7.3. The Faraday Screen: Magnetic Fields in Cluster Gas

Slysh (1966) and Mitton (1971) first pointed out the `anomalous' behavior of the polarized emission from Cygnus A, including changing fractional polarization with observed wavelength (at low resolution), essentially random projected electric field vectors, and a large difference between the magnitude of Faraday rotation measure (RM) towards the two lobes. This problem was delineated in detail in the study of the rotation measure distribution towards Cygnus A by Alexander et al. (1984). They find that the `rotation measure values are very noisy, and no overall pattern can be discerned.' They suggest a Galactic origin for the rotation measures, based simply on the fact that Cygnus A has a fairly low Galactic latitude (b = 5.8°).

The question of the large anomalous Faraday rotation towards Cygnus A was resolved by the sensitive, multi-frequency, spatially resolving polarization observations of Dreher et al. (1987b). These authors find that Cygnus A lies behind a deep `Faraday screen', with rotation measures varying from -4000 rad m^-2 to +3000 rad m^-2 across the source. Gradients in RM exceed 300 rad m^-2 arcsec^-1. In general the distribution is not random, but displays structure with coherence over spatial scales of order 10 kpc. Perhaps most importantly, Dreher et al. (1987b) find that the total rotation of the position angle of the polarization vector exceeds 600° in many regions, without departure from a ² dependence, where = observed wavelength, and without depolarization demonstrating conclusively that the Faraday rotation cannot be internal to the source but must be by an external screen (Burn 1966). This conclusion has been further strengthened by the recent 8 GHz results of Perley and Carilli (1996). The 8 GHz data set severe limits on deviations from a ² dependence of position angle throughout the observed wavelength range.

Dreher et al. (1987b) consider, and reject, a Galactic origin for the Faraday screen. They propose that the origin of the large RM's is in the hot intracluster gas in which Cygnus A is embedded. The implication is that the large-scale cluster gas is substantially magnetized. Using the cluster gas density radial profile derived from X-ray observations, Dreher et al. calculate cluster magnetic field strengths between 2 µG and 10 µG, depending on geometry. Also, from the lack of wavelength dependent fractional polarization at fixed resolution, Dreher et al. derive an upper limit to the thermal electron density in the radio lobes of 2 x 10^-4 cm^-3. This density limit depends on the minimum energy assumption for the radio source fields and assumes a uni-directional field through the lobes. Any field reversals along the line of sight would lead to a less stringent electron density limit.

An alternative model for the large rotation measures towards Cygnus A has been proposed by Bicknell et al. (1990), in which a thin mixing-layer exists along the contact discontinuity, where shocked cluster gas mixes with the large fields in the radio source. Support for this idea comes from the `striated' appearance of the RM distribution across the radio lobes, suggestive of large-scale K-H instabilities at the contact discontinuity. However, since the discovery of extreme rotation measures towards Cygnus A spatially resolving observations of many radio galaxies at the centers of dense, X-ray emitting cluster atmospheres have revealed large rotation measures (for a summary, see Taylor et al. 1994). Taylor et al. show a clear correlation between the cluster core thermal density and the magnitude of RM's observed to cluster center radio sources. This includes both luminous edge-brightened (FRII) sources, such as 3C 295 (Perley and Taylor 1991), and less luminous edge-darkened (FRI) sources, such as M 87 (Owen 1989). iven that the physical interaction between the source and its environment is thought to be very different for the two types of sources (DeYoung 1993), we conclude that the majority of the large RM's observed must be the result of substantially magnetized, large scale cluster gas, and not simply the result of the interaction between the source with its environments.

The discovery of the bow shock in the RM distribution in the northern hotspot region in Cygnus A has relevance to this debate. Carilli et al. (1988) show that the observed RM change at the bow shock implies pre-shock intracluster fields of 8 µG. They propose a simple model in which the large-scale RM distribution towards Cygnus A (amplitudes up to 4000 rad m^-2, typical scale-sizes 10 kpc) is caused by the unperturbed cluster atmosphere, while small scale fluctuations (amplitude 1000 rad m^-2, scale-sizes 5 kpc) can result from the interaction of the source and the ambient medium.

In summary, the detection of extreme rotation measures towards Cygnus A, and subsequently in other cluster-center radio galaxies, shows that the thermal cluster atmospheres must be substantially magnetized, with fields a few µG. In most cases the pressure in the intracluster fields is below the thermal energy density (e.g. the plasma -factor = ratio of thermal to magnetic pressure is between 20 and 50 for the Cygnus A cluster), implying a minor dynamical role for the fields. Even dynamically unimportant fields alter substantially the thermal conductivity of the plasma, and hence are very important when considering the cooling and heating of the ICM (Sarazin 1986, 1988). Lastly, the detailed study of the Faraday screen towards Cygnus A provides vital supporting evidence for physical models explaining the depolarization asymmetries towards extragalactic radio sources and their implication for quasar-powerful radio galaxy unification schemes (Laing 1988, Garrington et al. 1988, 1991).

Models for the origin of intracluster fields include: the dynamo action of turbulent wakes of galaxies, injection of fields by previous outbursts of the radio source, ram pressure stripping of fields from galaxies, and amplification of a primordial field (Jaffe 1980, Ruzmaikin et al. 1989, Jafelice and Opher 1992; see Sarazin 1986, 1988 for reviews). Another possible mechanism for generating large scale cluster fields is amplification of seed fields during the merger of two clusters. Eilek (1995) presents a detailed `Zeldovich rope dynamo' model for field generation in a cluster with net-helical turbulence, presumably driven by galaxy motions. She finds that the fields eventually evolve to equipartition strengths. She predicts RM distributions towards cluster center sources which compare well with observations, both in amplitude and structure.