Diffuse H in a Fractal Interstellar Medium

Intercloud Medium

A turbulent interstellar medium should generate holes, tunnels and other empty structures as a result of fast clearing action by random supersonic motions. The origin of this energy might be supernova explosions (Norman & Ferrara 1996), but the clearing can occur even far away from the explosions as the turbulent energy propagates around, perhaps on magnetic field lines. The source of turbulent energy might also be connected with galactic rotation because the HI velocity dispersion seems to remain fairly high even in the outer disks of some galaxies (Dickey, Hanson & Helou 1990), where the star formation activity is low.

An analogy to the turbulent structures envisioned here might be with the surface of highly agitated water in a pan. The height of the water corresponds to the density of the gas in a compressible fluid. Regions of strong convergence in both the interstellar gas and the pan make dense or high peaks, and regions of strong divergence make holes or valleys, respectively. The resulting gradients then drive more motions and the compression and splashing create new structures elsewhere at a later time. This structuring continues as long as there is turbulent energy in the fluid.

There are several implications for the formation of random holes by turbulence. First, the McKee & Ostriker (1977; "MO") model of a supernova dominated ISM is not necessary and may even be inappropriate for several reasons (E97a):

1. Supernovae are not as large as MO proposed, and therefore they may not overlap, because the interstellar boundary pressure for each remnant is 10 times higher than MO assumed.

2. The O VI from hot gas is not pervasive but is confined to only a few regions per kpc (Shelton & Cox 1994).

3. The filling factor of real shells is only around 10% in our Galaxy disk, not enough to fill the midplane intercloud medium with hot gas (Oey & Clarke 1997).

4. Supernovae are highly clustered into OB associations because of the fractal cloud structure, so the rate of explosions that have access to low density gas is much lower than MO assumed.

5. Most clouds have intricate small-scale structure along their periphery, much smaller than the lower limit of 1 pc for the McKee-Ostriker clouds in a pervasive hot intercloud medium.

6. Clouds hit by a supernova shock are shredded and swept back (Klein, McKee, & Colella 1994) thereby absorbing the shock momentum. The radiation from the dense clouds also removes the supernova energy. Thus supernovae do not simply go around the clouds and expand into the pure intercloud medium, as the MO model assumed.

7. Supernova remnant shells are arched in a concave fashion; in the MO model, these visible shells were supposed to be only the shocked ambient clouds, while the actual hot remnant extended far beyond these shocked clouds. Shocked clouds are now known to be comet-shaped, however, with tails pointing outward, not arched parallel to the shock front. Such comets are observed in the Gum nebula, for example, but not generally in supernova shells.

A viable replacement for the McKee-Ostriker model is one based largely on turbulent motions, especially since the turbulent and fractal structure is observed almost everywhere anyway. Similar ideas were expressed by Passot, Vazquez-Semadeni, & Pouquet (1995), without specific concern for the fractal aspect that is emphasized here.

Of course, the turbulence interpretation for the intercloud medium would not apply to the holes inside of giant shells and chimneys that surround massive star-formation sites - only to the low-density part of the ISM that is between these sites.

© Copyright Astronomical Society of Australia 1997