Missing metals, lost gas, and inadequate dwarfs.

To follow on from the last post, this is another round up of recent papers or preprints related to galactic winds. This time I'll provide much less commentary.

I will start by noting that it is a commonly held view that if galaxies do lose metals via outflows, then these outflows must be most effective in the lowest mass galaxies, largely based on the plausibility argument that low mass galaxies has shallower gravitational potential wells and hence less energy is required for gas to escape. In this view metal ejection and winds are only effective in dwarf galaxies (M_galaxy Dekel & Silk 1986, Ferrara and Tolstoy 2000).

Although I initially believed this idea while I was a grad student, there really wasn't any good physical reason to believe it. My theoretical work with Ian Stevens confirmed what others (See eg. De Young and Heckman 1994, or Suchkov et al 1994) had already pointed out - what controls superwind dynamics are the starburst energy and mass injection and the ambient ISM the ejecta interacts with, not the galactic gravitational field. Observationally the most spectacular, either largeest, brightest or fastest winds all appeared in large, M_galaxy > 1e10 Msun, galaxies and not in the dwarf starbursts. Worse still, the dwarf starbursts have the lowest optical outflow velocities and in both optical and X-ray typically show only small bubble or shell like structures rather than clear minor-axis outflows.

Having dispensed with background information, let us proceed to the new results:

Bouche et al (2007, Monthly Notices of the Royal Astronomical Society, Volume 378, Issue 2, 525, astro-ph/0703509) find that at redshift z~2.5 about a third of all the heavy element created by stars up to that point had been ejected from their original birth sites within galaxies. They find that much of this ejection must have occurred in galaxies with Blue-band luminosities between 1/10 and 1/3 of the characteristic galaxy L_B (L_B^*) at that redshift.

By way of comparison, L_B^* ~ 7x10 Lsun at z~2 and not much different now, with L_B^*~6e10 Lsun at z~2 (Lilly et al, 1995, ApJ, 455, 108). The classic local superwind galaxies, which are not in dwarf galaxies but in small to medium mass late-type galaxies, have 0.3e10 < href="http://arxiv.org/abs/0707.1345">astro-ph/0707.1345), using a very different (and rather more model-dependent) method using galaxies selected the SDSS. They calculate the fraction of baryons (note: not just metals) lost by galaxies as a function of look-back time and galaxy mass. Although they interpret their results as being consistent with the "classic dwarfs are more important" scenario, these results are actually contradict it in several ways. (1) Ejection fractions are both non-zero and non-negligible (10-20%) in galaxies up to 1e12 Msol, (2) the galaxy luminosity function will ensure that the net contribution from all galaxies will be largest from objects in the M ~ 1e10 range, not the tiny dwarfs, (3) the Mac-Low and Ferrara simulations of outflows from dwarf galaxies (not actually of starburst strength) showed that gas ejection was difficult and inefficient. Note that metal ejection and gas or mass ejection are distinctly different things.

Finally, still on superwinds, Cooper et al (2007, Ap&SS, currently online only) have reported on their initial 3-D hydrodynamical simulations of a superwind. At present the region covered is only the starburst region itself, although at good resolution. They generate extensive H-alpha filamentation by initially introducing (by hand) a large population of dense clumps within the starburst region. In other respects the models appear to reproduce many of the structures, in particular regions of soft X-ray emission, seen in the older 2-D cylindrically symmetric simulations by myself and others, which is nice.