Annu. Rev. Astron. Astrophys. 2005. 43:
xxx-xxx Copyright © 2005 by Annual Reviews. All rights reserved |
A number of conclusions are now clear from the analysis of the identified sources in the CIB:
The comoving energy produced in the past that makes up the CIB at different wavelengths is more uniform that what is suggested by its spectral energy distribution. This is due to the fact that the CIB at long wavelengths ( 400 µm) is dominated by emission from the peak of the SED of galaxies at high z. More quantitatively, the ISOCAM surveys reveal that about two-thirds of the CIB emission at ~ 150 µm is generated by LIRGs at z ~ 0.7. At 850 µm, more than half of the submillimeter CIB is generated by SMGs. The brightest SMGs (S850 > 3 mJy, ~ 30% of the CIB) are ULIRGs at a median redshift of 2.2. The energy density at 150 µm, which is ~ 20-25 times larger than the energy density at 850 µm requires a comoving energy production rate at z = 0.7 roughly 10 times the energy production rate at z = 2.2.
The evolution exhibited by LIRGs and ULIRGs is much faster than for optically selected galaxies. The ratio of infrared to optical, volume-averaged output of galaxies increases rapidly with increasing redshift.
Luminosity function evolution is such that the power output is dominated by LIRGs at z 0.7 and ULIRGs at z 2.5.
The energy output of CIB sources is dominated by starburst activity.
AGN activity is very common in the most luminous of these galaxies even though this activity does not dominate the energy output. The rate and fraction of the energy produced increase with the luminosity.
LIRGs at z 0.7 are dominated by interacting massive late-type galaxies. Major mergers become dominant in ULIRGs at z 2.5.
SMGs show rather strong correlations with correlation lengths larger than those of other high redshift sources.
LIRGs and ULIRGs cannot be identified with any of the distant populations found by rest-frame ultraviolet and optical surveys.
Although these findings are answering the basic questions about the sources that make up the CIB, there are still observational difficulties to be overcome to complete these answers. The SEDs of LIRGs and ULIRGs are quite variable and often not very well constrained in their ratio of far-infrared to mid-infrared or to submillimeter wavelengths. The far-infrared, where most of the energy is radiated, requires cryogenically cooled telescopes. These have small diameters and, hence, poor angular resolution and severe confusion limits for blind surveys. Establishing proper SEDs for the different classes of infrared galaxies detected either in mid-infrared (with ISOCAM at 15 µm or MIPS at 24 µm) or in millimeter-submillimeter surveys is one of the challenges of the coming decade. Making sure that no class of sources that contribute significantly to the CIB at any wavelength has been missed is an other observational challenge. The submillimeter galaxies not found through the radio-selected sources and the question of the warm submillimeter galaxies are also two of those challenges.
Multiwavelength observations of high-z infrared galaxies give a number of new insights on the galaxy formation and evolution problem. As an example, the gas masses and total masses of SMGs are found to be very high. There is a first indication that the number of such high-mass object at redshifts between 2 and 3 is uncomfortably large compared to semianalytical models of galaxy formation based on the standard hierarchical structure-formation frame. The evolution of the luminosity function is dominated by more luminous sources as redshift increases. This is surprising because the mass function of the collapsed structure is expected to be dominated by smaller and smaller objects as redshift increases.
The populations of infrared galaxies concentrated at z 1 and at z 2.5 studied so far reveal rather different type of sources. The lower redshift ones seem to be starburst phases of already-built massive, late-type field galaxies accreting gas or gas-rich companions forming the disks. We see today a rapid decrease of this activity probably associated with a dry out of the gas reservoir in their vicinity. The larger redshift ones, which are also more luminous, seem to belong to more massive complex systems involving major merging. These systems could be located in the rare larger amplitude peaks of the large-scale structures leading to massive elliptical galaxies at the center of rich clusters. The redshift distribution of these seems quite similar to the redshift distribution of quasars.
Interesting problems that are central to the understanding of galaxy formation and evolution have to be solved in the next decade:
Determining the role of the large scale environment (nodes, filaments and sheets of the large-scale structures) on star formation;
Find the relative rates of accretion of gas and smaller galaxies in the growth of massive objects;
Establish the cycle of bulge versus disk formation, as a function of the ratio between stars and gas in the accreted material;
Identify the different types of starburst (in a disk or in the nucleus, interaction or merger driven); and
Estimate the fraction of time spent in the starburst phase and the duration of this phase
Current observations all point in the direction of a possible strong effect of the large-scale environment and the need for models of hierarchical formation and evolution that include properly star versus gas ratio in the accreted material.
Finally the connection between the starburst phenomenon and the AGN activity is an old question still largely unresolved. Recent observations of infrared/submillimeter galaxies have reinforced the link but have not much improved our understanding of the physical link.
ACKNOWLEDGMENTS
We are very grateful to Alexandre Beelen, Karina Caputi, David Elbaz, François Hammer, and George Helou for very useful discussions during the writing of this manuscript. We also thank Pierre Chanial for providing us the M82 SED. Finally, we warmly thank the scientific editor, who found a large number of typos and mistakes in our use of English. The reading of the paper has been significantly improved by his detailed corrections.