Annu. Rev. Astron. Astrophys. 1992. 30: 575-611 Copyright © 1992 by Annual Reviews. All rights reserved

1. INTRODUCTION

If the radio energy source in a galaxy is not a supermassive black hole or related nuclear ``monster'', then that galaxy is called ``normal'' in this review, regardless of its optical morphology. Nearly all of the radio emission from normal galaxies is synchrotron radiation from relativistic electrons and free-free emission from HII regions. Thermal reradiation of starlight by dust quickly overwhelms these components above ~ 200 GHz ( ~ 1.5 mm), defining a practical upper bound to the frequencies of ``radio'' observations. Typical relative intensities of synchrotron radiation, free-free emission, and dust reradiation are shown in the radio/far-infrared (FIR) spectrum of M82 (Figure 1). The radio continuum from a normal galaxy is clearly just a tracer, accounting for < 10^-4 of its bolometric luminosity.

Figure 1. The observed radio/FIR spectrum of M82 (Klein et al. 1988, Carlstrom & Kronberg 1991) is the sum (solid line) of synchrotron (dot-dash line), free-free (dashed line), and dust (dotted line) components. The H II regions in this bright starburst galaxy start to become opaque below ~ 1 GHz, reducing both the free-free and synchrotron flux densities. The free-free component is largest only in the poorly observed frequency range 30-200 GHz. Thermal reradiation from T ~ 45 K dust with opacity proportional to ^1.5 swamps the radio emission at higher frequencies. Lower abscissa: frequency (GHz). Upper abscissa: wavelength (cm). Ordinate: flux density (Jy).

Only stars more massive than M ~ 8 M produce the Type II and Type Ib supernovae whose remnants (SNRs) are thought to accelerate most of the relativistic electrons in normal galaxies (see Section 4.3), and these massive stars ionize the H II regions as well. Such massive stars live 3 x 10 ⁷ yr, and the relativistic electrons probably have lifetimes 10⁸ yr. Radio observations are therefore probes of very recent star-formation activity in normal galaxies, and they are especially valuable for three reasons: (a) The radio emission from normal galaxies is not overwhelmed by stellar populations older than about 10 ⁸ yr. (b) Radio maps can be made with subarcsecond position accuracy and resolution, unambiguously identifying the most luminous star-forming regions within galaxies and resolving even the most compact ones. (c) Only at radio and FIR wavelengths are the most intense ``starbursts'' transparent, so that observed flux densities are accurately proportional to intrinsic luminosities.

On the other hand, existing radio data alone are poor constraints for quantitative models of star formation. The free-free emission emerges directly from H II regions containing the ionizing stars and its intensity is proportional to the production rate of Lyman continuum photons, but isolating the free-free component and measuring its flux density is difficult observationally because the flat-spectrum ( ~ +0.1, where the spectral index is defined by S ^-, S being the flux density and the frequency) free-free emission is usually weaker than the steep-spectrum ( ~ +0.8) synchrotron emission below ~ 30 GHz (Figure 1). Most of the easily observed synchrotron radiation in a typical galaxy arises from fairly old ( 10 ⁷ yr) relativistic electrons that have propagated significant distances ( 1 kpc) from their short-lived ( 10 ⁵ yr) and now defunct parent SNRs. Consequently the original sources of the relativistic electrons have disappeared, and their detailed spatial distribution has been smoothed beyond recognition. Finally, the steps between star formation and synchrotron emission (supernova explosion, acceleration of relativistic electrons in the SNR, propagation of cosmic rays throughout the galaxy, energy loss, and escape) are poorly understood, impeding quantitative interpretation of the observed synchrotron spectra and brightness distributions.

Fortunately, the remarkably tight and ubiquitous correlation found between the global FIR and (predominantly nonthermal) radio luminosities of normal galaxies (see Figure 8 later) is emerging as an almost miraculous constraint for models relating radio emission to massive star formation. Except perhaps for galaxies with very low star-formation rates, the FIR luminosity appears to be a good measure of the bolometric luminosity produced by fairly massive (M 5 M ) young stars. The FIR / radio correlation suggests that a one-parameter model specifying the FIR and radio luminosities in terms of the recent star-formation rate can describe most normal galaxies. Secondary parameters apparently do not vary significantly from galaxy to galaxy or are irrelevant. For example, the magnetic energy densities in galaxies obeying the FIR / radio correlation span four orders of magnitude, so the total synchrotron energy produced per SNR must be almost independent of magnetic field strength.

This review emphasizes those aspects of radio emission and recent star formation that appear to be shared by most normal galaxies and their application to a simple model consistent with, if not uniquely required by, the data. The range of source properties found in the population of normal galaxies is outlined in Section 2. The characteristics of free-free emission and synchrotron radiation that are needed to interpret observations of normal galaxies are outlined in Sections 3 and 4. Frequently used equations are presented in astronomically convenient units, and their consequences for real observations are stressed. The textbooks by Pacholczyk (1970) and Longair (1981) are recommended for broader coverage of these topics. Section 5 is devoted to the FIR / radio correlation. Population-synthesis models relating the FIR / radio emission to recent star formation are covered in Section 6. Section 7 summarizes the current situation and future hopes for understanding the radio emission from normal galaxies.