3.1. Identification with IRAS sources
The positive and tentative identifications, non-detections, serendipitous detections, and galaxies are listed in Table 1, together with the IRAS flux densities and NIR magnitudes, where applicable. The numbers of the stars refer to the LI-LMC numbers in the Schwering & Israel (1989, 1990) catalogue of IRAS point sources in the direction of the LMC. A suffix b, c, or d was added in the case of field stars or galaxies, neither of which were considered to be the counterpart of the quoted LI-LMC source. Positive and tentative identifications were found by first selecting the object in the field of view with the reddest J-K colour. Stars with J-K mag were considered to be the counterpart of the IRAS source, as no two such red objects ever appeared in the same field of view. In the absence of objects with J-K mag, objects were considered to be the counterpart of the IRAS source, if they either had J-K mag (i.e. cool and/or reddened stars, see e.g. Bell 1992) or were very bright (e.g. LI-LMC1821), and not too distant from the position of the IRAS source (i.e. within ). The counterparts with IR colours similar to those of mass-losing AGB stars have been enumerated under positive identifications, as this can be considered a confirmation of their identification with the IRAS source. The other counterparts have been enumerated under tentative identifications, and probably are stars with detached CSEs.
Table 1. Names (see text), NIR positions, separations of IRAS and NIR source (in arcsec), J- and K-band magnitudes and J-K colours, IRAS 12- and 25 µm fluxes (in Jy) and - colours for the objects in our sample. We adopt - , with and the flux density in Jy in the IRAS 12 and 25 µm bands, respectively (IRAS Explanatory Supplement 1988). 1- error estimates are given for the NIR photometry.
LI-LMC0109 was first detected in November 1995, using the NIR photometer at SAAO. IRAS05003-6712 was already known to be a mass-losing AGB star in the LMC (paper II). LI-LMC0530 is identified with SHV0510004-692755, an LPV with a period of 169 days (Hughes & Wood 1990, who measured J mag and K mag). We identify LI-LMC1721b with SHV0547489-704450, an LPV with a period of 264 days (Hughes & Wood 1990, who measured J mag and K mag): LI-LMC1721 is not associated with this LPV, because the NIR counterpart that we detect for this IRAS source is much redder than the LPV. The same is true for LI-LMC1721c, whereas LI-LMC1721d is too faint to be the LPV. We identify LI-LMC0937b with SHV0523536-700128, an LPV with a period of 381 days (Hughes & Wood 1990, who measured J mag and K mag). The IRAS source LI-LMC1759 lies at the edge of a small open cluster of perhaps a dozen stars within diameter, and may therefore not be related to any individual star. The galaxies are discussed below.
3.2. Serendipitous detections
We have calculated the separations of the IRAS point sources and the NIR sources detected with IRAC2 (Fig. 2: accumulated starting at infinity, and normalised to the total number of sources). The IRAS position is generally accurate to 5-15 arcsec, depending mainly on the 12 µm flux level, but can be off by more than half an arc-minute (paper II).
The reddest sources with J-K (bold line) are undoubtedly the IRAS counterparts. The combined area of our 31 fields covers square degrees. The area of the LMC covered by the selection in Paper I measures square degrees. Only if there were of the order of such NIR-red sources would one expect to have detected one serendipitously. There are of the order of IRAS detected (post-)AGB star candidates (paper I). Hence the population of stars with J-K that are not detected by IRAS would have to be at least as many as the IRAS-detected stars in the LMC. The number of planetary nebulae (PNe) in the LMC is estimated at 1100 (Pottasch 1984). The PN lifetimes of years are very similar to the lifetimes of the mass-losing AGB phase (less than years). Assuming that all mass-losing AGB stars will eventually form PNe, we estimate the number of mass-losing AGB stars in the LMC to be at most, arguing against the possibility of a population of LMC stars with J-K .
The bluer sources (J-K ) that we consider as positive or tentative identifications (thin line in Fig. 2) could be contaminated by NIR stars that are not associated with the IRAS source. But these cannot be many, as seen from the similarity of the separation distribution to that of the reddest sources. The serendipitously detected NIR sources (field stars) and the galaxies have very similar separation distributions, much broader than the separation distributions of the positive identifications. This is indicative of their not being the IRAS counterpart.
A field of view of (the deep area only) implies an expected mean separation of for a serendipitous detection. The mean separation for our serendipitous detections is , and for the galaxies. We suspect that the detection probability near the edges is somewhat lower. Essentially all our detections are situated in the deep area. The mean separation is for the reddest, and for the bluer positive and tentative identifications. Replacing 2 or 3 of the reddest stars with separations of by the same number of stars but with separations of , we generate a separation distribution with a mean of . Hence we expect there may be 2 or 3 serendipitous detections in the group of positive and, more likely, tentative identifications with J-K . This is too small to affect the conclusions that we reach in the present study.
3.3. Galaxies as a probe of the interstellar extinction inside the LMC
We discovered a few galaxies, which are probably not related to the IRAS point sources. Since the galaxies are located behind the LMC, they could, in principle, be used as probes of the interstellar reddening through the LMC. If the J-K colour excess of mag is due to extinction by dust inside of the LMC, a visual extinction of mag is indicated; the extinction through the entire LMC would vary between and mag. This could have severe consequences for the observation of stars inside of the LMC. Although Oestreicher et al. (1995) showed the foreground reddening towards the LMC to be only on average, stars inside the LMC could suffer from visual extinction of a few magnitudes.
Are the measured colours of the galaxies intrinsic to them, or have they been severely affected by interstellar reddening through the LMC? The intrinsic colours of a galaxy depend on its kind. The magnitudes for the galaxies approximately represent the total integrated light. There is no sign of steep colour gradients, although the bulges or nuclei appear somewhat redder. LI-LMC1818d (Fig. 3) is a bright edge-on spiral galaxy. LI-LMC0603c (Fig. 4) is a face-on spiral galaxy (A) interacting with another galaxy (B), and possibly with a third galaxy (C). All other objects in the images of LI-LMC1818d and LI-LMC0603c are unresolved. LI-LMC1759d is probably a spiral galaxy seen under a small inclination angle, but its position close to the edge of the field severely degraded the quality of its image. LI-LMC1803b is small and barely resolved.
Normal galaxies are confined to a colour J-K 0.8-0.9 mag (e.g. Glass 1984; Silva 1996). Only LI-LMC1759d has a colour consistent with a normal galaxy. LI-LMC1818d may suffer from extinction by dust in its disk because of its edge-on orientation, but the red colour of the near-face-on LI-LMC0603c cannot be explained easily by extinction inside of a normal galaxy.
Galaxies with anomalous IR colours include emission line galaxies. Whitelock (1985b) presented NIR data on a sample of thirteen IRAS galaxies, probably H II galaxies. Their mean J-K colour was 1.25 mag, with a standard deviation of only 0.13 mag. But the J-K colours of our reddened galaxies are still consistent with the reddest H II galaxies, while both Seyfert 1 and 2 galaxies can reach J-K colours in excess of 2 mag (Glass & Moorwood 1985; Almudena et al. 1996; Kotilainen & Ward 1994). Active galaxies usually have strong colour differences between the nucleus and the rest of the galaxy, something which we do not see in our galaxies.
There is less doubt about the question whether our galaxies could be the counterparts of the IRAS sources in the field. The large distances between the galaxies and the IRAS positions and/or the presence of a NIR-redder point source in the field already suggest that the detection of the galaxies was a mere coincidence. Also, the mid-IR colours of galaxies detected by IRAS are different from those of our sources. The IRAS galaxies discussed by Whitelock (1985b) have IRAS colours typical of cold dust, with S60 /S25 ratios . This would have lead to 60 µm flux densities for our IRAS sources of -6 Jy, whereas the measurements indicate they can only be Jy at most. Whitelock's galaxies have K magnitudes of -12 mag, which is considerably brighter than our galaxies, whereas their IRAS 25 µm flux densities are comparable. Hence we cannot exclude the possibility that our reddened galaxies have mid-IR excess emission below the sensitivity of IRAS, but they are not the counterparts of the IRAS sources.
The B and R-band photometry for the red sources in the fields of LI-LMC0603 and LI-LMC1818 is presented in Table 2. The stars that we identified as the NIR counterparts of the IRAS point sources LI-LMC0603 and LI-LMC1818 were not detected down to mag in the R-band. The other NIR-red objects do indeed have optically red counterparts, both the stars and the galaxies. Moreover, we discovered two very red stars in the Dutch telescope field of LI-LMC0603c, and a red galaxy in the Dutch telescope field of LI-LMC1818d, all three which are outside the corresponding IRAC2 fields. The B-R colour of mag for the two red stars LI-LMC0603d and LI-LMC0603e implies severe inter- or circumstellar extinction (Whitelock et al. 1996). The red B-R colour of galaxy LI-LMC1818e is remarkably similar to that of the galaxy LI-LMC1818d. Active galaxies are often optically blue (Véron-Cetty & Véron 1996 adopt B-R = 0.57). They can be optically red, though, as is the case for IRAS galaxies (P. Véron, private communication; see also Duc et al. 1997).
Table 2. Positions, R-band magnitudes, and B-R colours for the red sources in the Dutch telescope fields of LI-LMC0603 and LI-LMC1818.
We have taken the Lauberts-Valentijn ESO catalogue of galaxies (Lauberts & Valentijn 1989) to investigate the variation in galaxy B-R colours across the sky in the vicinity of the LMC. We had to sample in four square degree bins in order to obtain useful statistics. We identified a region of 64 square degrees centred at Right Ascension and Declination, that was relatively well populated (148 galaxies), and that appeared to be representative for the colours of galaxies unaffected by extinction through the Magellanic Clouds system. This region yielded a mean B-R mag, with minimum and maximum B-R of 0.70 and 1.38 mag respectively. We can compare this to the bins containing our galaxies LI-LMC0603c and LI-LMC1818d. Each of these bins contained one galaxy from the above mentioned catalogue. Although this is statistically very poor, it is interesting to note that the galaxy near LI-LMC0603c has B-R = 1.74 mag, and the galaxy near LI-LMC1818d has B-R = 1.44 mag, both considerably redder than the estimated typical colour. Comparing the B-R colours of the galaxies LI-LMC0603c and LI-LMC1818d with the canonical value of B-R , we arrive at extinctions of mag.
If we compare the positions of our galaxies with the dust column density maps by Schwering (1989), we notice that the dust distribution in the LMC is rather patchy as compared to the SMC. Two of our reddened galaxies are situated at the East side. Although they lie outside of the area covered by the dust map, it is at the East side that there appears to be a massive dust complex. Some of the serendipitously discovered red stars may thus be normal stars, situated behind a local, but large amount of dust in the LMC. Alternatively, they may be the nuclei of normal galaxies as seen through the LMC. In that case, the dust causing the reddening of the galaxy colours may be situated behind the main body of the LMC, leaving the colours of the stars in the LMC unaffected.
The colours and morphologies of the galaxies, the fact that we detect only a few galaxies of which most are red, and the presence of other red objects in their projected vicinities all provide circumstantial evidence for severe interstellar extinction inside or behind the LMC. However, it is based on very little data, and should be confirmed by the systematic study of the colours of galaxies seen through the LMC.
© European Southern Observatory (ESO) 1997
Online publication: April 28, 1998