3 Cluster parameters

The CaII triplet lines ($\lambda\lambda8498,8542,8662\mbox{\,\AA}$) have been used as metallicity indicators for star clusters in the Galaxy and the LMC, both in integrated spectra (BA87, AZ88) and individual stars (Olszewski et al. 1991).

  

Table 2: Equivalent widths (Å)

In order to determine metallicities we measure equivalent widths of the CaII triplet and compare them to those of the high Signal/Noise Galactic globular cluster templates G5 to G1, ranging from $\mbox{$[Z/Z_{\odot}]$}=-2.0$ to nearly-solar metallicity, from Bica (1988). In the case of G1, the Bica (1988) template has been complemented with the M31 globular clusters of comparable metallicity G222 and G170 (Jablonka et al. 1992) in order to further improve the Signal/Noise ratio. A comparison of an observed cluster spectrum with that of the most similar (reddening-free) template gives the reddening value. The metallicity, in turn, is determined by means of a calibration of the Ws in terms of $[Z/Z_{\odot}]$.

For the continuum tracing we chose one which minimises TiO contamination, similar to the low one in BA87. CMDs of bulge clusters often present underpopulated giant branches (GB), e.g. Terzan2 (Ortolani et al. 1997a) and Tonantzintla2 (Bica et al. 1996). As a consequence, stochastic effects in sampling the cool giants often occur, so that the near-infrared integrated spectrum may present strong (Tonantzintla2) or weak (Terzan2) TiO bands (Fig. 6). On the other hand, CaII arises in all spectral types from F to M (Alloin & Bica 1989), so that it is nearly independent of stochastic effects due to temperature, which makes it a reliable metallicity indicator in the integrated spectrum. Nevertheless, stochastic effects on the giant branch may affect the Caii triplet significantly, due to the strong gravity dependence (Jones et al. 1984; Alloin & Bica 1989).

The adopted continuum points are at $\lambda\approx8372\mbox{\,\AA}$ and $\lambda\approx8838\mbox{\,\AA}$,with a pivot point at $\lambda\approx8584\mbox{\,\AA}$, if necessary. For the three CaII lines the windows are from BA87: 8476 - 8520Å, 8520 - 8564Å and 8640 - 8700Å. The resulting Ws are listed in Cols. 2-4 of Table 2, together with their sum (Col. 5). Typical errors were estimated from uncertainties in the continuum level and resulted in the range 5 - 10%. For comparison and calibration purposes, CaII triplet Ws were also measured in the same way for the templates G1 to G5 (Table 2).

3.1 Reddening

Ideally, the best spectroscopic method to determine reddening for star clusters is to have available reddening-free template spectra of similar properties, and to use a wavelength baseline as wide as possible, as in the recent study of globular clusters in NGC5128 by Jablonka et al. (1996). Although the present wavelength range is limited to the near-infrared, it can be used to constrain the reddening values.

Guided by the sum of Ws of the CaII triplet ($\Sigma W$) we identify among the templates that resembling most the cluster spectrum. Subsequently, a Seaton (1979) law was applied to derive E(B-V). This method is illustrated in Fig. 4 for Terzan5 and Terzan9 with corresponding templates G1 and G3, respectively. The templates and derived E(B-V) for each cluster are given in Table 3. We present in Figs. 5 and 6 reddening-corrected spectra for the bulge sample. Notice that in the very reddened clusters UKS1 and Liller1 (respectively E(B-V)=3.1 and 2.8) a strong absorption feature occurs at $\lambda\approx7550-7850\mbox{\,\AA}$. We emphasise that the present spectra are corrected for the telluric A band (Sect. 2). A similar strong feature was observed in the extremely reddened (E(B-V)=4.4) open cluster Westerlund1 (Piatti et al. 1998). This feature appears to be a diffuse interstellar band (Sanner et al. 1978) and the present results suggest that the band is detectable in spectra of globular clusters more reddened than E(B-V)=1.5, and becomes prominent for ${E(B-V)}\geq2$ (see Figs. 4, 5 and 6). The diffuse interstellar band occurs in the same spectral region as TiO stellar bands and they are certainly blended in several spectra. However, the TiO absorption at $\lambda7100-7300\mbox{\,\AA}$ might be used to infer the relative proportions of the absorptions in the region $\lambda\approx7550-7850\mbox{\,\AA}$.

  

Figure 4: Top panel: comparison of the reddening-corrected Terzan9 spectrum to the $\mbox{$[Z/Z_{\odot}]$}\approx-1.0$ G3 template. Bottom panel: comparison of the reddening-corrected Terzan5 spectrum to the $\mbox{$[Z/Z_{\odot}]$}\approx0.0$ G1 template, including Galactic and M31 globular clusters

  

Figure 5: Reddening-corrected spectra for bulge clusters

  

Figure 6: Same as Fig. 5 for the remaining bulge clusters in the sample

The strong-lined clusters projected on the disk result of $\approx$ solar metallicity (Sect. 3.2), and their reddening-corrected spectra are shown in Fig. 7. The templates used in the matchings were the metal-rich globular cluster templates G1 and G2 (Table 3). Similar to CMD studies, which so far have not been able to establish the real nature of these objects as globular clusters or very old open clusters, owing to crowding and reddening effects, the present spectroscopic results are not conclusive, but the spectra are consistent with those of very old clusters.

  

Figure 7: Same as Fig. 5 for clusters projected on the disk

The bulge clusters HP1 and NGC6717 with 2 spatial extractions are shown in Figs. 8 and 9 respectively (see Sects. 4.5 and 4.20).

3.2 Metallicity

The correspondence between the metallicity scales $.\!\!^{\rm h}$ and $[Z/Z_{\odot}]$ is not a trivial issue, rather it is a fundamental matter of debate (for a recent discussion of metal-rich globular clusters see Bruzual et al. 1997). For the sake of simplicity, we adopt in what follows the original scales given by each author. To calibrate our data we employed a $[Z/Z_{\odot}]$ scale.

We calibrated $\Sigma W$ as a function of metallicity adopting $\mbox{$[Z/Z_{\odot}]$}=0.0, -0.5, -1.0, -1.5\ {\rm and} -2.0$ respectively for the templates G1, G2, G3, G4 and G5. This calibration is based on individual $[Z/Z_{\odot}]$ of clusters in each template and a grid of Ws as a function of metallicity (Bica & Alloin 1986a,b; BA87; Bica 1988). The metallicity scale is similar to that of ZW84 and the ones in Jablonka (1992) and Jablonka et al. (1996).

The control clusters NGC6528 and NGC6624 (Table 1), as well as 47Tucanae, have been used in the calibration, adopting metallicity values from ZW84. The near-infrared spectrum of 47Tuc is from Bica et al. (1992). The resulting calibration curve is shown in Fig. 10, and the derived $[Z/Z_{\odot}]$ values for the sample clusters are given in Col. 4 of Table 3.

The abundance calibration of the metal-rich end is a fundamental problem in stellar populations. The calibration problem has been recently addressed by means of high dispersion spectra of individual stars by Barbuy et al. (1992, 1997), deep colour magnitude diagrams by Ortolani et al. (1995), and integrated spectra (Santos et al. 1995), in the study of the nearly-solar metallicity key globular clusters NGC6553 and NGC6528 (which are part of the G1 template). The iron abundance $.\!\!^{\rm h}$ in such clusters appears to be somewhat under solar ($\hbox{$.\!\!^{\rm h}$}\approx-0.2/-0.3$), whereas [$\alpha$-elements/Fe] are enhanced, resulting in an overall metallicity $\mbox{$[Z/Z_{\odot}]$}\approx0.0$, as herein adopted for the G1 template.

Typical errors in $[Z/Z_{\odot}]$ implied by uncertainties in the Ws are $\approx\pm0.15$ dex. However, owing to the fact that these bulge regions are very crowded, contamination effects by field stars on the cluster and background regions, uncertainties may be larger for some clusters. The present values should be taken as indicative, and for definitive results for each cluster complementary information from deep CMDs and spectroscopy of individual member giants would be necessary.

  

Table 3: Derived parameters

• $[Z/Z_{\odot}]$ obtained from the calibration of $\Sigma W$ of the CaII triplet (Fig. 10). $\dagger$ - important contamination by metal-rich bulge stars.