Astron. Astrophys. 325, 585-600 (1997)

2. Observations

2.1. Sample

For our search for NIR counterparts, we reduced the list of nearly 200 best-candidate sources from paper I to a list of 68 sources, excluding stars that were studied in paper II (except for one source). Our selection criteria are either IRAS flux densities , or and Jy. Limited by the available observing time, we finally observed 31 sources. They were selected mainly amongst the sources that were relatively bright at 12 µm, had red [12]-[25] colours, and were not affected by cirrus at 60 µm. We adopt [12] and [12]-[25] , with and the flux density in Jy in the IRAS 12 and 25 µm bands, respectively (IRAS Explanatory Supplement 1988). In the IRAS [12] versus [12]-[25] magnitude-colour diagram shown in Fig. 1, these 31 sources are represented by shaded circles, whereas the remaining 37 sources are plotted as open circles. Several sources coincide in this diagram, because of the discrete values for the IRAS flux densities. The diagram shows that we mostly selected the brighter and slightly redder sources. We also plotted the AGB star candidates (dots) and red supergiants (crosses) from paper II. The present sample is fainter at 12 µm and has a redder [12]-[25] colour on average than the AGB star candidates from paper II, that are themselves fainter at 12 µm and have redder [12]-[25] colours than the supergiants.

Fig. 1. IRAS 12 µm magnitudes versus IRAS [12]-[25] colours for the sample of IRAS sources that we selected as targets for observation in the NIR (open circles), the targets that we actually observed (shaded circles), the AGB star candidates from paper II (dots), and the red supergiants from paper II (crosses)

2.2. J and K-band imaging photometry

We observed on the clear nights 1/2 and 2/3 January 1996, using the NIR camera IRAC2 at the ESO/MPI 2.2m telescope at La Silla, Chile. Lens C was chosen to cover a field of view of , with a pixel scale of . This combines the advantages of a large field of view and a reasonable sampling of the point spread function for doing photometry. For each IRAS point source field we did a sequence of 12 images in the K-band filter, each consisting of ten 3-second exposures, shifted by in right ascension with respect to the previous image. We repeated this procedure backwards, using the J-band filter. The search is deepest in the centred at the IRAS point source. We compared the IRAC2 fields with the Digitized Sky Survey (http://archive.eso.org/dss/dss), to determine the actual field centres. The absolute pointing of the telescope was found to be accurate to about , from a comparison of the IRAC2 field centres and the IRAS positions used to point the telescope.

We constructed frames that represent a good approximation of the background emission, by median-averaging the shifted images within each sequence, rejecting the brightest pixels to avoid contamination by stars. The background-subtracted images were flat-field corrected using flatfields obtained by taking an image of a screen in the dome, illuminated by a lamp dedicated for this purpose, and subtracting a similar image with the lamp off. The individual images were shifted to bring the position of the IRAS point source back in the centre, by integer number of pixels to ensure flux conservation. Finally these images were added together. In this way, we could detect stars down to limiting magnitudes of mag, and mag. The limiting magnitude varied from field to field by about a magnitude, mainly due to differences in the background.

The standard stars were observed by taking an image consisting of an average of thirty 0.6-second exposures, and another image shifted by . Their difference yields a background-subtracted frame, which we then flat-field. The standard stars were observed regularly during the night, covering the same air masses as the LMC.

The J and K-band images were blinked to identify the J-K reddest sources. On these, circular aperture photometry was done with an increasing aperture radius, to create a radial magnitude profile of the star. The same was done for the standard star. From the differential magnitude profile the magnitude of the program star can be estimated, as well as its accuracy. The standard stars used are HD38150 ( mag, mag), SA94-702 ( GSC 00048-00918; mag, mag), and HD52467 ( mag, mag). These magnitudes are in the SAAO system (cf. Carter 1990), and the IRAC2 magnitudes have been converted to the SAAO system using relations derived by Lidman (1995) on a single test night:

The 1- errors are 0.012 and 0.015 on the coefficients for the and magnitudes, respectively. The conditions during the observations were photometric. Extinction corrections were found to be 0.05 mag air-mass^-1 in the J-band, and 0.03 mag air-mass^-1 in the K-band on the first night, and twice as large on the second night. The relative humidity, dome temperature, and seeing were about 60-70%, C, and 0.7- on the first night, and about 40-60%, 15- C, and 0.7-1 on the second night.

2.3. BVRi-band imaging photometry

We observed on the clear night 24/25, and the partially cloudy night 30/31 December 1996, using the direct imaging camera at the Dutch 0.9m telescope at La Silla, Chile.

On the first night, we imaged two fields centred on the galaxies that we detected in the NIR close to the IRAS point sources LI-LMC0603 and LI-LMC1818 (see below). The night was photometric, although the Moon was close to opposition. We took six images in the Bessel B-band, and three images in the Bessel R-band, all 200 seconds integration time per frame. Stellar images on these frames had FWHM of typically - . The images were calibrated by stars in the Landolt (1992) standard star field SA98 that were observed close in time and airmass to the galaxy fields. Hence the B and R magnitudes are on the photometric systems of Johnson and Kron-Cousins, respectively. We obtained integrated magnitudes for the galaxies and the few redmost stars in the fields, using aperture photometry and applying aperture corrections derived from the much less crowded standard star fields.

On the second night, we measured the bright stellar counterpart (see below) of LI-LMC1821 in the Bessel B, V, and R-bands, and the Gunn i-band. The night was cloudy, and useful photometry could only be obtained by rapidly switching back and forth between the star and Dor (= HR2194), an A0 Main Sequence star of V and B-V mag at 1.5 degree separation, until stable and well-defined magnitudes could be determined. The B, V, and R magnitudes that we obtained for LI-LMC1821 are on the Johnson photometric system, whereas the i magnitude is on the Gunn photometric system. We estimated the Gunn i-band magnitude for Dor to be i mag. Integration times were 0.3, 0.2, 0.2, and 0.5 seconds for BVRi respectively. To avoid problems with shutter delay times for these short integration times, we used the same integration times for both stars, and they were always positioned at the same place on the CCD.

© European Southern Observatory (ESO) 1997

Online publication: April 28, 1998

helpdesk.link@springer.de