7 Intermediate age population

In contrast with the conclusions of HST studies (Harris et al. 1999; Harris & Harris 2000) of stellar populations in NGC 5128, we detect not only old population II stars, but also a significant number of stars with magnitudes brighter than the tip of RGB ( $M_K^{\rm Tip} = -0.64\,(\pm0.12){\rm [M/H]} - 6.93\,(\pm0.14)$; Ferraro et al. 2000).

We took bolometric corrections from Bessell & Wood (1984) and the empirical fit of (V-K) vs. $T_{\rm eff}$ from Bessell et al. (1998) to transform our K-(V-K) CMDs to the theoretical plane (Fig. 16). Overplotted on the H-R diagrams are the Padova tracks from Girardi et al. (2000) for the first-ascent giant branch stars with masses M=0.6, 0.8, 1.2 and 1.6 $M_\odot$ and metallicities Z=0.004 (full lines) and Z=0.008 (dashed lines). The sharp cut-off on the right side of the H-R diagrams is due to incompleteness in V-band.

In discussing the H-R diagram, we should consider blending. According to theoretical predictions (Renzini 1998), at the fiducial galactocentric distances of 9, 12.8 and 17.9 kpc, the number of blends consisting of 2 stars belonging to the tip of RGB is $\sim$1300, 300 and 55 stars (per $2.2\hbox{$^\prime$ }\times 2.2\hbox{$^\prime$ }$; i.e. field of view of ISAAC), respectively. In this calculation,

Figure 16: The HR diagram for old and intermediate stellar populations in Field 1 and 2. Overplotted are evolutionary tracks from Padova (Girardi et al. 2000) for stars with masses M=0.6, 0.8, 1.2 and 1.6 $M_\odot$ and Z=0.004 (full line) and Z=0.008 (dashed line).

the B-band surface brightness measurements from Mathieu et al. (1996) were used. Obviously, the innermost regions of the galaxy, at the distance <10 kpc from the nucleus are too crowded to give accurate photometry for RGB stars and most of the stars above the tip of RGB.

We attempted to use the NIC3 HST images overlapping with our Field 2 (galactocentric distance $R_{\rm gc}\sim9$ kpc) data to improve the resolution. However, we found out that many faint stars visible on ISAAC images are completely within the noise of NIC3 data. The resolution of our ISAAC data in the best seeing is as good as that of NICMOS images, with the advantage of having much better S/N and a more stable PSF. The confirmation of the number of AGB stars in Field 2 can be obtained only through the analysis of the LPVs (Rejkuba et al. in preparation).

In Field 1 ( $R_{\rm gc}\sim14$ kpc), our outer shell field, crowding is not that severe. The expected number of two-star blends at the RGB tip in this field ranges from 300-50, as the surface brightness drops across the field. The total number of such blends is therefore less than $\sim$200 in the most pessimistic calculation. On the other hand, the number of stars above the tip of RGB ( $M_{\rm bol}\sim-3.7$; Girardi et al. 2000) is 768. Subtracting the number of possible blends and allowing for a few of the brightest and bluest stars ($\sim$30 stars with M_K<9) to be the remaining foreground contamination, there are still more than 500 stars whose position in the H-R diagram and CMD is consistent with an intermediate-age AGB population.

In the inner halo field, Field 2, the number of stars above the tip of RGB is 2844. After subtracting the number of expected blends ( 1500), the number of AGB stars is 2.5 times larger than in Field 1. This confirms the presence of gradients in the intermediate-age population within the halo of NGC 5128, as suggested by Marleau et al. (2000).

The intermediate-age AGB population could have easily been missed in V and I-band HST studies due to small field of view and small (<0.5 mag) optical magnitude difference between the tip of the RGB and the tip of AGB. The AGB stars are up to $\sim$2 mag brighter in M_K and $M_{\rm bol}$ than RGB stars and thus are easily detectable in near-IR. Thanks to this, Marleau et al. (2000) could detect some AGB stars in the much smaller NICMOS field.

The brightest stars in M 32, in the Galactic bulge and the bulge of M 31 have similar brightnesses, reaching $M_{\rm bol} = -5.5$(Freedman 1992; Elston & Silva 1992; Frogel & Whitford 1987; Rich & Mould 1991). Due to this similarity, Davidge & van den Bergh (2001) suggested that the tip of AGB could be used as a standard candle for determination of distances. In Fig. 17 we present the $K_{\rm s}$-band and bolometric magnitude luminosity functions.

Figure 17: K-band (left) and bolometric magnitude luminosity functions (right) for stars redder than $(V-K_{\rm s})>2$ in Fields 1 and 2.

The tip of the AGB is observed at a bolometric magnitude of -5 in both fields (in spite of the crowding in Field 2), consistent with the adopted distance modulus for NGC 5128. Most of the stars brighter than this magnitude (as well as stars brighter than M_K = -9; see left panels in Fig. 17) probably belong to our Galaxy.