 Walsh J. R., Clegg R E. S., 1994, MNRAS, 268, L41 Walton N. A, Pottasch S. R., Reay N. K, Taylor A R, 1988, A&A, 200, L21