Online material

Appendix A: J050343.02-664456.7

This section is dedicated to a more detailed analysis of the extremely red source J050343.02-664456.7. It is one of the reddest sources in our sample, and one of two objects detected only in the mid-IR, with no counterpart in the VMC catalogue – the other one being a YSO.

J050343.02–664456.7 is one of the 13 sources identified by Gruendl & Chu (2009) and generically classified –on the basis of their photometric SED– as EROs. Subsequent Spitzer/IRS spectroscopic observations of seven of them revealed that they are extreme carbon stars (Gruendl et al. 2008). The conclusion is based on the detection of SiC and C₂H₂ absorption and on the presence of a broad MgS feature in two cases.

Assuming that these objects are AGB stars, the derived mass-loss rates are higher than those of any known carbon-rich AGB star in the LMC. Gruendl et al. (2008) estimated a mass-loss rate Ṁ = 1.5 × 10^-4   M_⊙   yr^-1 for J050343.02–664456.7. From empirical relations based on mid-IR colours, Matsuura et al. (2009) found a lower value, viz. 6.4 × 10^-5   M_⊙   yr^-1. This object alone provides a contribution equal – or bigger – than the integrated mass-loss rate of all stars in our sample. In the following we present some indications that J050343.02–664456.7 may not be a normal AGB star.

J050343.02–664456.7 is not present in the VMC catalogue because it is detected only in the K_s-band. It is in fact completely invisible in the J band as well as in the bluer Y band, as shown in Fig. A.1. The vmcSource catalogue is built with sources detected in the three VMC bands (Cioni et al. 2011) and therefore J050343.02-664456.7 is not included. The K_s magnitude was obtained from the vmcDetection table, i.e. the catalogue corresponding to individual observations. Its value is K_s = 18.75 mag.

The photometric data points and the best-fitting model of J050343.02-664456.7 obtained using the standard procedure described in Sect. 3 are presented in Fig. A.2. The model substantially underestimates the flux in the bluest part of the SED, in the K_s and in the IRAC 3.6 μm bands. The predicted K_s magnitude is  ~9 mag fainter than the observed one. We tested our model also on the other EROs identified by Gruendl & Chu (2009) – all located outside the VMC tile and hence not included in our sample – and found that in most cases our AGB models could describe quite well the observed SED. As an example, in Fig. A.2 we show the photometric data and our best-fitting SED model for J055026.08695603.1, an EROs with a SED similar to the one of J050343.02-664456.7. In this case the model SED seems to generally better reproduce the observed data. We are therefore led to consider the possibility that J050343.02-664456.7 could be something different from an AGB star.

The SED of J050343.02-664456.7 could be explained assuming that this is a post-AGB star with a detached shell. In this case the flux in the near-IR would be due to the contribution of the central star. In our case this contribution is extremely low, and produces only a flatter SED in the near-IR, rather than a secondary peak in the near-IR (see, e.g., Lagadec et al. 2011). This implies that the central star stopped losing mass – i.e. it ended the AGB phase – very recently (see, e.g., van der Veen et al. 1989).

To explore the post-AGB hypothesis, we used our models setting the condensation temperature of the dust grains as a

free parameter. The resulting best-fitting model is shown in the lower panel of Fig. A.2 as a dotted line. It is compatible with the upper limit for optical and near-IR magnitudes corresponding to the MCPS and VMC detection limit and it shows a much better agreement with the observed SED than the standard AGB model. In the best-fitting model the condensation temperature is T_c = 330 K, corresponding to an inner radius of the dusty shell of  ~5 × 10⁴   R_⊙ or  ~3.5 × 10¹⁰ km. This value is extremely small, just 10 times bigger than the inner radius of the dusty shell obtained for the standard model. This implies that the mass-loss production in J050343.02-664456.7 has dropped to zero extremely recently – of the order of  ~100 years ago assuming a shell expansion velocity of 10 km s^-1.

To conclude, we showed that the observed SED of J050343.02-664456.7 is not fully compatible with a standard model for AGB stars, showing a flux excess at wavelengths shorter than  ~4 μm. We proposed that this source could belong to the – rare – class of objects in transition from the AGB to the planetary nebula stage. Further evidence supporting this idea could be found in the fact that some of the SEDs of the 13 EROs identified by Gruendl & Chu (2009) show hints of a secondary peak at near-IR wavelengths. We hence point out that it may be not so straightforward to classify all EROs as AGB carbon stars.

Appendix B: SEDs and best-fitting models for all stars

© ESO, 2012