Annu. Rev. Astron. Astrophys. 1994. 32: 277-318 Copyright © 1994 by Annual Reviews. All rights reserved

6. COOLING FLOWS AND GALAXY FORMATION

We have seen that the most massive galaxies observed at z > 0.5 have many properties that can be interpreted as due to a surrounding cooling hot medium. Often t_cool ~ t_grav and is maximized (hence maximal cooling flow). Where is a few thousand M yr^-1 a very large galaxy can then form in a few billion years:

(8)

Cooling flows therefore must play some part in the formation of the most massive galaxies, i.e. the central cluster galaxies. Indeed, any theory of galaxy formation in which gas falls into potential wells, is heated to the virial temperature, and then cools with the possibility that t_cool t_grav (e.g. Rees & Ostriker 1977, Silk 1977, White & Rees 1978, White & Frenk 1991) requires cooling flows.

In most hierarchical models for structure formation, mass overdensities begin with t_cool < t_grav and form "normal" stars of which the more massive become supernovae. The energy feedback from these leaves most of the gas uncooled. The supernova ejecta also enriches the gas in metals. It is then incorporated into the next stage of the hierarchy. Many small perturbations do not proceed beyond this condition and appear as "normal" galaxies and loose groups. In larger perturbations, however, the total mass increases such that the object passes the cooling flow condition. This is where maximal cooling flows occur, which are thus expected in young clusters and massive groups before they merge to form richer clusters. The exact mass level at which this occurs depends on the role and fraction of any nonbaryonic dark matter (Figure 9).

Figure 9. The domain of cooling flows in galaxy formation occurs above the line where tcool = shown as the dotted line (zero metal abundance) and the solid curve (0.4 solar abundance). The mass scale shown is for self-gravitating gas clouds, but reduces by a factor of at least 10 in a universe dominated by cold dark master. In a simple hierarchical model, structures grow in successive stages proceeding from the lower left of the figure to the upper right: the metallicity of the gas increases as the structure grows, so the point at which an object passes into the cooling flow state depends upon the details of its evolution.

What happens at this stage can be deduced from our studies of nearby cooling flows. The gas is multiphase at all radii and lays down cooled gas according to M(< r) r, ( r^-2). Much of the cooled gas is in the form of very cold clouds which may efficiently form low-mass stars. Star formation may therefore switch from the "normal" IMF which occurred during the earlier phases of the hierarchy (giving rise to the observed galaxy) to an almost exclusively low-mass mode (Thomas & Fabian 1990). A massive isothermal dark halo is thereby assembled (see Thomas 1988). While a cooling flow persists, the IMF of any star formation can only be "normal" for a small fraction of the cooled gas, or massive galaxies would appear even more luminous. The dark mass of the central galaxy is continuing to increase from the cooling flow.

Some such process is required in order to account for the upper luminosity of the largest galaxies. These are inferred to occur at the boundary where t_cool = t_grav and some physical process is required to prevent more luminous galaxies from forming. Certainly more massive structures occur in the Universe, it is just that the most massive objects are mostly dark. Cooling flows are a mechanism for making the cooled gas dark. We do not know how or why, but the observations of nearby cooling flows reviewed here do show that this is so. Such a physical limit to the luminosity may help to explain the tight infrared K-band magnitude-redshift correlation noted for distant radio galaxies (Lilly & Longair 1984). If there is some alternative process that counteracts the cooling in some manner undetectable to X-ray observations, then it represents a major energy flow and determines the upper luminosity of galaxies.

The role of cooling flows in galaxy formation has also been studied by Ashman & Carr (1988, 1991). They consider both the cooling flows discussed above, and ones that can occur at very low mass scales when the virial temperature is below 10⁴ K. They investigate the total mass fraction that can be processed through cluster cooling flows and again argue that it is small unless significant heating occurs due to supernovae at the early stages of the hierarchy.

Note that it has not been argued here that most dark matter is baryonic and from cooling flows or that indeed any of it has been formed in this way except that around central cluster galaxies If, however, it is found that much of the dark matter in lower mass systems (e.g. normal galaxies and loose groups) is baryonic, perhaps through the detection of gravitational microlensing, and similar to that in the cores of clusters, then a more general connection can be pursued.

The most massive structures in the present Universe, from giant galaxies and groups upward, should thereby have passed through a cooling-flow phase. Observationally this implies a soft X-ray background from the radiation of the cooling gas, although absorption in cooled gas may reduce the observed flux by large factors. Line absorption of background (or embedded) quasars by the dense cold clouds in the flow may explain some of the many absorption lines commonly seen in the spectra of distant quasars. The hypothesis predicts that massive protogalaxies and subclusters will be a turbulent, extended mess of rapidly cooling hot gas with a massive embedded population of dense cold clouds. The smallest clouds, or cloud fragments, may mix into the hot gas and produce absorption lines of high ionization, whereas the large clouds may be predominantly neutral and create damped Ly lines. Lines from individual clouds may be narrow but the spread from all clouds may be up to 1000 km s^-1.

The formation of galaxies, and in particular massive galaxies, is therefore seen to be a complicated process with a stellar IMF that varies and depends in an indirect way on magnetic fields. Cooling-flow conditions must be common during the formation of massive galaxies and it is worth taking a careful look at nearby cooling flows to learn how they operate. These nearby examples show that much of the action is not directly detectable at visible wavelengths.