|Annu. Rev. Astron. Astrophys. 1996. 34:
Copyright © 1996 by . All rights reserved
7.2. Transport Out of the Disk
Evidence for the existence of galactic winds, with speeds U of hundreds of kilometers per second, is seen in some galactic halos, notably NGC 4631 and M82 (Section 3.6), implying turbulent magnetic Reynolds numbers RM = UL / t of order 100. Strong field freezing will thus occur and, since the wind advection time (L / U) is much shorter than the dynamo growth time, the wind will markedly affect the near-disk fields. For halo magnetic fields that are strong enough for their energy density to be comparable with the kinetic energy density of the wind, the dynamical effect of the field on the wind needs also to be considered, as in the analogous stellar wind problem, although such studies are in their infancy (see, e.g. Breitschwerdt et al 1993). With typical values of ~ 1 µG and ~ 10-27 g cm-3, a kinematic treatment will be valid for winds of speed in excess of about 100 km s-1. This outward advection of magnetic field may be partially offset near the disk by turbulent diamagnetism, which gives an effective velocity of field transport of a few kilometers per second towards the disk (if the diffusivity increases outwards), but for the larger wind velocities wind advection will dominate.
These problems were addressed in detail in the weak field approximation by Brandenburg et al (1993) and, with a rather different emphasis, by Elstner et al (1995). Brandenburg et al demonstrated that winds of plausible strength and geometry could drag out poloidal field lines almost radially into the halo and also move toroidal flux away from the disk. Moreover, by using realistic disk rotation curves for well-observed systems, and choosing appropriate (predominantly radial) wind velocity fields, solutions resembling the rather different halo fields of NGC 891 and NGC 4631, for example, can be generated without any careful "tuning." However, the halo field strengths are somewhat too low, and the field far from the disk makes too small an angle with the disk plane to provide a completely satisfactory model for NGC 4631.
However, a simple wind structure that is axisymmetric and varies smoothly with spherical polar angle may be inadequate; real galactic winds probably have considerably more structure, with streamers causing both azimuthal and latitudinal shear. Elstner et al (1995) presented a preliminary axisymmetric model (without azimuthal shear), with a wind velocity perpendicular to the disk and varying sinusoidally with distance from the rotation axis. They show that a short wavelength modulation (1.5 kpc) can markedly affect the field geometry and that odd parity "dipolar" fields may even be stable for some parameter values. Further work with a more realistic model is needed to elucidate the relation between such calculations and real galactic flows.
A priori, a quasi-radial or z-wise shearing flow is unlikely to produce a halo field that is predominantly parallel to the disk plane, although such fields are observed in some "edge-on" galaxies (e.g. NGC 253). A problem concerning mechanisms that advect field from the disk is that the gas dragging it into the halo belongs to the rarefied, hot phase of the ISM, where the field strength is typically about 0.1 µG (Kahn & Brett 1993), and so additional amplification outside the disk is necessary. Shearing by localized outflows can only amplify the vertical component. Brandenburg et al (1995b) pointed out that galactic fountain flows, especially in active starburst galaxies, may have the correct topology (upflows connected in horizontal cross section and isolated downdrafts) for a topological pumping mechanism to produce a strong mean horizontal field high in the halo. With realistic parameters, they showed that this mechanism might produce horizontal fields at a height of several kpc above the disk that were of comparable strength to those in the disk. As yet, this mechanism has not been included in a global dynamo calculation. Magnetic buoyancy in the disk may also play a role in moving field into the halo, but this mechanism has not yet been adequately quantified.
In general, an outflow that is symmetric both azimuthally and with respect to the disk plane will preserve in the halo any global parity or symmetry properties of the disk field. Clearly, if the outflow lacks such symmetries (as seems quite possible, a priori), then this connection between disk and advected halo fields will be lost.