|Annu. Rev. Astron. Astrophys. 1996. 34:
Copyright © 1996 by . All rights reserved
From observations of external galaxies, magnetic fields are inferred in halos of spiral galaxies to distances of at least 5 kpc and maybe even 10 kpc from the disk plane, significantly beyond a synchrotron scale height (cf Section 3.6). Recently, dynamo models have directed some attention to out-of-disk fields. Here we address the two logical possibilities (while noting that they are not mutually exclusive): that such fields are generated in situ in the halo or that they are generated in the disk and then transported into the halo.
7.1. In Situ Generation
Interpretations of observations in the Milky Way suggest the presence of turbulent velocities of at least 50 km s-1 in galactic halos, compared to estimates of 10 km s-1 in disks. If we assume a length scale of order 0.5 kpc and that halo angular velocities are comparable with those in the disk, we get canonical estimates of ~ 3 km s-1 and t ~ 5 × 1027 cm2 s-1, to be compared with t ~ 1026 cm2 s-1 in the disk. [See, e.g. the discussion in Poezd et al (1993). Note that Schultz et al (1994) adopt halo turbulent velocities that are much smaller than those in the disk: This may be a direct consequence of their turbulence model with <v2> / z.] Taking L ~ 10 kpc gives standard dynamo numbers R = L / t ~ 2 and R = 0 L2 / t ~ 200. These are large enough for a dynamo to be excited (Ruzmaikin et al 1988a, Section VIII.1; Kahn & Brett 1993). Note that such a dynamo would operate in a quasi-spherical volume, rather than a thin disk, that standard spherical dynamos preferentially excite fields of dipolar (A0) topology, and that these are then often the only stable solutions of the full nonlinear equations. In contrast, S0 fields are usually preferred in thin disks. This situation immediately suggests the interesting possibility of simultaneous excitation of dynamo fields of opposite parity types in the two subsystems (halo and disk) (see Sokoloff & Shukurov 1990). A priori, the possible existence of magnetic structures asymmetric with respect to the midplane, of neutral sheets, and of other nonstandard phenomena cannot be dismissed, as has been shown in some detail by Brandenburg et al (1992). Growth times in the halo are substantially longer than in the disk, and the halo field may still be in a transient state after a Hubble time. Detailed integrations show that, starting from a seed field of mixed parity, the overall field is initially dominated by S0 topology and concentrated in the disk. This phase can persist for order a Hubble time, but the final configuration is usually of A0 type, and may even be oscillatory. Given the long-lived transient phase with mixed parity fields present, observers today may be presented not with the eventual stable configuration, but rather an intermediate state of quite arbitrary geometry. Note that magnetic fields in the disk and halo of M51 are oppositely directed (EM Berkhuijsen et al, in preparation): This argues for in situ generation. More satisfactory halo models will need better data than is currently available on the dependence of the angular velocity in the halo on z, but these results seem qualitatively robust. To summarize, in some circumstances, dynamo theory may not be able to make detailed predictions about field geometries in specific galaxies.
A largely unexplored possibility is that some sort of Ponomarenko ("screw") dynamo (e.g. Ruzmaikin et al 1988c) might operate in the halo, if large-scale quasi-radial outflows ("winds") are twisted into helical form by the galactic rotation. Such dynamos excite nonaxisymmetric fields. If we take a simple model investigated by Ruzmaikin et al and use their definitions, then a wind velocity of 100 km s-1 and a typical galactic angular velocity gives a magnetic Reynolds number Rm large enough for the dynamo to work. Naively, the minimum e-folding time would be about 109 yr, but this increases as RM1/2 for larger Rm, because the screw dynamo is "slow." These estimates suggest that the mechanism might be of marginal importance in halos, but real galaxy velocity fields are likely to be less efficient dynamos than the idealized forms considered by Ruzmaikin et al. We note in passing that Spencer & Cram (1992) have discussed models of field amplification in which meridional flows ("winds") appear to play a central role. However, they solve the problem purely in the disk region; moreover, their solutions do not represent dynamo generation but rather local compression of field and hence the relevance to field generation processes in galaxies is unclear.