Astron. Astrophys. 336, 637-647 (1998)

7. H emission line profiles

7.1. H line profile observations

Our CAT spectra are not flux calibrated. We used Fe i absorption lines in the underlying M-star continuum in order to scale our H spectra to measure the H line strength. We assumed that the flux contribution from the M-star is constant in time, but allowed for a variable nebular continuum contribution. This is a reasonable calibration criterion, considering that the available IR magnitudes of BX Mon show no significant variations. This procedure allows us to determine H line fluxes on a relative scale. One spectrum cannot be directly scaled, due to the noisy continuum. The H equivalent widths and these relative H line fluxes are listed in Table 1.

The H equivalent widths and line fluxes are strongly reduced at phase , giving further support to a high
inclination i.

Fig. 9. Normalized H line profiles shifted according to their phase . Velocities are given with respect to the center of mass. The spectrum taken on 12 Sep 1991 () has been scaled according to the neighbouring spectra. The short horizontal lines mark the continuum of each spectrum and the zero flux-level. The intersections of the sine-curve (dashed) with the horizontal lines mark the radial velocity of the hot component at the time of observation.

The emission line consists of two principal components. One is a narrow absorption that is approximately at rest with respect to the red giant. It has a full width half maximum (FWHM) of about . The other is a broad emission, with full width at zero intensity of the order of . It is clearly too broad to be due to the undisturbed red giant wind. It may arise from a turbulent zone which is not understood, but it is unlikely to be due to electron scattering or other non-dynamical broadening mechanisms.

7.2. Synthesized H profiles

To gain some insight into the origin of the line variations, we have computed a schematic kinematic model of the absorption by the giant wind as a function of orbital phase. The model uses the procedure outlined by Shore et al. (1994). We arbitrarily assume that the absorption is produced by a screen seen against a stable emission line. This focuses attention on the variable column density of the matter in front of the line forming region, while ignoring the problem of formation of the emission line, which is assumed to be formed in the hot gas around the hot component. The absorption line forming region was assumed to be thermodynamically and dynamically decoupled from the ionized region. The only two parameters that are required to produce the observed profile changes are the relative radial velocity of the two stars and the line of sight optical depth toward the emission line producing region. We used a terminal velocity for the red giant wind of . The models are computed assuming an intrinsic Gaussian absorption profile with a FWHM of with a turbulent broadening of (Shore & Aufdenberg 1993). Unlike the "iron curtain" models, where column densities and ionization structure were explicitly computed, we used only the velocity gradient and optical depth as the input parameters. The velocity of the red giant wind was specified by:

where is the turbulent velocity and is the terminal velocity for a radial distance r and a stellar radius . For the computation shown in Fig. 10, we employed .

Fig. 10. Synthesized H profiles for (upper) and (lower).

One clue to the origin of the absorption component is that it never displays a truly Gaussian profile. The line formed by a simple absorbing wind is always scewed toward the terminal velocity.

At inferior conjunction, the hot component suffers minimum circumstellar extinction and the absorption should be at nearly the terminal velocity, depending on the size of the accelerating region for the wind. At the quadratures, the absorption should extend from the center of mass velocity, , to the terminal velocity, but only on the negative side of the profile in both cases. However, the relative motion of the two components has opposite signs at the two quadratures, so the absorption is shifted with respect to the emission line in opposite directions. At superior conjunction, or eclipse, if the system has a sufficiently high inclination, the absorption should extend over the whole range and be at its strongest. The profile should be more flat-bottomed at this phase.

If this picture is correct, it is possible to predict the orbital properties from the profile variations alone. Small relative velocities dominate in the long period, nearly circular systems. Therefore, the absorption line should generally be seen only on the blueward portion of the emission profile. High eccentricity and relatively large radial velocity amplitude can combine to shift the absorption to the red side of the emission peak. Absorption of the red giant continuum may mean one of two things. Either the hot component contributes substantially to the continuum at 6563 Å and the H line opacity is great enough to absorb, or the optical depth is always large at that wavelength; this would imply a chromosphere-like layer in the red giant atmosphere. At present, our procedure is too crude to permit detailed modeling of the line formation. But it provides a heuristic guide to the resolution of the origin of the diversity of line profiles observed in the S-type symbiotics. An alternative model based on non-LTE calculations for an expanding red giant's atmosphere ionized from the outside by the radiation of the nearby hot radiation source, can be found in Schwank et al. (1997).

© European Southern Observatory (ESO) 1998

Online publication: July 20, 1998
helpdesk.link@springer.de