3.1. Individual shell systems
A number of individual superbubbles exhibit multi-phase ISM, as is qualitatively predicted by the adiabatic shell model. DEM L152 (N44) in the Large Magellanic Cloud (LMC) is a beautiful example where the nebular (104 K) gas in the shell clearly confines the hot, X-ray-emitting (106 K) gas within (Magnier et al. 1996; Figure 4). Chu et al. (1994) also confirmed the existence of C IV and Si IV absorption in the lines of sight toward all stars within LMC superbubbles. These tracers of intermediate temperature (105 K) gas are expected in interface regions between hot and cold gas. Quantitatively, however, the detected X-ray emission from LMC superbubbles has been an order of magnitude higher than predicted by the adiabatic model. It is therefore thought that the anomalous emission results from impacts to the shell wall by internal SNRs (Chu & Mac Low 1990; Wang & Helfand 1991), a scenario which is supported by other signatures such as enhanced [S II] / H and anomalous kinematics (Oey 1996; see below). Many other superbubbles have not been detected in X-rays, although the upper limits tend to be high, and remain within the model predictions. It is hoped that the capabilities of Chandra and XMM will be applied to these objects.
Since most early-type stars are found in OB associations, wind-driven bubbles of individual massive stars are rare, and consequently few quantitative studies of these objects exist. In principle, O stars offer the most straightforward test of the standard shell evolution, since their wind histories are simple and relatively well-understood. One of the few such studies was carried out by Oey & Massey (1994) on two nebular bubbles around individual late-type O stars in M33. They found crude consistency with the model predictions, but the constraints are limited by lack of kinematic information. Cappa and collaborators (e.g., Cappa & Benaglia 1998; Benaglia & Cappa 1999; Cappa & Herbstmeier 2000) have studied a number of H I shells around Of stars, which are presumably evolved O stars. They generally find a significant growth-rate discrepancy such that the shells appear to be too small for the assumed stellar wind power. Wolf-Rayet (W-R) stars are well-known to have the most powerful stellar winds, and a number of W-R ring nebulae have also been examined kinematically. Optical studies include those by Treffers & Chu (1982), García-Segura & Mac Low (1995), and Drissen et al. (1995); while, e.g., Arnal et al. (1999) and Cappa et al. (2002) examine the neutral and radio continuum properties. The W-R nebulae are also apparently too small, but owing to the complicated stellar wind history and associated environment, it is more difficult to interpret the W-R shell dynamics.
Superbubbles around OB associations are much larger, brighter, and easier to identify than single star wind-driven bubbles, and consequently have been much more actively studied. They are especially prominent in the LMC, where the proximity and high galactic latitude offer a clear, detailed view of the objects. The most comprehensive study of superbubble dynamics was carried out by Oey and collaborators (Oey 1996; Oey & Smedley 1998; Oey & Massey 1995) on a total of eight LMC objects. Saken et al. (1992) and Brown et al. (1995) also examined two Galactic objects. All of these are young, nebular superbubbles having ages 5 Myr. These studies again consistently reveal a growth-rate discrepancy equivalent to an overestimate in the inferred L / n by up to an order of magnitude. About half of the objects also show anomalously high expansion velocities, implying a strong, rapid shell acceleration from the standard evolution.
A number of factors could be individually, or collectively, responsible for these dynamical discrepancies. The first possibility is a systematic overestimate in L / n: stellar wind parameters remain uncertain within factors of 2 - 3. The ambient density distribution is also critical to the shell evolution; as shown by Oey & Smedley (1998), a sudden drop in density can cause a "mini-blowout" of the shell, whose kinematics can reproduce those of the high-velocity LMC shells. In the case of those objects, however, the presence of anomalously high X-ray emission favors the SNR impact hypothesis discussed above. In any case, the critical role of the ambient environment motivated us to map the H I distribution around three superbubbles in the LMC sample (Oey et al. 2002). The results were surprisingly inhomogeneous, with one object essentially in a void, another with significant H I in close proximity, and a third with no correspondence at all between the nebular and neutral gas. Thus it is virtually impossible to infer the ambient H I properties for any given object without direct observations. This heterogeneity suggests that a systematic underestimate of n is not responsible for the universal growth-rate discrepancy. However, a related environmental parameter is the ambient pressure. If the ISM pressure has been systematically underestimated, then the superbubble growth would become pressure-confined at an earlier, smaller stage. We are currently exploring this possibility (Oey & García-Segura 2003, in preparation).
Finally, if the superbubble interiors are somehow cooling, then the shells will no longer grow adiabatically. While this possibility has been explored theoretically from several angles, there is as yet no empirical evidence, in particular, radiation, that the objects are cooling. Meanwhile, mass-loading has long been a candidate cooling mechanism (e.g., Cowie et al. 1981; Hartquist et al. 1986), either by evaporating material from the shell wall, or by ablating clumps that are overrun by the shell. The enhanced density would then increase the cooling rate of the hot interior. More recently, Silich et al. (2001) and Silich & Oey (2002) suggested that the metallicity increase expected from the parent SN explosions can significantly enhance the cooling and X-ray emission, especially for extremely low metallicity systems.