3 Discussion

The results of the optical campaigns reported in this letter led to the unambiguous identification of the optical counterpart of RXJ0806.3+1527, and provide also important new constraints on the source.

Firstly, the possibility that RXJ0806.3+1527 is a nearby isolated neutron star accreting from the interstellar medium can be ruled out, because for a distance of $\geq$100pc (as implied by the upper limit on the proper motion) the blackbody emission radius inferred from the optical continuum (or the optical continuum plus the RASS point) would be several hundred km at least, too large a value for any neutron star model.

Two additional results are especially relevant for assessing the nature of RXJ0806.3+1527; these are: (a) the absence of optical periodicities other than the 321s modulation (up to periods of $\sim$5hr); (b) the He emission line spectra. The absence of a second periodicity argues against models involving a non-synchronous magnetic accreting white dwarf (such as the intermediate polars, IPs). In addition to their orbital and spin periodicities (and/or their beat), IPs display optical spectra with intense Balmer emission lines with EW of tens of Å and widths of a few hundred kms^-1. Moreover the X-ray amplitude of the spin modulation is comparatively low (Hellier 1999). The 100% X-ray modulation and broad and weak emission lines that we observed from RXJ0806.3+1527 are also very much at variance with these IP properties. Moreover a hot blue thermal continuum similar to the one we revealed from RXJ0806.3+1527 is simply not observed in magnetic cataclysmic variables of any class. We note that if the blue thermal continuum were attributed to the accreting white dwarf (and/or the accretion stream) the optical spectra exclude the presence of a companion star earlier than a M7V for a (maximum) distance of 1kpc. The condition that such a (hypothetical) companion star fills its Roche lobe translates into an orbital period of <2hr, well within the range of periods sampled by our photometric studies. Moreover it would be very difficult the explain the observed He emission lines within such a scenario.

The most natural interpretation of our results is that RXJ0806.3+1527 is a double degenerate binary system, in which mass is transferred from a Roche lobe filling white dwarf to another more massive white dwarf. Such binaries belong to the AM CVn class and comprise only a few objects, although recently two new members have been proposed (RXJ1914+24, period of 9.5min; Ramsay et al. 2000, and KUV01584-0939, period of 10min; Warner & Woudt 2002). AM CVns are intrinsically blue objects with thermal optical continua, orbital period in the 10-50min range (usually detected in the optical band), fairly broad He emission and absorption lines ($FWHM \sim$ 15-35 Å; for a review see Warner 1995). The presence of flickering in the optical light curve testifies that these systems are powered by accretion. AM CVns are expected to originate from normal composition binaries which experienced two phases of mass exchange, exposing the helium cores of the original stars (e.g. Warner 1995). During their evolution, they reach a minimum orbital period of $\sim$4min and thereafter evolve to longer periods. Double-peaked emission line spectra clearly testify to the presence of an accretion disk mediating the flow of matter. However in the X-ray bright AM CVn candidate RXJ1914+24, the accretion disk is likely not present (Marsh & Steeghs 2002; Wu et al. 2002). In one model for such a short orbital period system the soft X-ray emission originates from direct impact accretion which occur when the minimum distance of the gas stream from the center of mass of the accretor is smaller than the accretor size (Marsh & Steeghs 2002).

We propose here that RXJ0806.3+1527 is a new member of the AM CVn class and in particular that: (i) The modulation at 321s corresponds to the orbital period of the system. The 100% amplitude in the X-ray modulation can be easily explained in terms of self-occultation of the stream impact point on the surface of the accreting white dwarf. The blue optical continuum and the small amplitude optical modulation likely results from X-ray reprocessing by a fraction of the donor star surface and/or the accretion stream. The similarities with RXJ1914+24 suggest that RXJ0806.3+1527 might also be a direct impact accretor. In this case the system may be non-synchronous, while no conspicuous modulation is produced at the accreting white dwarf spin period. The possibility that the system is magnetically locked, as in AM Her binaries, cannot be excluded at present, although the observed line EWs and shape of the optical continuum argue against this. (ii) Emission lines from He are expected given the temperatures implied by the optical continuum and the presence of H depleted accreting gas. Note that He emission lines were detected also in KUV01584-0939 (Wegner et al. 1987), whereas the study of the optical spectrum of RXJ1914+24 is hampered by the high absorption in the direction of the source (Ramsay et al. 2000, 2002). The presence of flickering in the RXJ0806.3+1527 optical light curves (and the X-ray ROSAT HRI light curves as well, see I99) further suggests that the optical emission is at least in part related to the accretion process, as expected in the irradiation scenario.

The condition that the mass donor white dwarf fills its Roche-lobe determines its mass M₂ = 0.12$M_{\odot}$, and radius $R_2 = 1.7 \times 10^{9}$cm = 0.02 $R_{\odot}$. The mass transfer rate M driven by angular momentum losses through gravitational radiation can then be calculated over a reasonable accretor mass range. We adopt $0.2 \leq M_1$/ $M_{\odot} \leq 0.5$, the lower limit being close to the stable mass transfer limit. This range translates into 1- $3\times10^{-7}~M_{\odot}{\rm yr}^{-1}$ (assuming a efficient tidal coupling), an accretion luminosity in the 2-5 $\times$ 10³⁵ergs^-1 range and a flux of 10^-9-4 $\times$ 10^-7ergs^-1cm^-2 (for a distance in the 0.1-1kpc range). The latter values are consistent with the peak flux inferred from the RASS (see Sect. 1). In order not to violate the (minimum) size of the blackbody emitting area derived from the normalisation of the optical continuum ( $200 \leq R_{\rm bb}$(km) $\leq 6000$ for a distance of 0.1-1kpc and the entire range of allowed blackbody temperatures) the optical continuum must come from a smaller region than the donor star, perhaps the accretion stream and/or the illuminated part of the donor star.

Within this interpretation the secondary's orbital velocity is expected to be in the 900-1500kms^-1 range, such that the observed emission line width might be dominated by the binary's Doppler velocity amplitude. Phase-resolved spectroscopy can ascertain this point unambiguously and provide important new information.

In summary our results provide compelling evidence that RXJ0806.3+1527 is a double white dwarf interacting binary system, with the shortest orbital period ever recorded. The source represents one of the most promising targets for gravitational wave detection from binary motion (Nelemans et al. 2001). Indeed a 321s orbital period falls well above $\log$$f \sim$ -2.7Hz, where the average galactic background hampers the gravitational wave detection. Moreover assuming reasonable values of the distance and mass of the accretor, a strain amplitude of up to ${\rm few} \times 10^{-21}$ is expected, which is well above the LISA sensitivity.

Acknowledgements

Based on observations made with the Italian Telescopio Nazionale Galileo (TNG) operated on the island of La Palma by the Centro Galileo Galilei of the CNAA (Consorzio Nazionale per l'Astronomia e l'Astrofisica) at the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias We thank the ESO director for the DDT allocated to observe RXJ0806.3+1527 and R. M. Athreya for the help in carrying out observations at the 3.6 m ESO. We thank G. Tessicini and G. Marino for the observations at the TNG. The Guide Star Catalogue-II is joint project of the Space Telescope Science Institute and the Osservatorio Astronomico di Torino. An anonymous referee provided useful comments on the first version of this paper. This work is supported through CNAA, ASI and MURST grants.