Shells and Supershells
                                   Astronomy 208

A 'first pass' survey over 208 square degrees of the Southern Galactic
Plane with the Parkes telescope has revealed a giant 'bubble', or hole, in the interstellar medium. Go here  for more information on the Southern Galactic Survey.

"H I shells and supershells."  Heiles, C. Astrophysical Journal, Part 1, vol. 229, Apr. 15, 1979, p. 533-537, 539-544. Full text here.

    Heiles apparently had a long-standing research interest in HI, having addressed the subject in different contexts in several earlier papers ( 19761979 ).  Here, he presented photographs in HI with additional selection in velocity intervals. He argued that filamentary structure was clear, and further thought that more structure would be visible with better resolution. His data is generally consistent with static shells, but there was some evidence for expanding shells. Almost all of the shells discussed in this paper had only a single hemisphere. There was no apparent correlation with other astronomical objects, despite Heiles' attempts. One of his most interesting conclusions is the existence of the so-called "supershells," structures on the order of kiloparsecs which have no apparent source.

    Heiles began work with the Weaver and Williams 1973 survey, and was interested in finding the HI equivalent of the HII/dust filaments of Brand and Zealey, 1975 (the  “cosmic bubble bath”). The “closed circles or ovals” were called shells. Many of his HI arcs looked like those produced by explosions, and some had the correct velocity differential for an expanding thin shell.  The majority of shells, however,  had no change in size with velocity, and were assumed to be static structures.

    He used a “confidence rating” from 1 to 3 to judge the “reality” of the shell. Most are given 2s, but the overall confidence was much higher for expanding shells (despite the admitted low velocity resolution). The alternative to a shell being "real" was that the alleged shell was simply a superposition of unrelated filaments. Presumably the author discarded results that did not meet his criteria, and  he asserts that the shells presented here truly exist. A salient concern for Heiles is the  general presence of only one hemisphere for each shell. He suggests that this could be explained by nonuniform distribution of interstellar gas, but remains uncomfortable.

    Distances to the shells were estimated using the object's velocity and a galactic rotation curve. The near/far ambiguity was a problem, and for simplicity he assumes all objects were located at the near point. Estimated error using the curve was about 35%. He considered several different models of galactic rotation for velocity dispersion purposes, and ended up arbitrarily choosing a value. Several objects could not be reliably located, and Heiles either assigned the maximum likely galactic radius or arbitrarily assumed a 500 pc distance.

    The paper presents the calculated energy necessary to create the structures, using the assumption that the energy was deposited instantaneously.
Stellar winds and supernovae were considered as likely candidates for energy sources, given the range of log E between 48 and 54.. Assuming a shockwave of some sort formed the shells, expanding shells were considered with to be in the expanding, constant momentum phase (as per Chevalier 1974). Conservative estimates for energy were used, resulting in likely underestimates.  Shell lifetimes were estimated, and likely total numbers of shells were calculated assuming that each Type II supernova created a shell. This yielded 100,000 shells currently existing, which was vastly different than the observed population. This suggested either a selection effect at work or a completely different production agent.

    In an attempt to explore the likely origins of the shells, attempts were made to associate them with known astronomical objects. No apparent association between HI shells and SNR was found. An attempt to correlate with weak radio features was difficult and also unsuccessful.  Correlations with HII regions was attempted, but the results were doubtful. There was, however, a possible correlation with stellar associations. This correlation could be explained by energetic stellar winds from OB stars. Unfortunately, the statistical results were insufficient to make the case.

    Even beyond the difficulties of HI shells, the so-called "supershells" were even more difficult to explain. Similar structures on the order of kpc had been found in external galaxies as early as 1964.  In other galaxies supershells seemed to be associated with Population I objects, but this association did not hold in the Milky Way. Heiles suggested that this might be due to selection effects. Some 11 objects with E > 1052 were found in this investigation. These energies were too large for supernovae, leaving no obvious explanation.

    Observations suggested that shells were produced infrequently, perhaps 1 every 107 years, making it unlikely that we would have the opportunity to see the production agent. Supershells were apparently ~1000 times more energetic than Type II supernovae. There was the possibility of multiple supernovae, but Heiles admits that the lack of any two being observed in the same spot makes this hypothesis marginal. Another possibility was a Type III supernova, which would theoretically have enough energy, but was seen as a shaky foundation on which to build a theory of supershells. Supershells were presented as a completely new object, and their agent might plausibly be a genuinely unknown object. Heiles closes with encouraging the enormous survey necessary to spot a supershell agent, which estimated as observing some 250 000 galaxies for 40 years.

So what they heck are they?

    Two papers appeared soon after Heiles published, presenting theories to explain the HI shells. Tenorio-Tagle (1980, and see 1987) calculated that a collision of a small neutral cloud with a galactic disk could result in the deposition of a large amount of energy in a small volume. This would result in a supersonic expansion that would lead to a large cavity whose characteristics were in agreement with observed supershells.  Collisions would also explain the tendency toward single hemisphere shells. Bruhweiler and Gull ( 1980 ) favored an interaction of stellar winds and supernovae in OB star regions to reproduce the observed characteristics. This model had the additional appeal of explaining the injection of heavy elements into the galactic halo. Similar work was done by McCray and Kafatos ( 1987 ), who also looked to link the gravitational instability of supershells with induced star formation.
    Elmegreen and Chiang's explanation ( 1982 ) was that radiation pressure from field stars exerted forces on large shells that increased with shell size, and therefore resulted in a "runaway expansion." Supershells were thus explained as small shells originating in OB regions that were inflated by radiation pressure over tens of millions of years, eventually providing energy equivalent to 100 supernovae.
    In  1984 , Heiles responded to some of these theories. He argued that supernovae and stellar winds were inadequate energy sources to explain supershells, except in rare cases of unusually active areas of star formation. He favored the high velocity cloud (HVC) explanation offered by Tenorio-Tagle. More recently,  the competing theories of collisions and stellar effects were addressed with a paper by Ehlerova and Palous ( 1996 ), which came down on the side of star formation as the more likely cause.  Their conclusions were based largely on the distribution of shells in the galactic disk, which seemed to closely match general stellar distributions.
    In the last few years, theorists have looked to gamma ray bursts, events which are both powerful and frequent enough to generate the observed supershell characteristics. Tentative explorations of this possibility are found in  Efremov, et. al. (1998) and Loeb and Perna ( 1998 ).