3.2. Star formation in high IR luminosity galaxies
A large amount of work has been devoted to the understanding of the physical processes at work in high IR luminosity objects, mostly interacting/merging systems. Although, as mentioned in Sanders (1999), 3-4 of the 5 closest examples of ultraluminous IR galaxies (ULIRGs) contain a powerful if not dominant AGN, most ISO studies on luminous IR galaxies emphasize the importance of the starburst process in the generation of the IR luminosity (but see sec. 5). A very interesting point comes from studies of luminous and ultraluminous IR galaxies (the frontier being located at L8 - 1000µm = 1012 L).
For their ISOCAM sample of Luminous IR galaxies (LIG), Hwang et al. (1999) show a clear anti-correlation between the compactness of the infrared source and the angular separation between the interacting objects. This seems at odds with the absence of any correlation in the MIR properties of ULIRGs with the angular separation (but note that little IR imaging is available for ULIRGs). Similarly, Gao & Solomon (1999) have shown that, in LIGs, there is a clear anti-correlation between the star formation efficiency (SFE) and the angular separation (i.e. higher SFE for closer pairs). No such correlation is seen in the ULIRG sample of Rigopoulou et al. (2000), but very interestingly, the maximum SFE reached by LIGs is of the order of the mean SFE of ULIRGs. This seems to place ULIRGs as a limit-case for interaction triggered star-formation and may explain the lack of clear correlations with interaction parameters for the ULIRG sample. It also supports the conclusions of e.g. Genzel et al. (1998) and Rigopoulou et al. (2000) that the ULIRG phenomenon is mostly related to individual properties of the interacting galaxies and not directly to the interaction itself.
Of interest then are the spatially resolved observations of starburst galaxies and ULIRGs. A common point of these studies is the discovery that in many cases, a significant part of the luminosity is produced by very compact, mostly extranuclear, sources. This was first seen in the Antennae ([Vigroux et al. 1996]), but is now observed also in Mrk 171 ([Gallais et al. 1999]), NGC 253 ([Keto et al. 1999]), NGC 5253 ([Crowther et al. 1999]), or Arp 220 ([Soifer et al. 1999]). A plausible interpretation of these sources is that they are buried super-star-clusters, that will later evolve in the blue super-star clusters seen in interacting galaxies (e.g. O'Connel et al. 1994]). Given the power output of these clusters, their infrared phase should be quite short, a fact that fits well in the scenario of starburst progression during the merging phase of interacting galaxies proposed by Rigopoulou et al. (2000).
3.3. The [CII] Deficit
One of the most surprising findings of ISO comes from ISOLWS: Malhotra et al. (1997) observed that galaxies with the highest [60 µm] / [100 µm] or star formation activity exhibited lower-than-expected [CII]-to-FIR luminosity ratios. This was unexpected given that [CII] is predicted to be strong in regions exposed to far-UV photons that abonds in these galaxies. This deficit was later confirmed by Luhman et al. (1998) in a sample of ULIRGs. Reasons for this deficit are still unknown. Extinction or self-absorption have been rejected as the very large AV required (~ 400-1000) are not confirmed by any other extinction measurements. Favored explanations are (1) a decreased efficiency of photoelectric heating in very high UV fields (grains become positively charged or are destroyed, thus reducing the number of photo-electrons), or (2) softer than expected UV radiation fields in ULIRGs due either to a troncated initial mass function or the presence of aging starburst regions.