It is widely recognized that globular clusters (GCs) are cornerstones for the
solution of a large variety of problems concerning the formation and evolution
of galaxies. They are among the oldest objects formed in the Galaxy and, as a
consequence, can be used as tracers of the early chemical evolutionary phases
of the galactic environment. In principle, as a first guess, one can consider
the total metal abundance (traditionally indicated by the ratio
[Fe/H]
) of a
globular cluster as representative of the original composition of the gas from
which it formed, and the variations in [Fe/H] among clusters as a fossil
recording of the chemical enrichment history occurred in the Galaxy. It is
nowadays well known, however, that this simple view is complicated by other
phenomena: the chemical composition of the atmospheres of stars can be altered
by processes occurring during their evolution (see
Smith 1987, and Kraft 1994
for reviews). Even if the actual mechanisms are not completely understood yet,
it seems that surface abundances of only the lightest element (C, N, O, Na, and
Al) are affected, while Fe abundances are unchanged. In a following paper we
will discuss in detail this aspect.
In the present paper we explore the possibility of building a new metallicity scale for galactic globular clusters using only [Fe/H] values obtained from fine analysis of high dispersion spectra. Although globular cluster distances restrict applicability of this technique to the brightest giants, it provides direct accurate and quantitative determination of the actual metallicity of stellar atmospheres.
It is rather surprising that despite the advent of CCD detectors and of sophisticated and efficient spectrographs, the increasing number of measures of cluster metallicities has been (and is!) almost totally ignored in a variety of astrophysical problems involving this parameter. The most widely used metallicity scale for globular clusters is in fact the one obtained by Zinn & West (1984; hereinafter ZW) and Zinn (1985) from a calibration of integrated parameters of globular clusters. The main advantage of using integrated parameters is that they can be easily measured even for the most distant objects in the Galaxy: homogeneous results can then be obtained for almost all known galactic clusters. However, integrated parameters are not directly related to metal abundances, and their use as abundance indices requires an accurate calibration in terms of the actual content of [Fe/H]. Reflecting uncertainties present at that epoch in abundances from high dispersion spectra, ZW attributed very low weight to direct abundance determinations for globular cluster stars when they constructed their metallicity scale for globular clusters.
Since ZW work, a number of accurate high-resolution spectroscopic determinations of metal abundances for stars in globular clusters appeared in the literature. However, only the work of Gratton and coworkers (Gratton et al. 1986; Gratton 1987; Gratton & Ortolani 1989: G86, G87, G89 respectively; collectively G8689) was aimed at a systematic determination of abundances for a large number of clusters (spectra for giants in 17 clusters were actually analyzed). However, since completion of the Gratton and coworkers survey, there have been significant progresses both in the analysis techniques and in observing facilities. In fact the new Kurucz (1992, hereinafter K92) model atmospheres allow an homogeneous comparison between solar and stellar abundances, a major drawback of former analysis of abundances for globular cluster stars (see e.g., Leep et al. 1987). Furthermore, improvements in high resolution spectrographs and detectors allow better spectra to be obtained for a larger number of stars: a major contribution has been done by observations with the Hamilton spectrograph at Lick by Kraft, Sneden and coworkers (Sneden et al. 1991; Kraft et al. 1992; Sneden et al. 1992; Kraft et al. 1993; Sneden et al. 1994; Kraft et al. 1995; hereinafter, SKPL1, SKPL2, SKPL3, SKPL4, SKPL5, SKPL6 and SKPL on the whole, for brevity). Regretfully, a homogeneous abundance scale based on high-resolution spectroscopic data does not exist yet, mainly due to inconsistencies in the model atmospheres and in the atomic parameters adopted in the various investigations. Moreover, the need for such an improved scale is continuously growing, due to the high degree of accuracy required by a variety of problems, one for all: the long-debated calibration of the absolute magnitude of the horizontal branch in terms of the metal abundance. Minor variations in the adopted metallicities could result, ultimately, in non-trivial changes in globular clusters ages.
Our goal is to exploit recent observational and theoretical progresses to construct a new, reliable metallicity scale for globular clusters, completely based on high-quality data for red giant stars in 24 clusters. To this purpose, we obtained new data for a few stars in three southern clusters using the Long Camera mounted on the ESO CASPEC spectrograph, and reanalyzed published equivalent widths (EWs), mainly from the Gratton and coworkers and Lick surveys, integrated by other sources of similar quality for clusters not included in those studies. These data were analyzed in a totally self-consistent way, allowing a modern calibration of the abundance indices considered by ZW.
In Sect. 2 we present the new data, and in Sect. 3 the data adopted from literature. In Sects. 4 and 5 we discuss respectively the atmospheric and atomic parameters required for the abundance analysis. Our results will be exposed in Sect. 6, together with a discussion of the error sources. After a comparison with previous works (Sect. 7), we present our conclusions in Sect. 8.
Table 1: Observed spectral regions
Table 2: Program stars observed in 47 Tuc, NGC 6397 and NGC 6752