![]() |
Annu. Rev. Astron. Astrophys. 1992. 30:
575-611 Copyright © 1992 by Annual Reviews. All rights reserved |
If the thermal and nonthermal radio sources in a normal galaxy are
coextensive, the total radio emissivity is everywhere proportional to the
free-free absorption coefficient . The radio
brightness-temperature spectrum of any such galaxy then follows from
Equation 5; it is approximately
where is the average free-free optical depth
along the line of sight and Te ~ 104 K is
the electron temperature. A set of
curves outlining the possible radio spectra of normal galaxies is shown
in Figure
4. Inhomogeneities in real sources will produce spectral peaks
somewhat broader than those plotted. In the limit [1 - exp(-
)] -> 1 Equation 6 predicts that
the maximum brightness temperatures of normal galaxies are
Te
105 K for
1 GHz, as indicated by the upper-left envelope
of the spectra in Figure 4. The
spectrum of our galaxy at high latitudes
turns over due to free-free absorption (
~ 1)
near
= 0.003 GHz, and the observed brightness
spectrum (Cane 1979) lies just
above the lowest curve in Figure
4. The median face-on disk surface
brightness of nearby spiral galaxies is < Tb > = 0.75
± 0.25 K at
= 1.4 GHz (Hummel 1981), so
the brightness spectra of most normal galaxies fall between the two
lowest curves (corresponding to
= 1 between
= 0.003 and 0.01 GHz) in Figure 4. Brighter
radio sources have higher turnover frequencies. For example,
the central component in the starburst galaxy M82 has a brightness
S /
~ 30 mJy arcsec-2 at
~ 1 GHz, the frequency
at which
~ 1. The integrated spectrum of M82
(Figure 1) flattens but does not
actually turn over below 1 GHz because
this source is quite inhomogeneous and synchrotron emission from the
surrounding halo (Seaquist & Odegard
1991) does not suffer significant free-free absorption.
Figure 4. Radio brightness spectra
(continuous curves) of normal galaxies containing mixed
thermal/nonthermal sources. Successive curves correspond to increasing
the frequency
1 at which the
free-free optical depth is
= 1:
1 = 0.003, 0.01, 0.03, 0.1, 0.3, 1, and
3 GHz. Dotted lines of constant brightness temperature
Tb (K) are alson shown. The brightness temperatures of
normal galaxies do not exceed Tb ~ 105 K at
frequencies above
~ 1 GHz. Abscissa:
frequency (GHz). Ordinate: brightness (mJy arcsec-2).
The low-frequency spectra of many normal spiral galaxies with only
moderate surface brightness (Israel &
Mahoney 1990), including our own
(Sironi 1974, Webster 1974), appear to flatten
slightly in the range 0.1-1
GHz, much higher than the
0.01 GHz predicted by
Equation 6 if Te ~ 104 K. Israel & Mahoney (1990) found that the
difference between flux densities extrapolated from higher frequencies
and their measured 57.5 MHz flux densities is greatest for highly
inclined disks, so they proposed free-free absorption by very cool
(Te < 1000 K) ionized gas filling a fairly large
fraction of the radio emitting volume to explain their low-frequency
spectra. Hummel (1991a)
reanalyzed the radio spectra, confirming a
spectral break with median <
> ~ 0.25 but questioning its
inclination dependence. His interpretation is that there is little absorption
at low frequencies, but the high-frequency nonthermal spectra steepen
owing to the effects of propagation (convection or energy-dependent
diffusion) on the relativistic electrons. However, Webster (1970) showed
that the spectral steepening from this process is too gradual to explain the
fairly sharp bend in the high-latitude spectrum of our own galaxy. Thus, the
low-frequency spectral break in typical spiral galaxies is still not
completely understood.
The brightest known radio sources in galaxies claimed to be ``normal''
are the compact starburst nuclei of ultraluminous IRAS sources. They
have 8.44 GHz brightness temperatures approaching 104 K and show
spectral flattening above 1.49 GHz (Condon et al. 1991c, Sopp & Alexander
1991). For example, the strong compact source in IC 694 has a flat
( ~ 0.3) spectrum above 1.49 GHz, leading
Gehrz et al. (1983) to conclude
that it was synchrotron self-absorbed
and hence much too bright (implied Tb >
1010 K) to be a starburst. However, an 8.44
GHz VLA map made with 0".25 resolution (Figure 5) shows that this
source is extended and its peak brightness temperature is only
Tb 104 K. Therefore, a flat spectrum even
at GHz frequencies is not sufficient to prove that a radio source is
powered by an AGN. Flat-spectrum sources produced by monsters in Seyfert
galaxies, classical radio galaxies quasars, etc. generally are synchrotron
self-absorbed. Thus, high-resolution maps capable of separating sources
with Tb
105 K from those with Tb >
1010 K (cf Norris et
al. 1990) should be able to distinguish
between flat-spectrum starbursts and monsters. But it is not enough to
show that Tb
1010 K at high frequencies (
~ 15 GHz) to reject starbursts (cf Carral et al. 1990, Chapman et al. 1990); starburst
galaxies much brighter than M82 do exist.
Figure 5. The nucleus of IC 694 is
typical of the brightest and most compact sources found in ``normal''
galaxies. Its 8.44 GHz brightness temperature is only ~ 104
K, so its relatively flat radio spectrum indicates free-free absorption
by thermal electrons with temperature Te ~
104 K, not synchrotron self-absorption by relativistic
electrons with kinetic temperatures Tr >
me c2 / k ~ 1010
K. The logarithmic contours are separated by factors of 21/2
in brightness, and the lowest contour is 0.1 mJy beam-1 ~ 28
K.