Annu. Rev. Astron. Astrophys. 2005. 43: xxx-xxx Copyright © 2005 by Annual Reviews. All rights reserved

5.3. Nature of the SMGs

Many LIRGs and ULIRGs at low redshifts have been identified with interacting or galaxy mergers. A substantial fraction show signs of AGN activity but it has been shown for the low-redshift LIRGs and ULIRGs that the starburst component dominates the energy output (Genzel et al. 1998; Lutz et al. 1998). The sources used for the redshift distribution by Chapman et al. (2003a) have been imaged with the HST. Most of them are multi-component-distorted galaxy systems (Conselice et al. 2003; Smail et al. 2004). They display irregular and frequently highly complex morphologies compared to optically selected galaxies at similar redshifts. They are often red galaxies with bluer companions, as expected for interacting, star-forming galaxies. They have higher concentrations, and more prevalent major-merger configurations than optically-selected galaxies at z ~ 2-3. Most strikingly, most of the SMGs are extraordinarily large and elongated relative to the field population regardless of optical magnitude (Chapman et al. 2003c). SMGs have large bolometric luminosities, ~ 10¹² - 10¹³ L, characteristic of ULIRGs. If the far-infrared emission arises from the star formation, the large luminosities translate to very high SFR 1000 M year^-1. Such high rates are sufficient to form the stellar population of a massive elliptical galaxy in only a few dynamical times, given a sufficient gas reservoir. SMGs are very massive systems with typical mass of 1-2 × 10¹¹ L (Swinbank et al. 2005), comparable to the dynamical mass estimates from CO observations. Genzel et al. (2004; and more recently Greve et al. 2005) have undertaken an ambitious program to study the nature of the SMGs in more details. They got CO spectra with the Plateau de Bure interferometer for 7 sources out of their sample of 12 for the CO 3-2 and 4-3 transitions redshifted in the 3 mm atmospheric window. They provide optical identifications and redshifts. The detection of these sources at the proper redshift confirms the usefulness of identification with the help of the radio sources. The median redshift of this sample is 2.4. In addition, one source was studied with the SPIFI instrument on the ESO/VLT. These observations are giving very interesting clues on the nature of the submillimeter galaxies. The gas masses obtained for these systems using CO luminosity/mass of gas determined from local ULIRGs is very large with a median of 2.2 × 10¹⁰ M (10 times larger than in the Milky Way). Using the velocity dispersion, they could infer that the dynamical median mass of these systems is 13 times larger than in Lyman-break galaxies (LBGs) at the same redshift or 5 times the mass of optically selected galaxies at this redshift. These SMGs with a flux at 850 µm larger than 5 mJy are not very rare and unusual objects, because they contribute to about 20% of the CIB at this frequency. Through multiwavelength observations, Genzel et al. (2004) get the stellar component in K band, and infer the star-formation rate and duration of the star-formation burst. They can then compare the number density of these massive systems with semiempirical models of galaxy formation. The very interesting result is that this number density is significantly larger than the predicted one, although the absolute numbers depends on a number of assumptions like the IMF. The comparison is shown in Figure 8. Such massive systems at high redshift are not easy to understand in current cold dark matter hierarchical merger cosmogonies. However, one must keep in mind that bright SMGs (S₈₅₀ > 5 mJy) that contribute 20% of the CIB may not be representative of the whole population. Gravitational lens magnification provides a rare opportunity to probe the nature of the distant sub-mJy SMGs. Kneib et al. (2005) study the property of one SMG with an 850 µm flux S₈₅₀ = 0.8 mJy at a redshift of z = 2.5. This galaxy is much less luminous and massive than other high-z SMGs. It resembles to similarly luminous dusty starbursts resulting from lower-mass mergers in the local Universe.

Figure 8. Comoving number densities of galaxies with baryonic masses 1011 M as a function of redshift. The triangle and open squares show densities of massive stellar systems at z = 0 and z ~ 1; The circle shows the density for massive SMGs at z ~ 2.7, with a factor of 7 correction for burst lifetime. Blue and red curves show the predictions of semianalytic models by the "Munich" and "Durham" groups, respectively. Dashed curves show the corresponding number densities of halos with available baryonic masses 1011 M. The two models use the same halo simulations but assume different b. From Genzel et al. (2004).

In order to link the different population of high-redshift objects, several LBGs at redshift between 2.5 and 4.5 have been targeted at 850 µm. The Lyman-break technique (Steidel et al. 1996) detects the rest-frame 91.2 nm neutral hydrogen absorption break in the SED of a galaxy as it passes through several broad-band filters. LBGs are the largest sample of spectroscopically confirmed high-redshift galaxies. Observing LBGs in the submillimeter is an important goal, because it would investigate the link, if any, between the two populations. However, the rather low success rate of submillimeter counterpart of LBGs (e.g., Chapman et al. 2000; Webb et al. 2003) argues against a large overlap of the two populations.