Next Contents Previous


Since it first was presented ([29]), the Lyman-break technique had stood out as one of the most powerful tools for identifing high-z galaxy candidates. As of this writing, more than 550 candidates have been spectroscopically confirmed to be at z ~ 3 over an area ~ 0.3 square degrees and about 50 at z ~ 4 over an area ~ 0.23 square degrees ([31]). In number density, the bright ends of the z ~ 3 and z ~ 4 populations have similar values to the local L > L* galaxies, for a flat cosmology. Even if merging may have played a role in changing these values over time, just the number of stars contained in Lyman-break galaxies at z > 2 accounts for ~ 20-30% of all the stars known today. The basic fact is that the Lyman-break galaxy populations are a significant fraction of the total galaxy population today. Thus, understanding the nature of the Lyman-break galaxies remains a gateway to understanding the evolution of galaxies.

The identification of high-z candidates is based on the detection of the Lyman break at 912 Å, which is the strongest discontinuity in the stellar continuum of star-forming galaxies. A galaxy at, say, z = 3 will have the Lyman break redshifted to 3648 Å. If a pair of filters is chosen to straddle the break, the galaxy will appear extremely red in this color. In order to avoid as much as possible low-z interlopers, one or more filters are generally added longward of the Lyman break to select only candidates which are blue in this(these) additional color(s). With a careful selection of the color criteria, the Lyman-break technique is extremely successful at identifying high-z candidates; spectroscopic confirmations give a ~ 95% success rate for the z ~ 3 sample and a ~ 80% success rate for the z ~ 4 sample ([30], [8], [31]). The lower success rate at z ~ 4 is due to the incidence of low-z interlopers, namely elliptical galaxies at z ~ 0.5-1 whose 4,000 Å break falls inside the selection window of the filters.

While the determination of the intrinsic nature of the Lyman-break galaxies, whether they are massive systems or galaxy fragments, and what kind of progenitors they are, is still a source of heated debate (e.g., [21], [13], [1]), the identification of their observational low-z counterparts appears less controversial.

By selection, Lyman-break galaxies are UV-bright, actively star-forming systems, with a preferentially blue spectral energy distribution (SED). Observed star formation rates (SFRs) range from a few to 50 Msmsun yr-1, for a Salpeter Initial Mass Function (IMF) in the range 0.1-100 Msmsun ([8]). This range of values is typical of what observed in Local, UV-bright starburst galaxies (e.g., [4]). The restframe UV and B-band half-light radii are around 0.2-0.3 arcsec, which correspond to spatial radii ~ 1-3 h50-1 kpc, depending on q0 ([12], [14]). The similarity of the half-light radii at both UV and B suggests that the UV is a reliable tracer of the full extent of the light-emitting body. Ground-based optical spectra, which correspond to the restframe 900-1800 Å range for a z ~ 3 galaxy, show a wealth of absorption features, and sometime P-Cygni profiles in the CIV(1550 Å) line (cf. the figures in [30]), typical of the predominance of young, massive stars in the UV spectrum. Currently limited ground-based near-IR spectroscopy (e.g. [27]) has revealed nebular line emission in these galaxies. Hybrid line equivalent widths constructed using the UV flux density f(UV) as denominator, namely, EW'(Å) = F(line)/f(UV), show that the observed values for the high-z galaxies fall in the loci observed for local starburst galaxies ([25]). In summary, the observational properties of the Lyman-break galaxy population fully resemble, in the restframe UV-optical range, those of low-redshift, UV-bright starburst galaxies ([25]).

The Lyman-break galaxies share another global characteristic with the Local starbursts. If we parametrize the observed UV stellar continuum with a power law, F(lambda) propto lambdabeta, Lyman-break galaxies cover a large range of beta values, roughly from -3 to 0.4, namely from very blue to moderately red (Figure 1, left panel). This range is not very different from that covered by the Local, UV-bright starbursts (Figure 1, right panel). Population synthesis models (e.g. [19]) indicate that a dust-free, young starburst or constant star-formation population have invariably values of beta < -2.0, for a vast range of metallicities. What does cause the UV stellar continuum of Lyman-break galaxies to be redder than expected for a young star-forming population?

Figure 1a
Figure 1b

Figure 1. (Left Panel) The distribution of the observed UV spectral indices beta of the z appeq 3 galaxies (M. Dickinson 1998, private communication). Only a small fraction of all the galaxies are blue enough (beta < -2) to be classifiable as dust-free, young star-forming populations. Different symbols corrsponds to different color selections. (Right Panel) The distribution of beta for a UV-bright sample of Local starburst galaxies ([18]). Note that the range of beta values is similar to the high-z sample. The UV spectral slopes of the local starbursts correlate with Es(B - V), the color excess of the optical stellar continuum due to dust reddening ([6], [4]). Blue UV spectra correspond to small values of the color excess, red UV spectra to large values of Es(B - V). The best fit line through the data is shown.

Next Contents Previous