Annu. Rev. Astron. Astrophys. 2003. 41: 191-239 Copyright © 2003 by Annual Reviews. All rights reserved

2. OVERVIEW OF HOT GAS FLOWS IN AND NEAR ELLIPTICAL GALAXIES

Optical evidence indicates that many or most massive elliptical galaxies formed at high redshifts (z 2) and there is strong theoretical evidence that they formed by mergers in group environments. Later, many ellipticals and their groups merged into massive galaxy clusters. Those that did not, the so-called isolated ellipticals, are surrounded by unmerged group companions and hot gas. As a consequence of their formation and galactic dynamics, big elliptical galaxies are often found near the centers of groups and rich clusters. It is natural that the diffuse thermal gas contained in these hot stellar systems is also hot.

There are two main sources of hot gas in elliptical galaxies: internal and external. Evolving stars inside the elliptical galaxy continuously eject gas at a rate ~ 1.3[L_B / (10¹¹ L_B, )] M yr^-1. It is generally assumed that gas ejected by orbiting red giant stars passes through shocks and is raised to the stellar kinematic temperature T_* T_vir µm_p ² / k ~ 10⁷ K ~ 1 keV where µ is the molecular weight and is the stellar velocity dispersion. Type Ia supernovae provide some additional heating. The large X-ray luminosities of massive E galaxies, L_x ~ 10⁴⁰ - 10⁴³ ergs s^-1 for L_B > 3 × 10¹⁰ L_B, , indicate that most of the internally produced gas is currently trapped in the galactic or group potential. But at early times, when most of the galactic stars were forming, Type II supernovae were frequent enough to drive winds of enriched gas into the local environment. Gas expelled in this manner from both central and non-central group or cluster galaxies has enriched the hot gas far beyond the stellar image of the central luminous E or cD galaxy. In time, some of this local (circumgalactic) gas flows back into the central galaxy, providing an external source of gas. Continued accretion from the ambient cosmological flow that is gravitationally bound to the group or cluster is an additional source of external gas. As diffuse external gas arrives after having fallen through the deeper potential well of the surrounding group/cluster, it is shock-heated to the virial temperature of the galaxy group/cluster. This more distant accreted and shocked gas is hotter than gas virialized to T_* deeper in the stellar potential of the E galaxy, and the two together form an an outwardly increasing gas temperature that is commonly observed. The accumulated mass of circumgalactic gas with T > T_*, bound to the dark matter halo, can extend far beyond the optical image of the luminous E galaxy.

The electron density of the hot gas in giant elliptical galaxies is typically n(0) ~ 0.1 cm^-3 at the center and declines with radius as n r^-1.25±0.25. Depending on its spatial extent, the total mass of hot gas in massive E galaxies varies up to at least several 10¹⁰ M or about 1 percent of the total stellar mass. This is only a few times less than the gas to stellar mass ratio in the Milky Way. The iron abundance in the hot gas in ellipticals increases from z_Fe ~ 0.2 - 0.4 beyond the optical image to z_Fe ~ 1 - 2 in the center where it is evidently enriched by Type Ia supernovae.

To a good approximation, the hot gas in and near elliptical galaxies is in hydrostatic equilibrium. Supersonic winds are not common in well-observed massive ellipticals (L_B 3 × 10¹⁰ L_B, ) because they would have low gas densities and much lower X-ray luminosities. A characteristic feature of the hot gas is that the the dynamical and sound crossing times are nearly equal, as expected in hydrostatic equilibrium, and both are much less than the radiative cooling time. Any cooling-induced flow is therefore highly subsonic, essentially in hydrostatic equilibrium. This equilibrium can be disturbed by mergers or by energy released in an active galactic nucleus (AGN) associated with a supermassive black hole in the core of the central elliptical. Nevertheless, by assuming hydrostatic equilibrium, the total mass distribution M_tot(r) has been determined for many galaxies and clusters from X-ray observations. If a galaxy group has been relatively undisturbed for many Gyrs, the metal enrichment in the hot gas may retain a memory of the (largely SNII-driven) galactic winds that occurred in the distant past. We are just beginning to exploit this gold mine of information.

In addition to the accretion shock, hot gas in ellipticals is heated further by Type Ia supernovae and by dissipation of mass lost from orbiting stars. As a consequence of the ~ 0.7 M of iron that each Type Ia is thought to contribute, the observed iron abundance in the hot gas is a measure of the supernovae heating for a given flow model. In low luminosity ellipticals, which have shallower gravitational potentials, current supernova heating may be sufficient to drive a galactic outflow. But in optically bright, X-ray luminous E galaxies, supernova heating is generally unable to balance the radiative losses from the hot plasma that produces the X-ray emission we observe. Paradoxically, the loss of radiative energy from the hot gas does not result in a proportional decrease in the local gas temperature. Instead, this loss of thermal energy is immediately compensated by Pdv compression in the gravitational potential of the galaxy/group that maintains the temperature ~ T_vir necessary to support the surrounding atmosphere of hot gas. However, near the center of the flow where there is no deeper galactic potential, catastrophic radiative cooling could in principle occur.

But something is wrong, perhaps radically wrong, with this simple "cooling flow" model. An estimate of the mass cooling rate required to generate the observed X-ray luminosity in a bright E galaxy, (2 µ m_p / 5kT)L_{x, bol} reveals that several 10¹⁰ M of gas should have cooled somewhere within the galaxy over a Hubble time. The cooling cannot be too concentrated because this mass exceeds the masses of known central black holes by factors of 10 - 20 and its gravitational attraction on the hot gas would produce an unobserved central peak in X-ray emission. The traditional solution to this problem is to invoke an ad hoc mass "dropout" assumption in which the gas somehow cools throughout a large volume of the flow. Unfortunately, such distributed cooling cannot result from thermal instabilities following small perturbations in the hot gas. Instead, larger perturbations formed in turbulent regions may be required, but the details remain uncertain.

Even more astonishing, detailed X-ray spectra taken with XMM show little or no emission from ions cooling at temperatures much below ~ T_vir / 2, implying that little gas completely cools or perhaps none at all! There are two possible explanations: (1) the gas is cooling, but it is not visible and/or (2) the gas is truly not cooling and the radiative losses are offset by some source of heating.

In principle, X-ray emission from cooling gas can be attenuated or hidden (1) by spatially distributed colder gas that absorbs softer X-rays or (2) if the cooling is somehow accelerated so that the X-ray line emission from the cooling gas is reduced. Although X-ray spectra provide some support for distributed absorption at energies 1 keV, the absorbing gas would need to be colder, T_abs 10⁶ K, spatially extended and very massive M_abs 10¹⁰ (r / 10 kpc)² M. Emission from this absorbing gas has not been observed. Alternatively, rapid cooling may be possible in localized regions of high metallicity such as remnants of Type Ia explosions or if cold gas rapidly mixes with the hot gas. Cooling may also be very rapid in the dust-rich gas recently ejected from evolving stars. These processes have not been studied in detail.

The currently most popular explanation for the absence of spectroscopic evidence for cooling is that the hot gas is being heated by active galactic nuclei (AGN) in the cores of flow-centered giant ellipticals. This is reasonable since almost all bright E galaxies have extended non-thermal radio emission at some level in their cores. Additional support for this heating hypothesis is provided by the spectacular X-ray images from Chandra in which the hot gas in the central regions of virtually all elliptical galaxies observed so far appears to be highly disturbed and irregular. In some cases bubbles (X-ray cavities) filled with relativistic or superheated gas have displaced the 1 keV thermal gas and appear to be buoyantly floating upward in the atmosphere. Many observers have noted with astonishment that the gas just adjacent to the bubbles is cooler than average, indicating that the holes have not been recently produced by strong shocks.

Certainly the luminosities of typical AGN are sufficient to offset the X-ray luminosities L_x of classic cooling flows and maintain their gas temperature near T_vir. However, the problem with the heating hypothesis, which is widely unappreciated, is that it is difficult to communicate this AGN energy to the hot gas at larger radii and still preserve the globally observed hot gas temperature and density profiles.