Imaging Polarimetry and OH Maser Observations of the Envelope of IRAS 19114+0002

Tim Gledhill¹, Jeremy Yates¹, Antonio Chrysostomou^1,2, Anita Richards³, Motohide Tamura⁴

1: Dept. of Physical Sciences, University of Hertfordshire, Hatfield, AL10 9AB, UK

2: Joint Astronomy Centre, 660 N. A'ohoku Place, Hilo, Hawaii 96720, USA

3: NRAL, Jodrell Bank, Macclesfield, Cheshire, UK

4: NAOJ, Osawa 2-21-1, Mitaka, Tokyo 181, Japan

 
 
 

Introduction Proto-Planetary Nebulae (PPN) represent the transition phase in the evolution of 1-10M¤ stars between the AGB stage and their emergence as Planetary Nebulae (PN). The star is still surrounded by a remnant dusty envelope resulting from mainly spherically symmetric mass outflow along the AGB. This produces a large excess at NIR through to millimetre wavelengths1. Towards the end of the AGB phase, or the begining of the PPN transition phase, the mass outflow can become anisotropic leading to the formation of a flattened envelope. The onset of a fast wind is then thought to carve out polar cavities in the envelope resulting in the escape of scattered light producing bipolar nebulae2. This is nicely illustrated by recent HST observations of IRAS17150-3223 which show a bipolar reflection nebula superimposed upon a series of 8 concentric shells, produced by an earlier phase of spherically symmetric mass outflow3.  In May 1998, we obtained high spatial resolution imaging linear polarimetry of a sample of PPN, using IRPOL2 and IRCAM3 on UKIRT, all of which we find to be linearly polarised with polarisations ranging from 10-40%. The distributions of polarised flux range from obviously bipolar nebulae with axes that must be close to the plane of the sky (e.g. 17150-3224, 16342-3814) to objects that are either spherical shells or pole-on dust discs (e.g. 19114+0002, 17436+5003). There are intermediate objects showing the "polarisation discs" seen in reflection nebulae around YSOs4 (e.g 19500-1709, 20000+3239).

Here we present some preliminary results for IRAS 19114+0002 which has been classified as a PPN candidate on the basis of double peaked optical and NIR spectra1. However, a large radial velocity (100 km s-1) suggests a kinematic distance of ~ 6 kpc5 which, coupled with a CO outflow velocity of 33 km s-1 implies a more massive object with luminosity ~ 105 L¤, such as a red supergiant. So the exact nature of this late-type star is still open to debate. IRAS 19114+0002 has undergone a period of intense mass loss resulting in a circumstellar envelope of gas and dust. Analysis of MIR (8.5 - 12.5 microns) images of thermal dust emission6 show a slightly elongated ring-like envelope with inner and outer radii of 1.1'' and 1.9''. At NIR (J,H,K) wavelengths coronographic imaging polarimetry has been used to detect a more extended nebula in scattered light7, with an outer radius of ~10'' at J. The derived total masses are 7 M¤ in the inner nebula and 1 M¤ in the outer. Velocity information from molecular line mapping8 imply an envelope dynamical age of between 5-7 x 103 years.  Amongst the molecular emission detected towards 19114+0002 are the 1667 MHz and 1612 MHz maser lines of OH which were mapped using the VLA9 with a 1'' beam and 1.1 km s-1 veocity channels, to show a ring-like structure in both lines.

 

Figure 1: A linear polarisation map superimposed upon a logarithmically scaled surface brightness image in the K-band. The centrosymmetric pattern of vectors shows that 19114+0002 is surrounded by a circumstellar reflection nebula illuminated from the centre. Possible nebula geometries are a spherical envelope illuminated isotropically or a flattened structure (e.g. an equatorial disk) viewed pole on. The degrees of polarisation (up to 35% after correction for the underlying photosphere) at K suggest a component of large (non-Rayleigh) particles (~ 1 micron) within the size distribution.

 

New Results We have used UKIRT with the polarimetry module IRPOL2 (designed and built by the University of Hertfordshire) and IRCAM3 to obtain high spatial resolution imaging polarimetry in the J-, H- and K-bands. These observations map the distribution of scattered light in the dusty circumstellar environment and will provide information on the distribution of dust, the scattering geometry and the nature of the dust particles themselves. This technique is particularly valuable for the detection of faint nebular structure around bright objects, such as AGB stars and PPN. When imaged in polarised flux the unpolarised photospheric emission, which dominates the total flux, disappears to reveal an image of scattered light in the envelope.  To investigate the distribution and kinematics of the gas around IRAS19114 we have used the MERLIN interferometer to map the 1667 MHz and 1612 MHz OH maser lines. A restoring beam size of 0.4'' was used to allow easy comparison with the UKIRT data. A total of 256 velocity channels (of 0.35 km s-1 per channel) were used to obtain velocity information. In addition these data contain full polarimetric information (all 4 Stokes parameters).   Near-Infrared Observations In Figure 1 we show a K band linear polarisation map superimposed upon a logarithmically scaled surface brightness image of the star. Pixel sizes are 0.14'' with seeing of 0.7''. The regular centrosymmetric pattern is typical of illumination from a central point source with a radially symmetric distribution of scattering angles around the source. Such a geometry could be provided

by a spherical nebula or shell, or by a flattened structure, such as an equatorial torus, viewed pole on. If the latter then the tilt of the polar axis must be less than 10o. A more accurate analysis, including detailed subtraction of the PSF, will constrain this limit further and provide details on the nebula geometry and dust grain model.  The extent of the nebula in scattered light is clearer in Figure 2, which shows images of linearly polarised flux in the J- and K-bands. The dust responsible for scattering in the NIR is distributed in a ring-like structure about the star. The central few pixels are contaminated due to image alignment residuals and fluctuations in the PSF during the observations. Although the ring appears quite regular, there is evidence for clumpy structure in both wavebands. The distribution of scattered light appears very similar to the 12.4 micron emission, although there is little evidence in these data for elliptical structure6. Figure 3 shows radial profiles of the K-band image obtained by averaging over circular annuli. The total flux and polarised flux are shown, along with an estimate of the total flux in the shell, obtained by subtracting a scaled PSF. The maximum of scattered emission from the ring, as defined by the polarised flux, occurs at a radius of 1.8'' with inner and outer radii of 0.5'' (obtained by extrapolating the downward trend in polarised flux towards the centre) and 4.0''. The estimated total flux in the shell follows approximately the same distribution (allowing for the over-correction by the PSF at radii less than 1.5'').

 

Figure 2: Images of 19114+0002 in linearly polarised light in the J (top) and K (bottom) wavebands that show a circumstellar reflection nebula, apparently detached from the star. The flux in the central few pixels is an artifact of the PSF and residuals in the image alignment. FWHM seeing is 0.7'' with 0.14'' pixels.

Figure 3: Azimuthally averaged radial profiles (averaged over circular annuli centred on the star). The cyan line shows the surface brightness of 19114+0002 and includes contributions from both the stellar photosphere and scattered light from the circumstellar shell. The blue line shows a PSF star profile, aligned and normalised to the 19114+0002 flux peak. The red line shows the difference between the target and PSF profiles, revealing the intensity profile of the circumstellar shell. At radii less than 1.5" the PSF has over-corrected the target profile. The green line shows the linearly polarised flux. At radii less than 1.0" the rise in polarised flux is spurious and due to alignment residuals in the central few pixels.

 
 

OH Maser Observations In Figure 4 we show plots of the OH maser emission at 1612 MHz and 1667 Mhz. In each case the data have been stacked over approximately 40 velocity channels centred on the systemic velocity of 100 km s-1. The distribution of emission appears ring-like at both frequecies, but particularly so at 1667 MHz. Rather surprisingly both frequencies trace similar spatial structure, the 1612 MHz emission being more clumpy, possibly indicating a range of physical conditions along lines of sight through the envelope. The distribution of OH emission is also remarkably similar to that of scattered light in the NIR, showing that the gas and dust are co-spatial, although the dust ring is more extensive. Although there is no detailed corespondence between the 1612 MHz maser clumps and the dust clumps seen in Figure 2, we note that both data sets show evidence for a break in the ring structure to the SSW.

Our fine sampling in velocity (0.3 kms-1 wide channels) allows us to investigate the detailed kinematics of this object. Figure 5 shows plots of identified masing features in radius-velocity space, showing that the two lines have quite different velocity structure. The plotted curves are the results of constant velocity shell models, the solid curves have radii chosen to encompass the identified features, whereas the dotted curve is a "best fit" through the data. Neither data set is well fitted by a shell expanding with constant velocity, but we can draw some general conclusions; the 1667 MHz emission appears to trace a thicker shell of greater angular extent (~ 3.5'' cf. 2.5'') and greater velocity extent (v-vo = 63 km s-1 cf. v-vo = 55 km s-1). In the 1612 MHz  data, the systemic velocities appear at smaller radii than the extreme velocities, which is rather puzzling. There is also a notable enhancement in the red shifted emission, which was also seen in CO8.

 

Figure 4: MERLIN interferometer observations of the 1612 MHz (top) and 1667 MHz (bottom) OH maser lines towards 19114+0002. Each image shows a stack of approximately 40 velocity channels (0.3 km s-1 per channel) distributed about the systemic velocity of 100 km s-1. Material at the systemic velocity traces a ring-like structure which may be interpreted as a circumstellar shell or pole on torus. These images may be compared with the NIR imaging data (Fig. 2) and show that the dust and gas are approximately co-spatial.

Figure 5: Radius-velocity plots of emission features obtained by gausian fits to the 1612 MHz (top) and 1667 MHz (bottom) data. Fits of constant velocity shell models are shown as solid and dotted ellipses. The solid ellipses attempt to encompass the range of velocities seen, whereas the dotted ellipses are a "best fit" through the data. Neither data set is particularly well fitted by a spherical shell of constant velocity. The 1667 MHz emission appears as a thicker shell than the 1612 MHz emission and has a broader velocity range showing evidence for acceleration.

 

Conclusions and Further Work We have detected the ring-like circumstellar envelope of IRAS 19114 +0002 in both scattered light in the NIR and in OH maser emission at 1612 and 1667 MHz. The similarity of the spatial scales of our UKIRT and MERLIN data allows us to easily compare the NIR and radio to show that the circumstellar dust and gas are largely co-spatial in the plane of the sky. The velocity structure of the gas derived from the maser kinematics cannot be readily fitted by a simple expanding shell model. However, the radially symmetric appearance of the scattered light (and hence the dust distribution) suggests that any non-spherical structure (such as a flattened equatorial disc) must be viewed almost pole on. Only a preliminary analysis has been presented here. Future work will include:

* An axisymmetric scattering model to simulate the NIR polarimetric data allowing us to fit dust density models and determine key parameters such as radial density dependence, degree of asphericity, inclination to the line of sight and dust grain properties. * Radiative transport modelling using an axisymmetric code to model the SED to mm wavelengths. We have 450 micron and 850 micron photometry to constrain this fit. * A detailed investigation of the maser kinematics and polarisations using SMMOL (level population code) and a magnetic beaming code. We are looking forward to imaging this target (and the others in our linear polarimetry survey of PPNs) in circularly polarised light this semester.

References

1. Kwok, S. 1993, Ann. Rev. Astron. Astrophys., 31, 63

2. Frank, A. & Mellema, G. 1994, Ap.J., 430, 800

3. Kwok, S., Su, K.Y.L. & Hrivnak, B.J. 1998 Ap.J., 501, L117

4. Chrysostomou, A., Clark, S.G., Hough, J.H., Gledhill, T.M., McCall, A., Tamura, M., 1996, MNRAS, 278, 449

5. Zuckerman, B. & Dyck, H.M. 1986, Ap.J., 311, 345

6. Hawkins, G.W., Skinner, C.J, Meixner, M.M., Jernigan, J.G., Arens, F., Keto, E. & Graham, J.R. 1995, Ap.J., 452, 314

7. Kastner, J.H. & Weintraub, D.A., 1995, Ap.J., 452, 833

8. Bujarrabal, V., Alcolea, J. & Planesas, P. 1992, A.&A., 257, 701

9. Clausen, M., 1992 "Astrophysical Masers" Springer-Verlag pg. 353
 
 
 

Back to the Contents