5.3. Modeling the detailed astrophysics of the submm galaxies

It is not possible to separate the modeling of the evolution of the population of submm galaxies fully from studies of the nature of the galaxies themselves. Their luminosities and masses (see Frayer et al., 1999) demand that the submm-luminous phase be short-lived as compared with the age of the Universe. The observational information for most of the submm galaxies is insufficient to be confident that their nature is understood at present. Of the galaxies with reliable counterparts, there are three bright Class-2 galaxies (Soucail et al., 1999; Ivison et al., 1998a, 2000a, 2001; Vernet and Cimatti, 2001) with optical redshifts and CO detections (Frayer et al., 1998, 1999; Kneib et al., 2002), and a total of four Class-1 galaxies (Smail et al., 1999; Bertoldi et al., 2000; Gear et al., 2000; Lutz et al. 2001), which are known to be either very red galaxies or formal EROs. Other submm-selected galaxies with accurate positions from radio observations (Smail et al., 2000) or mm-wave interferometry (Downes et al., 1999b; Bertoldi et al., 2000; Frayer et al., 2000; Dannerbauer et al., 2002) remain enigmatic. All that can be said about these galaxies is that they all appear to have thermal dust spectra, are all very faint at optical wavelengths, and most also appear to be very faint at near-IR wavelengths.

The Class-2 galaxies are all clearly undergoing mergers or interactions. Much less is known about the morphology of the faint, but typically extended Class-1 galaxies (Figs. 16, 17 and 20). It is certainly possible that they too are involved in interactions, which appear to trigger the dramatic luminosity of almost all the low-redshift ULIRGs (Sanders and Mirabel, 1996). Programs of ultradeep near-IR imaging on 10-m-class telescopes should soon test this idea.

Hydrodynamical simulations of gas-rich mergers by Mihos and Hernquist (1996), Bekki et al. (1999) and Mihos (2000) show the formation of very dense concentrations of gas, which could be associated with short-lived, very-intense bursts of star formation. However, at present it is not possible to simulate a representative sample of mergers with the range of geometries likely to be encountered, the necessary time resolution, and a sufficiently accurate treatment of the detailed astrophysics of star-formation to make a reliable connection between the limited observations and the underlying galaxy properties. The spatial extent of the three bright Class-2 galaxies in the optical waveband appears to be considerably greater than that of most low-redshift ULIRGs. It is thus difficult to be sure that simulations of well-studied low-redshift ULIRGs adequately represent the properties of the high-redshift submm galaxies. Note, however, that the precise spatial relationship between the optical and submm emission in these objects is still unclear (Ivison et al., 2001); the submm emission could be more compact than the optical galaxy.

Larger submm galaxy samples will be available over the next few years, boosting the likelihood that examples of the full range of submm galaxies will be available to be studied in detail. More sensitive observations of the properties of the known galaxies will also improve our knowledge of their astrophysics. One key question is the relationship of the submm galaxies to the formation of elliptical galaxies (Lilly et al., 1999). Whether the bulk of submm galaxies are high-redshift low-angular-momentum gas clouds, forming elliptical galaxies in a single episode by a `monolithic collapse' (Eggen et al., 1962), as advocated by for example Archibald et al. (2002), or galaxies observed during one of a series of repeated mergers of gas-rich, but pre-existing galaxy sub-units, likely to take place at relatively lower redshifts, as discussed by Sanders (2001), and which might ultimately yield elliptical merger remnants, is an important question that future follow-up observations will address. Existing observations of extended and disturbed counterparts to submm galaxies (Ivison et al., 1998a, 2001; Lutz et al., 2001) tend to favor the second explanation, in which well-defined pre-existing stellar systems merge. However, in either scenario it is likely that the bulk of the stellar population in the resulting galaxies form during the submm-luminous phase.