3. Line doubling phenomena
The most important observationnal characteristic of BW Vul is the presence of two line doubling phenomena which occur at each velocity discontinuity. Fig. 1 represents a series of spectra before and after the stillstand.
To interpret these line profiles, it is important to compute the spectra in the stellar rest frame i.e., the systemic velocity of the star must be calculated. This is usually done by an integration of the velocity curve over one pulsation period. But this supposes that the shape of the radial velocity curve is well determined i.e., the number of spectra is large enough. Because at some phases a line doubling appears, three kind of velocities can be measured. First, when they are visible, we can fit each line component (the blueshifted and the redshifted ones) by a gaussian to obtain their velocity (Fig. 2a) or a single gaussian fit over the whole profile whatever its shape (Fig. 2b).
Contrary to the double gaussian fit, only the single one provides a mean velocity of the motion of the atmospheric layers. The physical meaning of this average velocity is weakly informative on the dynamics of the atmosphere. Thus, the -velocities which can be deduced from these three velocity curves (Fig. 2a and b) are quite different. It is around -20 km.s-1 for the blue component, 4 km.s-1 for the red one and -11 km.s-1 for the whole profile. For the second night (August 8th), we respectively find -14 km.s-1, 1 km.s-1 and -10 km.s-1.
At phase , the Si iii line profile has the more symmetrical and narrow shape and hence can be interpreted as the phase of the largest atmospheric extension, when the velocity field within the line formation region may be negligeable. Thus, its associated radial velocity (- 7.6 km.s-1) can be considered as close to the systemic velocity. For the night August 8th, we obtained -10.8 km.s-1. Thus, we have assumed hereafter that the systemic velocity of BW Vul can be estimated by the average of these two evaluations over our two observation nights. The adopted value km.s-1 was used to compute the spectra in the stellar rest frame. This value is not very different from the average (-10.5 km.s-1) of the -velocities for the whole profile.
Our spectra follow the same general pattern as previous observations. During the inward atmospheric motion, the profile becomes slightly asymmetric () on the blue side and then more and more complex, until two components can be clearly distinguished (). In the meantime, the red component decreases until disappearing (). Note that the blue component is slightly redshifted until , while it is close to a zero-velocity at , 0.93 and 0.96. If the red component is considered alone, it seems to be more and more redshifted, regularly, during the whole spectra set. One can easily imagine a straight line joining the cores, at the different phases.
This behavior is similar after the stillstand, except that the doubling is not resolved and is much shorter (between and 1.07). However, this time, the blue component is really blueshifted and the red one is at zero-velocity. Then, from to , the profile is symmetric, and blueshifted. Finally, from until , the profile slowly moves to the red and becomes more and more sharper. This is well illustrated on Fig. 2a: from to 0.64, the velocity curve is smooth. Then, the asymmetric profile can be fitted with two gaussians, providing for both components an increasing velocity, the red component being in the continuity of the velocity curve, while the blue component decelerates.
When the two components are visible, the blue curve undergoes the first discontinuity which shifts the velocity to zero by 70 km.s-1, while the red curve vanishes at , inducing a gap of about 180 km.s-1. Moreover, it appears that the stillstand is not really constant, the velocity, after a very short expansion, being slightly positive. Then the second doubling induces the second velocity discontinuity, affecting first the blue component, with a gap around 80 km.s-1. After this violent expansion, the velocity seems to follow a ballistic motion.
This behaviour is nearly, but not exactly, the same in the upper atmosphere where H is formed. Indeed, because the Si iii line has a larger ionization and excitation potential compared to that of H, it is thought to be formed lower in the stellar atmosphere (see Sect. 4). Hence, the physical conditions may be different between the two line formation regions. Fig. 3 represents, for the same phases as Fig. 1, the H spectra.
Of course, the H profile being very broad, it is not as easy as in the case of the Si iii line to appreciate at which phase the profile becomes asymmetric, and even to distinguish the line doubling components. Only spectra at and 0.93 show such an evidence. Furthermore, the second doubling phase can only be suspected at . However, it seems that the amplitude of the doubling phase is comparable for both Si iii and H lines. The only difference between them is that the doubling discussed above for the Si iii line happens slightly later for the H one ().
We have compared these spectra with those obtained on night August 8th which are represented, for the Si iii line, on Fig. 4. The most striking difference between the two nights is that the line doubling is poorly seen during the second night. This is particularly true on phase . Also, during the first discontinuity, the spectra obtained on August 8th are much more symmetric (until ). The same velocity curves as in Fig. 2a and b are displayed in Fig. 5a and b.
One can see that the red curve is not a straight line as in Fig. 2a and b but decelerates at nearly the same amount as the blue one. During the first velocity discontinuity, the velocity jump associated to the red curve is about 130 km.s-1, and that associated to the blue curve is around 70 km.s-1. As for the second velocity discontinuity, the gap is larger for the red component, being around 100 km.s-1.
© European Southern Observatory (ESO) 1998
Online publication: October 21, 1998