2001AJ....122.1896F

We present a new method for the determination of [Fe/H] for globular clusters. This method is based on moderate-resolution (∼1500) near-IR spectroscopy in the K band of six to 10 of the brightest giants in a cluster. Our final [Fe/H] calibration is derived from spectra of 105 stars in 15 globular clusters with [Fe/H] ranging from near solar to -1.8. We measure the equivalent widths of three features in these spectra, EW(Na), EW(Ca), and EW(CO). Our calibration reproduces the Zinn & West abundance scale as updated by Harris to better than ±0.10 dex. A quadratic fit to the data is somewhat better than a simple linear fit. Three advantages of this new method are that it can be used for metal-rich, heavily reddened globular clusters in crowded fields; it does not require any knowledge of any other cluster or stellar parameter, such as reddening, distance, or luminosity; and it does not require much telescope time.

If stellar (J-K)₀ and M_K values are available in addition to the near-IR spectra, the accuracy of the [Fe/H] value derived is further improved. If observations–either near-IR spectra alone or spectra plus colors and magnitudes–of only three stars in a cluster are available, the resulting value of [Fe/H] will be nearly as reliable as the application of our new technique to 2 to 3 times as many stars. The accuracy of an [Fe/H] value based on observations of CO absorption by itself is significantly less than that which results from the three spectroscopic indices, in spite of the fact that the CO feature is by far the strongest feature in the spectrum. There are two probable reasons for this. First, the CO bands approach saturation for bright giants in high-metallicity clusters. Second, within a cluster there is considerable intrinsic star-to-star scatter in EW(CO) due to differences in mixing histories on the giant branch. In contrast, we do not find evidence for any intrinsic scatter in the Na or Ca indices. Nevertheless, we predict that observations from space of CO absorption in the integrated light of stellar systems will prove to be of great value for abundance determinations at least as distant as the Coma cluster of galaxies. Finally, a comparison of globular cluster abundances based on high-resolution spectroscopy with the Zinn & West/Harris [Fe/H] scale–and by inference ours–leads us to conclude that the two are closely linearly related over the entire range of globular cluster abundances, although there might be a small, constant offset between the two.