Next Contents Previous


In the previous sections, I have attempted to discriminate SFR calibrations applicable to whole galaxies from those applicable to regions within galaxies. As already mentioned at the beginning of this chapter, whole galaxies are, in first approximation, isolated systems. As long as they are calibrated and used in a self-consistent manner, most SFR indicators should yield similar values, and reflect the actual rate of recent star formation in a galaxy.

Regions within galaxies, instead, are emphatically not isolated. Most young star clusters disperse quickly (infant mortality due to gas expulsion), and they continue to disperse as they evolve, due to both stellar evolution and a variety of dynamical effects that include two-body relaxation, tidal stripping, large-scale shocks, etc. (Lamers et al. 2010). Models agree that as many as 80%-90% of clusters dissolve within the first 10-20 Myr of life, and their stars become part of the diffuse field, although the same models tend to disagree on the level of evolution at later stages (Fall et al. 2005; Lamers et al. 2005; Chandar et al. 2010). Cluster evolution and dispersal within the first few tens of Myr is what this review is mostly concerned with, since this can influence the derivation and interpretation of local SFRs.

Analysis of the HST UV spectra of star clusters and of the intracluster diffuse stellar light in the starburst regions of nearby galaxies has revealed marked differences between the two stellar populations. Star clusters show clear signatures of the presence of O star wind features, signalling the existence of stars more massive than ~ 30 Modot in the clusters. Conversely, the intracluster population systematically lacks those features (Tremonti et al. 2001; Chandar et al. 2005). Sufficient area is covered in each galaxy that stochastic sampling in the diffuse light regions is not an issue. The significant difference in the spectral features of cluster and intracluster spectra excludes the possibility that the intracluster UV light is scattered light which originates from the clusters themselves. Only two scenarios, thus, appear in agreement with the data: (1) stars form locally in the diffuse field, but either they have a different IMF than that of the clusters or they form in small clusters that systematically lack massive stars (Meurer 1995; Weidner et al. 2010); (2) most stars form in clusters, and the clusters dissolve over 7-10 Myr (Tremonti et al. 2001; Chandar et al. 2005).

Option (2) may be the most viable of the two scenarios, in light of what has been discussed earlier in this section. An important consideration is that option (1), also referred to as in-situ star formation, necessarily implies either a different IMF or a different cluster mass function between the clusters and the field.

GALEX has surveyed many local star-forming galaxies in the UV, at 0.153 µm (FUV) and 0.231 µm (NUV), across their entire disk. One of the most striking results is that the FUV-NUV colours of the arm regions are in general significantly bluer than those of the interarm regions (D. Thilker, private communication), despite the arms usually containing more dust than the interarms, roughly 1 mag more in the I-band (White et al. 2000; Holwerda et al. 2005). These red interarm colours, if not attributable to dust reddening, can result from evolving stellar populations that are diffusing from the arm regions (e.g., Pellerin et al. 2007). For example, in the galaxy NGC 300, located only 2 Mpc away, the UV colours are consistent with a dominant interarm population that is devoid of stars younger than 10-20 Myr. Another scenario for the UV light in the interarm regions of disk galaxies is dust-scattered light diffusing from the spiral arms (Popescu et al. 2005), although the significantly redder colours in the interarm regions could pose a challenge to this interpretation. Unambiguous evidence for scattered UV photons by dust has been found in the starburst-driven outflows of the two nearby galaxies M 82 and NGC 253 (Hoopes et al. 2005).

The UV light at 0.16 µm of a stellar population undergoing constant star formation for the past 100 Myr is contributed for ~ 70% and ~ 30% by stars younger and older than 10 Myr, respectively. We assume for simplicity that all stars younger than 10 Myr populate spiral arms, and those older than that are located in the interarm regions. Adopting extinction values of AI = 1.5 mag and 0.5 mag in the spiral and interarm regions, respectively, as determined by Holwerda et al. (2005), and a very simple dust geometry, the observed 0.16 µm UV emission becomes 40% and 60% contributed by stars younger and older than 10 Myr, respectively, almost reverting the intrinsic ratios. The observed ratio is about 50%-50% in NGC 5194 and NGC 3521 (Liu et al. 2011). Dust geometry plays a crucial role in this case, and almost about any value of the observed UV fraction from the two components can be obtained by varying the dust geometry, within a reasonable range for the geometrical distributions of dust and stars (see section 1.4).

Whether due to dust scattering and/or ageing stellar populations diffusing from the spiral arms and/or some in-situ star formation, or a combination of all three, the UV light in the inter-arm regions of spiral galaxies displays a more complex nature than that of the arms. This consideration suggests caution in using standard calibrations of the SFR(UV) in spatially-resolved studies of disks, especially when the targeted regions include inter-arm areas that have not been independently confirmed to be actively star-forming.

The escape of ionising photons from Hii regions has been already discussed in Section It is at the level of 40%-60% for the observed Halpha emission (e.g., Ferguson et al. 1996; Thilker et al. 2002; Oey et al. 2007), but gets reduced to ~ 30% when differential extinction in the Halpha within and outside the Hii regions is accounted for (Crocker et al. 2012). The nature of the diffuse Halpha in galaxies has been the subject of extensive studies by a number of authors, with still some open questions (e.g., Witt et al. 2010). For the purpose of measuring SFRs, we need to consider two effects. Firstly, the leakage of ionising photons from star-forming regions will impact SFR(Q(Ho)), reducing it roughly by 30% (trends with luminosity, gas density, etc, are only now starting to be investigated, see Pellegrini et al. 2012). Secondly, weak recombination line emission will appear in regions that are not star-forming; in sensitive surveys, this emission could be mistaken for faint in-situ star formation. This second effect should not be underestimated, as ionising photons have been shown to travel as far as about 1 kpc from their point of origin.

The non-discriminating nature of dust in regard to the sources of heating poses another challenge for local SFR measurements, if the IR is used as an indicator, either alone or in combination with a UV/optical one. The UV/optical photons produced by the stellar population of the diffuse field are generally sufficient to heat the dust in a galaxy (Draine et al. 2007). As already mentioned in Section, the 8 µm emission from a galaxy may be a better tracer of B stars than recent star formation (Boselli et al. 2004; Peeters et al. 2004), and about 30% of the 8 µm emission from the nearby galaxy NGC 628 is unrelated to star formation more recent than 100 Myr (Crocker et al. 2012). A similar fraction, ~ 30%, is recovered at 24 µm, when comparing the local with the global SFR indicators (see Section Direct attempts to quantify the fraction of 24 µm emission unrelated to recent star formation provide a minimum conservative value of about 20% (Leroy et al. 2012), with as much as 50% of the emission heated by populations older than ~ 10 Myr (Liu et al. 2011). Intermediate (0.2-2 Gyr) age stars can be bright in the 20-45 µm region, significantly contributing to the emission in this wavelength region (Verley et al. 2009; Kelson & Holden 2010).

In a study of the Triangulum Galaxy (M 33), Boquien et al. (2011) identify a threshold of SFR/area = 10-2.75 Modot yr-1 kpc-2, above which the IR emission reliably traces the current SFR; in M 33, this threshold delineates the spiral arms and the central ~ 1.2 kpc region, and excludes most of the interarm regions, which are dominated by evolved star heating. Dust can also be heated by UV photons leaking out of the arms into the interarm regions (Popescu et al. 2005; Law et al. 2011). In this case, the heating is not produced by in-situ star formation, but by photons that have originated a large distance away. In summary, care should be taken when attempting to measure SFRs in faint galaxy regions using the IR emission, since this emission may be dominated by heating by old stars and/or UV photons leaking out of Hii regions a large distance away, instead of tracing in-situ star formation.

The general conclusion to be taken away from this section is that SFR measurements at any wavelength could be tricky in regions that are not obviously dominated by recent star formation, such as the interarm regions of galaxies.

Next Contents Previous