Annu. Rev. Astron. Astrophys. 1994. 32: 115-52 Copyright © 1994 by Annual Reviews. All rights reserved

Far Infrared Emission

The IRAS all-sky survey has provided measurements, or their upper limits, of the far infrared (FIR) flux in the 12, 25, 60 and 100µ bands. FIR flux measurements of normal galaxies show a clear distinction between elliptical and spiral galaxies. Ellipticals have a much poorer detection rate and when detected are generally lower in both FIR luminosity and in the ratio L_FIR/L_B (de Jong et al. 1984; Bothun et al. 1989; Sauvage & Thuan 1993). Exceptions are found in those early-type systems apparently involved in interactions, e.g., NGC 1275 and NGC 1316.

The type dependences of FIR surface density _FIR and luminosity L_FIR are illustrated in Figs 3b and 4d. Many of the discussions in the literature focus on FIR-bright systems. These tend to be prominent starburst galaxies (e.g., M82), often peculiar in their morphological appearance and/or clearly involved in interaction with another system (Sanders 1987, but see Haynes & Herter 1988; Kennicutt 1990). In this review we avoid discussion of such galaxies.

Figure 4. Same as Figure 2, for (a) log total HI mass MHI, (b) log HI mass-to-blue luminosity ratio MHI / LB, (c) log HI mass fraction MHI / MT, (d) log FIR luminosity LFIR. The dashed lines indicate significantly fewer data for these types.

Three possible origins of this FIR radiation are commonly identified:

1.	Dust heated by nearby young, massive stars and reradiating in the infrared. The dust-molecular cloud complexes in the plane of the Milky Way are an example.
2.	Dust reradiating the heating by the general interstellar radiation field, e.g., the Galactic cirrus cloud population.
3.	Thermal and/or nonthermal radiation from active galactic nuclear regions, e.g., Seyfert galaxies.

Reviews of IR and FIR radiation are given by Soifer et al. (1987), Telesco (1988), Cox & Mezger (1989), Rowan-Robinson (1990).

If the massive-star population, i.e., O-B stars, are the dominant dust heating source then the FIR is a good measure of star formation (Devereux & Young 1991, 1992). The cirrus clouds in our Galaxy as well as the obvious presence of a general stellar radiation field have given rise to a two-component model to describe a galaxy's FIR radiation (e.g., Lonsdale Persson & Helou 1987; Buat & deHarving 1988) with proponents supporting the importance of one model over the other.

In a study of the luminosity ratio L_FIR/L_H, Sauvage & Thuan (1992) find a systematic decrease from early- to late-type spirals and propose that the cirrus fraction responsible for the FIR luminosity decreases from ~ 86 percent for Sa's to ~ 3 percent for Sdm's, a result which would require a large, systematic, type-related correction to the use of L_FIR as a measure of star formation. They call attention to an alternative explanation for the L_FIR/L_H type correlation: that the initial mass function (IMF) changes with type. They are reluctant to accept this possibility because of the proposal of a ``universal IMF'' (Scalo 1986), although there are many instances in our own Galaxy of differing IMF's (Gilmore & Roberts 1988). As noted in the discussion on H II regions, there is a strong dependence of number and luminosity of H II regions with galaxy type. This increase in both number and luminosity of H II regions with later type implies a type dependence of the IMF (Kennicutt 1988, 1989; Kennicutt et al. 1989) similiar to that suggested above.

The strong correlation of FIR and radio radiation for spirals is difficult to explain in a model that does not invoke massive stars as a heating mechanism (Xu 1990; but see Devereux & Eales 1989). This interpretation does not appear to hold for normal elliptical galaxies where no FIR-radio correlation is found (Bregman et al. 1992). Though visible patches of dust are frequently found in elliptical galaxies, here again this is no correlation with FIR radiation, and the evaluation of the amount of dust in elliptical galaxies from their (weak) FIR luminosity is at best uncertain.

The type dependence of the 1. FIR detection rate, 2. FIR luminosity, and 3. the luminosity ratio FIR/Radio (Fabbiano et al. 1988, Condon et al. 1991) is most impressive in separating ellipticals from spirals, with S0's somewhat intermediate. As illustrated in Figure 3b, the distinction in these properties within the spiral classes Sa-Sc, if present at all, is only slight (Bothun et al. 1989; Sauvage & Thuan 1993).