Dwarf galaxies, by definition of their class, are galaxies of low mass and size. This directly implies that they have a much weaker gravitational potential well than typical spiral galaxies and fill a smaller volume with gas and stars. All dwarf galaxies, even very low mass systems, show quite complicated star formation histories (e.g. Mateo 1998). From the color-magnitude diagrams one derives the presence of some low level star formation activity over most of the lifetime, sometimes interrupted by short intervals of strongly enhanced star formation rate. These results and the presence of blue compact dwarf galaxies, where the current star formation rate is so high that these dwarf galaxies appear as isolated giant HII regions, tells us that bursts of star formation are indeed a natural process in dwarf galaxies, as predicted by models of stochastic self propagating star formation (Gerola et al. 1980). Bursts of star formation, meaning spatially and temporally correlated energy input from massive stars and supernova explosions inside physically small systems, lead to a strong response of the gas in the host galaxy.
Analytical and numerical modeling of the reaction of the gas to the energy input from stellar winds and supernovae is an active topic since the papers of Castor et al. (1975) and Weaver et al. (1977). Recent examples are e.g. Freyer & Hensler (2000), Strickland & Stevens (1999), Mac Low & Ferrara (1999), and Tomisaka (1998). For a review of the basic ideas, see Tenorio-Tagle & Bodenheimer (1988). The result of the energy input into the interstellar medium (ISM) is basically an expanding bubble of hot gas inside the substrate of cool gas of the host galaxy. The hot bubble is enclosed by a dense cool shell, which has ionized gas at the inner boundary layer between the hot gas and the shell. If the bubble grows to a linear diameter comparable to the neutral gas scale height of the galaxy, the expansion speeds along the z-axis (e.g. Mac Low & McCray 1988) and the shell expands into the lower halo of the host galaxy while starting to deform and break due to Rayleigh-Taylor instabilities (e.g. Mac Low et al. 1989).
Since massive stars and supernova explosions are the dominant source of heavy elements, the newly processed material is located inside the shells. Whether and to which degree it is moved upward out of a galaxy and what happens to this gas in the lower halo, is of crucial importance for the understanding of the chemical evolution of dwarf galaxies (e.g. Hensler & Rieschick 1999). In the work Mac Low & Ferrara (1999) for the first time a set of hydrodynamical simulations incorporating a relatively detailed model of the dwarf galaxy potential (including dark matter) was used and the simulations for a whole dwarf galaxy was run for more that 100 Myr. Still, the authors had to compromise e.g. by relatively basic treatment of the cooling processes and ignoring magnetic fields.
The hot gas inside the bubbles is predicted to be in the temperature range of 105 and 107 K (e.g. Weaver et al. 1977), which implies that the plasma will radiate in the extreme UV and soft X-ray regime. The ionized gas at the boundary layer between the hot interior and the cool shell wall should be visible in optical and UV emission lines, while the cool shell itself can be observed in 21cm emission. Since density of the substrate medium and especially the size of the energy depositing stellar association (from few O or even B stars to many thousands of OB stars as e.g. inside giant HII regions like 30 Dor or NGC 5471 in M101) span a large parameter space, the size of the bubbles can vary a lot, from pc to kpc scale (Chu 1995). The (more abundant) small bubbles do not break out of the disk of a galaxy. These bubbles do still structure the interstellar medium and should lead to ionized filaments, as observed in the Magellanic Clouds (e.g. Kennicutt et al. 1995) and the Milky Way (Haffner et al. 1999). They also lead to large regions of hot gas observed in the LMC (e.g. Chu & Mac Low 1990, Bomans et al. 1994) and the Milky Way (e.g. Snowden et al. 2000).
The observation of warm and hot gas in dwarf galaxies allows therefore to study the mechanism shaping the topology and phase structure of the ISM as well as the processes responsible for the chemical and (at least partly) the dynamical evolution of the host galaxy. Dwarf galaxies are supremely suited for this task. They present the most extreme environment for feedback of massive stars on the interstellar medium due to their shallow potential wells, their small sizes, and the absence of complicating other factors like density waves.