4 The color-magnitude diagrams

   
4.1 Optical CMDs

Figure 9: UV color-magnitude diagram for Field 1 (left) and Field 2 (right). The arrow represents the reddening vector of E(B-V)=0.1. Typical photometry uncertainties are plotted on the right side. The solid line indicates the 50% completeness.

Optical CMDs probe young and intermediate age populations. Theoretically, the U and V light is dominated by young main sequence stars (Buzzoni 1995).

Figure 9 shows (U-V) - V color-magnitude diagrams for stars in both observed halo fields of NGC 5128. Due to better seeing the saturation magnitude of the V-band is $\sim$0.3 mag fainter for Field 2. Most of the stars redder than $(U-V)\sim0$ belong to our own Galaxy (see Fig. 7). The most important characteristic of the UV CMD of Field 1 is the upper main sequence, visible as the blue plume at $(U-V)\sim-1$ mag. By contrast, there are no such young massive stars in Field 2.

Figure 10: CMDs with foreground stars statistically subtracted. Overplotted are isochrones from Bertelli et al. (1994) for Z=0.004 with log(age) indicated with the number on the right side. The size of the reddening vector corresponds to E(B-V)=0.1 mag.

The Besançon Galaxy model (see previous section) was used to "clean'' the CMDs of foreground star contamination. In Fig. 10 we show UV CMDs after the subtraction of foreground stars. Overplotted are isochrones from Bertelli et al. (1994) for Z=0.004 and for log(age) = 7.0, 7.3, 7.5 for Field 1 and log(age) = 7.0, 7.5 and 7.8 for Field 2. The isochrones were shifted to the distance of NGC 5128, assuming a distance modulus of (m-M)_V=27.8 and reddening of E(B-V)=0.11 mag (Schlegel et al. 1998). The difference between the two diagrams is striking: the blue main sequence containing stars as young as $\sim$10 Myr that is present in Field 1 is completely absent from the CMD of Field 2. Well separated from the main sequence in Field 1 is the sequence of blue core-helium burning (BHeB) stars at $(U-V)\sim0.5$. The width of the gap and the tightness of the main sequence indicate low differential extinction in the field. In Field 2, the stars with 0.5<(U-V)<2.8 and brighter than have no corresponding main sequence stars and thus cannot be young stars migrating from blue to red during their core-He burning phase, as is the case for most of the objects in Field 1 with colors 0.5<(U-V)<2.5. These stars in Field 2 are most probably the remaining foreground contamination, indicating that the foreground contamination may affect the numbers of blue and red HeB stars in Field 1 (see next paragraph). Only a few stars lie along the isochrone of log(age)=7.5 (Fig. 10), while most of them are consistent with much older ages. We conclude that there are no stars younger than $\sim$40 Myr in Field 2.

Note that isochrones for metallicities higher than Z=0.004 extend on the red supergiant edge to redder values of U-V than the reddest stars in Field 1 and thus do not fit well our observations. The ratio of blue to red supergiants strongly depends on metallicity (Langer & Maeder 1995; Maeder & Meynet 2001) and in principle could be used to constrain the metallicity of the youngest population in NGC 5128. Counting the number of blue and red supergiants for stars more massive than $\sim$12 $M_\odot$, we find their ratio to be B/R<0.7-0.8 (although with high uncertainty due to the possible foreground contamination), in good agreement with the observed B/R value in the SMC cluster NGC 300 (see discussion by Langer & Maeder 1995). The metallicity of Z=0.004 (corresponding to [Fe/H]=-0.7 dex) is appropriate for the SMC. However, since the B/R value depends also on other parameters such as stellar mass, rotation and degree of overshooting (Maeder & Meynet 2001), a more detailed comparison with models and the determination of the metallicity of this youngest stellar population in NGC 5128 is warranted (Rejkuba et al., in preparation).

The low metallicity implied by the fit of the isochrones on UV CMDs probably reflects the metallicity of the gas left in the halo of NGC 5128 by the accreted satellite. Atomic H I (Schiminovich et al. 1994) and molecular CO gas (Charmandaris et al. 2000) present in Field 1 are slightly offset from the position of the diffuse stellar shell, the obvious remnant from the accreted galaxy.

   
4.2 Optical-near IR CMDs

Figure 11: VK color-magnitude diagram for Field 1 (left) and Field 2 (right). The dashed line identifies the 50% completeness level.

Figure 12: VK CMDs with foreground stars statistically subtracted. Overplotted are fiducial RGB sequences of Galactic globular clusters (from left to right): M15, 47 Tuc, NGC 6553 and NGC 6528, from Ferraro et al. (2000).

Optical-near IR CMDs probe old and intermediate-age stellar populations. Theoretically, more than two thirds of the light in K-band is dominated by cool stars on the red giant branch (RGB) and asymptotic giant branch (AGB), and by red dwarfs (Buzzoni 1995). The red dwarfs are too faint to be detected at the distance of NGC 5128, and thus our VK CMDs are entirely dominated by RGB and AGB stars (Fig. 11).

In Fig. 12 we show VK CMDs of Fields 1 and 2 after the statistical subtraction of foreground stars. Overlaid are fiducial RGB sequences of Galactic globular clusters (from left to right: M15, 47 Tuc, NGC 6553 and NGC 6528; Ferraro et al. 2000) spanning a large range of metallicities ($-2.17\le[$Fe/H$]\le-0.23$ dex). As before, we used a distance modulus of 27.8 and reddening corresponding to E(B-V)=0.1( E(V-K)=0.274 and A_K=0.0347; Rieke & Lebofsky 1985) to adjust the magnitudes and colors of RGB fiducials to those of NGC 5128. Obviously, most of the stars in Fig. 12 belong to the RGB. The right edge of the RGB is quite sharp, with most of the stars being more metal-poor than 47 Tuc ([Fe/H]=-0.71 dex) and none appearing to be as metal-rich as NGC 6553 ([Fe/H]=-0.29 dex). However, the latter is due to incompleteness in V-band photometry.

Figure 13: The histogram of the observed color distribution along the RGB (lower panels) is plotted accounting for the actual slope of the RGB (thick line in upper panels) in Field 1 (left) and 2 (right). The 1$\sigma$ spread in color of the RGB is indicated in the upper left corner. Overplotted with the dashed line is a Gaussian with the $\sigma =0.3$, typical of the distribution from the photometry errors.

The spread in color of the RGB is larger than the photometric uncertainties (Fig. 13), indicating the presence of spread in metallicity and/or age. The most metal-poor stars have metallicities of -2 dex if their ages correspond to those of Galactic globular clusters. The population we probe with VK photometry is more metal-poor than -0.7 dex. Our V-band images are not deep enough to detect more metal rich giants. Walsh et al. (1999) measured the mean oxygen abundance of five planetary nebulae in NGC 5128 to be [O/H $]={-0.5 \pm 0.3}$ dex, consistent with the presence of the large population of stars with metallicities below solar, as we observe in the VK CMDs.

There are 1830 stars in Field 1 and 1197 in Field 2 which have good K-band photometry ( $\sigma_K<0.5$, $-0.7<{\rm sharp}<0.7$ and $\chi<2.0$) above the respective K-band completeness limits, but which have not been detected in optical bands. They are uniformly distributed over all the ISAAC images. The large number of very red stars with good photometry in $K_{\rm s}$ and no counterpart in the V-band suggests that stars more metal-rich than [Fe/H]=-0.7 dex are present, as expected for a luminous giant elliptical galaxy. This is in good agreement with the results of Harris et al. (1999) and Harris & Harris (2000).