Metal Abundances and the History of Chemical Enrichment of Early-type Galaxies

The abundance ratios obtained from application of our method described in Section 4.4 to estimate stellar population parameters from Lick indices are displayed in Figure 30. As expected from Figure 29, we find all galaxies to have iron abundances slightly below solar. There is a correlation between [Fe/H] and $M_r$, where [Fe/H] ranges between $\sim$ -0.15 for the faintest and just below solar for the brightest bin. Abundance ratios of all the elements studied relative to iron are solar or above solar, and are all correlated with luminosity to different degrees. That all elemental abundances are correlated with luminosity is an expected result, which derives from the more fundamental relation between mass and metallicity (e.g., Tinsley 1978). The interpretation of our results for abundance ratios is more subtle.

First and foremost, the most striking result in Figure 30 is the behavior of nitrogen abundances, both in absolute terms and as a function of luminosity. We caution that this result is sensitive to the abundance of carbon, which might be subject to systematics due to unknown oxygen abundances and/or theoretical uncertainties in the sensitivity of CN formation to carbon abundance variations (see Korn et al. 2005 for a discussion). Taking our results at face value, we find nitrogen to be enhanced in this SDSS sample, with [N/Fe] varying from just above solar, in the low luminosity end, to $\sim$ +0.2 for the highest luminosity bin. In the Galaxy, [N/Fe] is essentially solar for stars in a wide range of iron abundance (c.f. Figure 5). The only stellar systems known where [N/Fe] departs strongly from solar are globular clusters, where its mean value can be as high as $\sim$ +0.8. (e.g., Cannon et al. 1998, Cohen et al. 2002, Briley et al. 2004, Carretta et al. 2005, Lee 2005, Smith & Briley 2006). In fact, stars in globular clusters present a wide range of nitrogen abundances, and the distribution of this parameter seems to be bimodal. Globular clusters in M 31 seem to be even more nitrogen-rich than those in the Galaxy (Burstein et al. 1984, Li & Burstein 2003). However, the leading scenarios attempting to explain those nitrogen abundances tend to invoke conditions that are only met in globular clusters (e.g. Cannon et al. 1998, Beasley et al. 2004, Carretta et al. 2005).

Nitrogen is one of the elements whose history of enrichment is the most uncertain. The main source of nitrogen enrichment seems to be mass loss by intermediate and low mass AGB stars (e.g., Timmes et al. 1995, Henry & Worthey 1999, Chiappini, Romano & Matteucci 2003, Gavilán, Mollá & Buell 2006), but explosive nucleosynthesis in high-mass stars can also contribute nitrogen, especially at early times (e.g., Chiappini, Matteucci & Ballero 2005). The strong correlation of [N/Fe] with luminosity and (presumably) metallicity, seems to be indicating a strong secondary contribution to the enrichment of nitrogen in the sample studied (e.g., Tinsley 1979). If this interpretation is correct, nitrogen abundances may pose a novel constraint on the timescale for star formation in early-type galaxies. Secondary contribution to nitrogen enrichment is predominantly due to stellar winds from AGB stars with zero-age-main-sequence masses in the 4-8 $M_\odot$ range (Chiappini et al. 2003), whose lifetimes, according to the Geneva evolutionary tracks (Lejeune & Schaerer 2001), are of the order of 40-200 Myr. If the strong dependence of [N/Fe] on galaxy luminosity (and, presumably, mass) is a signature of secondary nitrogen enrichment, star formation in early-type galaxies must have lasted for at least 40-200 Myr in order for nitrogen contributed by these intermediate mass stars to be incorporated into new generations of stars. Therefore, our result for the run of nitrogen abundances as a function of galaxy luminosity may be setting a lower limit for the duration of star formation in early-type galaxies. This is a new constraint on the timescale of star formation in these systems. It is clearly possible to obtain tighter constraints on the basis of calculations from chemical evolution models, taking into consideration up-to-date stellar yields as a function of mass and a realistic IMF.

We find that all galaxies in the sample under study are magnesium-enhanced ([Mg/Fe]$>$0), and that more luminous galaxies are more enhanced than their fainter counterparts. As discussed in Section 6.1, this is a well known result, commonly interpreted as being due to the fact that the bulk of the stars in these galaxies were formed in a major event which lasted no longer than $\sim$ 1 Gyr, so that supernova type Ia could not contribute significantly to chemical enrichment (e.g., Wheeler et al. 1989). The correlation between [Mg/Fe] and luminosity has also been found by other authors (e.g., Trager et al. 2000, Denicoló et al. 2005, Thomas et al. 2005, Mendes de Oliveira et al. 2005) and is usually interpreted as being due to shorter star formation time-scales in more massive galaxies. This result is in sync with our finding that star formation in lower luminosity galaxies seems to have lasted longer than in their more luminous counterparts (Section 6.2.3), based on the $H\delta _F$-based ages. IMF variations as a function of galaxy mass could also account for these trends, but this hypothesis is more difficult to test.

As in previous studies (e.g., Vazdekis et al. 1997, Henry & Worthey 1999, Saglia et al. 2002, Thomas et al. 2003b, Prochaska et al. 2005) we find that calcium is not as enhanced relative to iron as magnesium. However, unlike most previous studies, we find that [Ca/Fe] seems to be well correlated with galaxy luminosity. As shown by Prochaska et al. (2005), and discussed in Section 4.3.2 the Ca4227 index is very strongly affected by CN. Once this effect is accounted for, either by redefining the index in order to minimize CN contamination (Prochaska et al. 2005) or by estimating the impact of CN lines on the index on the basis of spectrum synthesis calculations (this work), calcium is seen to be as correlated with galaxy luminosity as magnesium, which is the other $\alpha$-element in our analysis.