2.1. Single star bubbles
Few studies exist of single star bubbles from isolated OB stars. Oey & Massey (1994) examined two nebular examples in M33, and spectroscopically classified the parent O stars. The inferred stellar masses and ages implied wind parameters that were consistent with the observed sizes and shell ages predicted by the adiabatic model. However, the parameters were loosely constrained. H i shells with radii of several tens of pc have been identified as wind-blown bubbles around a number of Galactic O and Of stars (Cappa & Benaglia 1998; Benaglia & Cappa 1999). These are largely consistent with the standard model, and probe a specific subset of fairly evolved stars with old shells that have essentially stopped expanding.
Studies of Wolf-Rayet ring nebulae suggest shells that are too small, equivalent to an overestimate in L / n by an order of magnitude (e.g., Treffers & Chu 1982; García-Segura & Mac Low 1995; Drissen et al. 1995). However, the progenitor star produces several wind phases, including both fast and slow winds, with extreme changes in L. Their cumulative effect on shell morphology is complex and poorly understood. It is therefore unsurprising to find significant discrepancies between the predictions and observations of shell parameters. Hence, more studies of the simpler, single OB star bubbles are needed.
Superbubbles around OB associations are more prominent than single-star bubbles, and thus have been studied more extensively. Soft X-ray emission has been detected within many objects, which is qualitatively consistent with the adiabatic evolution model. Two classes of X-ray emission have been identified: objects with X-ray luminosity Lx in excess of the model's prediction (Chu & Mac Low 1990; Wang & Helfand 1991), and objects that remain undetected in X-rays (Chu et al. 1995). The X-ray-bright objects are thought to be overluminous because of SNR impacts on the shell walls. Upper limits on the X-ray-dim objects remain consistent with Lx predicted by the adiabatic model. It will thus be of great interest to determine Lx for these objects with XMM-Newton or Chandra. Also, an interface region between the hot gas and cooler shells should generate intermediate temperatures and ions. Chu et al. (1994) searched a number sightlines through LMC superbubbles and confirmed the existence of C iv and Si iv absorption in all cases.
A stringent test of the adiabatic model is to compare the predicted and observed shell kinematics in cases where the input mechanical power and other parameters are well-constrained. This was carried out for eight, young, wind-dominated LMC superbubbles by Oey & Massey (1995), Oey (1996), and Oey & Smedley (1998). The predicted growth rate for the shells was higher than implied by their observed R and v, equivalent to an overestimate in L / n by an order of magnitude. However, even after adjusting L / n in the models, over half the objects still showed observed expansion velocities that were typically a factor of two higher than predicted for the given R. Similar discrepancies were reported for Galactic objects by Saken et al. (1992) and Brown et al. (1995).
SNR impacts on the shell wall are the favored explanation for the high-velocity shells, since these also exhibit the anomalously high X-ray emission and elevated [S II] / H ratios. However, a sudden drop in the ambient density can induce a "mini-blowout" with shell kinematics that can easily reproduce the anomalous velocities (Oey & Smedley 1998; Mac Low et al. 1998; Silich & Franco 1999). Indeed, were it not for the X-ray and nebular diagnostics, it would be impossible to distinguish the shell acceleration mechanism from the kinematics alone.
Thus we see that the ambient properties are critical in determining the shell evolution. An underestimate in n could, for example, contribute to the growth rate discrepancy described above, that is seen in all the objects. To clarify the ambient gas distribution, Oey et al. (2001) mapped the H i distribution within a ~ 40' radius of three nebular LMC superbubbles at 30" resolution. The results show neutral environments that vary to an extreme, despite morphologically similar optical nebulae. It is therefore essentially impossible to infer properties of the ambient material without direct, multi-wavelength observations.
Another vital parameter for shell evolution is the ambient pressure, P0, which determines whether and when the superbubble growth becomes pressure-confined. While P0 is usually unimportant in young, high-pressure superbubbles like the nebular objects mentioned above, it is of vital importance in the mid- to late-stage evolution. It may also be relevant in high-pressure, ionized environments like dense star-forming regions (e.g., García-Segura & Franco 1996). Ultimately, P0 determines the final size of the shells, and conditions relative to blowout. The uniformity and distribution of P0 in the multiphase ISM is therefore especially relevant to a global understanding of superbubbles in galaxies (see below).
Finally, if the hot gas within superbubbles does not blow out and merge into the hot, ionized medium (HIM), it is likely that the objects will cool and depart from energy conservation. Indeed, whether and how the hot interior cools has long been a major question for superbubble evolution and the fate of the hot gas. Thermal conduction at the interface between the cool shell wall and hot gas should cause a high rate of mass-loading into the interior. The evaporated shell material dominates the mass in the hot region, which could be further supplemented by evaporation and ablation from small clouds overrun by the expanding shocks (e.g., Cowie & McKee 1977; McKee et al. 1984; Arthur & Henney 1996). If the interior density is sufficiently increased, radiative cooling will dominate, and the shells will no longer grow adiabatically. In addition, Silich et al. (2001) point out the importance of enhanced metallicity in the superbubble interiors, caused by the stellar and SN yields. Preliminary investigation for individual objects by Silich & Oey (2001) shows enhancement in Lx by almost an order of magnitude for low-metallicity (Z = 0.05 Z) objects. This increase in the cooling rate could facilitate a transition from adiabatic to momentum-conserving evolution.
2.3. Supergiant shells
The very largest H i shells, having sizes of order 1 kpc, emphasize some of the problems with the mechanical feedback model, and also highlight possible alternative shell-creating mechanisms.
The existence of infalling high-velocity clouds (HVCs) suggests that the impact of these objects could be an important contributor to supergiant shell populations. This suggestion is consistent with galactic fountain models for disk galaxies (e.g., Shapiro & Field 1976), which are ultimately also powered by mechanical feedback in the disk. A number of hydrodynamical simulations of infalling HVCs confirm that these impacts result in shell-like structures (e.g., Tenorio-Tagle et al. 1986; Rand & Stone 1996; Santillán et al. 1999).
In addition, tidal effects, which dominate energetics and structure formation at the largest length scales, could also create H i features that resemble shells. Note that many SN-driven shells will not exhibit expansion velocities if they have become pressure-confined by the ambient medium, thus a lack of observed expansion velocities cannot distinguish between the feedback model and other models. It has been suggested that some of the largest holes in, e.g., M33 are simply morphologically-suggestive inter-arm regions (Deul & den Hartog 1990). The same may be true of the giant hole identified by de Blok & Walter (2000) in NGC 6822. Simple self-gravity effects have also produced shell- and hole-like structures in numerical simulations (Wada et al. 2000), although morphologically these structures appear more filamentary than the observations.
While such alternative mechanisms for creating shell-like structures undoubtedly contribute to the supergiant shell population, the conventional mechanical feedback model nevertheless also appears to apply in many situations. Meaburn's (1980) LMC-4 is a well-known example that is unambiguously linked to Shapley's Constellation III, a large, extended complex of young stars. Kim et al. (1999) are able to identify an evolutionary sequence for supergiant shells in the LMC, based on the relative loci of H and H i emission. In addition, Lee & Irwin (1997) considered formation mechanisms for supergiant shells in the edge-on SBc galaxy NGC 3044. They found no evidence of HVCs, and since the galaxy is isolated, tidal interactions are also unable to explain the supergiant shells. They therefore conclude that the active star formation seen in NGC 3044 most likely explains its supergiant shell structures.
Thus, probably both mechanical feedback and other mechanisms form supergiant shell structures. Presumably different processes dominate under different circumstances, and these remain to be understood.