Next Contents Previous

2. CHEMICAL EVOLUTION OF THE MILKY WAY

The Galactic halo, bulge and disk(s) are all relevant to early times, only the thin disk being younger than the bulge and thick disk. The respective roles of hierarchical clustering, mergers and monolithic collapse are still not very clear; probably all play a role, but the halo and bulge share a low specific angular momentum while the thick and thin disks share a high one and may result from later accretion of gas by the bulge, which would then resemble an E-galaxy. However, it is also possible that the bulge evolved from the disk by way of a bar.

Figure 2

Figure 2. Schematic cross-section through the Galaxy.

In any case, the stellar dynamics of the halo favour what Thomas, Greggio & Bender (1999) refer to as a "fast clumpy collapse", basically the old idea of Eggen, Lynden-Bell & Sandage (1962) placed in the context of modern hierarchical clustering scenarios. One point of interest is the metallicity distribution function (MDF), recently extended to very low metallicities by Beers et al. (1998). The MDF is essentially the modified Simple-model type distribution originally noted by Hartwick (1976), with a peak at about 1/10 of the true yield, down to [Fe/H] appeq - 3. Below that it begins to fall off and there are virtually no stars (compared to a predicted number of about 10) below -4, which could represent enrichment either from a hypothetical Population III or from contamination of low-mass stars by a nearby supernova.

A significant clue to early Galactic chemical evolution comes from the relation between oxygen and alpha-particle elements, thought to come exclusively or mainly from type II supernovae, and iron, more than half of which in the Solar System comes from Type Ia. Fig 3 suggests that there is a plateau in O,alpha / Fe at low metallicities (assumed to represent early times), but there is currently a controversy in the case of oxygen. Abundances derived from the forbidden [OI] line, which is probably the most reliable source when it is not too weak, suggest a plateau, but from measurements of the near UV OH bands in dwarfs and subgiants, both Israelian, García Lopez & Rebolo (1998) and Boesgaard et al. (1999) have derived a rising trend with diminishing [Fe/H] more or less following the open squares in the top panel of Figure 3. In contrast, Fulbright & Kraft (1999) have studied the [OI] spectral region in two of the extreme cases and find lower O/Fe ratios fitting the plateau. There are technical difficulties in both methods: the OH bands are subject to uncertainties in UV continuum absorption (cf. Balachandran & Bell 1998 on solar beryllium abundance) and effective temperature, while the forbidden line in the relevant cases is so very weak that the definition of the continuum becomes a crucial source of uncertainty.

Figure 3

Figure 3. Abundance ratios of oxygen and alpha-elements to iron, plotted against [Fe/H] for stellar samples from the Galactic disk and halo, after Pagel & Tautvaisiene (1995). The solid line and curve represent a simple analytical Galactic chemical evolution model.

However this controversy comes out, the O,alpha enhancement is not universal, as has been shown, e.g. by Nissen & Schuster (1997); there are "anomalous" halo stars which have more solar-like element ratios even at quite low metallicities, a feature that is also found in the Magellanic Clouds and can be explained on the basis of slower star formation rates and effective yields diminished by outflows (e.g. Pagel & Tautvaisiene 1998). However, within the halo the presence of "anomalies" shows no obvious relation with extreme kinematic properties that might be signatures of a captured satellite (Stephens 1999).

Within the thick disk, the alpha / Fe ratio is remarkably uniform, even up to quite high metallicities, indicating an old "get rich quick" population. This is well brought out by the work of Fuhrmann (1998) on Mg, shown in Figure 4, and in a still unpublished study of oxygen by Gratton et al. (1996), and it may be that this trend is continued in the bulge (cf. Rich 1999). The data cast an interesting light on the formation of the thick disk, since they indicate a hiatus in star formation during which Fe/alpha increased but overall metallicity diminished, maybe from inflow of relatively unprocessed material, e.g. in a merger, before the stars now belonging to the thin disk were formed.

Figure 4

Figure 4. [Mg/Fe] vs [Fe/H] and [Fe/Mg] vs [Mg/H] for stars of the Galactic halo, thick disk and thin disk, after Fuhrmann (1998). Courtesy Klaus Fuhrmann.

Returning to the earliest stage of evolution of the Population II halo, when we consider a regime in which [Fe/H] < - 2.5 or so, we reach a stage where pollution by a single supernova becomes significant over a region the size of a globular cluster or superbubble of the order of 105 Msun. Metallicity (however defined) then becomes a poor clock and strange patterns appear, accompanied by significant scatter (McWilliam 1997). There are marked changes within the iron group, with Cr, Mn (and Cu) going down relative to iron and Co going up. Ryan, Norris & Beers (1996) suggest that at these low levels [Fe/H] is an increasing function of the mass of an individual supernova, and Tsujimoto & Shigeyama (1998) have estimated revised stellar yields as a function of progenitor mass on this basis. Most yields increase, with the conspicuous exception of the r-process, whose representative Eu/Fe has a large scatter and may be anti-correlated with [Fe/H]. Ba and Sr also mainly come from the r-process at these low metallicities and have even more scatter because the s-process can also contribute in evolved stars or stars with evolved companions. In a model recently put forward by Tsujimoto, Shigeyama & Yoshii (1999), stars form in superbubbles dominated by a single supernova, so that their composition is a weighted mean of the interstellar medium (with [Eu/Fe] appeq [alpha/Fe] = constant) and supernova ejecta. Fe/H increases with the mass of the supernova while Eu/Fe decreases, leading to an anti-correlation with scatter superimposed until the ISM is sufficiently enriched to take over and normal Galactic chemical evolution proceeds.

Further evidence for inhomogeneity comes from the abundances of the light elements 6Li, beryllium and boron, which show an unexpected "primary" behaviour - at least relative to iron - down to very low metallicities. This cannot be understood on the basis of spallation of interstellar CNO nuclei by primary cosmic ray protons and alpha-particles; these give a reasonable explanation for their abundances in the Sun and Population I stars in general but led to an expectation of secondary behaviour (Be,B/O propto O/H) with diminishing metallicity. (1) There are also energetic problems with the production by interstellar spallation at low metallicity (Ramaty et al. 1997). Thus various inhomogeneous processes have been proposed, beginning with the hypothesis of Duncan, Lambert & Lemke (1992) that fast CNO nuclei in primary cosmic rays are reponsible, and that their abundance is dominated by supernova ejecta rather than the interstellar medium. A more detailed model by Ramaty & Lingenfelter (1999) postulates an origin of of cosmic rays from acceleration of ions sputtered off dust grains in supernova ejecta by shocks within a superbubble. Thus the composition of cosmic rays is more or less constant and they dominate light element production at early times in the way suggested by Duncan, Lambert & Lemke.



1 With the large increase in O/Fe claimed by Israelian et al. and Boesgaard et al. there could be some semblance of secondary behaviour of the light elements after all, along with iron, magnesium, calcium etc; the likelihood of this depends on how the oxygen debate comes out. Back.

Next Contents Previous