At the time of this writing, only the morphology of the far-UV light of Lyman-break galaxies has been studied with HST in some detail in both the ground-based and HDF sample (Giavalisco et al. 1996a; Steidel et al. 1996b; Lowenthal et al. 1997). These studies have revealed a dispersion of properties. Galaxies from the ground-based sample are characterized by rather compact and relatively regular morphologies, while those from the HDF, which are on average considerably fainter, include a larger number of more diffuse, irregular and apparently "disturbed" cases.
Figure 6 shows the HST + WFPC2 image of 0000-263-D6 at z = 2.961, an example of Lyman-break galaxy with compact morphology from the ground-based sample (Giavalisco et al. 1998 in prep.). The image has been taken through the F702W passband and samples the rest-frame spectrum at ~ 1700 Å. We have displayed it with two different levels of contrast to show the compact "core" of the galaxy and the surrounding diffuse nebulosity or "halo" that extends up to a few arcsec. This is representative of the morphology of the compact Lyman-break galaxies. The cores segregate ~ 90% of the total light and tend to have regular morphologies, characterized by a relatively steep light profile that is often well approximated by a r1/4 law, but exponential profiles have also been observed (Giavalisco et al. 1996b). Surrounding the core there is a diffuse "halo" with an irregular light distribution, sometimes harboring faint, compact sub-structure. The half-light radii of the cores are typically ~ 0.2-0.3 arcsec or ~ 2.2-3.4 (~ 1.5-2.2) h-150 kpc at z = 3, while the observed isophotal diameters (lower limits to the intrinsic sizes due to the extreme (1 + z)4 surface brightness dimming), are typically 1.5-2 arcsec or ~ 17-22 (~ 11-15) h-150 kpc.
Figure 6. WFPC2/PC image through the F702W passband of galaxy 0000-263-D6 at redshift z = 2.961. a) The contrast is adjusted to show the compact and regular "core" of the galaxy, which is very well modeled by a r1/4 radial profile, with re = 0.25 arcsec. b) A different contrast shows the diffuse and structured "halo", where disturbances are observed as a blob to the NE (actually merged with the core isophotes, but clearly visible in the left panel), a plume to the S and an extended nebulosity to the W. The compact source to the NW of the galaxy is at a significantly lower redshift. Both panels are 3.13 arcsec in size.
The Lyman-break galaxies from the HDF have UV morphologies that tend to be less regular and more fragmented than their ground-based counterparts. In part, this can be due to the increased depth of the HDF over the WFPC-2 images of the ground-based sample. This has certainly made fainter features being detected with a relatively high S/N ratio and these can now contribute to the overall appearance of the structures. However, we must remember that most of the HDF galaxies belong to a fainter portion of the luminosity function and have, on average, a smaller redshift. Only a few of them have luminosity that can be directly compared with the ground-based ones (Steidel et al. 1996b). More importantly, as mentioned above and discussed by Dickinson, the HDF data seem to suggest that some amount of evolution in the number density of Lyman-break galaxies has taken place in the redshift range 2 z 3. Because we do not know the redshift distribution of the fainter HDF U-band dropouts - this could plausibly be dominated by galaxies at z 2 - it is not entirely clear that the HDF and ground-based Lyman-break galaxies belong to the same "population". It would be quite plausible that a insurgent population at z ~ 2 be characterized by a more irregular UV morphologies than the one at z ~ 3.
Overall, the morphology of the UV light of the Lyman-break galaxies is suggestive of relaxed structures. In particular, the fact that these morphologies are relatively regular for galaxies with high absolute luminosity (large star-formation rates) and become more irregular and fragmented at fainter absolute luminosities reinforces the idea that star formation takes places in virialized dark halos with an efficiency that is directly related to their mass and/or the degree of relaxation. The presence of relaxed structures at these redshift is certainly not surprising and it is actually consistent with the expectations of most theories of galaxies formation, which predict that the most actively star-forming regions are located in correspondence of the bottom of the potential well of virialized dark halos (e.g., Baugh et al. 1998). What it is interesting here is that the observed sizes are relatively large. As discussed by Giavalisco, Steidel & Macchetto 1996) the half-light radii and isophotal diameters of LBGs are similar to the effective radii re and sizes of present-day elliptical galaxies of intermediate luminosities and bulges of spirals. At z ~ 3 these are about one order of magnitudes larger than the expectations from the semi-analytical models based on the CDM theory (e.g., Baugh et al. 1998). Furthermore, the theory predicts systems dominated by a disc morphology, while the observations show the predominance of spherically symmetric structures, with very few cases of highly elongated systems (an example of these rather rare cases is 0000-263-C10 whose optical and near-IR images are shown in Figure 8, top row).
In this respect it is interesting to compare the UV morphologies of high- and low-redshift galaxies (Giavalisco et al. 1996). Figure 7 shows local galaxies, observed with UIT and the SW camera at ~ 1550 Å, as HST would have observed them in the F606W passband if they were placed at redshift z = 3 (first and third column of the figure except the top panel of 3rd column. See labels), compared against HST LBGs. The images are all photometrically consistent, however we had to boost the surface brightness of the local galaxies by a factor of 100 to detect them with the same exposure time of the HST observations (5 hours). This exercise shows that 1) the sizes of the inner regions (i.e., where the oldest stellar population are generally thought to be located) are similar to those of the LBGs; 2) the local galaxies have significantly less star formation per unit volume (proportional to the UV surface brightness) than their distant counterparts and, as the images show, this is distributed in the disc, while the bulges or central regions contribute very little.
Figure 7. WFPC2+F606W images of LLGs and local galaxies as HST would observe them if placed at z = 3. We had to boost their surface brightness 100x to detect them in 5 hr except NGC 1068, shown before and after the boost. Panels are 7 arcsec in size and q0 = 0.1.
In local galaxies star formation is generally localized in small regions
that involve a
small fraction of the stellar mass. Thus, the UV morphology can be
form the optical one, and the effects of the morphological
k-correction can be relatively
large, shifting the position of the galaxy in the Hubble sequence
towards later types at
bluer wavelengths. The LBGs, on the other hand, are systems that are
bulk of their stars, and star formation is propagated to a much larger
fraction of their structure. Therefore, in this case it is plausible
that their UV morphology more closely
traces that of the final structure. This seems directly supported by the
the rest-frame UV and optical morphologies of the Lyman-break galaxies
recent NICMOS images (Giavalisco et al. 1998, in prep.), as shown in
Figure 8. Comparison between the rest-frame
UV and optical morphologies of 2 Lyman-break
galaxies, 0000-263-C10 at z = 3.238 (top) and 0000-263-D6 at
z = 2.961 (bottom). The UV
images have been obtained with WFPC-2/PC and the F606W passband
(rest-frame 1500 Å).
The optical images are from NICMOS-2 and the F160W passband (rest-frame
4000 Å). The
NICMOS images have been resampled to the PC resolution of 0.046
arcsec/pix. Panels are
4.6 arcsec in size. In both cases, sizes and overall morphology at UV
and optical wavelengths
are quite similar. Galaxy 0000-263-C10 is an example of the relatively
rare cases with
elongated morphologies, suggestive of a disc system. The UV light
profile of this galaxy is well
approximated by an exponential law
Figure 8. Comparison between the rest-frame UV and optical morphologies of 2 Lyman-break galaxies, 0000-263-C10 at z = 3.238 (top) and 0000-263-D6 at z = 2.961 (bottom). The UV images have been obtained with WFPC-2/PC and the F606W passband (rest-frame 1500 Å). The optical images are from NICMOS-2 and the F160W passband (rest-frame 4000 Å). The NICMOS images have been resampled to the PC resolution of 0.046 arcsec/pix. Panels are 4.6 arcsec in size. In both cases, sizes and overall morphology at UV and optical wavelengths are quite similar. Galaxy 0000-263-C10 is an example of the relatively rare cases with elongated morphologies, suggestive of a disc system. The UV light profile of this galaxy is well approximated by an exponential law (Giavalisco et al. 1996b).