Indications of magnetic activity have been detected in a variety of late type stars in close binary systems, including (detached) RS CVn systems and contact systems of W UMa class. Although the solar field is predominantly axisymmetric, there is evidence even there for some nonaxisymmetric large-scale features (solar `sector structure', and see Jetsu et al. 1996, preprint), and there is much stronger evidence for nonaxisymmetry in some other single stars and RS CVn-type binaries, both from the analysis of long time series of photometric data (eg Jetsu et al. 1993, 1994; Jetsu 1996), and from photometric modelling (eg Bradstreet 1985; Zeilik et al. 1989, 1990a, 1990b; Henry et al. 1995). Snapshots of this kind of spot structure have been derived by surface imaging techniques (eg Piskunov, Tuominen and Vilhu 1990; Piskunov, Ryabchikova and Tuominen, preprint). There is also the possibility of using eclipses to image binaries (Vincent, Piskunov and Tuominen, 1993; Vincent et al., in preparation). The conventional view is that magnetic activity in late-type stars is the result of dynamo action in their deep convective envelopes, and this has been studied in single stars by a large number of authors. However, for some time large-scale nonaxisymmetric structures of the kinds indicated by the above-quoted and other papers seemed difficult to explain in the context of mean field dynamo theory. More recently a number of mechanisms have been identified that can result in the excitation of stable nonaxisymmetric fields, provided that the differential rotation is not too strong (Rädler et al 1990; Moss, Brandenburg and Tuominen 1991; Moss & Brandenburg 1993; Rüdiger & Elstner 1994; Moss et al 1995). If both stars of a close binary are dynamo-active, then it is easier to envisage the excitation of large-scale nonaxisymmetric fields - the geometry of the system is intrinsically nonaxisymmetric and tidal interactions can be expected to reduce severely the differential rotation and, indeed, to lock the spin and orbital frequencies.
Thus, in this paper we study dynamo action in two corotating spheres that may be separate, touch or partially overlap. Two touching or overlapping spheres provide only a poor geometric approximation to the Roche geometry of contact or over-contact binary systems, but we feel that the essentially nonaxisymmetric geometry is represented to an adequate first approximation. We recognize that dynamo action can be expected to be limited to a thick outer envelope, but for reasons of computational convenience and economy we assume the dynamo-active regions to occupy the entire volumes of the spheres. The two dynamo-active spheres are embedded in a `computational sphere', at the surface of which vacuum boundary conditions are applied. Within this computational sphere, the diffusion coefficient is assumed everywhere uniform, but the -coefficient is only non-zero within the dynamo-active region(s), thus implying that is a function of azimuthal coordinate measured about an axis parallel to the rotation axis (cf Moss et al 1991). We ignore the effects of any large-scale circulation: this approximation might be especially inappropriate for common envelope systems. Overall, the computational philosophy has some similarity to that of the `embedded disc' galaxy codes (Stepinsky & Levy 1988; Elstner, Meinel and Rüdiger 1990; Moss & Tuominen 1990). The code used is essentially that described in Moss et al (1991).
© European Southern Observatory (ESO) 1997
Online publication: June 30, 1998