Proceedings of the Workshop
"The Magellanic Clouds and Other Dwarf Galaxies"
of the Bonn/Bochum-Graduiertenkolleg

Starbursts in dwarf galaxies:

Blown out or blown away?

Mordecai-Mark Mac Low1 and Andrea Ferrara2

1Max-Planck-Institut für Astronomie, D-69117 Heidelberg, Germany
2Osservatorio Astrofisico di Arcetri, I-50125 Firenze, Italy

Received 15th March 1998
Abstract. We numerically model the impact of superbubbles produced by starbursts on the gas in dwarf galaxies, using realistic gravitational potentials including the contributions from dark matter haloes, and galaxy radii based on empirical laws. We explore supernova rates from one every 30 000 yr to one every 3 million yr, equivalent to steady mechanical luminosities of Lm=0.1-10·1038 erg s-1, occurring in dwarf galaxies with gas masses Mg=106-109 Msun. We give quantitative results for when blowout will or will not occur in galaxies with 106≤Mg≤109 Msun. Surprisingly, we find that the mass ejection efficiency is very low in such outflows for galaxies with mass Mg≥107 Msun. Only galaxies with Mg≤106 Msun have their interstellar gas blown away, and then virtually independently of Lm. On the other hand, metals from the supernova ejecta are accelerated to velocities larger than the escape speed from the galaxy far more easily than the gas. We find that for Lm=1038 erg s-1, about 97% of the metals are retained by a 109 Msun galaxy, but this fraction is already only 40% for Mg=108 Msun and decreases to 0.27% for Mg=107 Msun.

1. Introduction

The stellar winds and supernovae from the massive stars formed in starbursts produce huge shock waves that can blow out of the interstellar medium of a dwarf galaxy, and might even be able to blow away all the gas from the galaxy (e.g. De Young & Heckman 1995). It has become clear that such supernova-driven winds play a crucial role in the evolution of dwarf galaxies since they regulate the mass, metal enrichment, and energy balance of the interstellar medium (ISM) in these galaxies. Theoretical work on this includes papers by Larson (1974), Dekel & Silk (1986), Silk et al. (1987), Vader (1986, 1987), De Young & Gallagher (1990), and Ferrara & Tolstoy (1998). The observational evidence in support of the existence of outflows from dwarf galaxies has grown rapidly in recent years, with examples including Meurer et al. (1992), Martin (1996), Della Ceca et al. (1996), Bomans et al. (1997), and the survey by Marlowe et al. (1995).

Here, we attempt to answer the following questions about the effects of starbursts on dwarf galaxies:

  1. What are the conditions for either blowout or blow away to occur?
  2. What fraction of gas escapes the galaxy when either occurs?
  3. What is the fate of metals ejected by the massive stars of the starburst?
We investigate these problems taking into account the full structure of dwarf galaxies, including their dark matter haloes. We set up the ISM density distribution in the galaxies using the rotation curves of Persic et al. (1996) and the mass-radius relationship of Ferrara & Tolstoy (1998). We combine analytic calculations with numerical hydrodynamic simulations using ZEUS-3D*, a code using the algorithms described by Stone & Norman (1992). Our work is described in greater detail by Mac Low & Ferrara (1998).
*Developed by the Laboratory for Computational Astrophysics at the National Center for Supercomputer Applications, and available for community use by registration at the email address lca@ncsa.uiuc.edu

2. Analytics

We distinguish between two potential results of a central starburst. First, in a "blowout", the central supernova explosions blow a hole through the galactic gas distribution, parallel to the steepest density gradient (usually along the rotation axis), accelerating some fraction of the gas and releasing the energy of subsequent explosions without major effects on the remaining gas. Second, in a "blow away", all, or nearly all, of the ambient ISM is accelerated above the escape velocity and is lost to the galactic potential well.

We derive analytic conditions for each of these to occur. The blowout condition can be derived by requiring that the blowout velocity vb exceeds the escape velocity of the galaxy ve. The mechanical luminosity Lm from stellar winds and supernovae required for blowout to occur is
Eq. (1)
where Mg is the visible mass of the galaxy, cs is the sound speed in the ISM, and h=H0/(100 km s-1 Mpc-1) is the scaled Hubble constant (which enters due to our calibrations of dwarf galaxy structure). This is a relatively easy luminosity to reach in typical dwarf galaxies.

To entirely blow the gas away, on the other hand, the momentum of the gas swept up in the plane of the galaxy at the time of blowout must be larger than the momentum required to accelerate the remaining ISM to a velocity greater than the escape velocity. This requires a luminosity of
Eq. (2)
a much more difficult luminosity to reach except in very small galaxies, or ones with hot, high sound speed ISMs that have large scale heights.

3. Numerics

For our numerical models, we use ZEUS-3D, including equilibrium radiative cooling, an implicit energy equation, and a tracer field to follow the metal-enriched gas ejected by the starburst. The supernovae of the starburst are modelled as a constant luminosity central wind driven by a thermal energy source that lasts for 50 Myr, the lifetime of the least massive star able to become a supernova (e.g., McCray & Kafatos 1987). We assume azimuthal and equatorial symmetry, and use outflow boundary conditions on the other two axes.

In Fig. 1 we show the density distributions of all the models in our parameter study at a time of 100 Myr. Most of the bubbles have begun to accelerate, and show strong Rayleigh-Taylor instabilities. After energy input ceases, the holes in the planes of the galaxies recollapse under the influence of gravity and the pressure of the disk gas, except in the extreme cases of low mass and high mechanical luminosity, where the disk gas escapes the potential of the dark matter halo, and is swept completely off the grid at late times. In cases of either blowout or blow away, low density, metal-enriched gas originating in the central winds and supernovae spreads over regions of tens of kpc. Much of it is travelling at high enough velocity to escape even from the halo potential.

We can directly compute the efficiency of ejection of ISM, ξ, and of metal-enriched gas, ξZ, from our models. The question we ask is how much of each is travelling at speeds higher than the local escape velocity due to the potential of the dark matter halo. In Table 1 we show the ejection efficiency ξ for each of our models. Only in our most extreme models, with masses of 106 Msun, is most of the mass ejected. In more massive objects, less than 7% of the mass is ejected, usually far less.

In Table 2 we give ξZ for each model. In the more massive galaxies and at lower supernova rates, a significant fraction of the metal enriched gas is retained in the gravitational well of the dark halo, and will eventually fall back on to the galaxy, while at lower masses and higher luminosities, virtually all of the metals escape the grasp of the halo and travel freely into the surrounding intergalactic space.

Table 1. Mass Ejection Efficiency, ξ
Luminosity Visible Mass Mg/Msun
(1038 erg s-1) 106 107 108 109
  0.1 9.4 (-2) 4.6 (-3) 0.0 0.0
  1.0 3.8 (-1) 1.4 (-2) 6.5 (-4) 1.0 (-6)
10.0 1.0 6.8 (-2) 4.8 (-3) 1.3 (-4)
Table 2. Metal Ejection Efficiency, ξZ
Luminosity Visible Mass Mg/Msun
(1038 erg s-1) 106 107 108 109
  0.1 8.6 (-1) 5.9 (-1) 0.0 0.0
  1.0 1.0 1.0 6.0 (-1) 2.6 (-2)
10.0 1.0 1.0 9.8 (-1) 5.5 (-1)

These results have several implications for the evolution of dwarfs. First, outflows from dwarf galaxies should be strongly metal enriched. As a consequence, it appears that dwarfs could be the major pollutors of the IGM, and certainly have major effects on the environment in which they live. Starbursts in dwarf galaxies may be traced and studied in detail by X-ray and optical emission-line studies. The results presented here show that huge gaseous halos, with sizes of dozens of kpc are produced, with regions of high X-ray emissivity close to the galactic disk. Relatively cool, dense filaments also occur near the galaxy, well within the external shock, due to shell fragmentation. High spatial resolution spectra of the filaments may be useful to investigate the radiation field in the halo of dwarfs and the escape fraction of ionizing photons from massive stars in the disk, since this cool gas should be predominantly photoionized and hence show up in optical emission lines. Since these filaments are surrounded by the hot gas, the X-ray emission may actually be strongest in regions close to the filaments, where evaporation takes place. Thus, the bulk of the observed X-ray emission may come from these conductive interfaces, in which the gas is far out of ionization equilibrium, and hence emitting strongly.

[Click here to see Fig. 1!]
Acknowledgments. We are grateful to E. Tolstoy, D. Bomans, R.-J. Dettmar, G. Golla, and L. van Zee for discussions of the observations. Computations were performed at the Rechenzentrum Garching of the Max-Planck-Gesellschaft. We each thank the other's institute for hospitality during work on this paper.

References

Links (back/forward) to:
First version: 09thAugust,1998
Last update: 28thSeptember,1998
Jochen M. Braun   &   Tom Richtler
 (E-Mail: jbraun|richtler@astro.uni-bonn.de)