The infrared region of the spectrum is a very interesting and important one for the study of extragalactic backgrounds, both galaxian and pre-galactic. This review concentrates on backgrounds from protogalaxies and galaxies. Detailed discussions of many possible pre-galactic backgrounds have recently been presented by Bond, Carr and Hogan (1986, 1991); see Carr (1982) for a recent review of these and Wright et al. (1993) for a summary of how well some of them have fared in light of the most recent COBE observational results. I will concentrate on the approximate wavelength range 10 to 500µm since both the near-infrared and the submillimeter range are discussed by others.
The infrared spectral region is a good place to look for detectable backgrounds from the integrated light of galaxies and protogalaxies for several reasons. First, young galaxies and protogalaxies may have been relatively much more luminous at far-infrared wavelengths, compared to the optical-UV, than galaxies at the present epoch. Metallicity can increase rapidly during the early evolution of galactic systems, and if dust formation follows suit the dust optical depth and the far-infrared luminosity can rise dramatically at the expense of the obscured optical-UV luminosity (Wang 1991a, b; Lonsdale 1992; Mazzei, De Zotti and Xu 1993).
Figure 1 illustrates a second reason for the importance of the infrared region to background studies of galaxies and protogalaxies: the prominent "windows" between the various foregrounds and the cosmic microwave background radiation (CMB). This figure is similar to several which have been shown already at this meeting, depicting the intensity I in W / cm2 / sr. The main foregrounds in the infrared spectral region shown in Figure 1 are the zodiacal light which peaks near 1µm and falls into the far-infrared, the interplanetary dust (IPD) emission peaking near 10µm, and the interstellar dust (ISD) peaking beyond 100µm. There are two main infrared "windows": one near 3µm and the second at about 300µm.
Figure 1. The infrared foregrounds and COBE limits. The solid curve is the cosmic microwave background (CMB) radiation, and the other three curves are various foregrounds, as derived by Beichman and Helou (1991; C. Beichman, private communication): the dotted curve is the reflected solar zodiacal emission, the dot-dash curve the thermal emission from interplanetary dust, and the dashed curve the thermal emission from interstellar dust scaled to a brightness of I (100µm) = 1 MJy/sr, which is representative of the typical sky brightness in regions of the weakest cirrus emission at high galactic latitude. The lower short heavy solid line illustrates the maximum deviation from the CMB measured by FIRAS (Mather et al. 1993), 3.4 x 10-8 erg cm-2 s-1 sr-1 cm, which is 0.03% of the peak of the CMB spectrum, while the upper line represents an estimate of this maximum deviation adopting a more conservative galactic foreground subtraction (see text). The solid squares depict the DIRBE limits at the north ecliptic pole, as discussed by M. Hauser - note that these do not include any subtraction of the galactic or zodiacal foregrounds.
It is not simply the existence of these windows that marks their importance, but also the fortunate chance that they happen to coincide very nicely with the two prominent peaks in the spectral energy distributions of moderate-to-high redshift galaxies: the stellar spectral energy distribution of nearby galaxies peaks near 1µm, thus moves into the 3µm window with increasing redshift, while the dust re-emission peak of ISM-rich galaxies peaks near 60 to 100µm, moving into the 300µm window with increasing redshift. Thus there is a rich hunting ground for the integrated stellar light of galaxies in the near-infrared window, and another one for the dust emission of galaxies in the far-infrared window. Conversely it will be difficult to ever measure the integrated light of galaxies or protogalaxies in the 5 to 30µm region unless spacecraft can be sent to the more distant reaches of the solar system where the interplanetary dust emission is much reduced.
Another reason why the infrared spectral region is one of the most valuable for studying the background light due to galaxies is that there is a strong positive far-infrared K-correction with redshift. Unlike the situation in the UV through near-infrared spectral region, the energy distributions of galaxies at 80µm fall with increasing wavelength with a very steep dependence on , therefore at longer wavelengths than this the K-correction can almost counter the cosmological effects of luminosity distance and surface brightness dimming so that the apparent flux density at a fixed observing frequency has little dependence on distance. The same effect holds to a more limited extent in the 3 - 20µm region.
Finally, it is possible that intervening galaxies may produce sufficient obscuration to eliminate optically-selected background quasars from flux-limited samples (Ostriker and Heisler 1984; Wright 1990; Fall and Pei 1993), especially if there has been strong evolution of the dust optical depth (Wang 1991a, b; Mazzei et al. 1993). It is therefore also possible that such an effect will obscure background young galaxies and proto-galaxies. At far-infrared wavelengths, not only will the obscuration be low enough to be insignificant, but the dust which is responsible for extinguishing the optical-UV light will re-emit this light in the far-infrared and submillimeter.
Mike Hauser has given us an excellent summary of the observational results on the infrared backgrounds. For the purposes of my discussion of the theoretical backgrounds expected from galaxies and protogalaxies, I summarize on Figure 1 the most recent observational limits from DIRBE and FIRAS on COBE. As Mike has described, the DIRBE data do not yet include any foreground subtraction, pending the very difficult task of modeling the galactic emission in detail. The spectacular FIRAS results of Mather et al. (1993) do include a galactic foreground subtraction, and have a maximum deviation from the CMB blackbody spectrum in the 2 - 20 cm-1 region of 0.03% of the peak of the CMB spectrum. However, as Wright et al. (1993) have noted, this foreground subtraction is not appropriate for modeling the cosmological backgrounds due to the integrated light of galaxies because the background itself is expected to have a spectral shape similar to that of the galaxy, thus the "galactic foreground" that has been subtracted could include some cosmological background. Wright et al. used a csc |b| method of galactic foreground subtraction to avoid this problem, and from their integrated galaxy light model fits I estimate a maximum deviation of about twice that inferred by Mather et al. This is illustrated by the upper of the two heavy solid lines in Figure 1.