Holes in the IGM, superwinds and mysterious winds from post-starburst galaxies

It has been a while since I have presented any posts about superwinds. June has been a busy month for a variety of reasons, and various duties have cut down on the time I get to spend on doing actual science.

Nevertheless my next paper, a theoretical investigation of diffuse hard X-ray emission from starburst regions and what this can tell us about the efficiency of massive star "feedback" is progressing nicely, but slowly. I've been forced to go back to looking at the X-ray data on M82 even more carefully yet again (which I thought I'd done to death in my last paper), but all the extra work may turn out to be useful for something else I'm planning... Anyway, the efficiency of supernova heating is the fraction of the explosion energy that ends up being able to create galactic winds, and is a crucial but poorly determined variable that goes into the models of galaxies used by numerical cosmologists.

There have been a variety of papers and/or preprints related to starburst-driven winds in the last few weeks, so I thought I'd collect them all together in a post and add a comment or two of my own.

We know, unambiguously, that winds exist in both local starburst galaxies and in the more-actively star-forming galaxies seen at high redshift because we can see the blue-shifted absorption from out-flowing gas in their spectra (see for example Heckman Armus & Miley 1990; Heckman et al 2000; Adelberger et al 2003 and references therein).

In nearby galaxies, like the classic starburst with superwind M82, we can also see the superwind in emission extending out to distances of typically 10 - 20 kpc from the host galaxy. The amount of emission is proportional to the square of the gas density, so as the wind expands and the density drops the wind becomes fainter very rapidly. Eventually, at the sort of distances mentioned above, the emission becomes too faint to trace further, so we don't know if that is the maximum size of the wind (although this is highly unlikely) or whether the winds extend out to much larger distances.

If the energy and newly-synthesized heavy elements in superwinds only travels less than or equal to 20 kpc then they (superwinds) are unlikely to be responsible for enriching and/or heating the inter galactic medium (IGM).

The starburst galaxies at high redshift (often called Lyman Break Galaxies) are too faint to detect winds in emission, so the only probe is absorption-line spectroscopy. The amount of absorption is proportional to the density of the appropriate ion present, and so does not drop as rapidly as emission does as the density decreases. Unfortunately these spectra do not tell you the location of the gas, or how far out the gas eventually gets, so assessing the impact of winds on the true inter-galactic medium is difficult.

Several years back Adelberger et al (2003) published evidence that appeared to show large, 100-kpc-sized, holes in the neutral hydrogen gas of the IGM around the Lyman break galaxies. They interpreted this as holes being blown by the superwinds these galaxies were known to have. This was a very important result, as it is was potentially the first direct measurement of how far out winds can go.

Kawata and Rauch have submitted a paper [which appeared on astro-ph on April 4th 2007] in which they argue that these HI holes (HI is neutral hydrogen) are not created by outflowing hot gas in superwinds, but rather the infall (accretion) of the IGM onto the galaxies heating the IGM up and hence ionizing the hydrogen. They further argue that the signature of true superwinds would be O VI absorption lines (from highly ionized oxygen at temperature of 300 000 to 1 000 000 K).

In a similar vein, but with somewhat different conclusions, is a preprint of an accepted paper by Fangano, Ferrara and Richter, which also uses SPH simulations and also investigates the expected absorption line signatures from a variety of metal lines including O VI. Their conclusion is that it would be hard to distinguish accretion from outflow.

This is interesting work, but must be interpreted with caution. Cosmological simulations lack the resolution to capture the multi-phase nature of superwinds, and hence model them in a very simple way that almost always artificially reduces their temperature (hence affecting the simulated observable absorption lines) and energy per particle (hence reducing the chance of metal-enriched gas of escaping the host galaxy).

Furthermore, getting the physics of the O VI absorbing gas right is difficult. The gas responsible for real O VI absorption in either the superwinds or the shock-heated accreted IGM is unlikely to be in the state of ionization equilibrium assumed in the simulations. In simulations which aim to model single superwinds at resolutions far superior to that achieved in cosmological simulation regions where O VI might be found (the green region in the center panel below) are the interfaces between hotter X-ray emitting gas and compact cooler clouds and clumps. These regions are physically tiny in volume, they are numerically unresolved even in these simulations, and are certainly not the huge volumes the preprints above identify with O VI absorption.

Admittedly here are a talking about findings from studies of local superwinds, studies specifically covering the material within 10-20 kpc of the host galaxy where the material driving the wind is still too hot to produce appreciable O VI absorption. At larger radii it may be that larger volumes start to reach the correct ionization state for O VI.

However, even if this is true and galaxies are surrounded by large low density regions of O VI absorbing gas it it not clear that any O VI absorption seen (I mean actually observed) toward LBGs could be unambiguously attributed to the structures produced in the Kawata and Fangano models. Physical processes not including in cosmological simulations, such a cloud stripping and thermal conduction, are in my mind the most likely the primary producers of O VI absorption in superwinds (e.g. Marcolini et al 2005). Basically the dense gas in the inner wind is more likely to produce most of the absorption column density.

The images are from our (Strickland & Stevens 2000) old 2-D simulations of superwinds. Left hand side image: Log 2-D gas number density. Black/dark blue denotes low density (log n <= -3), red high density (log n >= 1). Central image: Log soft X-ray (blue), O VI 1032+1038 (green) and H-alpha (red) emission. The 2-D model has been converted into 3-D using the assumed cylindrical geometry and the emission projected onto the plane of the sky, assuming that the minor axis of the galaxy (i.e. the major axis of the wind) is inclined by 15 degrees into the plane of the sky. Right hand side image: Soft X-ray energy-color coded image, log intensity. Red: 0.3-0.6 keV, Green: 0.6-1.0 keV, blue 1.0-2.0 keV.

Moving on, let us turn from bona-fide starburst-driven winds to something more mysterious. Tremonti Moustakas and Diamond-Stanic (2007, ApJL, 663, L77, or read astro-ph/0706.0527) have found a class of object showing fast (1000 km/s) absorption-line outflows that aren't starbursts (in fact they're post starburst) and don't seem to host significant AGN. So what is driving the winds in these objects? And why are the velocities higher than in typical starburst-driven winds? Again, the unknown size of the flow hampers interpretation.

Tremonti et al interpret these as physically-large (100 kpc-scale) fossil winds, a left over from earlier starburst activity, with the high (cool gas) velocity perhaps attributable to some (now dormant?) AGN activity.

This is plausible, although unsatisfying. Alternative include the absorption arising closer into a weak AGN in few kpc scale flow, or perhaps a SN Ia-driven wind associated with the prompt SN Ia channel.

I was also going to discuss the issue of The Missing Metals, but lunch is over and I need to get back to my M82 paper...