5.1. A Global View of Metal Enrichment in the Universe Two Billion Years after the Big Bang
We could briefly summarise everything we have covered so far as follows.
(a) The intergalactic medium at z = 3 does not consist entirely of pristine material. At least the regions we have been able to probe so far show traces of metals at levels between 1/1000 and 1/100 of solar metallicity. Matter in these regions, however, is still at a higher density than the mean density of the universe at that epoch, and it is unclear at present whether a pre-galactic episode of star formation (often referred to as Population III stars), is required to explain the large scale distribution of elements in the IGM.
(b) Damped Ly systems have abundances similar to those of Population II stars in our Galaxy. Perhaps they represent an early stage in the formation of spiral galaxies, before most of the gas had been converted into stars. It is also clear that a wide variety of galaxy morphologies, including low surface brightness galaxies, share the common characteristic of providing a large cross-section on the sky at high surface densities of H I.
(c) Finally, Lyman break galaxies strongly resemble what we call Population I stars in the Milky Way. They are the sites of vigorous star formation which (i) has produced a relatively high level of chemical enrichment at early epochs, (ii) has built up stellar masses of 1010 M in a sizeable fraction of the population, and (iii) drives large-scale outflows of gas, metals and dust into the intergalactic medium. All of these characteristics point to LBGs as the progenitors of today's spiral bulges and elliptical galaxies, observed during the most active phase in their lifetimes.
The connection between LBGs and DLAs is currently the subject of considerable discussion, as astronomers try and piece together these different pieces of the puzzle describing the universe at z = 3. Table 2 summarises some of the relevant properties. Lyman break galaxies have systematically higher star formation rates and metallicities, and their interstellar media have been stirred to higher velocities, than is the case in most DLAs. These seemingly contrasting properties can perhaps be reconciled if the two classes of objects are in fact drawn from the same luminosity function of galaxies at z = 3. Since they are selected from magnitude limited samples, the LBGs are preferentially bright galaxies - the data in Table 2 refer to galaxies brighter than L* which corresponds to = 24.5 (Adelberger & Steidel 2000). If, on the other hand, the H I absorption cross-section decreases only slowly with galaxy luminosity, as is the case at lower redshifts (Steidel, Dickinson, & Persson 1994), the DLA counts would naturally be dominated by the far more numerous galaxies at the steep ( = - 1.6) faint end of the luminosity function.
|SFR (M yr-1)||~ 50||< 10|
|Z (Z)||~ 1/3||~ 1/20|
|v (km s-1)||~ 500||200|
Such a picture finds theoretical support in the results of hydrodynamical simulations and semi-analytic models of galaxy formation (e.g. Nagamine et al. 2001; Mo, Mao, & White 1999). In the coming years it will be tested by deeper and more extensive searches for DLA galaxies, by comparing the clustering of LBGs and DLAs (Adelberger et al. 2002, in preparation), and by more reliable measurements of the star formation activity associated with DLAs. This last project is best tackled in the near-infrared, by targeting the H emission line with integral field - rather than slit - spectroscopy, as provided for example by the Cambridge Infrared Panoramic Survey Spectrograph (CIRPASS - see Figure 31).
Figure 31. (Reproduced from Bunker et al. 2001). Left: An H image of the local spiral galaxy NGC4254 as it would appear at z = 1.44 with the CIRPASS integral field unit overlaid (using 0.25 arcsec diameter fibres). Right: A spiral galaxy at z 1 from the Hubble Deep Field B-band. The star-forming H II regions are prominent in the rest-frame UV. CIRPASS will accurately determine the true star formation rates, since (1) the compact knots of star formation are well-matched to the fibre size, reducing the sky background and increasing the sensitivity; (2) the large area surveyed by the integral field unit covers most of a spiral disk and (3) the H line is a much more robust measure of the star formation rate than the dust-suppressed UV continuum and resonantly-scattered Ly.
When we combine the available abundance determinations for Lyman break galaxies, damped Ly systems and the Ly forest with those for the inner regions of active galactic nuclei (from analyses of the broad emission lines and outflowing gas in broad absorption line QSOs - see Hamann & Ferland 1999), a `snapshot' of metal enrichment in the universe at z 3 emerges (Figure 32). The x-axis in the figure gives the typical linear dimensions of the structures to which the abundance measurements refer, from the 10-100 pc broad emission line region of QSOs, to the kpc scales of LBGs revealed by HST imaging (Giavalisco, Steidel, & Macchetto 1996), to the 10 kpc typical radii of DLAs deduced from their number density per unit redshift (Steidel 1993), to the 100 kpc dimensions of condensations in the Ly forest with N(H I) 1014 cm-2 indicated by the comparison of the absorption along adjacent sightlines in the real universe (e.g. Bechtold et al. 1994) and in the simulations (e.g. Hernquist et al. 1996).
Figure 32. Summary of our current knowledge of abundances at high redshift. The `metallicity' is plotted on the y-axis on a log scale relative to the solar reference; the latter is shown as the broken horizontal line at 0.0 and corresponds to approximately 2% of the baryons being incorporated in elements heavier than helium. The x-axis shows the typical linear dimensions of the strucures to which the abundance measurements refer, from the central regions of active galactic nuclei on scales of 10-100 pc to the intergalactic medium traced by the Ly forest on Mpc scales. Generally speaking, these typical linear scales are inversely proportional to the overdensities of the structures considered relative to the background.
These different physical scales in turn reflect the depths of the underlying potential wells and therefore the overdensities of matter in these structures relative to the mean density of the universe. Even from such an approximate sketch as Figure 32, it seems clear that it is this overdensity parameter which determines the degree of metal enrichment achieved at any particular cosmic epoch. Thus, even at the relatively early times which correspond to z = 3, the gas in the deepest potential wells where AGN are found had already undergone considerable processing and reached solar or super-solar abundances. At the other end of the scale, condensations in the Ly forest which correspond to mild overdensities contained only traces of metals with metallicity Z 1/100 - 1/1000 Z. The dependence of Z on the environment appears to be stronger than any age-metallicity relation - old does not necessarily mean metal-poor, not only for our own Galaxy but also on a global scale. This empirical conclusion can be understood in a general way within the framework of hierarchical structure formation in cold dark matter models (e.g. Cen & Ostriker 1999; Nagamine et al. 2001).