2.3. Heating Sources

A long standing question has been the heating source for the IR emission, and in particular, the extent to which this emission can be considered a good star formation tracer (see e.g. [Kennicutt1998]). ISO now allows us to re-address this question. Examining the global luminosity of disks of spiral galaxies, Roussel et al. (1999b) establish a clear correlation between the MIR and H luminosity of disks (see fig. 1a). This correlation is linear and implies that in the disks of normal galaxies, the energy source for the MIR emission is the young stellar population and therefore that it can be used as a star formation tracer. There are however a few caveats: (1) in the disk of normal spirals, the broad-band [6.75 µm] / [15 µm] color is ~ 1, as expected if there is little if any contribution from very small hot grains such as observed in HII regions (see sec. 1.2). This will not be the case in stronger star forming galaxies (see sec. 5). (2) The central regions of spiral galaxies typically have a lower [6.75 µm] / [15 µm] ratio (e.g. [Sauvage et al. 1996]; [Roussel et al. 1999a]; [Dale et al. 1999]), possibly indicating again a higher local star formation rate. Therefore, in the case of an unresolved galaxy, Roussel et al (1999b) argue that their calibration should provide a lower limit to the actual star formation rate.

Figure 1. (a-left) The linear correlation between the 15 µm and H luminosities in the disk of spiral galaxies (normalized by the disk's surface to cancel scale effects). A similar linear correlation is observed with 6.75 µm luminosities. (b-right) The correlation between the FIR and H total fluxes for galaxies that have little contribution of the central regions in the 15 µm maps. On this sample, the correlation is linear, revealing a tight link between the two quantities (figures adapted from Roussel et al., 1999b).

The possibility to separate the nuclear region from the disk of the galaxy in the MIR now allows a re-investigation of the FIR-H correlation. This correlation is known to be non-linear, implying that more than one process is present and that the FIR cannot be directly used to infer star-formation rates (e.g. [Sauvage & Thuan1992]). Roussel et al (1999b) have selected in their sample those galaxies with the smallest contribution of the central region in the MIR maps and show that for this sample the FIR-H correlation is very good and linear (fig. 1b). This indicates that the non-linearity present in the general FIR-H relation is induced by the nuclear component and that in the disk of normal galaxies, the FIR emission collected by IRAS is mostly from dust heated by the young stellar population.