Astron. Astrophys. 348, 418-436 (1999)

4. Colors

4.1. The photometry

Performing accurate photometry on the NGC 5128 GC candidates is difficult. These GC candidates are extended objects, not point sources, so it is not possible to determine their magnitudes simply by fitting a scaled PSF to each candidate. Since each GC candidate has a unique size and radial profile (parameterized by the concentration, c) aperture photometry with a single aperture will not return accurate magnitudes. Therefore, we elected to use the total flux in the best-fitting Michie-King model as the best estimate of its flux.

We converted the total instrumental fluxes to the standard Johnson-Cousins magnitudes using the prescription of Holtzman et al. (1995). The calibration equations we used are:

where C is the count rate in ADU/second after correcting for charge transfer efficiency (CTE) effects. We applied the CTE corrections of Whitmore & Heyer (1997) for a five pixel aperture. G is the gain ratio between the 14 e^- gain state (which was used for the calibration observations) and the 7 e^- gain state (which was used for the NGC 5128 observations). As discussed in Holtzman et al. (1995), reddening corrections need to be applied before the instrumental magnitudes are calibrated to the standard Johnson-Cousins system. This is because the WFPC2 filters have different band passes and effective wavelengths than those of the standard V and I filters. We adopted a foreground reddening in the direction of NGC 5128 of from the reddening maps of Burstein & Heiles (1982) and assumed (Taylor 1986; Fahlman et al. 1989). We used the extinction corrections for K0III stars from Table 12 of Holtzman et al. (1995) to obtain and .

Fig. 16 shows the distribution of colors for our 21 GC candidates, and for 62 spectroscopically confirmed GCs in NGC 5128 (HG92). We converted the HG92 colors to colors using (Geisler 1996). The VI magnitudes determined this way should be treated with caution since Geisler (1996) found that colors do not reproduce colors particularly well. There is only one object (#12, , ) which appears significantly bluer than the other GC candidates. Based on its color this object may be a young GC. Visual inspection of the WFPC2 images (see Fig. 5) shows that this object appears to be a normal GC in NGC 5128. The V- and I-band photometry of all of our GC candidates are listed in Table 7. The R-, J-, H-, and K-band photometry from AM97 are also given.

Fig. 16. The upper panel shows the distribution of colors for our 21 GC candidates in NGC 5128 while the lower panel shows the distribution of colors for 62 spectroscopically confirmed GCs in NGC 5128 from HG92. The colors have been corrected for reddening in the Milky Way in the direction of NGC 5128, but not for internal reddening in NGC 5128 itself.

Table 7. Photometry for the GC candidates in NGC 5128.

4.2. Reddening within NGC 5128

In order to estimate the amount of internal reddening within NGC 5128 in the direction of each GC candidate we determined the color of the background near each object. This was done by letting the mean background color be a free parameter during the Michie-King model fitting process. The expected unreddened color at the location of each GC candidate was then subtracted from the observed mean background color to get an estimate of the internal reddening. The expected color of the background that each GC candidate sits on was determined as follows.

From Fig. 5 of van den Bergh (1976) we derived

where D is the projected distance from the center of NGC 5128 in arcminutes. We estimate that the uncertainty in the values from Eq. 10 is . The colors were converted to colors by subtracting the adopted Milky Way reddening of and using

which we derived from the Galactic Globular Cluster Catalogue of W. Harris (1996). For each GC candidate in NGC 5128 we computed the expected background color using Eqs. 10 and 11 and subtracted this from the mean color of the unresolved background around each GC candidate. The internal reddening due to dust in NGC 5128 was then computed by subtracting the expected color from the observed color for each GC candidate. The internal reddenings, as well as the dereddened and values for each GC candidate, are listed in Table 8. The and values are corrected for both the Burstein & Heiles (1982) reddening and our estimate of the internal reddening in NGC 5128. The uncertainty in each reddening estimate is and negative internal reddenings were set to .

Table 8. The estimated internal reddening within NGC 5128 along the line of sight, the dereddened V-band magnitude, and the dereddened color for each GC candidate.

In order to check these differential reddenings, we repeated our calculation using the color of a typical galaxy with the same morphological type as NGC 5128. The Third Reference Catalogue of Bright Galaxies (de Vaucouleurs et al 1991) lists NGC 5128 as being an intermediate S0 galaxy with a morphological type of . Buta & Williams (1995) find a mean color for galaxies of based on a sample of 55 galaxies. This is within of the expected background colors derived from the observed color gradient in the outer regions of NGC 5128. If we assume that the unresolved light from NGC 5128 is the same as that from a typical galaxy, but has been reddened due to the presence of dust from the merger, then the excess reddening can be estimated by subtracting the mean color of a galaxy from the observed color of the unresolved background around each GC candidate. The differential reddenings that we obtain by assuming that NGC 5128 is a galaxy are consistent with those obtained using the van den Bergh (1976) color gradient.

The first problem with this method of estimating the internal reddening in NGC 5128 is that NGC 5128 is not a normal elliptical galaxy, but an elliptical galaxy that has undergone a merger with a small late-type spiral galaxy. Therefore, the unresolved light near the center will be a combination of light from the original galaxy and from stars in the captured spiral galaxy. In our derivations of the internal reddening we have assumed that the only changes in the color of the central regions of NGC 5128 are those due to dust, and ignored changes in color due to star formation and differences in the underlying stellar population. Buta & Williams (1995) list mean colors for late spiral galaxies of to 0.7, depending on the morphological type of the spiral. Therefore, the contribution from the merged spiral galaxy will cause the actual unreddened color of the unresolved background in NGC 5128 to be as much as a few tenths of a magnitude bluer than we have assumed. This will lead us to underestimate the internal reddening within NGC 5128 by a few tenths of a magnitude. It is possible that some of the GC candidates are being seen against regions of recent star formation in NGC 5128. This would also lead us to underestimate the internal reddening by a few tenths of a magnitude.

A second problem is that the structure of the dust features in NGC 5128 changes on spatial scales of less than . The mean tidal radius of the GC candidates is (se), and the surface brightness of the background was determined while fitting the Michie-King models by taking the mean value of the pixels that fell within the fitting box ( pixels) but beyond the tidal radius. Therefore, small-scale spatial structure in the dust lane could introduce errors of several tenths of a magnitude in our estimates of the internal reddening in NGC 5128.

Finally, the reddening corrections are not being made in the WFPC2 photometric system. This may introduce small errors in the reddening corrections that we determine. We believe, however, that these errors will be small compared to other uncertainties in the method. In light of these uncertainties we believe that it is possible that our estimates of the differential reddening for each GC candidate are only accurate to mag.

Fig. 17 shows the color distribution of the GC candidates after correcting for internal reddening within NGC 5128. There appear to be three populations in the upper panel of Fig. 17. The largest population is centered at and contains 16 of the 21 GC candidates. This population appears to be similar to the spectroscopically confirmed GCs of HG92 and probably represents a genuine old population of GCs in NGC 5128. The similarity between the colors of these GC candidates and the dereddened colors of the Milky Way GCs suggests that this population is not heavily reddened and therefore probably lies on the near side of NGC 5128. The second component is the four objects with . These objects are probably GCs that are being seen through large amounts of dust in NGC 5128. Alternately, they could be background galaxies that have been misidentified as GC candidates. The third component is the single blue object with . An examination of the image of this object (Fig. 5) suggests that it is a legitimate GC candidate so its blue color makes it the best candidate for being a young GC in our sample.

Fig. 17. The upper panel shows the distribution of colors for the GC candidates in NGC 5128 after correcting for the estimated internal reddening due to dust in NGC 5128. The middle panel shows the distribution of colors for the GC candidates without any correction for internal reddening within NGC 5128 (but with a correction for Galactic reddening in the direction of NGC 5128). The lower panel shows the distribution of dereddened colors for the Milky Way GC system.

4.3. Iron abundance

The iron abundance for each GC candidate was estimated using

which was determined from 45 Galactic GCs with low reddenings and good metallicity estimates, and from 13 metal-rich GCs in NGC 1399 (Kissler-Patig et al. 1998).

If the blue color of object #12 is due solely to its iron abundance, then it has . This implausibly low iron abundance suggests that age is primarily responsible for at least some of the blue color, supporting the notion that #12 is a young object. The mean iron abundance of the 16 GC candidates with is (se), which is within of the value found by HG92. This suggests that the colors of these objects are consistent with them being old GCs, or moderately reddened intermediate-age GCs. The four objects with all have iron abundances of greater than the Solar value (). This suggests that the red colors of these objects are primarily due to dust along the line of sight within NGC 5128, not high iron abundances, although we can not rule out the possibility that at least some of these GC candidates have high iron abundances.

© European Southern Observatory (ESO) 1999

Online publication: July 26, 1999
helpdesk.link@springer.de