It is often stated that UltraLuminous InfraRed Galaxies were discovered by the InfraRed Astronomy Satellite (IRAS), which was launched in 1983. Strictly speaking this may be true since the first objects to meet the now generally acknowledged criteria which define a ULIRG - L8-1000µm > 1012 L and, optionally, exceptionally large ratios of Linfrared / Loptical - can indeed be considered to have been discovered with the publication by [Houck et al. 1985] of a sample of 9 IRAS sources invisible or exceptionally faint on the Palomar Sky Survey plates, and exhibiting Lir / Lopt ratios over 50. We must go back further, however, to appreciate the origins of the ULIRG phenomenon. Galaxies with an unusually high amount of mid- or far-infrared emission compared to their optical output had been discovered by the pioneers of infrared astronomy in the 10 to 15 years preceding IRAS, and had been recognised to represent luminous, probably short-lived, active events of some sort within a galaxy.
Several excellent reviews have described the IRAS and pre-IRAS legacy, including [Rieke and Lebofsky (1979, Soifer et al. 1987, Sanders & Mirabel 1996, and Mirabel 2002] so we present here only the brief highlights relevant to ULIRG discovery and ULIRG samples.
The first infrared observations of galaxies were undertaken in the late sixties [Low & Kleinmann 1968, Kleinmann & Low 1970], and a population of galaxies with infrared-dominant emission from the nuclear regions was discussed by [Rieke & Low 1972]. The well known radio-infared relation [de Jong et al. 1985, Helou et al. 1985] was originally suggested by [van der Kruit 1971. An early debate ensued as to the thermal vs non-thermal origin of the infrared emission from galaxies [Rees et al. 1969, Burbidge & Stein 1970] with the conclusion that most systems are powered by re-radiation of starlight by dust ([Rieke & Lebofsky 1979] and references therein). The other key contribution of the pre-IRAS days was the role of interactions in triggering nuclear and starburst activity [Toomre & Toomre 1972, Larson & Tinsley 1978], which was confirmed for the strongly interacting system Arp 299 by [Gehrz et al. 1983], and several other interacting and merging systems by [Lonsdale, Persson & Matthews 1984, Joseph & Wright 1985].
IRAS scanned almost the entire sky in the thermal infrared 3, observing in four bands centered at 12, 25, 60 and 100 µm and opening up an unprecedented volume of space to study in this wavelength regime. This allowed IRAS to find exceedingly rare objects and to demonstrate the importance of this new class of exceptionally infrared-luminous objects (Figure 1). One of the most exciting early discoveries was that IRAS would not be limited to exploring only the very local Universe due to its relatively bright flux limits, but it could reach out to significant distances thanks to the existence of large numbers of exceptionally infrared-luminous sources. Strong evolution was demonstrated for the IRAS deep north ecliptic hole field by [Hacking, Houck & Condon 1987] and confirmed over the sky by [Lonsdale et al. 1990] and [Saunders et al. 1990]. Extensive redshift surveys [Oliver et al. 1996] led to the discovery of the first z > 2 IRAS HLIRG, IRAS FSC 10214+4724 at z = 2.86 [Rowan-Robinson et al. 1991] which is a lensed system. The extremely high evolution rates, modeled, for example, as Lir ~ (1 + z) ~ 4, were actually anticipated, based on the previously known strong evolution of starburst-related sub-mJy radio sources [Hacking, Houck & Condon 1987].
ULIRGs were found to have comparable space densities to those of PG QSOs of similar luminosity by [Soifer et al. 1987]; however, significant incompleteness in the PG QSO sample has since been demonstrated [Wisotzki et al. 2000] so this result needs re-visiting. The best known samples of IRAS luminous and ultraluminous galaxies, all selected at 60 µm, are the Bright Galaxy Sample of [Soifer et al. 1987], recently significantly updated into the Revised Bright Galaxy Sample (RBGS; Sanders et al. 2005), and the complete flux-limited IRAS 1 Jy sample [Kim & Sanders 1998]. Also notable are the 2Jy sample of [Strauss et al. 1990] and the FIRST/IRAS sample of [Stanford et al. 2000]. The RBGS contains 629 IRAS 60 µm galaxies brighter than 5.24 Jy and Galactic latitude > 5 degrees and contains 20 ULIRGs; the most luminous being Mrk 231 with Lir = 3.2 × 1012 L, the highest redshift being IRAS 07251-0248 at z = 0.0876, and the closest being Arp 220 at z = 0.018. The 1 Jy sample consists of 118 ULIRGs drawn from the IRAS Faint Source Catalog, with declination > -40 degrees and Galactic latitude |b| > 30 degrees. This sample also has a warm 60/100 µm > 0.3 colour selection, which introduces a bias against cooler objects.
Whilst dramatic in nature, LIRGs and ULIRGs in the local Universe are rare, and they contribute only ~ 6% of the total infrared luminous energy density [Soifer & Neugebauer 1991], and about ~ 3% of the total energy density. It was found that both the IRAS 60/100 µm colour and the Lir / Lopt ratio increased with luminosity [Soifer et al. 1987], reaching ~ 100 in the most extreme systems. This indicates that the more luminous systems must have an increasing contribution from an additional warm source compared to the relatively cool emission from modest levels of star formation seen in spiral disks [de Jong et al. 1985]. IRAS ULIRGs have been separated into warm and cool subsamples based on the 25/60 and 60/100 µm IRAS colours [Sanders et al. 1988b, Surace et al. 1998], with the warmer objects more likely to host an AGN [de Grijp et al. 1985]. The spectral energy distributions of IRAS ULIRGS were reviewed by [Sanders and Mirabel 1996], illustrating these trends with colour and luminosity. These reviewers also compared the SEDs of QSOs and Blazars with those of ULIRGs.
It should be remembered that although the IRAS datasets are now nearly a quarter of a century old, they are very under-explored. As of publication date there are > 37,000 infrared-bright galaxies in the IRAS Faint Source Catalog (FSC) that have never been observed with any other instrument and reported in any journal article. Only 43% of the 64,606 IRAS extragalactic FSC sources 4 have been included in any sort of publication (J. Mazzarella, priv. comm.).
3 defined as the wavelength region over which dust grains can thermally re-radiate emission, ranging from the dust sublimation temperature on the short wavelength side ( ~ 1 µm) to ~ 1000 µm on the long wavelength side where non-thermal processes often begin to dominate Back.
4 As reported by NED, the NASA/IPAC Extragalactic Database Back.