References: a: Keenan & Boeshaar (1980), Keenan & McNeil (1989); b: Jorissen et al. (1998); c: Van Eck et al. (1998); d: Jorissen et al. (1996); e: Flux densities at 160 nm from Johnson et al. (1993) combined with distances from Van Eck et al. (1998) or Eggen (1972); f: Johnson et al. (1993), Ake (1997); g: Ake et al. (1991); h: Brown et al. (1990), Shcherbakov & Tuominen (1992).
Remarks: x: 1(-3) stands for ; y: HD 49368 = V613 Mon; z: HIPPARCOS parallax very uncertain; for HD 35155, distance from Eggen (1972) adopted instead.
Binary S stars share many properties with symbiotic stars, since both families consist of a cool red giant and a WD companion in systems with orbital periods of a few hundred to a few thousand days (Jorissen 1997). Some kind of symbiotic activity should thus be expected among binary S stars as well. Table 4 lists those systems where the usual signatures of binary interaction have been probed. In that table, refers to the luminosity in the ROSAT hard band (0.5 - 2.4 keV) taken from Jorissen et al. (1996), and adapted to the new HIPPARCOS distances (Van Eck et al. 1998). Hard X-rays have been observed in HR 363 and HD 35155, and appear strongly variable. In the UV domain, several stars exhibit strong emission lines of highly ionized species typical of interacting binary systems (like CIV nm). In the column labelled "UV em. lines'', "IB'' stands for "interacting binary'', "no'' stands for no emission lines seen, and "WD'' indicates that the UV spectrum fits that of a clean WD. The continuum UV luminosity in the 125.0 - 195.0 nm band is often larger than would be expected from an isolated WD [see column "'']. In Table 4, , where the average flux density in the 125.0 - 195.0 nm domain is taken from Johnson et al. (1993) and the distance d from Van Eck et al. (1998) or Eggen (1972). The UV luminosity is often strongly variable, in which case the different observed values are listed. The HeI 1083.0 nm triplet generally confirms the UV diagnostics. The three binary S stars flagged as "interacting binaries" from their UV features are also those showing strong and variable HeI nm lines, whereas the two stars (HR 363 and HD 191226) with no UV emission lines exhibit only a weak HeI triplet in emission.
Two important conclusions may be drawn from the observations
summarized in Table 4:
(i) the level of binary interaction does not appear to be correlated with the orbital period. HD 49368 (=V613 Mon) for instance exhibits a much higher level of activity than the shorter-period system HD 191226, whereas the shortest-period system HD 121447 does not show any sign of interaction at all. Moreover, the maximum X-ray luminosities of HR 363 and HD 35155 are comparable despite very different orbital periods;
(ii) the activity appears to be strongly variable.
Orbital modulation is likely a major cause of the activity variations, as shown by Shcherbakov & Tuominen (1992) and Ake et al. (1994) in the well-documented case of HR 1105. However, at a given phase, important cycle-to-cycle variations remain (Ake et al. observed a variation by a factor of 3 in the nm flux of HR 1105 at phase 0.3 in two different orbital cycles), suggesting the existence of yet another cause of (secular) variability.
Various physical processes are able to produce hard photons
by the orbital motion in a binary system:
1. Heating of the red-giant hemisphere facing a hot WD;
2. Accretion-powered hot spot;
3. Stream of gas from the red-giant wind, heated when funneled through the inner Lagrangian point.
A strong sensitivity of the activity level to the binary separation is expected in the first and second cases. In the first case, the orbital separation directly controls the dilution suffered by the hot radiation when it reaches the giant atmosphere. In the second case, the mass accretion rate by the secondary roughly scales as (where is the wind mass-loss rate of the giant, and k is the ratio between the orbital and the wind velocities) in the case of supersonic Bondi-Hoyle accretion in a detached system (see Theuns et al. 1996). The previous relation reduces to (where A is the orbital separation) when k << 1 (i.e. ).
From a detailed analysis of the variations with orbital phase of the UV flux level and the HeI nm line shape, Shcherbakov & Tuominen (1992) and Ake et al. (1994) favour the third process as the origin of the hard photons, i.e. funneling of the red-giant wind through the inner Lagrangian point. The existence of such funneled streams is moreover predicted by smooth particle hydrodynamics simulations of mass transfer in detached binary systems (Theuns & Jorissen 1993; Theuns et al. 1996). Different viewing angles of the stream during the orbital cycle account for the orbital modulation, whereas long-term fluctuations of the mass-loss rate account for the secular variations (like those observed in Mira variables, and associated with a clumpy and non-spherically symmetric wind; e.g. Whitelock et al. 1997; Lopez et al. 1997; Olofsson 1997). The wind mass-loss rate of the red giant, rather than the orbital separation, is expected to be the dominant factor controlling the activity level in this case. The absence of any correlation between the orbital periods and the activity levels in the sample of S stars listed in Table 4 therefore suggests that streams like the one observed in the system HR 1105 might in fact be responsible for the activity observed in other S stars as well. The absence of any activity observed in the system HD 121447, despite the fact that it is the closest system in the sample, may then be attributed to its low luminosity (=-1.4), and therefore low mass-loss rate. Among the more luminous S stars (), differences in their mass-loss rates may account for their different activity levels (compare e.g. HD 35155 and HR 1105 having different activity levels despite similar periods and spectral types, or HD 35155 and HR 363 having the same X-ray flux at very different orbital periods).
Future detailed studies of this class of mass-losing, binary red giants may thus be expected to shed light on the mass-loss process, as well as on the physics of interacting binaries.Acknowledgements
We wish to express our thanks to Tom Ake for communicating us results in advance of publication. Data and bibliographic references made available by the Centre de Données Stellaires (Strasbourg) were of great help in the present study. This work was supported in part by the Fonds National de la Recherche Scientifique (Belgium, Switzerland).
Copyright The European Southern Observatory (ESO)