Next Contents Previous

6. STAR FORMATION IN EXTRAGALACTIC GMCs

The evolution of GMCs substantially influences the evolution of galaxies. In particular, star formation in GMCs is a central event that affects galactic structure, energetics, and chemistry. A detailed understanding of star formation is therefore an important step for a better understanding of galaxy evolution.

6.1. Identification of Star Formation

In Galactic molecular clouds, we are able to study the formation of stars from high mass to low mass including even brown dwarfs. In all external galaxies, even those in the Local Group, such studies are limited to only the highest mass stars as a result of limited sensitivity. It is nonetheless worthwhile to learn how high-mass stars form in GMCs because high-mass stars impart the highest energies to the ISM via UV photons, stellar winds, and supernova explosions.

Young, high-mass stars are apparent at optical/radio wavelengths as the brightest members of stellar clusters or associations or by the Halpha and radio continuum emission from H2 regions. The positional coincidence between these signposts of star formation and GMCs is the most common method of identifying the star formation associated with individual clouds. Such associations can be made with reasonable confidence when the source density is small enough that confusion is not important. When confusion becomes significant, however, conclusions can only be drawn by either making more careful comparisons at higher angular resolution should or by adopting a statistical approach.

6.2. The Large Magellanic Cloud

The most complete datasets for young stars are available for the LMC, which has a distance of 50 kpc. They include catalogs of clusters and associations (e.g., Bica et al., 1996) and of optical and radio H2 regions (Henize, 1956, Davies et al., 1976, Kennicutt and Hodge, 1986, Filipovic et al., 1998). The colors of the stellar clusters are studied in detail at four optical wavelengths and are classified into an age sequence from SWB0 to SWB VII, where SWB0 is the youngest with an age of less than 10 Myr, SWB I in a range 10-30 Myr, and so on (Bica et al., 1996). The sensitivity limit of the published catalogs of star clusters is 14.5 mag (V); it is not straightforward to convert this into the number of stars since a stellar mass function must be assumed. The datasets of H2 regions have a detection limit in Halpha flux of 10-12 ergs cm-2 s-1, and the radio sensitivity limit at 5 GHz thermal emission corresponds to 20 mJy. The faintest detectable H2 regions correspond to the ionization by an O5 star if a single ionizing source is assumed. We note that the detection limit of H2 regions is quite good, L(Halpha) = 2 × 1036 ergs s-1, corresponding to one-fourth the luminosity of the Orion Nebula.

Using the first NANTEN CO survey (Fukui et al., 1999; Mizuno et al., 2001b), the GMCs in the LMC were classified into the three categories according to their associated young objects (Fukui et al., 1999; Yamaguchi et al., 2001c):

  1. Starless GMCs (no early O stars); "starless" here indicates no associated early O star capable of ionizing an H2 regions, which does not exclude the possibility of associated young stars later than B-type
  2. GMCs with H2 regions only; those with small H2 regions whose Halpha luminosity is less than 1037 erg s-1
  3. GMCs with H2 regions and stellar clusters; those with stellar clusters and large H2 regions of Halpha luminosity greater than 1037 erg s-1.

The new NANTEN GMC catalog (Fukui et al., 2006) is used to improve and confirm the statistics of these three classes (Kawamura et al., 2006). For the updated sample of 181 GMCs in Fig. 1, Fig. 11 shows the frequency distribution of the apparent separation of young objects, i.e., optical H2 regions and stellar clusters, measured from the lowest contour of the nearest GMC. Obviously, the youngest stellar clusters, SWB0 and H2 regions, exhibit marked peaks within 50-100 pc, indicating their strong concentrations towards GMCs. Comparisons of these distributions with a purely random distribution is shown by lines. The differences between these peaks are significant. The correlation with young clusters establishes the physical association of the young objects with the GMCs. On the other hand, clusters older than SWB I show almost no correlation with GMCs.

Figure 11

Figure 11. Histograms of the projected separation from the H II regions (Top left Davies et al. 1976) and clusters cataloged by Bica et al. (1996) to the neareset CO emission; (Top right) clusters with tau < 10 Myr (SWB 0), (Bottom left) clusters with 10 Myr < tau < 30 Myr (SWB I), and (Bottom right) clusters with 30 Myr < tau (SWB II - VII), respectively. The lines represent the frequency distribution expected if the same number of the clusters are distributed at random in the observed area.

In order to look for any optically obscured H2 regions we have also used the Parkes/ATNF radio continuum survey carried out at five frequencies 1.4, 2.45, 4.75, 4.8, and 8.55 GHz (Filipovic et al., 1995, 1998). The typical sensitivity limits of these new datasets are quite good, allowing us to reach flux limits equivalent to those in Halpha. The radio continuum results are summarized in a catalog of 483 sources, and the spectral information makes it possible to select H2 regions and eliminate background sources not related the LMC. By comparing these data with the GMCs, we found that all of the starless GMCs have no embedded H2 regions that are detectable at radio wavelengths (Kawamura et al., 2006).

Table 4 summarizes the results of the present comparison between GMCs and young objects, SWB0 clusters and the H2 regions including radio sources. It shows that ~ 25% of the GMCs are starless in the sense that they are not associated with H2 regions or young clusters. Fig. 12 shows mass histograms of the three classes, I, II and III. These indicate that the mass range of the three is from 104.5 Modot to a few times 106 Modot. It is also noteworthy that class I GMCs tend to be less massive than the other two in the sense that the number of GMCs more massive than 105 Modot is about half of those of class II and class III GMCs, respectively.

Table 4. Association of the young objects with GMCs

Class of GMC Number of GMCs a Time scale b Association
    (Myr)  

Class I 44 (25.7 %) 7 Starless
Class II 88 (51.5 %) 14 H2 regions
Class III 39 (22.8 %) 6 H2 regions and clusters c

a GMCs with M > 104.5 Modot; mass is derived by using XCO = 5.4 × 1020 cm-2 (K km s-1)-1 (Table 2).

b A steady state evolution is assumed. The absolute time scale is based on the age of stellar clusters; the age of SWB0 clusters, a half of which are associated with the GMCs, is taken to be 10 Myr.

c Young clusters or associations, SWB 0, by Bica et al. (1996).

Figure 12

Figure 12. Histograms of the mass of class I (Top), class II (Middle), and class III (Bottom), respectively. Mass is derived by using XCO = 5.4 × 1020 cm-2 (K km s-1)-1 (Table 2).

6.3. The Evolution of GMCs in the LMC

The completeness of the present GMC sample covering the whole LMC enables us to infer the evolutionary timescales of the GMCs. We assume a steady state evolution and therefore time spent in each phase is proportional to the number of objects in Table 4. Fig. 13 is a scheme representing the evolution suggested from Table 4. The absolute time scale is based on the age of stellar clusters: the age of SWB 0 clusters is taken to be 10 Myr. The first stage corresponds to starless GMCs, having a long time scale of 7 Myr. This is followed by a phase with small H2 regions, implying the formation of a few to several O stars. The subsequent phase indicates the most active formation of rich clusters including many early O stars (one of such an example is N 159N). In the final phase, the GMC has been more or less dissipated under the strong ionization and stellar winds from O stars. The lifetime of a typical GMC in the LMC is then estimated as the total of the timescales in Table 4: ~ 27 Myr, assuming that the GMC is completely disrupted by the star formation. As noted earlier (Section 6), the mass of class I GMCs tends to be smaller than the rest. We may speculate that class I GMCs, and possibly part of class II GMCs, are still growing in mass via mass accretion from their surrounding lower density atomic gas. In addition, the lifetime of GMCs likely varies with cloud mass, so 27 Myr is only a characteristic value and is probably uncertain by about 50%.

Figure 13

Figure 13. Evolutionary sequence of the GMCs in the LMC. An example of the GMCs and illustration at each class are shown in the left panels and the middle column, respectively. The images and contours in the left panels are Halpha (Kim et al., 1998) and CO integrated intensity by NANTEN (Fukui et al., 2001; Fukui et al., in preparation); contour levels are from K km s-1 with 1.2 K km s-1 intervals. Crosses and filled circles indicate the position of the H2 regions and young clusters, SWB 0 (Bica et al., 1996), respectively. The number of the objects and the time scale at each class are also presented on the right.

6.4. Star Formation in M 33

None of the other galaxies in our sample has as complete a record of interstellar gas and star formation as does the LMC, which is due, in part, to its proximity. Nevertheless, it is possible to draw some conclusions about the star formation in M33. Engargiola et al. (2003) correlated the H2 regions cataloged by Hodge et al. (1999) with the 149 GMCs in the M33 catalog. For reference, the completeness limit of the Hodge et al. (1999) cataloge is L(Halpha) = 3 × 1035 erg s-1; a similar range of H2 regions is cataloged in the LMC and M33. Engargiola et al. (2003) assumed that an H2 region is associated with a GMC if its boundary lies either within or tangent to a GMC; 36% of the flux from H2 regions can be associated with the cataloged GMCs. Correcting for the incompleteness of the GMCs cataloged below their sensitivity limit suggests that > 90% of the total flux of ionized gas from M33 originates from GMCs. Within the uncertainties, essentially all of the flux from H2 regions is consistent with an origin in GMCs. Apparently, about half of the star formation in M33 originates in GMCs below the the sensitivity limit of our survey.

A related question is to ask, what fraction of GMCs in M33 is actively forming stars? Engargiola et al. (2003) counted the fraction of GMCs with at least one H2 region having a separation Deltar. They defined the correlation length, such that half the GMCs have at least one H2 region within this distance. The correlation length for the GMCs and H2 regions is 35 pc; a random distribution of GMCs and H2 regions would return a correlation length of 80 pc. They assumed that a GMC is actively forming stars if there is an H2 region within 50 pc of the centroid of a GMC. With this assumption, as many as 100 GMCs (67 %) are forming massive stars. Of the 75 GMCs with masses above the median cataloged mass, the fraction of clouds actively forming stars rises to 85%. They estimate that the number of totally obscured H2 regions affect these results by at most 5%.

Thus the fraction of GMCs without star formation is estimated to be about 1/3, a fraction similar to that in the LMC. The M33 study estimated the lifetime of GMCs to be ~ 20 Myr, also similar to that found for the LMC. The fraction of clouds without active star formation is much higher than that found in the vicinity of the Sun where only one of all of the GMCs within 2 kpc is found to be devoid of star formation. It is unclear whether this difference is significant. Neither the LMC analysis, nor the M33 analysis would detect the low-mass star formation which it is proceeding in the Taurus molecular clouds. In any event, both the LMC and M33 studies suggest that the fraction of clouds without star formation is small. Thus the onset of star formation in GMCs is rather rapid not only in the Milky Way, but in at least some lower mass spiral and irregular galaxies.

Next Contents Previous