Astrophysicists have long wondered as to what confines gas clouds within the Galaxy. In 1956, Spitzer speculated that a rarefied (~ 10-3 cm-3), hot gas (~ 106 K), extending a kiloparsec or more above the galactic plane, would confine the diffuse gas clouds observed far above the plane and would prevent their expansion and dissipation. Confirmation of the predicted corona came 17 years later with the advent of space astronomy (Table 1). Present day satellites allow for the direct detection of diffuse x-ray emission and ultraviolet/x-ray spectral lines from the highly ionized trace elements within the hot gas (see the review by Spitzer 1990). The hot gas is thought to arise from the action of supernovae as we discuss below (see Fig. 4).
The existence of widespread hot gas in the disk of the Galaxy comes from observations of O+5 ions at ultraviolet wavelengths, and the direct detection of soft x-ray emission. A clear demonstration of coronal halo gas far from the disk has been harder to come by. A wide range of ions is observed (Si+3, C+3, N+4) towards the halo. Current models suggest that ultraviolet light from disk stars can only account for some of the ionization. The N V absorption lines appear to require collisional heating from a pervasive hot corona (Sembach & Savage 1992).
Figure 2. A simulation of the breakout from a galactic disk of a superbubble driven by multiple supernovae. The density of gas in the disk and superbubble are shown in cross-section, with only the upper half of the disk shown. Red is high-density gas, while blue is low-density gas, with other colors of the rainbow intermediate. Each side of the image is approximately 800 pc long. (Courtesy of Mordecai-Mark Mac Low, American Museum of Natural History).
From a theoretical standpoint, the expectation is that hot young stars and supernovae punch holes or blow bubbles (Fig. 2) in the surrounding gas, and the diffuse hot component escapes into the halo through buoyancy. In fact, there are spectacular examples of bubbles seen at 21 cm in the Galactic interstellar medium (see Fig. 3). But there are many outstanding problems with these models, not least of which are complications imposed by the magnetic field. Since the gas is thermally unstable, it is equally probable that the gas undergoes a `cooling flow' or `fountain flow' back towards the disk.
Figure 3. An image of a large Galactic HI supershell (white region at center). The empty supershell has a central brightness temperature of about 3~K; the shell edges have a brightness temperature around 60~K (black). The shell also shows narrow channels which appear to extend to the Galactic halo, forming a "chimney" above and below the plane. The shell lies at a distance of about 6.5 kpc, has a diameter of roughly 600 pc and extends more than 1.1 kpc above the Galactic plane. The data were obtained at the Parkes Radio Telescope as part of the Southern Galactic Plane Survey. (Courtesy of N.M. McLure-Griffiths & J.R. Dickey, University of Minnesota).
Filling the space between clouds are two additional components of the intercloud medium. By mass, most of the intercloud medium is in the form a `warm neutral' or a `warm ionized' medium. These phases extend far beyond the thin disk of cold gas (Table 1). The existence of the warm ionized medium, was firmly established by 1973 from three independent observations: (i) low frequency radio observations by Hoyle & Ellis, (ii) time delays in radio pulses from pulsars (see below), and (iii) through direct observation of H+ recombination emission by wide-field Fabry-Perot interferometers. This gas has a density of roughly 0.1 cm-3 and a temperature near 104K. The dominant source of ionization appears to be dilute ionizing flux from young stars in the disk, although some models suggest that the cooling radiation in old supernova remnants can be important. The deepest optical spectra to date show that the ionized gas extends to at least 5 kpc into the halo in some cases, and extends even further in radius than the HI disk.
Figure 4. These images show ROSAT false-color images of the Corona Australis dark molecular cloud. The contours show the 100 micron emission from dust in this cloud measured by the IRAS infrared satellite. The self-scaled images are for two different energy bands: (a) 100 eV - 300 eV (C band), and (b) 500 - 1100 eV (M band). These soft (low energy) x-rays are absorbed by interstellar dust and gas. Because the M band x-rays are more penetrating than the C band x-rays, they are absorbed more strongly in the core of the cloud than in the periphery, while the C band x-rays from beyond the cloud are completely absorbed over the entire cloud. These images demonstrate that much of the x-ray flux originates from beyond the cloud. (Courtesy of D.N. Burrows, Penn State University.)
Roughly half of the interstellar HI appears to be located in the `warm neutral' component of the intercloud medium. This intercloud HI was first identified in 1965 as the source of the ubiquitous, relatively broad (velocity dispersion ~ 9 km s-1) 21-cm emission features that had no corresponding absorption when viewed against bright background radio sources. The large velocity dispersions and the absence of absorption imply temperatures of 5000 to 10,000 K. Observations of the Ly absorption line of HI toward bright stars show that this gas has a mean extent from the midplace of 500 pc, i.e., much thicker than the cold neutral disk. If the warm neutral medium is in pressure equilibrium with the cold component, then it would be clumped into regions occupying 35% of the intercloud volume with a density of 0.3 cm-3 at the midplane (Table 1), although these numbers are highly uncertain.