5. A change in the ionizing continuum
If the hydrogen was already present around the star in a partly neutral steady-state envelope, a change in ionizing flux from the star could, in principle, be the cause of the observed emission.
As the Lyman continuum constitutes a tiny fraction of the total energy output of a B star, it might undergo dramatic changes without requiring much adjustment of the ultraviolet or optical flux. Only two B stars have been observed in the Lyman continuum, and CMa (Cassinelli et al. 1996 and Cassinelli et al. 1995, respectively). The first object shows Cepheid pulsations. Cassinelli et al. (1996) detected changes with a total amplitude of more than 30% in the extreme ultraviolet (between 504 and 700 Å) of CMa, while the object is variable in the V band with an amplitude of only 0.02 magnitude. Interestingly, the period of these variations in this star - classified as a giant - is 6 hours.
One should bear in mind that the Lyman continuum of B stars is not well known, having only been observed directly in the two cases just mentioned. Cassinelli et al. (1995) found that the observed continuum radiation from CMa (B2II) is an order of magnitude stronger than model atmospheres predict. This discrepancy is smaller or perhaps absent for CMa (B1 II-III, Cassinelli et al. 1996), yet illustrates the uncertainty of the current knowledge of the ionizing continuum of B stars.
It is thus possible that slight atmospheric changes due to some pulsation mechanism are capable of increasing the ionizing continuum significantly, whilst having a small effect on the optical spectrum. Although the period of the activity in HD 76534 is not known, it is likely that it is of order hours. Such timescales have been observed both in Cepheid variables and, photometrically, in several Be stars, where the periods range from several hours to days, and are often attributed to non-radial pulsations (Peters, 1991). Unfortunately, the number of photometric measurements of HD 76534 is low, and it is unclear whether the V band is variable at all.
A useful exercise is to use existing model atmosphere predictions to estimate the number of ionizing Lyman continuum photons emitted by an early B dwarf. If the ionization of the envelope is entirely attributable to Lyman continuum photons, then we would hope to find this number to be at least comparable to the number of recombinations associated with the H emission measure we have derived ( s-1). Taking a stellar effective temperature of 22000 K for a B2 star from the main sequence scale of Schmidt-Kaler (1982), a surface gravity , and stellar radius 7.5 , the predicted Lyman continuum photon luminosity is s-1 according to the Kurucz (1992) grid. Hence, the numbers are in the right neighbourhood. Furthermore, in this spectral type range, the Lyman continuum flux is a very sensitive function of stellar parameters (e.g. roughly doubling with a 1000 K increase in effective temperature). Given this sensitivity we can only conclude that the early B spectral range is that in which atmospheric variability is most likely to be betrayed by changes in circumstellar ionization.
There is a caveat. If the H emission arises in a circumstellar disk presenting sufficient optical depth to require ionization of hydrogen via the Balmer rather than the Lyman continuum, it becomes highly improbable that the H variability is attributable to an ionization change. This is because the factor of several change in H EW would demand almost as big a change in the Balmer continuum and Lyman line fluxes. Assuming radiative excitation to n=2, a crude argument would be that the Balmer continuum is used twice (see Drew, 1989). Since this is where the Planck maximum for this star falls, it is obviously implausible.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998