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

5. ORIGIN AND EVOLUTION OF LUMINOUS INFRARED GALAXIES

Table 3. IRAS galaxy properties versus L_ir

		10.5-10.99	11.0-11.49	11.5-11.99	12.0-12.50
log (Lir / L)
No. of objects a		50	50	30	40
Morphology	merger	12%	32%	66%	95%
	close pair	21%	36%	14%	0%
	single (?)	67%	32%	20%	5%
Separation b	[kpc]	36.	27.	6.4	1.2
Opt Spectra	Seyfert 1 or 2	7%	10%	17%	34%
	LINER	28%	32%	34%	38%
	H II	65%	58%	49%	28%
Lir / LBc		1	5	13	25
Lir / L'CO c	[L (K km s-1 pc2)-1]	37	78	122	230
a Objects in the IRAS BGS plus additional ULIGs from Kim & Sanders (1996). b Mean projected separation of nuclei for mergers and close pairs only. c Mean values

The data presented in Section 4 are sufficient to present a fairly detailed picture of the morphology of LIGs and to investigate how their properties change as a function of infrared luminosity. Table 3 summarizes some of the main results using observations of the complete sample of LIGs in the BGS supplemented by data for a subsample of less luminous BGS objects and data for a larger sample of ULIGs from the survey of Kim (1995).

5.1 Strong Interactions and Mergers

It now seems clear that strong interactions and mergers of molecular gas-rich spirals are the trigger for producing the most luminous infrared galaxies. The images of ULIGs from the BGS (see Figure 4) show clearly that maximum infrared luminosity is produced close to the time when the two nuclei actually merge. At L_ir < 10¹¹ L the vast majority of infrared galaxies appear to be single, gas-rich spirals whose infrared luminosity can be accounted for largely by star formation. Over the range L_ir = 10¹¹-10¹² L there is a dramatic increase in the frequency of strongly interacting systems that are extremely rich in molecular gas; at the low end of this range the luminosity appears to be dominated by starbursts with Seyferts becoming increasingly important at higher luminosities. Those objects that reach the highest infrared luminosities, L_ir > 10¹² L, contain exceptionally large central concentrations of molecular gas; because of heavy dust obscuration it is hard to distinguish the relative roles of starburst and AGN activity, although the conditions are clearly optimal for fueling both enormous nuclear starbursts as well as building and/or fueling an AGN.

The enormous build-up of molecular gas in the centers of the most luminous infrared objects plays a fundamental role in LIGs and is best illustrated by showing data for several relatively nearby well-studied objects. The ultimate fate of these mergers, once their infrared excess subsides, is not completely clear. Examples are also shown below for a small but important subclass of objects which show both strong optical and infrared emission, and which plausibly represent a transition stage in the evolution of LIGs into optically selected AGN (e.g. Sanders et al. 1988b). Finally we mention a few objects initially identified as HyLIGs that illustrate why caution needs to be taken when identifying objects at higher-z.

5.2 Case Studies

Figure 8. Well-studied mergers: (a) NGC 4038 / 39 (Arp 244 = ``The Antennae''); (b) NGC 7252 (Arp 226 = ``Atoms for Peace''); (c) IRAS 19254-7245 (``The Super Antennae''); (d) IC 4553 / 54 (Arp 220). Contours of H I 21-cm line column density (black) are superimposed on deep optical (r-band) images. Inserts show a more detailed view in the K-band (2.2 µm) of the nuclear regions of NGC 4038 / 39, NGC 7252, and IRAS 19254-7245, and in the r-band (0.65 µm) of Arp 220. White contours represent the CO(1->0) line integrated intensity as measured by the OVRO millimeter-wave interferometer. No H I or CO interferometer data are available for the southern hemisphere object IRAS 19254-7245. The scale bar represents 20 kpc.

5.2.1 MERGERS OF MOLECULAR GAS-RICH SPIRALS   NGC 4038 / 39 (Arp 244 = VV 245 = ``The Antennae'') This classic, nearby ``early merger'' system is composed of what appear to be two overlapping, distorted, late-type spiral disks (Figure 8a). The total infrared luminosity is the minimum required by our definition of LIG. The K-band image shows two nuclei ~ 15 kpc apart and what appears to be a large ring of bright H II regions in the northern disk. CO interferometer observations (Stanford et al. 1990) show that ~ 60% of the 3.9 x 10⁹ M of molecular gas in the system (Sanders & Mirabel 1985, Young et al. 1995) is concentrated in the two nuclei and in a large off-nuclear complex where the two disks strongly overlap. No CO has been detected in the extended tails. In contrast, about 70% of the total H I mass, M(H I) ~ 10⁹ M, is associated with the long (total extent of ~ 100 kpc) tidal tails (van der Hulst et al., in preparation). About one quarter of the total H I mass is located at the tip of the southern ``antenna''. In this early merger the total ratio M(H₂) / M(H I) is ~ 4.

NGC 7252 (Arp 226 = ``Atoms for Peace'') NGC 7252 has been considered the prototype of a late-stage merger (Schweizer 1978). This object has L_ir ~ 5 x 10¹⁰ L and therefore is currently not a LIG. The K-band image shows a single nucleus inside what has been described as a relaxed elliptical body. The optical image shows relatively large symmetric tails, which rotate in opposite directions and have a total extent of ~ 130 kpc (Figure 8b). NGC 7252 has been described as a decaying merger of two massive gas-rich late-type spirals caught in the act of forming a single elliptical galaxy (Schweizer 1978, Casoli et al. 1991, Fritze-von Alvensleben & Gerhard 1994). However, it is not yet clear whether this system has passed through a LIG phase. Single-dish (Dupraz et al. 1990) and interferometer (Wang et al. 1992) CO observations reveal a nuclear molecular gas disk at r < 1.5 kpc, which contains nearly 75% of the total H₂ mass of 4.7 x 10⁹ M. Hibbard et al. (1994) and Hibbard & van Gorkom (1996) detect a slightly smaller total mass of H I (3.6 x 10⁹ M), but all of the H I is located in the tidal tails. Until most of the H I and H₂ is either consumed, expelled, or ionized, it would appear that NGC 7252 still has too much cold gas to resemble most present-day ellipticals.

IRAS 19254-7245 (``The Super Antennae'')   IRAS 19254-7245 is a remarkable ULIG in which two distinct rotating merging disks can still be identified (Mirabel et al. 1991). The colossal tidal tails have a total extent of ~ 350 kpc (Figure 8c). Among the 20 ULIGs in the IRAS BGS this object has the largest projected nuclear separation (~ 10 kpc; Melnick & Mirabel 1990). If the IRAS luminosity has the same spatial distribution as the observed 10-µm luminosity, then 80% of the total luminosity, L_ir = 1.1 x 10¹² L, originates in the southern, heavily obscured Seyfert nucleus. The total molecular gas content of the system is M(H₂) ~ 3.0 x 10¹⁰ M (Mirabel et al. 1988).

IC 4553 / 54 (Arp 220)   At a distance of 77 Mpc, Arp 220 is the nearest ULIG (by a factor of ~ 2). In contrast to ``The Super Antennae'', it shows two relatively short and wide tails (Figure 8d). The nuclear region is completely obscured in the optical, but two distinct nuclei separated by 0.8" (~ 300 pc) are detected at K-band (Graham et al. 1990). The radial brightness profile at 2.2 µm is closely approximated by a r<sup>-1/4</sup> de Vaucouleurs' profile (Wright et al. 1990, Kim 1995). The total mass of molecular gas is M(H<sub>2</sub>) ~ 3.6 x 10<sup>10</sup> M, with ~ 2/3 of it inside a projected radius of ~ 300 pc. Compact OH megamaser emission ( 10 pc in diameter) has recently been discovered at the center of Arp 220 (Londsdale et al. 1994). H I is detected in absorption against the nuclear radio continuum source (Mirabel 1982). In emission, all of the ~ 2.3 x 10<sup>9</sup> M of H I is located outside the main body (Hibbard & Yun 1996). Although as luminous as ``The Super Antennae'', the shorter tidal tails, smaller nuclear separation, and more relaxed central body of Arp 220 suggest that it is a more advanced merger. Because of the strong H I absorption it is not possible to estimate the total mass of H I in the merger disks; however, most of the cold gas there is likely to be in molecular form, and the overall ratio M(H₂) / M(H I) is probably close to 15.

5.2.2 ENCOUNTERS IN CLUSTERS   Most LIGs detected by IRAS appear to involve strong interactions/mergers of molecular gas-rich spirals where the pairs are either isolated or part of small groups. In these interactions/mergers the relative mean velocity of the individual members is typically 200 km s^-1. These conditions are not expected to be typical of cluster environments where the relative velocities are often much higher, and many galaxies may be either gas-poor spirals (from ram-pressure striping) or already transformed into ellipticals (perhaps due to past mergers). Although multiple high-speed encounters in clusters may produce some LIGs (e.g. Moore et al. 1996), the relatively small number of IRAS LIGs found in clusters suggests that lower-speed mergers in pairs or loose groups may be a more efficient way of enhancing infrared activity. As examples of the types of strong interactions involving relatively large ( L*) galaxies in low-z clusters, we discuss here observations of an elliptical-spiral encounter (Arp 105) and a relatively high-speed lenticular-spiral encounter (NGC 5291A/B).

Arp 105 (Figure 9a) consists of an infrared luminous (L_ir ~ 10¹¹ L) spiral galaxy (NGC 3561A) being torn apart by a massive elliptical (NGC 3561B) in the X-ray rich cluster Abell 1185. Duc & Mirabel (1994) find that the elliptical has already accreted gas-rich objects and may be the precursor of a cD galaxy. At the tip of one of the colossal tidal tails that emanate from the spiral are a compact dwarf galaxy and an irregular galaxy of Magellanic type. The H I and CO interferometer observations by Duc (1995) show that the collision has caused a marked spatial separation of the cold interstellar gas: Whereas all of the ~ 10¹⁰ M of molecular gas (H₂) is found within the central ~ 6 kpc radius of the spiral, a similar mass of atomic gas (H I) is found outside that region, most of it far from the spiral galaxy near the tips of the tidal tails. H I clouds infalling at high velocities toward the nucleus of the elliptical are detected in absorption against a central compact radio source.

Figure 9. Encounters in clusters: (a) NGC 3561A/B (Arp 105); (b) NGC 5291A/B (``Sea shell''). Contours of H I 21-cm line column density (black) are superimposed on deep optical (r-band) images. Inserts show a more detailed view in r-band of the spiral galaxy NGC 3561A (Duc & Mirabel 1994), and of the interacting pair NGC 5291A/B. White contours represent the CO(1->0) line integrated intensity as measured by the IRAM millimeter-wave interferometer. CO emission has not been detected in NGC 5291A/B. The scale bar represents 20 kpc.

Figure 9b shows an optical image of NGC 5291A/B (Duc & Mirabel 1996, in preparation) and H I column density map (Malphrus et al. 1996, in preparation). The nuclei of the lenticular galaxy NGC 5291A and the spiral NGC 5291B (the ``Sea shell'' galaxy) are at a projected separation of 12 kpc and are moving with a relative radial speed of ~ 450 km s^-1 (Longmore et al. 1979). Figure 9b reveals a multitude of intergalactic H II regions superimposed on an arc-like distribution of H I, which has a diameter of ~ 200 kpc and a total H I mass of ~ 10¹¹ M (Simpson et al., private communication). However, optical spectroscopy of the spiral that is being torn apart shows no signs of recent star formation, which is consistent with its apparently low infrared luminosity ( L_ir < 10⁹ L); this source was not detected by IRAS.

5.2.3 MOLECULAR GAS-RICH MERGERS IN QSOs AND PRGs   Infrared and millimeter-wave observations of QSOs and PRGs show that a significant fraction of these objects have values of L_ir and M(H₂) that overlap values found for infrared-selected ULIGs. These data have been used as evidence for an evolutionary connection between QSOs, PRGs, and ULIGs (e.g. Sanders et al. 1988b, Mirabel et al. 1989). Figure 10 shows high-resolution optical images of the optically-selected QSO Mrk 1014, the radio-selected PRG Pks 1345+12, and the radio-selected QSO 3C 48, three objects that exhibit several properties in common with ULIGs.

Figure 10. Optical images of QSOs and PRGs with strong infrared emission: (a) Mrk 1014 (PG 0157+001), (b) 4C 12.50 (Pks 1345+12), (c) 3C 48. The ``+'' sign indicates the position of putative optical nuclei. Tick marks are at 5" intervals, and the scale bar represents 10 kpc.

Figure 10a shows the UV-excess QSO Mrk 1014 at a redshift of 0.163. The deep CCD image shows two tidal tails extending from a large distorted disk with a total diameter of ~ 90 kpc (e.g. Mackenty & Stockton 1984). This warm infrared object has L_ir ~ 3 x 10¹² L and M(H₂) ~ 4 x 10¹⁰ M (Sanders et al. 1988c).

Figure 10b shows a deep optical image of the PRG 4C 12.50. This object exhibits a double nucleus embedded in a distorted disk ~ 85 kpc in diameter (see also Heckman et al. 1986). IRAS observations imply L_ir ~ 1.6 x 10¹² L. CO emission (Mirabel et al. 1989) and H I absorption (Mirabel 1989) with widths of ~ 950 km s^-1 have been detected from this object. The CO data implies an extremely large mass of molecular gas, M(H₂) ~ 6.5 x 10¹⁰ M. Mirabel (1989) proposed that objects like 4C 12.50 (which is classified as a compact steep-spectrum radio source) are relatively young PRGs that have evolved from progenitors like Arp 220, where a powerful compact source of nonthermal radio emission is still confined by a large nuclear concentration of gas and dust.

Figure 10c shows a recent deep optical image of 3C 48, the first radio source to be identified as a QSO (Matthews & Sandage 1963) and the second QSO to have its redshift determined (Greenstein & Matthews 1963). 3C 48 appears to have a double nucleus and at least one prominent tidal tail (Stockton & Ridgway 1991). Neugebauer et al. (1985) used the strength and shape of the far-infrared continuum, and Stein (1995) used the f₆₀ / f₂₅ ratio versus the radio spectral index at 4.8 GHz to argue that the dominant luminosity source in this classic QSO is star formation in the host galaxy rather than the central AGN. IRAS observations of 3C 48 imply L_ir ~ 5 x 10¹² L, and the recent detection of CO emission by Scoville et al. (1993) implies a molecular gas mass M(H₂) ~ 4 x 10¹⁰ M (see also Evans et al. 1996b).

5.2.4 GRAVITATIONAL LENSING IN HIGH-Z OBJECTS   The identification of the IRAS Faint Source 10214+4724 with an emission-line galaxy at a redshift of 2.286 (Rowan-Robinson et al. 1991) has attracted considerable interest. The apparent enormous infrared luminosity, L_ir ~ 2 x 10¹⁴ L, the detection of 10¹¹ M of molecular gas (Brown & Vanden Bout 1991; Solomon et al. 1992b, c; Radford et al. 1996), and the detection of strong submillimeter continuum emission from dust (Clements et al. 1992, Downes et al. 1992) led many observers to propose that this object was a prime candidate for a ``primeval galaxy''. However, IRAS 10214+4724 exhibits at least one unusual property for an object with such extreme luminosity and mass; the CO line width of ~ 250 km s^-1 is small for such a massive object.

Figure 11. HST Planetary Camera (F814W) image of the gravitational lens object IRAS 10214+4724 (Eisenhart et al. 1996).

The idea that IRAS 10214+4724 might be gravitationally lensed by a foreground galaxy or group of galaxies (Elston et al. 1994, Trentham 1995, Broadhurst & Lehár 1995) gained support from sub-arcsecond K-band images obtained with the Keck telescope (Matthews et al. 1994, Graham & Liu 1995), and now appears certain from the image obtained with the Hubble Space Telescope (HST) (Figure 11). The geometry of the lensed arc suggests an amplification factor of ~ 30 for the infrared emission (Einsenhart et al. 1996). CO interferometer maps indicate that the molecular gas may be somewhat more extended than the infrared emission (Scoville et al. 1995, Downes et al. 1995), and Downes et al. (1995) suggest that the CO may be amplified by a factor 10. Thus, the intrinsic infrared luminosity and molecular gas mass of this object may only be L_ir ~ 7 x 10¹² L and M(H₂) ~ 2 x 10¹⁰ M, respectively, suggesting that IRAS 10214+4724 is not a unique infrared selected object but is better described as a high-redshift analogue of ULIGs found in the local Universe.

At present, the only other IRAS object at z > 2 that has been detected in CO is the Cloverleaf QSO at z ~ 2.5, which is also a lensed object (Barvainis et al. 1994). After correcting for amplification, the Cloverleaf appears to have a similar infrared and submillimeter spectrum as for IRAS 10214+4724 (Barvainis et al. 1995), again suggesting that this object is yet another high-redshift analog of local ULIGs.

However, several objects at z > 4 not detected by IRAS have recently been shown to be strong far-infrared (rest-frame) emitters (Omont et al. 1996a), and one of these sources - the radio-quiet QSO BR1202-0725 at z = 4.69 (Irwin et al. 1991) - was recently detected in CO (Ohta et al. 1996, Omont et al. 1996b) implying M(H₂) ~ 6 x 10¹⁰ M. This object has been shown to have a submillimeter spectrum remarkably similar to IRAS 10214+4724 (McMahon et al. 1994, Isaak et al. 1994). An HST image (Hu et al. 1996) shows no obvious evidence that this object is lensed, but it may be premature to rule out this possibility.

If one extrapolates the local luminosity function of ULIGs assuming relatively strong evolution (e.g. Kim & Sanders 1996), then the probability that a ULIG detected at z 2 is lensed by at least a factor of 10 can be as high as ~ 30% (Trentham 1995, Broadhurst & Lehár 1995). Future studies of LIGs at high redshifts will almost certainly require high-resolution imaging to test whether these objects are lensed.