One of my on-going research interests concerns the dust shells of proto-planetary nebulae or PPNs (also called pre-planetary nebulae, transition objects, and post-AGB objects in the literature; the latter term also applies to a wider selection of objects such as RV Tauri stars), especially the carbon-rich ones that show the unidentified "21 micron' emission feature. A good general review of the topic of proto-planetary nebulae is given in Sun Kwok's 1993 Annual Review of Astronomy and Astrophysics article, although its getting a bit dated.
These stars are thought to be caught in a brief evolutionary phase between the asymptotic giant branch (AGB) and planetary nebula (PN) phases of evolution. During the AGB phase the star is an M-type giant, is a pulsating variable star similar to Mira, and is losing material from its surface in stellar wind. During the PN phase the atmosphere of the star has all been ejected in the stellar wind, the stellar core has been exposed as a hot white dwarf star, and the core is ionizing the previously ejected material to form the nebula. One of the things that I find interesting about these PPNs is that the stars are evolving on rather short timescales, and we can hope to observe changes in the objects over the period of a human lifetime.
To find PPNs we look for objects with properties that are intermediate between AGB stars and PNs: the AGB stars are very cool stars, M-type stars or carbon stars, often have signiicant dust shells around them, and are long period variable stars; PNs have hot stars illuminating the surrounding material, have only cool dust around them, and the stars are generally not variable. So we look for stars of intermediate spectral type (K-type to B-type), with evidence of a remnant cool dust shell and which are not as strongly variable as Mira variables. Such stars are candiate PPNs. A list of possible PPNs can be found here.
The tricky part is to distinguish the candidate objects from either massive evolved stars or young stellar objects, both of which may have some similar characteristics to PPNs. The clearest examples of PPNs are carbon-rich objects that have elemental abundance analysis indicating that they have had elements produced in nuclear reactions mixed into the atmosphere from the stellar core. Carbon stars show such evidence of nuclear reactions which are not expected in other types of stars. Without such analysis we cannot be certain of which stars are really PPNs.
Below I show some ground-based thermal infrared images (taken at Gemini North or Gemini South) of a small subset of these PPNs which have an unidentified dust feature, the 21 micron feature (actually it peaks at 20 microns, but originally I thought it peaked at ~21 microns in the low resolution discovery spectra), of which only 14 are clearly identified in the Galaxy. The Gemini images use the colour to code the intensity. The images were taken at a wavelength of 11.7 microns, about 20 times the wavelength that people's eyes can see.
These images sometimes show resolved structure, which is decidedly not spherically symmetric. It appears to be in this phase of the evolution where the circumstellar shell is altered from a roughly spherical spherical shape to a rather bipolar shape. The resolved objects in the above images mostly appear to have a shape like an incomplete toroid, just observed at different orientations.
The images below are each normalized to peak intensity:
Its more interesting in some ways to plot all the images on a common intensity scale (in Jy/square arc-second, in the case of the images presented here) since the sky surface brightness is a quantity that is independent of distance:
There is a surprising range of brightnesses among these objects in this filter, which is dominated by the UIR features, commonly attributed to small polycyclic aromatic hydrocarbon (PAH) molecules. To even see the fainter objects in the figure one has to use a square-root intensity scale:
The differences in brightness are a function of the dust temperature and the dust optical depth along the line of sight for the different objects.
We can model the dust shells around these objects, but adding the images of thermal dust emission adds a lot of constraints to the modeling process. Given only photometry, or only photometry and spectroscopy, there are always a number of ways to fit the available data depending on how one goes about the calculations. This is a result of redundancies among various of the input parameters.
Recently we have discovered ~15 similar objects in the Large and Small Magellanic Clouds. This is due to the SAGE project. The properties of this small group of sources are a bit different than those for the Galactic objects shown above. I am working to find a complete sample of proto-planetary nebulae in these satellite galaxies from the millions of detected stars. This is something we cannot do for the Milky Way Galaxy because the distances to the objects are generally either totally unknown or at best very poorly known.