3. THE GLOBULAR CLUSTERS

An important factor in the confusion about the nature of NGC 5128 was its apparent lack of globular clusters. By the late 1970s studies of galaxies beyond the Local Group had found globular clusters (GCs) in every galaxy searched, and shown that large E galaxies contained thousands of GCs (cf. Harris & Racine 1979 and references cited therein). But in the case of NGC 5128 no such GC population had been uncovered.

van den Bergh (1979) carried out star counts in annuli between 45" (~ 800 pc) and 484" (~ 9000 pc) (cf. Figure 2), finding an excess of only 15 ± 60 objects above background. An obvious explanation for the apparent lack of GCs in NGC 5128 was the dust lane and its signalling of a cataclysmic history. We now know that the GCs are present in numbers expected for an E galaxy in a small group (e.g. Harris et al. 1984, Harris et al. 2004b). But they are hard to separate unambiguously from the noise of foreground Milky Way halo stars and compact background galaxies. The two main reasons for this are proximity and low Galactic latitude. At a distance of 3.8 Mpc, 1' corresponds to ~ 1100 pc and, consequently, a survey to discover a widely distributed population like globular clusters must cover an area of more than two square degrees. Observationally this is a major challenge, but that is not the worst.

Figure 2. Star counts in NGC 5128 based on data from van den Bergh (1979).

But, for NGC 5128 this is nearly useless. Its brightest GCs should appear at an apparent magnitude of V ~ 17 while the peak of the GC luminosity function (GCLF) would be seen at a magnitude of V ~ 21. Because of NGC 5128's low Galactic latitude (19°) we see it through the Milky Way halo and disk and its mainly late type stars whose colours and apparent magnitudes will fall in the same range as GC candidates. Compact background galaxies which are fairly easily seen through the dilute halo may also have GC-like colours and surveys show that their numbers begin to increase rapidly at V ~ 20-21, just at the GCLF peak (Metcalfe et al. 2001). In short, the sample contamination problem is extraordinarily difficult to solve.

The GC mystery began to disappear with the discovery by Graham & Phillips (1980) of a globular cluster candidate ~ 8.2' northeast of the galaxy centre. This was followed quickly with the identification of six more (van den Bergh et al. 1981). All seven were initially selected by their slightly non-stellar appearance and then confirmed spectroscopically as having radial velocities consistent with membership in NGC 5128. Using Schmidt plates with a much larger field of view than the 4-m prime focus plate used by van den Bergh (1979), van den Bergh et al. (1981) also revisited the question of star counts. Their results, reproduced in Figure 3, show a clear excess of objects within ~ 375" of the nucleus; based on this they estimated the total cluster population at 600 ± 61. Later wide field surveys have estimated the total GCS population at ~ 1000-2000 (Harris et al. 1984, 2004b)

Figure 3. Surface density of objects plotted vs. distance from the centre of NGC 5128, adapted from van den Bergh (1979) (Figure 2); note the line labeled 4m indicating the area counted previously by van den Bergh (1979).

3.1. GCS Total Population

Once the existence of globular clusters in NGC 5128 was established, increased efforts were made to understand this system. But the observational issues described above have made it extremely difficult to build an overall picture. To illustrate this I have plotted a historical overview in Figure 4. Shown plotted vs. year are: estimates of the total GC population (van den Bergh 1979, van den Bergh et al. 1981, Harris et al. 1984, 2004b) and this paper); number of confirmed GCs (Graham & Phillips 1980, van den Bergh et al. 1981, Hesser et al. 1986, Rejkuba et al. 2001, Peng et al. 2004b, Woodley et al. 2005, 2007, Beasley et al. 2008, Woodley & Gomez 2010b); and, for comparison, the number of confirmed planetary nebulae (Jacoby, Ciardullo, & Ford 1988, Ford et al. 1989, Hui et al. 1993, Peng et al. 2004a, Walsh et al. 2010). Note that the number of known GCs reaches a plateau at ~ 100 in the late 1980s and remains virtually unchanged for more than a decade. This reflects the fact that the "easy" GC candidates which could be identified by a slightly non-stellar appearance had pretty much been found. And remember, adding more to the sample by photometry alone is an almost impossible task for ground based seeing, while the large angular scale of the system precludes HST imaging due to its small field of view. In their wide field imaging study Harris et al. (2004a) found ~ 10⁵ objects which fell in the colour and magnitude range populated by globular clusters in NGC 5128. If the total cluster population is ~ 2000, it is embedded in a S/N of 0.02; i.e. only about 2% of the detected star-like objects in the field encompassed by the halo of NGC 5128 are likely globular clusters. Recent increases in the number of known clusters have been achieved by the brute force approach of obtaining radial velocity spectra for the brightest candidates (Beasley et al. 2008, Woodley et al. 2005, 2007, 2010a) and the number of confirmed GCs is now 607. (Woodley et al. 2010a) fit a Gaussian to the luminosity function (LF) of their sample, finding a turnover of T₁ (to) = 19.44 mag. This is almost 1 mag brighter than the expected value and a clear sign that this sample is missing most of the fainter clusters. Because both the distance to NGC 5128 and the absolute magnitude of the standard globular cluster luminosity function (GCLF) turnover are well known we can force a fit to the cluster sample, assuming it is complete for the brightest clusters, and estimate the total population of the cluster system, N_t. With (m - M)₀ = 27.9 (Harris et al. 2010), A_V = 0.35 (Schlegel et al. 1998), and M_V (to) = -7.3 for giant ellipticals (Harris 2001) the GCLF turnover should occur at V = 20.95 or T₁ = 20.47. Fitting a Gaussian function with = 1.4, typical of gE's (Harris 2001) yields the result shown in Figure 5, from which we estimate N_t = 1300 ± 300.

Figure 4. Plotted are estimates of total GC population (blue), GCs confirmed by radial velocity and/or resolved by HST imaging, and confirmed planetary nebulae. If the estimated total number of GCs in NGC 5128 is correct (cf. Figure 5 and accompanying text), then we have now identified ~ half of the GC population.

This value is consistent with previous estimates, but is based on a more secure data set, especially for the brightest clusters, and stronger constraints on the distance to NGC 5128. From Figure 5 it is also clear that roughly half the GC population is now known, and that almost all of the undiscovered clusters will be fainter than V ~ 20 or T₁ ~ 19.5. Finally, we can calculate the specific frequency S_N = N_t 10^{0.4(M_VT + 15)} (Harris & van de Bergh 1981). Combining N_t = 1300 ± 300 and M_V^T = -22.1 (Harris et al. 2004b), gives S_N = 1.9 ± 0.5. Disk galaxies typically have low S_N 1.5, E galaxies have S_N ranging from 1-10 and even higher, while field E galaxies are on the lower end of this range with S_N 2 (Harris 2001). Thus S_N ~ 2 for NGC 5128 is consistent with a normal GCS in a small group.

Figure 5. Gaussian function superimposed on the NGC 5128 globular cluster luminosity function using data from Woodley et al. (2010a).

3.2. GCS Metallicities and Ages

Harris et al. (1992) obtained Washington photometry to derive metallicities for 62 GCs based on C - T₁ colours. The results (Figure 6) showed a broad, bimodal metallicity distribution with a mean [Fe/H]_{C - T1} = -0.8 ± 0.2. None of the clusters showed the extreme, greater than solar abundances found by Frogel (1984) based on JHK photometry and it appears likely that these values were due to the difficulty of calibration of colour-metallicity relations at such high abundances.

As the number of confirmed GCs has grown the "simple" bimodality of the MDF has become less clear (e.g. Harris et al. 2004a, Peng et al. 2004b, Beasley et al. 2008, Woodley et al. 2010a) and references therein). But the breadth of the GCS colour distribution and its high mean metallicity have not changed. In addition, we now have chemical abundances and ages based on high S/N spectra which are helping to disentangle the frustrating age/metallicity degeneracy endemic to purely photometric studies. Beasley et al. (2008) and Woodley et al. (2010a) have found that the majority of clusters, both metal-rich and metal-poor, are old (> 8-10 Gyr). The GCS has a subpopulation of younger clusters (< 3-5 Gyr) but virtually no extremely young members. The clusters are -enhanced above solar, but not as strongly as seen for Milky Way globulars.

3.3. Structure of GCs

The structures of Milky Way globular clusters are well described by single mass, isotropic King models (King 1962) and occupy a narrow range of parameter space similar to the fundamental plane for elliptical galaxies (Djorgovski 1995). McLaughlin (2000) has shown that the clusters have a common core M/L and that there is a strong correlation between core binding energy and cluster luminosity. Barmby et al. (2007) and McLaughlin et al. (2008) have shown that the GCs in both M31 and NGC 5128 follow the same fundamental plane for clusters as massive as ~ 3 × 10⁶ M, suggesting that old GCs have similar structural properties in all galaxies. Studies of the most luminous GCs in other systems (Hasegan et al. 2005, Rejkuba et al. 2007) suggest that they bridge the gap between lower mass (normal) clusters and compact dwarfs (UCDs) (Mieske et al. 2008). With a total population ~ 10 × that in the Milky Way, the NGC 5128 GCS is an important resource for exploring how structures and M/L ratios change for the most massive clusters.