H Emission From Point Sources In The Magellanic Clouds

Use of New Filter

It is clear from the foregoing that an H survey of the LMC is needed to identify the many emission-line stars fainter than the 14.5mag limit of the BE survey, to detect the low excitation planetary nebulae not seen in the published [OIII]5007-based surveys, and to extend the planetary nebula searches to fainter limits. Some, but not all, of this is in progress with the CCD surveys mentioned above. The UKST is still the best telescope for obtaining complete surveys of the Magellanic Clouds because of its large areal coverage.

An H survey for emission-line objects cannot be done with the UKST and its objective prisms because the best dispersion available at H is 2050Å/mm which is many times poorer than that used in the other surveys. Nor does combining the new filter with this objective prism help because that would produce almost circular images of 2 arcsec in diameter in which no useful continuum would be discernible. What can be done is to use SuperCOSMOS or another plate measuring machine to measure direct on/off-band images in H and to search the data for objects that are bright on the H image. With the very crowded fields found in the MCs, the best method of analysis is not to use the object catalogues created by standard image deblending software, but to subtract one film or plate from another in pixel space using software similar to that used by Hawkins (1997) to coadd plates. This is now a practicable proposal with the large amount of computing power available. One difficulty is that there is no suitable wide-angle off-band filter available at the UKST. Nevertheless, the work can be done using the SERC SR Sky Atlas plates which include H but have a passband ten-times wider than the new filter. With this method, the SR plate is not at the same epoch as the H plate and comparison of the two will reveal variable stars as well as emission-line objects. This problem can be minimised by using the other UKST SR plates that are available for each Magellanic Cloud field but are not part of the official SR Sky Atlas, and also by using the SERC I Sky Atlas plates which will in turn be useful for selecting the planetary nebulae which are usually invisible in I and for identifying red stars. Ideally, a contemporaneous red exposure should be made on TechPan emulsion. This will achieve a better match in depth and resolution to the H exposure and will eliminate the problem caused by variable stars.

The increase in brightness caused by adding an emission line to a star with a flat spectrum (in -space) will be much greater on a plate taken through the new H filter than on the SR Sky Atlas plate, and this difference increases as the equivalent width (W) of the line increases, reaching a limit of 2.33mag. This is shown in Figure 2. Thus, if 0.1mag is a realistic photometric detection limit, it is expected that stars with W < 10Å will be detected. However, real spectra are not flat and smooth. The point at 0.2 on the abscissa of Figure 2 is the mean and range of magnitude differences between real stars in the SR and H bands, similarly normalized to a flat spectrum, for 16 spectra taken with FLAIR of stars in the spectral type range B2-M3. These differences are small and mostly negative (i.e. in the opposite sense to the effects of adding an H emission line). Typical H equivalent widths for emission-line stars are 30-60Å, with a significant number rising to several hundred Å (see Morgan, Watson & Parker 1992) and are clearly much larger than the limit set by the expected photometric errors and the effects of spectral type in normal stars.

 
Figure 2: Change in difference in magnitude between an H-emitting star and a neighbour seen when comparing an H film with an SR Survey plate, assuming both spectra are flat and smooth. The point at 0.2 on the abscissa is the mean and range of magnitude differences between real stars in the SR and H bands, also normalized to a flat spectrum (see text).

Once the survey is photometrically calibrated it can be used for providing large numbers of H fluxes for the planetary nebulae. To date, most of the brighter planetary nebulae have published H and [OIII]5007 photometry, but this is just a third of the whole population. Leisy et al. (1997) have used on the SERC J Sky Atlas to provide B broad-band magnitudes for all the LMC planetary nebulae, but these objects are extremely faint (14-21mag) in the broad B survey waveband, with respect to the neighbouring population. In relative terms they are many times brighter and consequently easier to measure in H light. These data could be used to extend many of the evolutionary diagnostic diagrams and provide a check on the interstellar reddening which could be subject to errors of wavelength dependent light loss in standard slit spectroscopy.

The principal uses of a large catalogue of early-type emission-line stars are twofold: (1) it will allow the delineation of structures of star-forming regions, and (2) it will be a source of many interesting objects that can be identified by comparing it with data from other wavebands (e.g. X-ray) or photometric colours. One example is the symbiotic star population which is still very small (Morgan 1992). Its importance in the Magellanic Clouds has been to provide absolute luminosities for the cool donors which, for the larger Galaxy population, are uncertain because of the unknown stellar distances. The Magellanic Cloud objects identified to date are bright giants probably on the AGB, but this could be a selection effect. It is important to search for fainter symbiotic stars. The MA survey yielded 59 candidate red stars with H emission, most of which still need to be observed in greater detail. An LMC survey would produce proportionately more.

A particularly useful approach would be to add the pixel maps of several exposures on the same centres to provide images that are much deeper than are available now, matching the depths reached by CCDs and extending over much larger areas.

© Copyright Astronomical Society of Australia 1997