3.4. Stellar Populations and radial distributions

The stellar content of the outer region of M 33 is poorly explored. It is interesting to investigate if the different kind of stars identified in our CMDs share the same spatial distribution. To check this point we defined three selection boxes that are depicted in the F1 CMD shown in Fig. 7. Note that the boxes are defined in a range of magnitude in which incompleteness effects should be weak or negligible (M_I < - 3.0, e.g. more than 2 mag above the limiting magnitude) and cover similar magnitude ranges.

Figure 7. Description of the adopted selection boxes. The box around V - I = 0.0 selects a sample of young MS stars, the large box above the TRGB selects AGB stars, the small box below the TRGB selects RGB stars. The continuous lines are isochrones of solar metallicity, the open symbols are isochrones at [M / H] = - 0.7, from the set by Girardi et al. (2000). The ages are reported in the plot. The small dots are the observed stars of F1.

As a guideline, we have superimposed three isochrones of solar metallicity and age = 20, 40, 100 Myr (continuous lines) and three isochrones with [M / H] = - 0.7 and age = 2, 6, 12 Gyr (empty symbols) from the set by Girardi et al. (2000). The bluest box samples the upper MS, e.g. young stars with age 100 Myr. The large box above the TRGB samples the bright AGB stars, the smaller box samples the brightest RGB stars. Both kinds of tracers are associated with intermediate to old age stars, but are not necessarily linked thogether. In particular, bright RGB stars trace populations older than 1 - 2 gyr (see Salaris, Cassisi & Weiss, 2002, and references therein). In the following, R must be intended as the projected angular distance from the center of the galaxy.

In the upper panels of Fig. 8 the adopted selection boxes are superposed on the CMDs of F1 (left panel) and F2 (right panel). The lower panels of Fig. 8 show the cumulative radial distributions of the stars falling in the boxes in the two different fields. In the radial range covered by F1 the distributions of RGB and AGB stars are indistinguishable. On the other hand, MS stars appear much more centrally concentrated since their distribution seems to end at R ~ 17' -18', e.g. around 2 disc scale-lengths (Van den Bergh, 1991). According to a Kolmogorov-Smirnov test, the probability that the MS and RGB samples are drawn from the same parent population is P 0.04%. This suggests that the RGB and AGB stars are not associated with the young disc component traced by MS stars. At least a (significant) fraction of them should belong to a more extended galactic component.

Figure 8. Upper panels: the selection boxes described in Fig. 7 are superposed on the CMDs of F1 (left) and F2 (right). The corresponding cumulative radial distributions are displayed in the lower panels.

The radial distribution of MS stars is not reported in the lower right panel of Fig. 8 since, in agreement with the above conclusion, F2 is virtually devoid of stars populating the MS box. It is surprising to note that in this field, AGB stars appear to follow a significantly different distribution with respect to RGB stars. The distribution of AGB stars is less centrally concentrated and it is quite similar to uniform distribution on the sky. The latter fact would be naturally explained if the AGB sample of F2 would be dominated by foreground contamination (see also Davidge, 2003). According to the predictions of the Galactic model by Robin et al. (2003) this seems to be the actual case. The model predicts that the number of Galactic stars falling in the AGB box is 48, less than compared to the 61 actually observed. Hence 80 % of the putative AGB stars in F2 are likely foreground stars. The impact is much smaller on the F1 stars where 584 stars are observed in the AGB box, hence the fraction of foreground contaminants is < 10 %. On the other hand the expected number of foreground stars falling in the RGB box is ~ 4, e.g. negligible in both fields.

Therefore, the difference of radial distribution shown in the lower right panel of Fig. 8 is completely spurious. On the other hand, if we consider star counts and take into account the corrections for foreground contamination, it turns out that while the number of RGB stars drops by a factor 27 going from F1 to F2, the number of AGB stars decreases by a larger factor, 42. This suggests that RGB stars may follow a more extended distribution with respect to AGB stars at large radii.

All the above considerations seem to indicate that populations of different characteristic ages follow different distributions on the sampled scales, the older stars having more extended distributions. This is suggestive of the presence of a weak "classical" old halo component in M 33. The only previous indication in this sense (from field stars) is provided by the discovery of a few candidate RR Lyrae variables by Pritchet (1988), while all other hints of the existence of an old spheroidal stellar component come from the study of globular clusters (Sarajedini et al., 2000; Schommer et al., 1991). Note, however, that an extended and old disc component is also compatible with our observations (see T04).

In the upper panel of Fig. 9 we report the radial profiles of RGB and MS stars over the range(s) sampled by the present study. The MS profile is reasonably reproduced by an exponential law with scale-length h = 9.2' up to R 17', in agreement with the results by Kent (1987). However, at R = 19' the sharp drop of the density already observed and discussed in Fig. 8 is clearly evident. On the other hand the RGB profile is well fitted by an exponential law with h = 4.9' over the whole radial range sampled by F1. The density of RGB stars falls significantly below the adopted exponential profile in F2, suggesting a break in the observed profile in the range 20' R 25'. In our view, the most interesting result of this comparison is that, even ignoring the observed density cut-offs, the RGB and MS distributions do have significantly different profiles, again suggesting a different origin.

Figure 9. Upper panel: stellar density profiles for MS (open circles) and RGB stars (open squares) compared with two different exponential laws. All the estimates with R 20' are computed over sections of concentric annuli 2' wide. The RGB points at R = 27.6' is the density estimate obtained from F2 as a whole. Lower panel: expansion of the RGB density profile. The continuous line is the same exponential law displayed in the upper panel, the dotted line is a R1/4 law with Re = 9.2'. The associated uncertainty is the Poisson noise of the star counts propagated to the adopted unities.

The lower panel of Fig. 9 shows that in the radial range covered by F1 (10' R 20') the observed density profile of RGB stars is equally well fitted by the exponential law described above and by a R^1/4 law having effective radius R_e = 2.7' (in fact, the R^1/4 law provides a marginally better fit with respect to the exponential). It is interesting to note that the same R^1/4 law was found to provide a good fit also to the central bulge of M 33 (Boulesteix et al., 1980; Bothun, 1992), suggesting a possible connection between the bulge and the putative halo component (but see Stephens & Frogel, 2002, for a detailed decomposition of the inner profile). While suggestive, the above result is limited to the considered radial range (10' R 20') where the contribution of disc stars to the RGB population may be low. To correctly disentangle the contribution of the disc from that of the more extended component identified here, a complete sampling of the density profile from the center to the outskirts of the galaxy is needed, e.g. covering also the regions in which the surface brightness should be dominated by the exponential disc. This kind of analysis is clearly beyond the reach of the present study.