Annu. Rev. Astron. Astrophys. 1981. 19: 77-113 Copyright © 1981 by Annual Reviews. All rights reserved

2.2. Age-Metallicity Relation

Stellar abundances in the solar neighborhood have been reviewed by Pagel (1979a), to which the reader is referred for a brief account of the methods used. Relatively unbiased surveys such as that of Bond (1970) or an inspection of the Bright Star Catalog (B.S.C.) indicate that the vast majority of stars have values of "metallicity," i.e. [Fe/H] (where [X] denotes the logarithm of any quantity X in a star minus log X in the Sun), that have a Gaussian distribution with mean - 0.15 and standard deviation 0.2, with only 3 stars out of about 10⁴ in the B.S.C. forming a separate population (extreme Population II) with -2.6 [Fe/H] - 1.3 and extreme kinematic properties, which may be shared by all late-type stars having [Fe/H] < - 0.6 (Eggen 1979). A catalogue of spectroscopic [Fe/H] determinations has been published by Cayrel et al. (1980). The existence of "super metal-rich" (SMR) stars (Spinrad & Taylor 1969), with very strong CN, Mg I, and Na I, of which the best-known prototype is the giant µ Leo, has aroused much controversy (Peterson 1976, Bonnell & Branch 1979 and references therein, Deming 1980, Pritchet & Campbell 1980). Our inclination is to believe the concordant results of high-dispersion spectroscopy in the infrared (Branch, Bonnell & Tomkin 1978) and photoelectric narrow-band spectrum synthesis studies (Williams 1971, Gustafsson, Kjaergaard & Andersen 1974, Edmunds 1976), which indicate 0.3 [Fe/H] 0.6 for this star. The existence of SMR stars is also supported by observations in visual binary systems (Deming & Butler 1979). However, they are neither extreme enough (in contrast to the halo stars) nor numerous enough to affect the approximate statistical description given above and they are not present (as was once suspected) in the galactic clusters M67 and NGC 188 (e.g. Pagel 1974 and references therein, Janes 1974, Griffin 1975, 1979, Cohen 1980a, b); their importance lies more in the implications of the inclusion of µ Leo-like stars in stellar population syntheses to account for the integrated spectra of the nuclear regions of large galaxies. The intrinsic metallicity dispersion of the nearby stars, after age effects, abundance gradients, and observational errors have been allowed for, is believed not to exceed 0.15 dex (Mayor 1976).

Eggen, Lynden-Bell & Sandage (1962) studied the kinematics of extremely metal-poor stars and showed that they have large eccentricities and motions perpendicular to the galactic plane, extending over a wide range. On the plausible hypothesis that these are the oldest stars in the solar neighborhood, this suggests that they were born during a phase of rapid collapse in which the galactic halo was formed, followed by a dissipative phase that eventually gave rise to the disk. Age dating of globular clusters (Demarque 1980) gives a nominal range of a few Gyr between the oldest (most metal-deficient) and youngest (least metal-deficient), but in view of various uncertainties about the systematic effects of chemical composition on age estimates, combined with recent doubts about the actual metallicity of the more metal-rich clusters, we cannot regard this result as conclusive (cf. van Albada, de Boer & Dickens 1979).

In contrast to the situation in the halo, star formation in the disk has been an ongoing process over a significant time of the order of 10¹⁰ years and the simplest ideas on galactic chemical evolution lead one to expect a steady enrichment with time, unless the results of nucleosynthesis are counteracted by some effect like inflow of unprocessed material (e.g. Audouze & Tinsley 1976). The resulting age-metallicity effect has proved difficult to establish because of problems of calibration and observational selection, combined with the facts that the metallicity range of disk stars is fairly small and that the relatively easily dateable galactic clusters cover only the latter half of the age span of the disk. [Ages and metallicities of clusters have been listed by Demarque (1980).] The best data on the age-metallicity relation come from studies of the Strömgren colors of field stars of spectral type F by Mayor (1976) and Twarog (1980), the latter, in particular, including significant corrections for helium abundance and selection effects not included in previous work. The resulting relations (Figure 1) differ considerably in detail owing to the use of different stellar evolutionary models, etc., but the general trend appears well established and can be regarded as being more or less linear in actual (not logarithmic) abundances.

Figure 1. Relation between age and geometric mean metallicity (actual value relative to Hyades), deduced from Strömgren uvby colors of F stars in the solar neighborhood. Filled circles: Twarog, with error bars denoting the standard error of the mean; boxes indicate widths of age bins and sample standard deviation within each bin. (Twarog's youngest age bin is omitted because he gives it little weight.) Open circles: Mayor; vertical error bars denote systematic calibration uncertainties while horizontal error bars denote widths of age bins. The position of the Sun is shown, following the Sun-Hyades calibration by Branch, Lambert & Tomkin (1980).