Annu. Rev. Astron. Astrophys. 2001. 39: 249-307 Copyright © 2001 by Annual Reviews. All rights reserved

2. HISTORICAL REVIEW

Recognition that the brightness of the night sky is an important astronomical datum dates back at least to Olbers (1826; for historical review, see Harrison 1990). Early calculations of the optical background radiation due to galaxies within the context of general relativity were carried out by Shakeshaft (1954), McVittie & Wyatt (1959), Sandage & Tammann (1964), Whitrow & Yallop (1964, 1965). These authors dealt primarily with integrated starlight. Whitrow & Yallop (1965) also took account of starlight absorption by intervening galaxies and by intergalactic dust. Reemission of energy at infrared wavelengths by the absorbing dust was not considered.

The discovery of the CMB (Penzias & Wilson 1965) provided strong support for the notion of a hot, evolving early universe and led to recognition of the significance of a CIB for cosmology. In such a universe, one would expect a CIB, distinct from the CMB, associated with the formation of structure and condensation of luminous objects from primordial neutral matter following the decoupling of matter and radiation at a redshift z ~ 1100. Peebles considered this infrared background in 1965 [see Peebles (1993), unpublished lecture, pp. 146-147] and noted the lack of direct knowledge of the sky brightness in the three decades of wavelength from 1-1000 µm. The only observational limit on the CIB, which suggested that it was not large enough to close the universe, was provided by the presence of 10¹⁹ eV cosmic-ray protons, which would have been attenuated by photo-pion production by such a large CIB (Peebles 1969).

Partridge & Peebles (1967a) recognized that in order to produce the present metal abundances, young galaxies must have been more luminous than more evolved systems, and they studied the possibility of detecting them individually for various epochs of formation. In a second paper, Partridge & Peebles (1967b) also calculated the integrated infrared background that would be produced by their model young galaxies in various cosmological scenarios. The effect of dust was ignored, so the background was generally brightest in the 1 to 10 µm range. They compared these predictions with estimates of the foreground radiation from solar system and Galactic sources and correctly concluded that the CIB is much fainter than the foregrounds. Their work was important in stimulating observational programs to measure the CIB.

Harwit (1970) reviewed the status of early attempts to measure the infrared sky brightness. He noted the importance of background measurements for understanding discrete classes of objects, such as QSOs, which had been found to be highly luminous in the far infrared (Kleinmann & Low 1970, Low 1970). Measurement of the far-infrared background would set limits on the number and duration of such luminous episodes. The high far-infrared luminosity of galaxies in the local universe had led Low & Tucker (1968) to predict an infrared background peaking at a wavelength longer than 50 µm, with a total energy 1%-10% of that in the recently discovered CMB. As we show (Section 4.1), this prediction was substantially correct. Harwit also reviewed the relationship between low-energy photon backgrounds and energetic cosmic-ray electrons, protons, and -rays. Through the processes of inverse Compton scattering, photo-pion production, and electron-positron pair production, respectively, the CIB provides a source of opacity to each of these cosmic radiations. At the time of Harwit's review, only high upper limits had been obtained on the CIB. Even the bright diffuse foreground radiations were poorly known because of the difficulty of such measurements. Longair & Sunyaev (1972) presented a similarly uncertain description of the infrared background in a comprehensive review of extragalactic background radiation across the entire electromagnetic spectrum.

The early theoretical estimates of the infrared background ignored the effect of partial thermalization of starlight by dust. Since the mid-1970s, many investigators have taken this into account in models of varying degrees of sophistication and complexity (e.g., Kaufman 1976, Stecker et al. 1977, Negroponte 1986, Bond et al. 1986, 1991, Hacking & Soifer 1991, Beichman & Helou 1991, Franceschini et al. 1991, 1994). The modeling efforts listed here were all carried out before the CIB had been detected. Current knowledge of the CIB provides important new constraints on models for its origin (Section 5).