|Annu. Rev. Astron. Astrophys. 1996. 34:
Copyright © 1996 by . All rights reserved
3.4. Intermediate-Age Clusters
3.4.1. THE AGE DISTRIBUTION It is now widely accepted that the distributions of cluster ages differ dramatically between the two Clouds (e.g. Olszewski 1993, Da Costa 1993, Feast 1995; see Figure 1). For example, we know of (almost) no LMC star clusters with precise age estimates obtained from deep, main-sequence photometry in the age range 4-12 Gyr. The one exception - ESO121SC-03 - has an age of approximately 6-9 Gyr (Mateo et al 1986, Sarajedini et al 1995). In contrast, the SMC contains many clusters with ages between 4 and 12 Gyr: L 113 (age 4 Gyr; Mould et al 1984), K 3 (age 7 Gyr; Rich et al 1984), L 1 (age 9 Gyr; Olszewski et al 1987), NGC 121 (age 12 Gyr; Stryker et al 1985) all have ages that are either within or close to the LMC age gap. This suggests that ESO121SC-03 may have originally formed in the SMC and was swapped to the LMC (see also Lin & Richer 1992).
Figure 1. (a) A plot of [Fe/H] vs age (in Gyr) for Magellanic Cloud clusters with precise metallicity and age determinations. The open triangles denote LMC clusters, while the filled squares denote SMC clusters. Results for 60 LMC and 11 SMC clusters are plotted. Following Da Costa (1992), we have plotted the age scale in linear units to emphasize the duration of the age gap in the LMC; no clusters seemed to have formed in that galaxy over < 50% of its lifetime. Note that the present-day metallicity of LMC clusters is considerably larger in the mean than for SMC clusters, whereas the metallicites were more similar 15 Gyr ago. (b) The age distribution histogram for LMC star clusters, now plotted as a function of log (Age). More clusters are represented in this panel than in panel (a) because of the considerable numbers of clusters with reasonable age estimates but only very crude metallicity determinations. (c) The same as panel (b), except for the SMC. In comparing panels (b) and (c), note that the intermediate-age and old populations reside to the right of the dotted lines. It is in this age range that the cluster age distributions show strikingly dissimilar behavior.
This dichotomy in cluster ages is not due to evolutionary fading of clusters. Hodge (1988a) has shown that the typical cluster destruction timescale is about 5-10 times longer in the LMC than in the Galaxy. In the LMC, there are numerous examples of 1-4 Gyr clusters that will remain prominent objects when they are older than 10 Gyr (Mateo 1993), including NGC 1978 (current age 2 Gyr; Bomans et al 1995), NGC 2155 (current age 3.5 Gyr; Mateo 1996b), and NGC 1831 (current age 1 Gyr; Vallenari et al 1992). There is also little doubt that the LMC age gap is real and not merely some statistitical fluke. Da Costa (1991) reported the preliminary results of a deep CCD survey designed to target clusters in the age gap exhibited by the LMC, selecting the clusters on the basis of their integrated colors. Of all the candidate clusters, none had ages in the 4-12 Gyr gap. Likewise, Mateo (1996b) reports the results of a study of clusters over a wide luminosity range in a small northern region of the LMC (see also Mateo 1988); none of these clusters are older than 4 Gyr. NGC 2155, noted in Section 2.2, is in fact one of the oldest intermediate-age clusters known. Olszewski et al (1991) note that the age gap is also a metallicity gap: The old clusters are metal poor; the younger ones are metal rich. This also strongly suggests a physical origin for the gap.
Girardi et al (1995), Girardi & Bica (1993) have recently analyzed the integrated UBV photometric properties of a sample of over 624 newly observed LMC clusters (Bica et al 1992). Even though they span a range of over 10 Gyr in age, the intermediate-age and old clusters in this sample span a range of only 0.25 mag in (B-V) and (U-B) centered at approximately (B-V) = 0.80, (U-B) = 0.90. Clusters younger than 1 Gyr span a range of 0.8 and 1.1 mag in (B-V) and (U-B), respectively. The color degeneracy of intermediate-age and old MC clusters is also apparent in broad-band UV and IR colors (Cassatella et al 1987, Meurer et al 1990, Testa 1994). Broad-band photometry can lead to a distorted picture of the distribution of cluster ages (e.g. Elson & Fall 1988). This is not due to any sort of intrinsic error in the photometry or because of shortcomings in the evolutionary models used to interpret the photometry. Rather, the age resolution of broad-band colors is poorest precisely in the age range of greatest interest for intermediate-age and old MC populations (Girardi et al 1995). The only way to study the detailed differences in the distribution of ages of intermediate-age and old MC clusters is to obtain precise main-sequence stellar photometry.
Existing data clearly suggest that the LMC has formed clusters in at least two distinct bursts (4 Gyr and younger plus the ancient clusters), whereas the SMC has formed clusters more uniformly over the past 12 Gyr. Nevertheless, some additional work is desirable to strengthen and extend these conclusions. First, clusters are rarely selected for age determination on the basis of any clear selection rules [one exception is the study of Da Costa (1992) mentioned above]. Before we can confidently rule out the existence of any 4-12 Gyr clusters in the LMC, reliable ages should be determined for a complete sample of clusters. The brute force approach to this problem - measuring main-sequence ages for all MC clusters - is still impractical at this time: The LMC contains about 4200 clusters and the SMC about 2000 (Hodge 1986, 1988b). Nonetheless, fewer than 2% of all MC clusters currently have reliable age determinations - a figure that can certainly be significantly improved upon. Some recent attempts at complete studies (e.g. Kubiak 1990a, b, Westerlund et al 1995, Mateo 1988, 1996b) have sampled only very small numbers of clusters and have been limited, so far, to the LMC. Such complete studies would also help to constrain the destruction rate of MC clusters and to obtain a better estimate of the absolute rate of cluster formation (cf Hodge 1988a).
A second difficulty in interpreting the age distribution of clusters is that in many cases one must combine age determinations based on very different assumptions about the relevant physical parameters (e.g. distance, reddening, metallicity) or based on comparisons with fundamentally different evolutionary models. For clusters younger than about 1-2 Gyr, modern models yield ages that can differ by a factor of up to two depending simply on how they treat convection at the core/envelope interface of intermediate-mass main-sequence stars (Mateo & Hodge 1987, Seggewiss & Richtler 1989). For the intermediate-age clusters discussed here, this effect is not too serious. But other effects are. Bomans et al (1995) have reanalyzed a number of CCD color-magnitude diagrams using a consistent set of evolutionary models and adopted astrophysical parameters. Although their new age determinations do not alter the basic conclusion above (no clusters are known older than about 2.5 Gyr in their reanalyzed LMC cluster sample) the ages of some of the clusters did change significantly from previously published values. Again, this does not reflect any sort of "error" in the initial studies, but rather the change in the age scale as one adopts different evolutionary models or astrophysical parameters. Until such time as all isochrones yield identical ages for a given cluster, it might be wise to establish a standard age scale based on a single set of models and parameters so that the relative ages of MC clusters can be compared with ease and confidence.
3.4.2. THE AGE-METALLICITY RELATION FOR STAR CLUSTERS Olszewski et al (1991) obtained abundances of approximately 80 LMC star clusters. This study represents by far the most extensive set of metallicities for intermediate-age clusters on a common scale, enlarging greatly the number of spectroscopic abundance determinations for these clusters (e.g. Cohen 1982, Cowley & Harwick 1982). Very few SMC clusters have spectroscopically determined abundances; their metallicities are generally derived from isochrone fits to the clusters' color-magnitude diagrams or from applications of empirical relations between the metallicity and the color of the RGB (e.g. Zinn & West 1984). Figure 1 illustrates the age-metallicity relations for clusters in both Magellanic Clouds. The data come from the compilations of Sagar & Panday (1989), Seggewiss & Richtler (1989), supplemented by more recent results. We wish to highlight two important differences between the Clouds that are apparent from this Figure.
First, the mean LMC cluster metallicity jumps by a factor of about 40 during a time when virtually no clusters or - as we saw above - field stars were forming. Gilmore & Wyse (1991), Köppen (1993) have shown that the chemical evolution of a galaxy characterized by a star-formation history punctuated by bursts can be very complex. In the case of the LMC, Gilmore & Wyse (1991) would argue that type I supernovae from the initial population steadily polluted the interstellar medium (ISM) during the long hiatus in star formation. When stars finally did begin to form again some 4 Gyr ago, the ISM was greatly enriched. One important implication of this idea - that O and other elements should be enriched relative to Fe compared to their present day abundances - has already been described in Section 2.7.
But can this idea account for the very large observed increase in metallicity (Figure 1) during the age gap? A simple model in which only 2% of the LMC's dynamical mass formed in a burst more than 10 Gyr ago cannot account for this huge increase in metallicity. Supernova (type I and II) yields from Weaver & Woosley (1993), Thielemann et al (1986) imply that the ancient LMC population could have only raised the mean iron abundance of the entire galaxy to [Fe/H] = -1 or less, far short of the value ~ -0.4 seen in 2-3 Gyr clusters. We do not seem to see sufficient numbers of field stars and clusters with ages between 4-12 Gyr to explain this discrepancy, unless the SN rates was much higher in the old MC populations than predicted by these models.
The second noteworthy feature of Figure 1 is that the two Magellanic Clouds have experienced quite distinct chemical enrichment histories (Da Costa 1992, Olszewski 1993, Feast 1995). This is clearly apparent in the shape of the age-metallicity relations, but also in the starting and ending values. Today the LMC is more metal rich than the SMC; 12 Gyr ago, these two galaxies may have had very similar abundances; it is not clear what the SMC abundance was 15 Gyr ago. In the case of the SMC, heavy-element enrichment (presumably from Type I SNe) may have lagged the star formation by a few Gyr. Currently, we see the metallicity rising as heavy elements are injected into the ISM. The lower mean abundance in the SMC simply reflects the fact that it has converted a smaller fraction of its gas to stars in accordance with a simple closed-box model of chemical evolution, and consistent with the present-day gas fractions of the two galaxies (MHI / Mtotal) ~ 0.35 and 0.03 for the SMC and LMC, respectively.
Of course, all the arguments above assume that the MC clusters reliably track the chemical evolution of the LMC and SMC as a whole. Richtler (1993) summarizes the empirical and theoretical evidence for why this may not be the case.
3.4.3. KINEMATICS OF INTERMEDIATE-AGE CLUSTERS Freeman et al (1983), Schommer et al (1992) obtained velocities for large numbers of intermediate-age clusters along with the results for ancient clusters described in Section 1. The intermediate-age clusters exhibit kinematics that are quite similar to those of young clusters and intermediate-age LPVs. Unlike the ancient clusters for which a disk rotation solution and small velocity dispersion was unexpected, the intermediate-age clusters appear to have kinematics that are "normal" for their ages. The orientation of the intermediate-age disk and the disk defined by the ancient clusters are statistically indistinguishable.
Feast (1995) has suggested that the commencement of the active star-formation epoch in the LMC some 4 Gyr ago may have coincided with the final collapse of the disk component of the LMC. The smooth transition of cluster kinematical properties is hard to understand in this scenario. Moreover, this idea does not offer any explanation why the old clusters also show disk kinematics: The LMC disk seems to have already been present when the oldest clusters formed.