1.1. The Biggest and the Brightest
In human endeavour there's a fascination with the "biggest and the best", or the "best and the brightest". It's a matter for the social and psychological sciences to speculate on the reasons we feel driven to give Oscars to the best movies and to climb the highest mountains, but being human (well, most of them anyway), astrophysicists are not immune to the desire to search for the Universe's own brand of the biggest and the brightest. We tend to give them extreme names such as ultra-this or hyper-that. In their rarity however, unusual objects and environments teach us about the most extreme physical processes in the Universe.
Which brings us to ULIRGs (or ULIGs to some) - Ultra Luminous InfraRed Galaxies. There are also HLIRGs (or HiLIRGs or HyLIRGs, or indeed HyLIGs): Hyper Luminous InfraRed Galaxies. These Oscar contenders have historically been defined simply in terms of luminosity: L8-1000µm between 1012 and 1013 L for the ULIRGs and > 1013 L for HLIRGs 1. What are these dramatic objects, and why are they among the brightest objects in the Universe? Does this also imply they are the biggest in term of mass or size, or are they just superficial fireworks that leave little lasting impression? What can understanding these enigmatic objects teach us about the evolution of galaxies and how the the Universe came to look as it now does?
The last few years have seen a dramatic shift in our perceptions of ULIRGs. Once believed by many to be a rare curiosity - certainly interesting, yes, but perhaps no more than a local oddity - they're beginning to see center stage much more frequently. Ironically, this increased interest in these rare objects has arisen because we now realize that ULIRGs were once not nearly so rare as we find them to be in the local Universe: pioneering submillimeter and millimeter surveys have demonstrated that ULIRGs are many hundreds of times more numerous at z > 1 than they are locally. This in turn suggests that they played a much more important role in galaxy formation and evolution than we imagined, and so understanding them becomes of prime importance. Fortunately, this rise in the fortunes of the ULIRG has occurred in an era when many new observing capabilities are coming on line. Foremost among these is the Spitzer Space Telescope, with its suite of deep imaging infrared cameras and its sensitive infrared spectrograph. Spitzer's extensive first results on ULIRGs are just now beginning to be published. Submillimeter and millimeter cameras are also improving dramatically; upcoming facilities include AzTEC, SCUBA-2, Herschel, and ALMA. We can also look forward to major new insights from mid- and near-IR facilities such as ASTRO-F, WISE, and JWST, as well as radio facilities such as the Square Kilometer Array, and new X-ray facilities with high hard X-ray sensitivities such as Con-X and XEUS. The timeliness of this subject is exemplified by recent reviews of The Cosmic InfraRed Background [Lagache, Puget & Dole 2005], Interacting Galaxies [van Gorkum & Hibbard 2005], Megamasers [Lo 2005], Galactic Winds [Veilleux, Cecil, & Bland-Hawthorn 2005], and High Redshift Molecular Gas [Vanden Bout & Solomon 2005], all of relevance to ULIRGs.
ULIRGs were first discovered in large numbers by the Infrared Astronomical Satellite in 1983, and were found to be comparatively rare locally, with a space density several orders of magnitude lower than that of normal galaxies, and possibly a factor of a few higher than QSOs. Followup observations show that most, if not all ULIRGs are found in major disk mergers, and that the central few hundred pc of their nuclear regions harbour very large masses of gas and dust. The power source behind the IR emission is some combination of a large population of hot young stars (a `starburst' 2) or a very massive black hole accreting matter at a rapid rate (which for the remainder of this review we refer to as an `AGN'). Though distinguishing between the two initially (and even now) proved to be very difficult, it is now thought that, at least locally, ULIRGs are mainly powered by a starburst, but frequently with a significant AGN contribution. Local ULIRGs reside in relatively low-density environments (not unexpectedly, since relative velocities are thought to be too high for mergers to occur in rich, virialized environments), and are expected to evolve into spheroidal systems as the galaxy mergers that appear to trigger ULIRG activity progresses.
Even IRAS was sensitive enough to determine that there has been very strong evolution in the ULIRG (and LIRG) population with redshift out to at least z ~ 0.5, with approximate form (1 + z)4. IRAS also found ULIRGs out to extremely high redshifts, including the famous, lensed, F10214+4724 at z = 2.286. This strong evolution was confirmed with results from the Infrared Space Observatory which, although covering much smaller areas than IRAS, could probe this evolution out to z ~ 1 due to its greater sensitivity (Figure 1). This evolution was later recast as the now ubiquitous `star formation history of the Universe' figures, which show that LIRGs rather than ULIRGs are responsible for the bulk of the evolution seen since z ~ 1 in IR galaxies. ULIRGs, however, did not slink into the shadows; on the contrary they returned triumphant with the advent of submillimeter imaging surveys, which came shortly after ISO and can in principle probe IR-luminous galaxies up to z ~ 7. These sub-mm surveys showed that ULIRGs are orders of magnitude more numerous at z > 1 than locally, outnumbering optically bright QSOs at those redshifts by a large margin. Followup observations showed that these distant ULIRGs bear many similarities to their local cousins, but also exhibit some key differences, and that they may signpost the obscured phases of the very dramatic events suspected of building the most massive galaxies seen in the local Universe.
Figure 1. Simulation of the ability of recent infrared surveys to discover ULIRGs, with shorter wavelength ( < 25 µm) surveys plotted in the upper panel and longer wavelength surveys below. The band in the upper figure at z ~ 1.5 is produced by the the 10 µm silicate absorption feature which falls in the Spitzer 24 µm band at that redshift. Wide shallow surveys have the largest volume for discovering the most luminous ULIRGs, while narrow deep surveys can of course find the most distant ones, though in much smaller number. Tiered ('wedding cake') surveys are thus required to construct complete luminosity functions within a given redshift interval (vertical slices). Based on the simulations of [Xu et al. 2003], these simulations can fit IRAS, ISO, Spitzer and submillimeter counts and redshift distributions. We plot 1/5 of the objects expected within the area and depth of each survey: ISO ELAIS 15 µm [Vaccari et al. 2005]; SWIRE 24 & 70 µm [Surace et al. 2005]; Spitzer Guaranteed Time Observer deep 24 µm [Pérez-González et al. 2005]; GOODS 24 µm [Chary et al. 2004]; IRAS 60 µm [Lonsdale et al. 1990]; SCUBA 850 µm [Scott et al. 2002]; GOODS 70 µm (D. Frayer, priv. comm.); Spitzer GTO deep 160 µm [Dole et al. 2004].
When considered within the framework of modern theories for the formation of galaxies and large-scale structure, it seems initially surprising that there are many more ULIRGs at high redshift than locally, because in early implementations of the `hierarchical buildup' paradigm, large galaxies build up slowly from the mergers of smaller systems. This is in contrast to the early `monolithic collapse' models [Eggen, Lynden-Bell & Sandage 1962], where ellipticals formed early in a dramatic burst of star formation, which had been largely supplanted in favour of hierarchical models. The discovery of so many ULIRGs at high redshifts caused hierarchical models of the time major difficulties in making enough distant systems with such high star formation rates. The basic dark matter halo growth theory, described by an extended/modified Press-Schechter formalism, does however allow for rapid baryon accumulation in very massive dark matter halos, and recent galaxy formation models are having greater success in producing the observed number of ULIRGs in sub-mm surveys, albeit with some stringent assumptions.
In this review, we will therefore focus on selected key topics: (1) our understanding of the astrophysics of local ULIRGs, and in particular the relative importance of star formation versus AGN in powering ULIRGs, (2) similarities and differences between local and high-redshift ULIRGs, and (3) the relationship between ULIRGs and the formation of large-scale structure and of galaxies as a function of redshift. Since the study of ULIRGs clearly connects to many major disciplines of observational and theoretical extragalactic astronomy we cannot hope to cover all topics of relevance to them in this review. Nor can we completely review all recent published studies of ULIRGs; excellent ULIRG papers simply abound. We therefore highlight the most recent advances and our perspective on the most important questions concerning their study within the framework of galaxy and structure formation.
In section 2 we provide brief historical context to the discovery of ULIRGs and their evolutionary role. In section 3 we review current understanding of the astrophysics of local ULIRGs by wavelength, and in section 4 we summarize local studies into a picture of ULIRG nature and evolution in the local Universe. In section 5 we review observations of ULIRGs at higher redshifts, based primarily on data from ISO, SCUBA, HST and Spitzer. Section 6 places these studies into the context of structure formation and reviews their role within galaxy formation scenarios. Finally, section 7 highlights the key open questions and our perspectives of where the answers are likely to come from.
For the remainder of this review we usually refer to ULIRGs and HLIRGs combined as ULIRGs, because many of the earlier works used the term "ULIRG" to refer to all objects above 1012 in L and the term HLIRG has been in only recent and inconsistent use. We assume H0 = 70 km s-1 Mpc-1, = 1, and = 0.7. Luminosities are quoted in units of bolometric solar luminosities, where L = 3.826 × 1026 Watts. Unless otherwise stated, the term `IR' or `infrared' luminosity refers to the integrated rest-frame luminosity over 1-1000 or 8-1000 µm (which differ very little for most SEDs).
1 The supporting cast consists of LIRGs (or LIGs), the much more common lower luminosity understudies of the prima donnas, with L8-1000µm between 1011 and 1012 L Back.
2 for our purposes defined as a star forming event with a gas exhaustion timescale very short compared to the Hubble time Back.