Our knowledge of the haloes of distant elliptical galaxies relies almost totally on measurements of their integrated light (cf. the classic study of NGC 3379 by deVaucouleurs & Capaccioli 1979). But the exceptional angular resolution and sensitivity of the HST changed all that. The first two-colour photometric study of an inner halo field in NGC 5128 (Soria et al. 1996) used images that were not very deep but did show the red giant tip and a broad colour distribution for the giant stars. The large angular size of NGC 5128 helps us in this kind of work, and we now have deeper photometry for four fields at distances of ~ 10, 20, 30, and 40 kpc or ~ 1.5 - 7reff (Harris et al. 1999, Harris & Harris 2000, Harris et al. 2002b, Rejkuba et al. 2005).
4.1. Metallicity of Halo Stars
The brightest stars in an old stellar population are the red giants which first appear at MI ~ -4.0. Figure 6, taken from Harris et al. (2002b), shows this in the form of fiducial red giant branches (RGBs) for Milky Way GCs covering a metallicity range of -2 [m/H] 0. From stellar model calculations (e.g. VandenBerg et al. 2000) we know the colour of the giant branch varies much less with age than metallicity. Therefore interpolating with RGB tracks calibrated onto the Milky Way giant branch (GB) grid provides a fast, efficient way to derive a first order metallicity distribution function (MDF). We use V, I photometry because it has a strong metallicity sensitivity for old stars; though not the best colour option, it represents a compromise between the best cameras available and the spectral energy distributions of these old stars.
Figure 6. Fiducial giant branches for Milky Way globular clusters ranging in [m/H] from -2.0 to ~ solar (Harris et al. 2002b).
The I, V - I colour magnitude diagram (CMD) for the outermost (~ 40 kpc) NGC 5128 field (Rejkuba et al. 2005) is shown in Figure 7. This is the deepest of the halo star datasets but shows much the same GB characteristics as do the other three: a well defined upper boundary, especially on the blue (metal-poor) side of the giant branch and a very wide range in colour. In Figure 8 are shown the MDFs for all four fields (Rejkuba et al. 2005), with the data for the ~ 20 and 30 kpc fields combined since they are essentially indistinguishable (Harris et al. 2002a). The RGB stars in NGC 5128's halo are predominantly metal-rich with a broad MDF that varies little with galactocentric distance. And, in all fields, there are almost no metal-poor stars!
Figure 7. Color-magnitude diagram for the outer halo of NGC 5128 (Rejkuba et al. 2005), dashed lines showing the TRGB measurement uncertainty of ± 0.1 mag; compare with GC giant branches plotted in Figure 6.
The lack of metal-poor stars out to almost 7reff was a surprise and naturally leads to the question: where is the classic metal-poor halo in this galaxy? A possible clue to this puzzle can be seen in the V, I CMD of a field centred ~ 30 kpc or ~ 12reff from the centre of the Leo elliptical NGC 3379. At a distance of ~ 10 Mpc the HST ACS field covers a wide range in galactocentric distance of 10 kpc or ~ 3 reff. Unlike what has been found in NGC 5128, the CMD for this field shows a significant metal-poor population along with the metal-rich stars that were expected (Harris et al. 2007b). When the image was divided into inner (closer to the galaxy centre) and outer (more distant), the CMDs for these two regions were distinctly different. As seen in Figure 9, the inner field shows a wide RGB colour range which almost disappears in the outer field CMD. In Figure 10 are plotted the radial distributions for the blue ([m/H] < -0.7) and red ([m/H] > -0.7] stars superimposed on the surface brightness data from deVaucouleurs & Capaccioli (1979); the blue population density falls off with radius as ~ R-1.2 and the red as ~ R-6.0.
Are we seeing here the transition to a classic metal-poor halo and, if so, why haven't we seen the same in NGC 5128? Possibly the reason is simply that the NGC 5128 data extend to only ~ 7 reff compared with the NGC 3379 field which covered a range of ~ 10.3-13.6 reff. Should we then expect the transition from a metal-rich to a metal-poor halo to occur at ~ 10 reff? Observational data which might help answer this question are limited, but a study by Kalirai et al. (2006) appears to have found such a transition for M31. Their V,I photometry for confirmed M31 halo stars shows a radial gradient in metallicity, with the metal-poor component dominating beyond r > 10reff. V,I photometry for a halo field in NGC 5128 beyond ~ 12 reff would be a simple way to test this.
4.2. Age of Halo Stars
The MDFs for the four halo fields described in the previous section assumed a constant and old (~ 12 Gyr) age. Elements of the CMDs such as the small numbers of stars brighter than the RGB tip and the lack of luminous blue stars suggest that this is a legitimate starting assumption. But we would like to have better age constraints for these data, and this was the primary motivation for the 40 kpc study (cf. Figure 7). A CMD that reaches the main sequence turnoff is out of the question at present, but it was possible to obtain data to the depth of the horizontal branch (I = 28-29). The resulting photometry is good enough to compare with CMD simulations and narrow the possible range in age and metallicity using red clump and asymptotic giant branch stars as well as those on the RGB. It now appears that the dominant population in this outer field has an age of ~ 10-12 Gyr and is probably combined with a small fraction of younger (~ 5 Gyr) stars.
A comparison between the GC and halo field stars (cf. Figure 11 of Harris 2010) shows some intriguing similarities. The metallicity peaks for the halo stars and the metal-rich GCSs are quite similar and current analysis indicates that both samples are mainly old with a small younger component. Simplistically this suggests that major star formation episodes at ~ 12 and 5Gyr may have occurred in NGC 5128.