AM CVn is the prototype of a small group of variable stars with remarkable properties. They are all very blue objects, showing only He I and, occasionally He II features, which can vary from absorption to emission, in their optical spectra. Their UV-spectra contain features from processed material as C, N, O and heavier elements (Solheim 1993a). Photometric variations with periods 15-45 minutes are interpreted as indicators of orbital periods far shorter than obtained for ordinary cataclysmic variables.
In addition to AM CVn, five other objects belong to this group: EC 15330-1403, CR Boo, V803 Cen, CP Eri and GP Com. They form a sequence with increasing photo- metric periods and changes in behaviour. The first part of the sequence includes AM CVn and EC 15330-1403, each with periods near 1000 s, constant mean magnitudes, and absorption spectra. The middle of the sequence is populated by CR Boo, V803 Cen and CP Eri, all of which undergo large amplitude outbursts upon which are superimposed periods ranging from 1600 to 1800 s, and whose spectra vary from absorption to emission. GP Com represents the end of the sequence, with the longest photometric period of 2800 s and an emission spectrum.
Only for GP Com is the orbital period spectroscopically detected. This sequence may indicate a dependence of mass transfer rate on orbital period. Smak (1983) calculated, for reasonable assumptions of mass ratios and orbital periods, that AM CVn has a high mass transfer rate and is in a perpetual high state, while GP Com has a low rate and is in a permanent low state. CR Boo, V803 Cen and CP Eri were found to have mass transfer rates which make them unstable to disk perturbations.
Our empirically derived dependence of mass transfer rate on orbital period implies that the AM CVn stars form an evolutionary sequence. The binary model for these objects, involving mass transfer from a low mass degenerate secondary to a more typical deg enerate white dwarf via an accretion disk, was first proposed by Faulkner et al. (1972). More recent calculations (Savonije et al. 1986) show that the donor secondary will be driven out of thermal equilibrium, whereby the star does not become fully deg enerate, but remains semi-degenerate. Warner (1995a) finds that a semi-degenerate secondary model gives reasonable primary masses for all the AM CVn objects.
The evolution of a pair of interacting degenerate stars with low mass ratio can produce white dwarf stars of different composition and masses than is possible with single star evolution. Nather et al. (1981) proposed that interacting binaries could evolve into single helium white dwarfs (DBs), which are normal (C/O core) white dwarfs showing He lines. Binary evolution may also explain the He core white dwarfs. Low mass single stars evolve too slowly to form the lighter helium core white dwarf stars in a Hubble time (Iben & Webbink 1989). It is therefore difficult to explain white dwarfs with masses less than 0.55 without binary star evolution (Iben & Tutukov 1986). Evolutionary calculations for white dwarfs with various compositions and a non-degenerate helium or hydrogen secondary star (Iben & Tutukov 1991) show that after several thermonuclear runaway flashes, during which helium rich shells are emitted, the secondary has a mass below 0.2 and no more flashes are expected. Mass transfer will strip the donor until a degenerate core of less than 0.1 remains, which will continue to transfer mass at a mass transfer rate /yr. In contrast to the short rapid transfer phases, the final slow transfer can last a few times years, when the orbital period is slowly increasing. The study of this process, where the internal layers of the donor star are exposed, can give important clues to its earlier history, and gives us means to test theories of stellar evolution (Nather 1985).
AM CVn's IUE and HST UV spectra reveal absorption features of Si III, Si IV, N IV, N V, C IV, O V and He II (Solheim & Kjeldseth-Moe 1983; Solheim et al. 1997). If these features originate from an accreting white dwarf, we require a temperature of 50, 000-150,000 K. The wide absorption profile of He II ( 1640 Å) is almost identical to a similar profile for the coldest DO star HZ 21, and may indicate a connection between the central star of the AM CVn system and the DOs (Solheim & Sion 1994).
Non-radial oscillations have been proposed as a possible explanation of AM CVn's photometric variations. A class of pulsating white dwarfs, the DOVs, do exist within the central white dwarf's temperature range. The identification of a pulsating white dwarf within a binary system would provide a unique opportunity to probe, using the techniques of astro- seismology (Nather et al. 1990), the interior of an accreting star. We may expect to observe changes in the pulsation spectrum as a function of the mass transfer rate and the changes in the temperature profile of the accretor's atmosphere (Nitta 1996). However, if AM CVn's central star is a DO pulsator, we may have serious problems in detecting its pulsations in the optical part of the spectrum where the flux from the disk will dominate.
Even if we detect only a few pulsations for AM CVn, it would be of great interest, and a challenge to theoreticians to model the pulsation spectrum of pulsating accretors in the AM CVn systems. These objects may be examples of white dwarfs which are dist urbed from the outside by accreting matter dumped on them and triggering pulsations, which otherwise would not be observable. Parametric resonance with the secondary l ow mass object and possible pulsations in the disk may complicate the interpretation of the modulations we observe.
AM CVn has not changed its mean magnitude significantly in the more than 30 years it has been observed 2 (Sect. 2). A study of its properties may be justified by the possibility of identifying the orbital period and its harmonics which probe the disk, and the period of rotation of the central star, which may tell us if a hot boundary layer is needed or not. g -mode pulsations, if detected, will give clues to the understanding of the structure, composition, and evolution of the accreting star. By combining spectroscopic and astroseismological information, we would be able to set firm limits for theories of stellar evolution.
Patterson et al. (1993) reported periodic absorption line profile modulation for AM CVn with a period of 13.4 hrs which they interpreted as a period of disk precession. In Sects. 4 and 5 we will show that this period also exists in the photo- metric temporal spectrum, indicating a connection between the spectroscopic and photometric observations.
We will give a summary of earlier photometric observations of AM CVn in Sect. 2, then we will describe the Whole Earth Telescope (WET) observations and the data reduction procedure in Sect. 3. In Sect. 4 we present an analysis of the data, and som e special problems they gave us. In Sect. 5 we discuss one model which may explain the observations. Finally, in Sect. 6 we draw conclusions and point out the direction for further observations and theoretical work to understand this system.
© European Southern Observatory (ESO) 1998
Online publication: March 30, 1998