Astron. Astrophys. 364, 455-466 (2000)

4. Metallicities

4.1. Metallicities for GK giants

For GK giants it is appropriate to define the metallicity calibration directly in terms of and instead of using the notation for F dwarfs of e.g. Crawford (1975). We use the calibration by Hilker (2000),

with , , and , valid for . When plotted in the diagram (i.e. after de-reddening), GK giant stars of the same metallicity will fall along a straight line, independently of luminosity. Compared to an earlier calibration of [Fe/H] as a function of and by Grebel & Richtler (1992), the Hilker (2000) calibration yields somewhat lower metallicities, especially at low metallicities. The difference amounts to about 0.1 dex for [Fe/H] and increases to dex for [Fe/H].

For a "typical" GK giant with and , errors in v, b and y of 0.01 magnitudes correspond to a total error in the derived metallicity of about 0.16 dex, with the most significant contribution to the total error arising from the b band error. Similarly, an error in the estimated reddening of 0.01 mag translates to about 0.05 dex in the derived [Fe/H]. Hence, the derived metallicities are quite sensitive to errors in the measurements as well as in the assumed reddenings, and great care must be taken to avoid systematic errors which may shift the observed metallicity distributions by significant amounts.

In this section we analyze the metallicity distributions derived for GK giants using the calibration given by Eq. (3). Following Hilker (2000) we have used the calibration for stars with . Additional selection criteria based on DAOPHOT parameters were err and . This resulted in about 100-200 stars for metallicity analysis in each field (see Table 4).

Fig. 8 shows the diagrams for the four fields and the derived metallicity distributions are in Fig. 9 (solid lines). The average reddenings given in Table 3 have been used. Fig. 9 shows a significant scatter in the derived metallicities, from [Fe/H] = -2.0 up to around [Fe/H] = 0. In some of the fields the derived metallicity distributions include a number of stars with [Fe/H] 0. However, as shown for the Milky Way globular cluster 47 Tuc by Dickens et al. (1979), stars with peculiar CNO abundances can mimic stars with a generally high metallicity. The effect of CN anomaly on metallicities derived from Strömgren photometry has also been illustrated by Richter et al. (1999). Furthermore, the metallicity calibration is valid only for subsolar metallicities. We therefore can not conclude with certainty from our data that stars with truly high metallicities exist in the LMC or SMC, while the possibility remains open. Spectroscopic studies will be needed in order to answer this question definitively.

Fig. 8. diagrams for the four fields. The straight lines represent constant metallicities according to Eq. 3.

Fig. 9. Metallicities for the four fields; see text. The error bars refer to systematic errors due to zero-point errors in calibration of the photometry.

Table 3. Basic reddening characteristics for the observed fields. Columns 2 and 3 give the minimum and average reddening for each field according to our investigation based on B stars.

Average, median and mode statistics for the metallicity distributions are listed in Table 4, with numbers in parantheses based on [Fe/H] values less than 0. The SMC fields generally come out more metal poor than the LMC fields, although the SMC HV11284 field appears to be nearly as metal-rich as the LMC HV12578 field. However, the uncertainty on the photometric zero-points translates to roughly 0.2 dex in [Fe/H], so within the error limits the metallicities are consistent with the results from spectroscopic studies of F and G supergiants, [Fe/H] = for the LMC and [Fe/H] = for the SMC (Westerlund 1997). Our metallicities for the LMC fields are also consistent with those obtained by Dirsch et al. (2000) who quote [Fe/H] values in the range to dex for young LMC clusters.

Table 4. Metallicity data for the four fields. Numbers in parantheses indicate the values when only stars with [Fe/H]0.0 are included. N is the number of GK giants used.

4.2. Investigating the effect of reddening

With the knowledge of reddenings obtained from B stars, it is possible to estimate how much of the apparent scatter in metallicity seen in Fig. 9 may actually be attributed to reddening variations.

In order to investigate how reddening variations affect the derived metallicity distributions, we carried out the following experiment: First, a set of data pairs were generated, corresponding to one single metallicity (the canonical values were used for this experiment). This was accomplished by using the list of observed indices and then generating the indices from the calibration equation. When plotted in the diagram these points would then per definition fall along straight lines. Next, for each data pair the reddening of a randomly selected B stars was added. The "metallicities" were then determined for these synthetic data using an average reddening, in the same way as for the real GK giants. This lead to the histograms drawn with dashed lines in Fig. 9.

The peaks of the simulated metallicity distributions are in quite good agreement with those of the actual observed distributions, at least to within the uncertainty arising from zero-point errors in the photometry. The scatter in the observed metallicity distributions remains somewhat larger than that of the simulated ones, and in particular, we note the presence of what might be interpreted as a metal-poor population, with metallicities extending down to [Fe/H] . In order to quantify to what extent some of the scatter in the metallicity distributions may be intrinsic, we compared the observed and simulated distributions using an F-test (Press et al. 1992). For all the fields, the hypothesis that the variances of the two distributions are similar is rejected at a very high (%) confidence level. This remains true even if the analysis is restricted to data with [Fe/H], except for the HV11284 field where no statistically significant difference is now found between the variances of the observed and simulated metallicity distributions.

It is instructive to consider what happens when one tries to compare e.g. the metallicity of cluster stars and the metallicity of surrounding field star populations: The cluster stars will all be located at the same depth in the LMC or SMC, so they will all be affected by the same amount of interstellar absorption. On the other hand, the field stars will be randomly distributed radially, and therefore their reddenings will vary accordingly. When metallicities are derived using Strömgren photometry one will indeed be able to confirm that the cluster stars all have the same metallicity, seemingly proving that the Strömgren photometry "works", while the field stars will seem to occupy a wide range in metallicity. However, a significant amount of the scatter in the metallicities derived for the field stars may not be real, but is instead due to differences in the reddening from star to star. Any observed difference between the average field star metallicity and the cluster metallicity may be partly real, but will also depend on the amount of reddening internally in the SMC or LMC to which the cluster is subject.

For LMC/SMC clusters that are sufficiently young for early-type stars to be present it may be worthwhile to consider including u band observations in future Strömgren photometry so that reddenings can be determined. With the new generation of UV-sensitive CCD chips this will not be very costly in terms of observing time.

4.3. Ages

Independent age determinations for the metal-poor stars in our sample could provide insight into the age-metallicity relation for field stars in the Magellanic Clouds. Such a relation is relatively well established for clusters (e.g. van den Bergh 1991), but the situation for field stars is more uncertain due to the inherent difficulties in obtaining independent metallicities and ages. Here we will not attempt to derive age information for GK giants, but we note that Dirsch et al. (2000) attempted to determine an age-metallicity relation and the star formation history of both red giant field stars and clusters in their six LMC fields, using Strömgren vby photometry. They found evidence for an increase in the star formation rate Gyr ago, along with a rapid enrichment. However, the results remain uncertain because of possible CN anomalies in the LMC GK giants and the problems discussed in Sect. 4.2.

© European Southern Observatory (ESO) 2000

Online publication: January 29, 2001
helpdesk.link@springer.de