The Magellanic Clouds have been used to determine how well this method works because it is there that one can measure both the Halpha luminosity of a region and its massive star content. Studies of individual regions have found that in some the stellar output of ionizing photons is comparable to that predicted by the Halpha luminosity while in others it is up to a few times greater implying that there is considerable leakage of photons from the region (Massey et al. 1989; Parker et al. 1992; Garmany et al. 1994; Oey & Massey 1995; Testor & Niemela 1998). Oey and Kennicutt (1997) estimate that 0-51% of stellar ionizing radiation escapes local nebulae in the Magellanic Clouds, providing the source of ionization of the warm diffuse ionized medium. In a study of the ionized gas in two spiral galaxies, Ferguson et al. (1996) find that photons can escape distances of order a kpc and Hunter and Gallagher (1997) deduce the same for other irregular galaxies. Thus, there is some local shuffling of the ionizing photons.
With these caveats in mind, we will, nevertheless, proceed to use the Halpha luminosities as tracers of the star formation activity, both integrated over the entire galaxy and locally within a galaxy. Figure 1 shows global star formation rates determined from Halpha luminosities in a large sample of irregular galaxies (Hunter 1997). The star formation rates of the Magellanic Clouds are also shown (Kennicutt & Hodge 1986). We see that the star formation rates cover a range of 4 orders of magnitude (Hunter & Gallagher 1989), from NGC 1569, which is undergoing a true global starburst (Gallagher et al. 1984; Israel 1988), to DDO 216, which has two H II regions. Thus, many irregular galaxies are rightfully claimed to be slow evolvers. The Magellanic Clouds sit just to the high side of the average star formation rates for this group. In Fig. 1 the star formation rate is plotted against the integrated MB of the galaxy. Although there is a trend of lower luminosity galaxies having lower star formation rates, as one might expect, there is also a very large spread in MB for a given star formation rate, particularly at the high end.
Not only is there an exponential size distribution, but Hodge & Lee (1990) and Strobel et al. (1991) see a correlation of the characteristic size of H II regions with the luminosity of the galaxy. This goes in the sense that lower luminosity galaxies have a smaller characteristic scale for H II region diameters. Hodge and Lee speculate that this could be due to a difference in the gas density size fluctuations, gas kinematics, or abundances. At this point there is no explanation, but one must consider that it indicates a real difference in the size of the star-forming unit. At this meeting Mendez has shown us that BCDs do not fit this relation, having D0 that are too large for their MB; this then could represent a physical difference in the star formation process in BCD's compared to Im galaxies.
Kennicutt and Hodge (1986) also examined the luminosity distribution of H II regions in the Magellanic Clouds. There they found the number distribution to be fit by a power law N(f)/df=Af-x. From this they conclude that, unlike the size scale, there is no characteristic luminosity scale for H II regions although why this would be is not clear. Kennicutt et al. (1989) have carried this further and examined the luminosity function of H II regions in disk galaxies of different Hubble type. They find that the normalization and shape change with Hubble type.
An intriguing suggestion by Kennicutt et al. (1989) was that irregulars have more of the supergiant H II regions than earlier type spirals. However, in their study the Im class was represented by 6 galaxies of which three - LMC, SMC, NGC 4449 - have been affected by external interactions. Thus, how true this conclusion is for the class as a whole is in question (Kennicutt, private communication; see, for example, Hunter & Gallagher 1985). However, Hodge (1983) also commented on 6 irregulars in his sample having anomalously large largest H II regions for their size distributions. Furthermore, Larson (1983) points out that thicker disks with less shear should allow the growth of larger gas clouds. The 30 Doradus region is a prime example of this since it is associated with an enormous complex of molecular (Cohen et al. 1988) and atomic (Luks & Rohlfs 1992) gas. In the LMC, 30 Doradus contains 1/4 of the total star formation activity of the galaxy (Kennicutt 1991). The 30 Doradus nebula itself is 370 pc in diameter (Kennicutt 1984), and, if it were placed at the distance of Orion, it would cover 47° on the sky. Another example is NGC 2366 which contains the supergiant H II region NGC 2363. NGC 2363 has the Halpha luminosity of 1.6 30 Doradus nebulae and contains over half of the total Halpha luminosity of an otherwise moderate star-forming galaxy.
An interesting twist on this is the finding by Elmegreen et al. (1996) in a study of stellar complexes in star-forming disk galaxies. They found that the characteristic diameter of the largest star complexes decreases as galactic MB decreases, implying that irregulars have smaller star complexes than spirals. This is consistent with gravitational instability models for the formation of complexes (the expected scale length), and they argue against its being a size-of-sample effect. Thus, they suggest that irregulars may form larger giant H II regions but that these are part of smaller star complexes.
If irregulars do have proportionally larger concentrations of stars of the same age (that is, more supergiant H II regions), one consequence is a greater impact of massive stars on the ISM of irregular galaxies as in the models that Mac Low has discussed at this meeting. An example of this is Constellation III, a star complex in the LMC that has blown a 1.8 kpc diameter hole in the ISM (Dopita et al. 1985).
Dolphin & Hunter (1998) have compared what is known about Constellation III with 30 Doradus (see references cited therein). They find that 30 Doradus in not strictly a young version of Constellation III, although it is likely that its evolution and final appearance, minus R 136, will be roughly similar. Both complexes formed stars out of ∼3·107 Msun of gas, over an area of diameter ∼0.9 kpc, with star formation extending over about ∼10 Myr. There is no relationship between age and distance from the centers of the complexes, except that R 136, the youngest cluster in 30 Doradus, has formed at its center. The mode of star formation, that is the spatial concentration of that star formation, has also been somewhat different in 30 Doradus. Outside of R 136, the spatial concentration in 30 Doradus has been about 5 times higher than the bulk of distributed star formation in Constellation III, but 3 times lower than in the little star clusters in Constellation III. R 136, of course, is much higher in spatial concentration than the star formation in Constellation III by a factor of 300.
A key question is why 30 Doradus produced an R 136 object while Constellation III did not. A possible explanation is that the distribution of the gas within the two complexes was not the same, and a higher concentration of gas, perhaps influenced by the proximity of 30 Doradus to the stellar bar potential (Elmegreen & Elmegreen 1980), enabled R 136 to form.
The spatial concentration of stars in R 136 is extreme. The concentration of stars is 200-350 times higher than those of normal OB associations (Hunter 1995). This is shown in Fig. 2 where concentrations of stars with MV≤-4 and with masses 6.5-15 Msun are shown for a variety of star-forming regions. The MV criterion is bothered by the fact that massive stars evolve quickly and only comparisons of regions with the same age have meaning. However, it allows a comparison of some regions for which that is the only information. The comparison of stars within the 6.5-15 Msun mass range is more reliable, at least for these young regions. We see that there are other star-forming regions as rich in stars as R 136, but they have spatial concentrations that are comparable to those of the smaller scale OB associations. Such regions include NGC 604 in M33, the 30 Doradus region outside of R 136, and the Constellation III star complex. Even two BCD's, galaxies known for their intense star formation activity, have concentrations like normal OB associations: I Zw 18 (Hunter & Thronson 1995) and VII Zw 403 (Lynds et al. 1988). Thus, most giant and supergiant star-forming regions are just scaled versions of OB associations, but not R 136.
However, application of this model to the Magellanic Clouds has not been successful. The Magellanic Clouds have been affected by external factors, so the applicability of this model is doubtful, but Kennicutt et al. (1995) have shown that in the LMC star formation is found where Sigmag/Sigmac is 0.5. Only in the SMC is Sigmag/Sigmac reasonably high and there star formation would be expected to be more extensive than it is.
Hunter et al. (1998) explored the applicability of several models to a small sample of irregular galaxies. These models considered thin pure-gas disks, thick gas+star disks (because irregulars are thicker than spirals), three dimensional systems including dark matter, the thermal properties of the gas, and shear-regulated cloud formation. The resulting critical gas densities for cloud formation for each of these models were compared with the observed gas densities and with the azimuthally-averaged current star formation rate as traced by Halpha. They found that the ratios of observed to critical gas densities are lower in irregulars than in spiral galaxies by a factor of ∼ in all cases but the shear-regulated model. This suggests that the gas in irregulars is closer to stability, even though star formation is occurring in these galaxies. However, no model was able to predict where the star formation actually occurs, and especially where it ends. These failures of the models suggest that other processes are important for cloud and star formation in irregulars.
An obvious mechanism is sequentially triggered star formation driven by the mechanical energy input from concentrations of massive stars. Another process is random gas compression from turbulence. Kinematics in Im systems, including the Magellanic Clouds (Spicker & Feitzinger 1988), are often dominated by random gas motions which can contribute significantly to the pressure (for example, Gottesman & Weliachew 1977; Tully et al. 1978; Huchtmeier et al. 1981; Sargent et al. 1983; Lo et al. 1993; Feitzinger et al. 1981), and in some cases rotation can be insignificant compared to random motions (Young & Lo 1996). Both of these processes have a sensitivity to the critical density like the gravitational instability model, but do not involve large-scale instabilities directly. They could in fact dominate the star formation process when spiral arms are not present or are weak, as in the inner part of the spiral NGC 2403 (Thornley & Wilson 1995).
Other evidence also points to the importance of star induced star formation in determining local star formation activity. The best correlation that Hunter et al. (1998) found with the current star formation activity was the stellar surface density (see also Hunter & Gallagher 1985), a correlation also seen in spirals (Hodge & Kennicutt 1983). The correlation of Halpha with µV implies a feedback mechanism. This feedback mechanism is probably the energy input of stars to the ISM. In addition in irregulars, BCD's, and some low surface brightness galaxies, H II regions are often found where locally Sigmag is higher even if azimuthally-averaged quantities are not (van der Hulst et al. 1993; Taylor et al. 1994; van Zee et al. 1996, 1997), suggesting that the gas has been locally rearranged as would be the case when the energy input from massive stars blows holes in the gas. Models for the evolution of dwarf galaxies by Mori et al. (1997) also come to the conclusion that feedback mechanisms are particularly important in these galaxies. On the other hand, theoretical considerations by Silk (1997) of the ISM porosity suggest that the feedback from massive stars would be negative and that the numerous holes seen in the ISM of irregulars make it even more difficult for irregulars to form stars overall. That this is a real issue was driven home by the stunning H I map of the LMC shown by Kim at this meeting, as well as the hole-filled H I maps of other irregulars shown by Brinks. Furthermore, as Larson (1996) has pointed out, the overall feedback from star formation has to be negative or the process would run away with itself. Constellation III may well provide an example of the limitations of this process: There has been a first and second generation, but will there be a significant third generation? The H I map that Kim has presented suggests that the ISM is too fragmentary (porous) to support this third generation. However, it could well be that star-induced star formation, as was proposed for the LMC by Feitzinger et al. (1981), could be relatively more important in irregulars than in spirals. If this feedback does play a key role in the formation of clouds in irregulars, this could also explain why irregular galaxies were slow to begin forming stars after they first coalesced out of the primordial gas.
Although my emphasis here is on ionized gas structures, they are a part of what is often referred to as the diffuse ionized gas in galaxies. In the Magellanic Clouds Kennicutt et al. (1995) have found that the diffuse ionized gas constitutes 30-40% of the total Halpha luminosity of these galaxies. Hunter & Gallagher (1997) estimate 25% for a group of irregulars, and Ferguson et al. (1996) find 30-50% for two spirals. Ferguson et al. go further and suggest that the fraction of the total ionized gas in a galaxy that is the diffuse ionized gas is more or less a constant.
In a survey of irregular galaxies Hunter et al. (1993) showed that large ionized gas structures are a common phenomenon of star-forming irregulars. In a sample of 51 galaxies, they found that 12% contained at least one ionized supershell (radius ≥300 pc), 24% contained supergiant (>0.5 kpc) filaments, and 27% contained one or more of these structures. Not surprisingly, the presence of these structures was correlated with the presence of at least one large concentration of young stars in the galaxy although that was not a sufficient condition for the presence of such structures. Very few irregulars seen at high inclination angles showed evidence for structures extending out of the plane of the galaxy although several (NGC 1569, IC 10, NGC 4449) have structures that are probably supershells that have cooled and fragmented above the plane of the galaxy. Hunter & Gallagher (1997) have shown that these supergiant H II regions are photoionized by stars (see also the models presented by Mac Low at this meeting), implying that ionizing photons can travel distances of order kpcs.
| First version: | 14th | February, | 1998 |
| Last update: | 08th | October, | 1998 |