4.4. Dating the Star Formation Activity

Realistically, the detailed spectroscopic analysis described above can only be applied to a subset of LBGs, at the bright end of the luminosity funtion. However, the coarser spectral energy distribution (SED) of Lyman break galaxies still holds important information on the star formation episodes. Broad-band photometry in the optical and near-infrared, spanning the wavelength interval 900-5500Å in the rest-frame, is now available for more than one hundred galaxies at z 3 (Papovich, Dickinson, & Ferguson 2001; Shapley et al. 2001). The colours over this range (typically four colours are used in the analysis) depend on the degree of dust reddening, E(B - V), and on the age of the stellar population, t_sf. The two can be decoupled with some degree of confidence provided that the SED includes the age-sensitive Balmer break near 3650Å, which at z = 3 falls between the H and K bands - hence the need for accurate near-IR photometry. A third parameter, the instantaneous star formation rate, (t_sf), determines the normalisation (rather than the shape) of the SED. The analyses by Papovich et al. (2001) and Shapley et al. (2001) deduced the best-fitting values of E(B - V), t_sf, and (t_sf) by ² minimisation of the differences between the observed SEDs and those predicted by the widely used population synthesis code of Bruzual & Charlot (1993 and subsequent updates). The results have turned out to be very interesting - some would say surprising (see Figures 28 and 29).

Figure 28. Histograms of best-fitting ages and reddening for the sample of 81 z 3 LBGs analysed by Shapley et al. (2001). There is a large spread of ages in the population; the median age is 320 Myr and 20% of the objects are older than 1 Gyr. The cyan (light grey) bin corresponds to galaxies with inferred ages older than the age of the universe at their redshifts (an indication of the approximate nature of the ages derived by SED fitting). The median E(B - V) = 0.155 for the sample corresponds to attenuations by factors of ~ 4.5 and ~ 2 at 1600Å and 5500Å respectively.

Figure 29. Histograms of assembled stellar mass and star formation rates from Shapley et al. (2001). By redshift z 3 a significant fraction of LBGs seem to be approaching the stellar mass of today's L* galaxies, mstar 4 × 1010 M.

Evidently, Lyman break galaxies span a wide range of ages. One fifth of the sample considered by Shapley et al. (2001) consists of objects which apparently have just collapsed and are forming stars on a dynamical timescale (~ 35 Myr). As we have seen, cB58 seems to belong to this class. At the other end of the scale, some 20% of the galaxies at z = 3 have been forming stars for more than 1 Gyr, placing the onset of star formation at much higher redshifts (z > 5 - 10). Furthermore, there appears to be a correlation between age and star formation rate, with the younger objects typically forming stars at about ten times the rate of the older ones and being more reddened on average. The mean SFR is <(t_sf)> = 210h^-2 M yr^-1 for galaxies with t_sf < 35Myr while, for the 20% of the sample with t_sf > 1 Gyr, <(t_sf)> = 25h^-2 M yr^-1.

This range of properties is further reflected in the total formed stellar masses m_star obtained by integrating (t_sf) over t_sf. A variety of star formation histories was considered (e.g. star formation which is continuous or decreases with time); in general m_star does not depend sensitively on this choice, although an older population of stars which by z 3 have faded at UV and optical wavelengths could remain hidden (Papovich et al. 2001). As can be seen from Figure 29, by redshift z 3 some galaxies had apparently already assembled a stellar mass comparable to that of an L^* galaxy today, m_star 4 × 10¹⁰ M, while 20% of the sample have values of m_star one order of magnitude smaller. These findings led Shapley et al. (2001) to speculate that we may be beginning to discern an evolutionary sequence in Lyman break galaxies, with the younger, dustier, more actively star-forming objects evolving to the older, less reddened, and more quiescent phase. It remains to be seen how this scenario stands up to the scrutiny of future observations, as we try to link the properties of the stellar populations of individual galaxies to other parameters, such as dynamical mass and metallicity.