Measuring age differences among globular clusters having similar metallicities - A new method and first results

Abstract

A new method is described for comparing observed color-magnitude diagrams to obtain accurate relative ages for star clusters having similar chemical compositions. It is exceedingly simple and straightforward: the principal sequence for one system is superimposed on that for another by applying whatever vertical and horizontal shifts are needed to make their main-sequence turnoff segments coincide in both V magnitude and B - V color. When this has been done, any apparent separation of the two lower giant branch loci can be interpreted in terms of an age disparity since, as is well known from basic theory, the color difference between the turnoff and the giant branch is a monotonic and inverse function of age. This diagnostic has the distinct advantage that it is strictly independent of distance, reddening, and the zero-point of color calibrations; and theoretical isochrones show it to be nearly independent of metallicity - particularly for [m/H] < - 1.2. (In fact, if the cluster photometry is secure and the metal abundance is accurately known, our technique provides an excellent way to determine relative reddenings.) Furthermore, computed models need be used only in a differential sense to calibrate the observed variations in the turnoff-to-giant-branch color difference in terms of relative age. We apply this technique to the principal sequences of 22 globular clusters divided into three abundance bins - [m/H] ≈ -2.1, -1.6, and -1.3 - and take the fiducials for M92, NGC 6752, and NGC 362 to be the standard sequences in each bin, respectively. More detailed star-by-star statistical analyses are undertaken for three groups of clusters of particular interest: M68/M92/NGC 6397, M3/ M13, and NGC 288/NGC 362. We find that the most metal-poor systems are extremely uniform in age, with no convincing evidence for differences as great as 0.5 Gyr. The [m/H] ≈ -1.6 clusters also seem to be nearly coeval, though the data are not yet of sufficiently high quality to rule out some age spread (at about the 1.5δ confidence level). However, the most metal-rich globulars that we have considered do appear to encompass a significant range in age. In particular, our analysis supports the conclusion reached by several others that NGC 288 is older than NGC 362 and M5 by ≈2 Gyr. We also find that most of the distant objects in our sample (e.g., Palomar 5 and NGC 7492) are not distinctly different in age from nearer globulars having the same metal content, suggesting that there is not a significant age - Galactocentric-distance relation in the outer halo. Palomar 12 still seems to be an anomaly in that its C - M diagram differs from the adopted fiducial for its metallicity - that of NGC 362 - by more than any other cluster in our sample. It seems younger than NGC 362 by about 4 Gyr, in agreement with previous findings. The main result of this work, then, is that the dispersion in globular cluster ages as a function of metal abundance appears to increase from near zero at [m/H] ≈ -2.1 to ∼2 Gyr at [m/ H] ≈ -1.3. This suggests that the collapse of the Galaxy was of prolonged rather than of brief (i.e., ≪ 1 Gyr) duration. A further implication of our analysis is that age is unlikely to be the "second parameter" in the globulars unless it can be demonstrated that age differences of ≲ 2 Gyr can cause the observed wide variation in horizontal-branch morphology among clusters of the same [m/H].