Annu. Rev. Astron. Astrophys. 1997. 35: 637-675 Copyright © 1997 by Annual Reviews. All rights reserved

4. HIGH-REDSHIFT BULGES

Direct searches for the progenitors of local bulges may be made by the combination of statistically complete redshift surveys of the field galaxy population, combined with photometric and especially with morphological data. As an example, the I-band-selected CFHT redshift survey contains galaxies out to redshifts of order unity, and these galaxies may be analyzed in terms of the evolution of the luminosity function of galaxies of different colors, presumed to correlate with morphological type (Lilly et al 1995). The data are consistent with very little evolution in the luminosity function of the red galaxies, over the entire redshift range 0 < z < 1, and substantial evolution in the blue galaxies' luminosity function, with the color cut dividing the sample into blue and red taken as the rest-frame color of an unevolving Sbc galaxy. This lack of evolution for red galaxies may be interpreted as showing that the stars of bulge-dominated systems - the red galaxies - were already formed at redshifts greater than unity, corresponding to a look-back time of greater than half of the age of the universe, or 5-10 Gyr (depending on cosmological parameters).

The high spatial resolution of the HST allows collection of morphological information. Schade et al (1995) obtained HST images for a subset (32 galaxies in total) of the CFHT redshift survey, mostly blue galaxies with z > 0.5. They found, in addition to the "normal" blue galaxies with exponential disks and spiral arms and red bulge-dominated galaxies, a significant population of high luminosity (M_B < -20) "blue nucleated galaxies" (BNG), with large bulge-to-disk ratio (B/T 0.5) - could these be bulges in formation, at look-back times of ~ 5 Gyr? Small number statistics notwithstanding, most of the blue nucleated galaxies are asymmetric and show some suggestions of interactions. Schade et al (1996) found similar results for a larger sample, using just CFHT images for morphological classification, and confirmed that red galaxies tend to have high bulge-to-disk ratios.

Extending these results to even higher redshifts, and hence studies of progenitors of older R present-day bulges, has been achieved by the identification of a sample of galaxies with z 3 based on a simple color criterion that selects systems with a Lyman-continuum break, superposed on an otherwise flat spectrum, redshifted into the optical (e.g. Steidel et al 1996a, b). Ground-based spectroscopy of 23 high-redshift candidates provided 16 galaxies at z > 3 (Steidel et al 1996b). The observed optical spectra probe the rest-frame 1400- to 1900-Å UV and provide a reasonable estimate of the reddening, and hence dust content, and of the star formation rate. The systems are inferred to be relatively dust-free, with the extinction at ~ 1600 Å typically ~ 1.7 mag, which corresponds to an optical reddening in the galaxies' rest-frame of E(B-V) ~ 0.3 mag. Whether the low dust content is a selection effect, perhaps due to fortuitous observational line of sight, or is a general feature of these high-redshift galaxies is not clear. The comoving space density of these systems is large - on the order of half that of bright (L > L*, with L* the knee of the Schecter luminosity function) galaxies locally, which suggests that not too many of them can be hidden. The star formation rates, assuming a solar neighborhood IMF, are typically ~ 10 M / year. There are interstellar absorption lines due to various chemical species; these lines may be interpreted as indicative of gas motions in a gravitational potential of characteristic velocity dispersion of ~200 km/s, which is typical of normal galaxies today.

Morphological information from optical HST images (Giavalisco et al 1996) for 19 Lyman-break candidates, of which 6 have confirmed redshifts, show that in the rest-frame UV (1400-1900 Å) these systems are mostly rather similar, in contrast to the wide range of morphological types seen at lower redshifts, z ~ 1, discussed above. Furthermore, the typical z ~ 3 galaxy selected this way is compact, at least in the UV, and has a half-light radius of ~2 kpc, which is reminiscent of present-day bulges in the optical. Some of these galaxies show faint surrounding emission that could be interpreted as "disks." The star formation rates inferred from the spectra build the equivalent of a bulge - say 10¹⁰ M - over a few billion years, which spans the redshift range from 1~z ~ 4. Similar results are obtained from z < 3 samples derived from the HST Deep Field (Steidel et al 1996a) and for one galaxy at a redshift of z = 3.43, the central regions of which do, in fact, fit a de Vaucouleurs profile (Giavalisco et al 1995).

Thus, there is strong evidence that some (parts of some) bulges are formed at z 3. However, it is hard to draw definite conclusions about all bulges on the basis of these results because the observations at these redshifts can be biased. If, for example, half of all bulges form at z 0.5, then we would simply not observe those at higher redshifts. At higher and higher redshifts, we would simply be selecting older and older bulges. Our conclusions would become strongly biased. This is very similar to the bias for early-type galaxies discussed by van Dokkum & Franx (1996).