Annu. Rev. Astron. Astrophys. 1996. 34: 749-792 Copyright © 1996 by Annual Reviews. All rights reserved

2. BACKGROUND

The fact that some galaxies emit as much energy in the infrared as at optical wavelengths was established with the first mid-infrared observations of extragalactic sources (Low & Kleinmann 1968; Kleinmann & Low 1970a, b). Observations of both optical- and radio-selected objects at wavelengths of 2-25µm uncovered several objects - including luminous starbursts, Seyferts, and QSOs - with ``similar infrared continua'' that appeared to emit most of their luminosity in the far-infrared. More accurate photometry of a larger number of sources (Rieke & Low 1972) provided further evidence for dominant infrared emission from Seyfert galaxies and the nuclei of relatively normal spiral galaxies and also singled out several ``ultrahigh'' infrared luminous galaxies whose extrapolated luminosity at far-infrared wavelengths rivaled the bolometric luminosity of QSOs. A tight correlation between the 21-cm radio continuum and 10-µm infrared fluxes was established for ``Seyfert and related galaxies'', although the relevance of this correlation for determining the nature of the dominant energy source in these objects was not discussed.

The first critical evidence that the infrared emission from Seyferts was not direct synchrotron radiation was provided by monitoring of the 10-µm flux from the archetypal Seyfert 2 galaxy NGC 1068, which failed to show evidence for variability (Stein et al. 1974), plus measurements that showed the infrared source to be extended at 10µm (Becklin et al. 1973). The infrared spectrum appeared to be better explained by models of thermal reradiation from dust (e.g. Rees et al. 1969, Burbidge & Stein 1970). More extensive mid-infrared photometry of larger samples of Markarian Seyferts and starbursts (Rieke & Low 1975, Neugebauer et al. 1976), Seyfert galaxies (Rieke 1978), and bright spirals (Rieke & Lebofsky 1978, Lebofsky & Rieke 1979), plus far-infrared (30-300µm) observations of nearby bright galaxies (Harper & Low 1973, Telesco & Harper 1980) showed that ``infrared excess'' was indeed a common property of extragalactic objects, and that the shape of the infrared continuum in most of these sources, with the possible exception of Seyfert1 galaxies and QSOs, could best be understood in terms of thermal emission from dust. Although star formation seemed to be the most obvious explanation for the dominant energy source in normal galaxies and starbursts, a dust-enshrouded active galactic nucleus (AGN) remained a plausible model for Seyferts and QSOs.

A class of objects that would prove to be particularly relevant to LIGs were those objects in catalogs of interacting and peculiar galaxies (e.g. Vorontsov-Velyaminov 1959, Arp 1966, Zwicky & Zwicky 1971). The classic papers by Toomre & Toomre (1972), and Larson & Tinsley (1978) called attention to the role of interactions in triggering extreme nuclear activity, as well as more widespread starbursts¹. Condon & Dressel (1978) and Hummel (1980) found that the 21-cm radio continuum in the nuclei of interacting galaxies was enhanced (by factors of 2-3) compared to isolated spirals. Condon et al. (1982) later interpreted the radio continuum morphology of a class of ``bright radio spiral galaxies'' as evidence for powerful nuclear starbursts, the majority of which seemed to be triggered by galaxy interactions. Heckman (1983), following a suggestion by Fosbury & Wall (1979) that systems identified as ``ongoing mergers'' by Toomre (1977) might be exceptionally radio-loud, found that these and similar systems identified from the Arp atlas (Arp 1966) were ~ 8 times more likely to be radio-loud than single spirals with the same total optical luminosity, although it was not clear whether this enhanced radio activity was due to an AGN or a starburst.

Extremely strong mid-infrared and radio continuum emission in the interacting galaxy system Arp 299 (NGC 3690/IC 694) (Gherz et al. 1983) was interpreted as evidence for ``super starbursts'' involving several regions, each forming 10<sup>9</sup> M of stars in bursts and lasting ~ 10<sup>8</sup>years (although the most luminous infrared source, associated with the nucleus of IC 694, appeared to be powered by an AGN). Surveys of interacting galaxies in the mid-infrared (Joseph et al. 1984a, Lonsdale et al. 1984, Cutri & McAlary 1985) revealed an enhancement of infrared emission in interacting systems (typically by factors of 2-3) compared to isolated galaxies. Joseph & Wright (1985) identified a subset of advanced mergers in the Arp atlas with extremely strong mid-infrared emission that they described as ``ultraluminous'' infrared galaxies; they argued that super starbursts may occur in the evolution of most mergers.

2.2 Early IRAS Results

IRAS was the first telescope with sufficient sensitivity to detect large numbers of extragalactic sources at mid- and far-infrared wavelengths (Neugebauer et al. 1984). IRAS surveyed ~ 96% of the sky, producing an initial IRAS Point Source Catalog (1988; hearafter PSC) with a completeness limit of ~ 0.5Jy at 12µm, 25µm, and 60µm, and ~ 1.5Jy at 100µm. It contained ~ 20,000 galaxies, the majority of which had not been previously cataloged. Table 1 lists the definitions that have generally been adopted as standards for computing the broad-band infrared properties of IRAS galaxies.

Table 1. Abbreviations and definitions^a

Ffir	1.26 x 10-14 {2.58 f60 + f100} [W m-2]
Lfir	L(40-500 µm) = 4 DL2 C Ffir [L]
Fir	1.8 x 10-14 {13.48 f12 + 5.16 f25 + 2.58 f60 + f100} [W m-2]
Lir	L(8-1000 µm) = 4 DL2 Fir [L]
Lir/LB	Fir / f (0.44 µm)
LIG	Luminous Infrared Galaxy, Lir > 1011 L
ULIG	UltraLuminous Infrared Galaxy, Lir > 1012 L
HyLIG	HyperLuminous Infrared Galaxy, Lir > 1013 L
a Throughout this review we adopt H0 = 75km s-11 Mpc-1, q0 = 0. A luminosity quoted at a specific wavelength refers to L(), and is given in units of solar bolometric luminosity (3.83 x 1033 ergs s-1). The quantities f12, f25, f60, f100 are the IRAS flux densities in Jy at 12, 25, 60, and 100µm. The broad-band far-infrared luminosity, Lfir, is computed using the prescription given in Appendix B of Cataloged Galaxies and Quasars Observed in the IRAS Survey (1985). The scale factor C (typically in the range 1.4-1.8) is the correction factor required to account principally for the extrapolated flux longward of the IRAS 100 µm filter. DL is the luminosity distance. Lfir has mostly been replaced by the quantity Lir, which better represents the total mid- and far-infrared luminosity. Lir is computed by fitting a single temperature dust emissivity model ( -1) to the flux in all four IRAS bands and should be accurate to ± 5% for dust temperatures in the range 25-65K (Perault 1987).

Although some previously cataloged objects would prove to have extreme infrared properties, the vast majority were more modest infrared emitters as typified by the results reported by de Jong et al. (1984) for galaxies in the Shapley-Ames catalog. In a sample of 165 SA galaxies, IRAS detected nearly all late-type spirals (Sb-Sd) and Irr-Am galaxies, approximately half of the early type, S0-Sa, galaxies and none of the ellipticals. For those galaxies detected, L_ir/L_B = 0.1-5, with a mean value of ~ 0.4. The few objects with L_ir/L_B > 2 were typically SBs or irregulars. Objects with higher L_ir/L_B ratios tended to have warmer f₆₀/f₁₀₀ colors. The classic starburst galaxies M82 and NGC 253 had L_ir/L_B ratios of 3 and 5, and L_ir = 10^10.3 and 10^10.8 L respectively. No objects were found with L_ir > 10¹¹ L.

The more extreme infrared properties of infrared-selected samples are typified by objects in the IRAS minisurvey (Rowan-Robinson et al. 1984). For a complete flux-limited sample of 86 infrared-selected galaxies from the minisurvey, Soifer et al. (1984a) found that virtually all had L_ir > 10¹⁰ L and ratios L_ir / L_B = 1-50, with the fraction of interacting galaxies being as high as one fourth. More intriguing were the 9 ``unidentified'' sources (L<sub>ir</sub> / L<sub>B</sub> > 50) which had no obvious optical counterparts in galaxy catalogs and often no visible counterpart on the Palomar Sky Survey plates (Houck et al. 1984). Initial cross-correlation of larger IRAS source lists with galaxy catalogs had produced only one or two objects with similar extreme ratios, most notably the ULIG Arp 220 (Soifer et al. 1984b) and NGC 6240 (Wright et al. 1984, Joseph et al. 1984b). Ground-based observations of the unidentified minisurvey objects quickly led to the discovery of faint galaxies, typically at redshifts 0.1-0.2 (Aaronson & Olszewski 1984, Houck et al. 1985, Antonucci & Olszewski 1985, Allen et al. 1985, Iyengar & Verma 1984), implying that these objects also had ``ultrahigh'' infrared luminosities, typically L_ir 10¹² L, and L_ir / L_B = 30-400. None of these objects showed obvious evidence for an active nucleus.

IRAS surveys of optically selected Seyfert galaxies (Miley et al. 1985) and QSOs (Neugebauer et al. 1985 1986) showed that active galaxies could be strong far-infrared emitters; most optically selected AGNs had ratios L_ir / L_B in the range 0.2 to 1.0 with higher values in only a small number of objects. However, the full range of infrared excess exhibited by active galaxies is indeed much larger (e.g. Fairclough 1986). de Grijp et al. (1985) found that searches based on ``warm'' (f<sub>25</sub> / f<sub>60</sub> 0.3) colors could be useful for discovering new infrared-luminous active galaxies in the IRAS database. This technique appeared to have been motivated by the shape of the infrared spectrum of the Seyfert2 galaxy NGC 1068 (Telesco & Harper 1980) and the discovery of a similar ``warm'' 25-µm component in the broad-line, infrared-luminous radio galaxy 3C 390.3 (Miley et al. 1984). Early statistics suggested that the true space density of AGNs could be a factor of two larger than previously assumed with the majority of the new infrared selected objects being a mixture of LINERS and Seyfert2s.