ARlogo Annu. Rev. Astron. Astrophys. 2002. 40:487-537
Copyright © 2002 by Annual Reviews. All rights reserved

Next Contents Previous

2. STELLAR AGE DATING

Nucleo-cosmochronology (or cosmochronometry), or the aging of the elements through radioactive decay, has a long history (Rutherford 1904, Fowler & Hoyle 1960, Butcher 1987). A related technique is widely used in solar system geophysics. Independent schemes have aged the oldest meteorites at 4.53 ± 0.04 Ga (Guenther & Demarque 1997, Manuel 2000). The small uncertainties reflect that the age dating is direct. Element pairs like Rb and Sr are chemically distinct and freeze out during solidification into different crystalline grains. The isotope 87Rb decays into 87Sr, which can be compared to 86Sr, a nonradiogenic isotope, measured from a control sample of Sr-rich grains. This provides a direct measure of the fraction of a 87Rb half-life (τ1/2 = 47.5 Ga) since the meteorite solidified.

It appears that, until we have a precise understanding of BBNS and the early chemical evolution history of the Galaxy, geophysical precision will not be possible for stellar ages. The major problem is that, as far as we know, there is no chemical differentiation that requires that we know precisely how much of each isotope was originally present. Modern nucleo-cosmochronology compares radioactive isotope strengths to a stable r-process element (e.g. Nd, Eu, La, Pt). The thorium method (232Th, τ1/2 = 14.0 Ga) was first applied by Butcher (1987) and refined by Pagel (1989). Other radioactive chronometers include 235U (τ1/2 = 0.70 Ga) and 238U (τ1/2 = 4.47 Ga), although Yokoi et al. (1983) have expressed concerns about their use (compare Cayrel et al. 2001). Arnould & Goriely (2001) propose that the isotope pair 187Re-187Os (τ1/2 = 43.5 Ga) may be better suited for future work.

With the above caveats, we point out that several groups are now obtaining exquisite high-resolution data on stars with enhanced r-process elements (Cayrel et al. 2001, Sneden et al. 2000, Burris et al. 2000, Westin et al. 2000, Johnson & Bolte 2001, Cohen et al. 2002, Hill et al. 2002). For a subset of these stars, radioactive ages have been derived (Truran et al. 2001) normalized to the heavy element abundances observed in meteorites.

There are few other direct methods for deriving ages of individual stars. A promising field is astero-seismology that relies on the evolving mean molecular weight in stellar cores (Christensen-Dalsgaard 1986, Ulrich 1986, Gough 1987, Guenther 1989). Gough (2001) has determined 4.57 ± 0.12 Ga for the Sun, which should be compared with the age of meteorites quoted above. The Eddington satellite under consideration by ESA for launch at the end of the decade proposes to use stellar oscillations to age 50,000 main sequence stars (Gimenez & Favata 2001).

It has long been known that disk stars span a wide range of ages from the diversity of main-sequence stars. Edvardsson et al. (1993) derived precise stellar evolution ages for nearby individual post-main-sequence F stars using Strömgren photometry, and showed that the stars in the Galactic disk exhibit a large age spread with ages up to roughly 10 Ga (Figure 2). Using the inverse age-luminosity relation for RR Lyrae stars, Chaboyer et al. (1996) found that the oldest globular clusters are older than 12 Ga with 95% confidence, with a best estimate of 14.6 ± 1.7 Ga (Chaboyer 1998). But Hipparcos appears to show that the RR Lyr distances are underestimated leading to a downward revision of the cluster ages: 8.5−13.3 Ga (Gratton et al. 1997); 11−13 Ga (Reid 1998); 10.2−12.8 Ga (Chaboyer et al. 1998). For a coeval population (e.g. open and globular clusters), isochrone fitting is widely used. The ages of the galactic halo and globular clusters, when averaged over eight independent surveys, lead to 12.2 ± 0.5 Ga (Lineweaver 1999).

Other traditional methods rely on aging a population of stars that are representative of a particular component of the Galaxy. For example, Gilmore et al. (1989) use the envelope of the distribution in a color-abundance plane to show that all stars more metal poor than [Fe/H] = -0.8 are as old as the globular clusters. Similarly, the faint end of the white dwarf luminosity function is associated with the coolest, and therefore the oldest, stars (Oswalt et al. 1996). The present estimate for the age of the old thin disk population when averaged over five independent surveys is 8.7 ± 0.4 Ga (Lineweaver 1999), although Oswalt et al. argue for 9.5+1.1-0.8 Ga.

For a world model with (ΩΛ = 0.7, Ωm = 0.3), the Big Bang occurred 14 Ga ago (Efstathiou et al. 2002) – in our view, there is no compelling evidence for an age crisis from a comparison of estimates in the near and far field. But the inaccuracy of age dating relative to an absolute scale does cause problems. At present, the absolute ages of the oldest stars cannot be tied down to better than about 2 Ga, the time elapsed between z = 6 and z = 2. This is a particular handicap to identifying specific events in the early Universe from the stellar record.

Next Contents Previous