GALAXIES, RADIO EMISSION ULRICH KLEIN HISTORY Almost six decades have passed since the first detection of cosmic radio waves by Karl Jansky in the early 193Os, the origin of which he first associated with the Galactic center (1933) and later with the plane of the Milky Way (1935). The first attempt to detect radio waves of the nearest external spiral galaxy similar to our own (M31 in the constellation of Andromeda) was reported by Grote Reber in 1944, but the first undoubted detection of M31 was that by Robert Hanbury Brown and Cyril Hazard in 1951. The limitation to observations at low frequencies at that time, however, did not allow detailed investigation of the distribution of radio emission, even in the closest spiral galaxies. This was partly due to the limited sensitivity of telescopes and receiver equipment, but was also due, in particular, to the low angular resolution, which as a rule of thumb is given by ***************************** where D is the telescope aperture and * is the wavelength (measured in the same units as D and related to the frequency * via the speed of light c, by *=c/v). In the case of Reber's observations, the beam width was approximately 12ø at wavelength 1.9 m, whereas the (optical) angular extent of M31 is about 2ø.5xOø.7. Hence, although the early radio surveys of the Milky Way galaxy already had provided some detailed information about the (projected) distribution in our home galaxy, such studies of external galaxies had to await modern high-resolution aperture synthesis and single-dish telescopes equipped with high-frequency receivers with good sensitivity and stability. Detailed measurements of galaxies other than those in the Local Group become feasible starting in the early 197Os with the operation of the Westerbork Synthesis Radio Telescope (WSRT) in the Netherlands, the Effelsberg 1OO-m telescope in western Germany, and the Very Large Array (VLA) in the United States (New Mexico). Measurements of global properties, such as flux densities and the positions of radio sources associated with galaxies had been conducted earlier with telescopes of various kinds, mainly interferometers. EMISSION PROCESSES THERMAL RADIATION Galaxies are continuously forming stars from the available gas reservoir. A small fraction of these newly born stars consists of stars with large masses (* 8 solar masses) and correspondingly high surface temperatures, so that the bulk of their emission is at wavelengths shorter than 912 *. This energetic ultraviolet radiation is capable of ionizing the surrounding gas, forming so-called HII regions in which free (i.e., unbound) protons and electrons are present. The electrons move on hyperbolic paths between the protons and the corresponding accelerations give rise to the emission of radiation (called thermal free-free radiation, or bremsstrahlung). According to the typical velocities involved, this emission is observed in the centimeter and decimeter wavelength range and its intensity ** decreases only slightly as a function of frequency v, **************************** This thermal emission (and the HII regions) is seen predominantly in the spiral arms of galaxies in which the current star formation rate is highest. NONTHERMAL RADIATION A massive star ends its life as a star (i.e., the hydrogen-burning phase) in the form of a supernova, which releases considerable energy (=10** ergs) into the interstellar medium, along with radiation, fast particles (e.g., protons and electrons), heavier nuclei, and probably also magnetic fields. The fast particles may eventually assume relativistic speeds (i.e., v=c) by continuous (re)acceleration in the interstellar medium, which is filled to a fairly large degree by supernova "bubbles," produced by the huge energy releases of these explosions. When these fast particles encounter a magnetic field, they are forced to follow helical paths around it so that they are continuously accelerated, giving rise to so-called synchrotron radiation, which also shows up in the radio spectral region. This radiation mechanism is substantially different from that of thermal radiation, in that it is caused (mainly) by relativistic electrons and its intensity decreases rapidly as a function of frequency, **************************** where B* is the component of the magnetic field perpendicular to the line of sight and **** is the spectral index. (The subscript "nth" denotes nonthermal radiation.) Furthermore, this emission is partly polarized, that is, the emitted waves are oscillating in a preferential direction that is perpendicular to the magnetic field that gives rise to this radiation. However, the polarization direction of the received wave is, in general, different from that of the wave when it left the source, due to the effect of Faraday rotation in which the wave plane is rotated while penetrating a magnetized plasma. The amount of Faraday rotation is wavelength-dependent and is given by ****************************** where * is the wavelength (in meters) and RM is the so-called rotation measure, given by **************************************, where n* is the thermal electron density of the medium (in cm-*), B** is the magnetic field component parallel to the line of sight, and the integral is taken over the whole line of sight L (in parsecs). We therefore have a powerful tool to detect magnetic fields in galaxies! Karl O. Kiepenheuer (1950) was the first to consider the process of synchrotron radiation as the predominant mechanism producing the observed low-frequency emission of the Milky Way. In general, the radio continuum emission in a normal spiral galaxy is a mixture of the two components described here (thermal and nonthermal), and it takes observations made at two or more frequencies to disentangle them, either globally or, if the quality of the radio maps is sufficient, even within different locations in the galaxy. A TYPICAL GALAXY IN THE RADIO WINDOW FACE-ON VIEW Figure 1 presents a typical example of what a spiral galaxy looks like in the radio window when seen face-on. Shown here are isointensity contours of the radio emission of the spiral galaxy NGC 6946 taken at *=2.8 cm (*=10.7 GHz) with the Effelsberg 100-m telescope and superimposed onto an optical photograph of the galaxy. Because the observations were taken at a fairly high radio frequency, thermal emission shows up as enhanced radiation in some locations that are coincident with giant HII complexes, whereas the underlying synchrotron emission decreases more or less smoothly from the center of the galaxy toward the outer (optical) edge. This decrease is best described by an exponential law of the observed intensity ** at frequency *, *****************************, where R is the galactocentric radius and R* is the scale length of the distribution. In general, R* is always larger than the scale length of the visible light. If mapped with high-resolution aperture synthesis instruments (e.g., WSRT and VLA), which emphasize the small-scale structure, such a galaxy exhibits enhanced emission in the spiral arms, where the young massive stars are located. However, the distribution of the synchrotron emission is always found to be smoother than that of the thermal emission, probably reflecting the diffusive propagation of the relativistic particles away from their sites of origin. Therefore, spiral galaxies in general are more extended in the radio continuum than at optical wavelengths. If we map the distribution of the spectral index ********************************** between any two frequencies **, ** across the galaxy, we can investigate the varying relative proportion of thermal and nonthermal emission as a function of location in the galaxy because we are looking everywhere at the sum of a thermal flux with a "flat" spectrum and a nonthermal flux with a "steep" spectrum, that is, ******************************, where S**(r) is the thermal contribution at a certain frequency at location r in the galaxy and S***(r) is the corresponding nonthermal flux. This tells us where the sites of recent formation of massive stars are and how the cosmic rays propagate away from these regions, assuming that they, too, are produced by the aftermaths of stellar death. Flat spectra (*=0.1,...,0.5), which indicate a high contribution of thermal emission, are found coincident with HII regions and central starburst regions of galaxies, whereas steeper spectra (*=0.5,...,0.9) are seen away from these, especially between the spiral arms and at the outer peripheries of galaxies. Observations of the linearly polarized component of the emission allows to trace the magnetic field in galaxies. If measurements are made at several wavelengths, these can be corrected for Faraday rotation, thus recovering the intrinsic field direction. After initial attempts, (commencing in the mid-197Os) to apply this technique to the spiral galaxies M51,M81,and M31, using the WSRT this kind of study was resumed with the 100-m telescope (in the early 1980s) and later with the VLA. This field of research is one of the most important in the investigation of galaxies, given the significance and impact of magnetic fields on various physical processes in the interstellar medium (and probably the intergalactic medium, too!). The large-scale structure of magnetic fields in spiral galaxies is known for about a dozen nearby objects, and two substantially different field configurations appear to prevail: an axisymmetric field, in which the field lines run inward everywhere in the galaxy's disk (note that these field lines must be closed above and below the galaxy's plane to avoid magnetic monopoles!), and a bisymmetric field, in which the field enters the disk on one side and leaves it on the opposite side, probably closing in intergalactic space. It is noteworthy that these two scenarios have been postulated by theoreticians modeling the galactic dynamo and that the researchers studying the dynamo mechanism in disk galaxies have become increasingly interested in the prolific observations of polarization in galaxies. Magnetic field strengths can be estimated by assuming energy equipartition between the magnetic field and the cosmic rays in a galaxy (which are coupled). Typical field strengths obtained are in the range of a few microgauss (**) up to 10 **, with higher values only encountered in the dense and energetic central regions of starburst galaxies like M82. EDGE-ON VIEW If we look at a spiral galaxy edge-on in the radio window, we can study the distribution of radio emission away from the galaxy's plane, in the so-called z direction. Figure 2 presents a typical example, the radio continuum map of the spiral galaxy NGC 4631, obtained with the VLA at *=20 cm and superimposed onto an optical picture. One striking difference between radio and optical emissions is immediately obvious: The radio emission extends much further from the plane than does the optical. If the same galaxy is investigated at a higher radio frequency, this "halo" of radio emission is found to have a much smaller extent in z. This behavior is explicable in terms of the processes of propagation and energy loss of relativistic particles during their lifetime. What we see here is the superposition of a thin radio disk with predominantly thermal emission and a thicker one with purely nonthermal emission. The relativistic particles lose energy via synchrotron radiation due to their interaction with the magnetic field that is also present outside the (optically visible) disk of the galaxy. They are evidenced by the presence of polarized radio emission at large * distances and are marked in Fig. 1 by the "vectors," which represent the direction of the electric field of the received radiation. This would be perpendicular to the magnetic field direction if there were no Faraday rotation. The latter can be assumed to be negligible outside the disk, so that in NGC 4631 the magnetic field seems to be oriented primarily away from the disk at larger z [a few kiloparsecs (kpc)]. More detailed measurements are needed to show what happens to the field at the interface between the disk and the halo. The existence of extended radio halos such as the one visible in fig.2 was postulated more than 30 years ago (e.g., by Iosif S. Sklovskij in 1952), yet it has been proven only for two or three nearby edge-on galaxies. Most galaxies possess what is usually referred to as a thick radio disk, with z-extents of a few kiloparsecs, whereas in NGC 4631 radio emission can be traced as far out as z=7.5 kpc. The spectrum of the radio emission steepens as a function of z, reflecting the energy and possibly the adiabatic losses of the relativistic electrons as they escape from the disk (note that there is hardly any reacceleration of such particles once they have left the disk and its violent interstellar medium). DWARf IRREGULAR GALAXIES Dwarf irregular galaxies, such as the Magellanic Clouds, exhibit a rather chaotic appearance at optical wavelengths. A similar structure is found at radio wavelengths, most likely due to the patchy distribution of star-forming complexes. These galaxies obviously lack the exponential nonthermal disk that governs the radio images of massive spiral galaxies. Furthermore, the synchrotron emission from such stellar systems is rather weak, which is one reason for the lack of detailed studies of its distribution in dwarf galaxies. The overall scenario that is presently favored to explain the radio properties of dwarf galaxies is that most of them are forming stars at lower rates than massive galaxies and that the containment of cosmic rays in them is also significantly reduced as compared to normal spiral galaxies. This could be due to a substantially different magnetic field configuration, but that possibility will have to be investigated by future studies (with superior sensitivity) of polarized radio emission in dwarf galaxies. This picture is supported by the properties of a special class of dwarf galaxies that are vigorously forming stars at present, the so-called blue compact dwarf galaxies. These galaxies are undergoing bursts of star formation, yet their nonthermal radio emission is comparatively low. If in these galaxies, too, cosmic rays are produced in supernova events, then the deficiency of synchrotron emission can be explained only in terms of different properties of cosmic ray propagation and magnetic field. GLOBAL RADIO PROPERTIES OF GALAXIES Normal spiral galaxies (as opposed to radio galaxies) emit monochromatic radio luminosities at 5 GHz of up to approximately 10** W Hz**, whereas for dwarf galaxies at the same frequency the lowest luminosities are of the order of approximately 5x 10** W Hs**. This emission accounts for only a small fraction of the total energy output of a galaxy, which is generally dominated by the far-infrared emission (i.e., thermal radiation from dust). The total radio emission shows a good correlation with that radiated in other spectral domains, such as blue light or x-ray, but the tightest correlation by far is that with the far-infrared emission from dust. The radio emission is of predominantly nonthermal origin down to even shorter (centimeter) wavelengths, whereas in the far-infrared we measure the reradiation of the energetic Lyman continuum photons that are produced by massive OB stars and that are efficiently absorbed by dust grains and transformed into longer-wavelength radiation. Because of the obviously entirely different and independent radiation processes (including their origins) this close correlation is still awaiting a thorough and comprehensive theoretical interpretation. The spectral indices of spiral galaxies, measured between 408 MHz and 10.7 GHz, show a remarkably small scatter around a mean value of <*>= 0.75. After separating the thermal content of the radiation, the mean value becomes somewhat higher, <****>=-0.88, representing the pure synchrotron spectrum of a typical galaxy. The dispersion of this quantity is not very large either, clearly showing that the same overall and unique process is controlling cosmic ray acceleration and/or propagation in all massive spiral galaxies. The frequency spectrum of synchrotron radiation in galaxies is connected with the energy distribution N(E) of the relativistic particles, **************************, where A is a constant and y is the so-called energy spectral index, both of which can be measured directly near the Earth. The latter is related to the frequency spectral index **** via ***************************** The value of y measured in the Earth's immediate vicinity is y=2.6, in remarkably good agreement with ***=0.88, the global value obtained for a sample of normal spiral galaxies. Dwarf galaxies appear to have global radio properties that are markedly different from those of normal spirals. Only a few of them have been thoroughly investigated, and their radio spectra hint at rather steep synchrotron spectra and a much smaller proportion of synchrotron emission relative to thermal emission. This probably reflects the lack of confinement of cosmic rays in dwarf galaxies, as described above. CURRENT AND FUTURE INVESTIGATIONS There are a great many open questions (partially addressed here) that require scrutiny, both observationally and by theoretical modeling. The following questions are intriguing issues: Why do some galaxies possess extended radio coronae and others don't? What gives rise to axisymmetric magnetic field configurations in some spiral galaxies and bisymmetric fields in others? How is this related to galaxian dynamics? What determines the confinement of cosmic rays (which appears to be so different in massive and dwarf galaxies)? Which parameters control the conspicuously tight correlation of nonthermal radio emission and thermal far-infrared emission? Related to this, where and how are relativistic particles produced, accelerated, and reaccelerated, and how do they propagate? (This issue, though discussed in many papers, is far from being thoroughly understood.) The key to the resolution of these questions lies in the development of radio astronomical techniques to facilitate improved observations, in the intercomparison of data obtained in various spectral domains, and in progress in the theories of magnetic fleld production and maintenance and of cosmic ray production and propagation (possibly facilitated by magnetohydrodynamical modeling of this constituent of galaxies). Additional Reading Beck, R.(1986). Interstellar magnetic fields. IEEE Trans. Plasma Science PS-14 740. Hummel, E., Lesch, H., Wielebinski, R., and Schlickeiser, R.(1988). The radio halo of NGC 4631: Ordered magnetic fields far above the plane. Astron. Ap. 198 L29. klein, U., Beck, R., Buczilowski, U.R., and Wielebinski, R.(1982). A survey of the distribution of 2.8 cm radio continuum in nearby galaxies. Astron. Ap. 108 176. Sofue, Y., Fujimoto, M., and Wielebinski, R.(1986). Global structure of magnetic fields in spiral galaxies. Ann. Rev. Astron. Ap. 24 459. van der Kruit, P.C. and Allen, R.J.(1976). The radio continuum morphology of spiral galaxies. Ann. Rev. Astron. Ap. 14 417. Verschuur, G.L. and Kellermann, K.I.(1988). Galactic and Extragalactic Radio Astronomy. Springer-Verlag, Berlin.