6 Emission lines

The spectrum of HD149404 exhibits many emission lines some of which display strong orbital variability.
- The unidentified $\lambda \lambda$4486, 4504 emission lines appear to be associated with the secondary star as shown by Nazé et al. (2000). These emission lines are a common feature in the spectra of the most luminous stars between spectral type O7 and O9.5 (Conti 1973; Walborn & Fitzpatrick 1990);
- NIII $\lambda \lambda$4634-41: the bulk of the emission seems to follow the primary whilst the lines are observed in absorption in the secondary's spectrum. The RVbehaviour of the NIII emissions itself is however rather complex and the lines are quite broad and display strong profile variability probably indicating that these lines are not purely stellar features;
- HeII $\lambda$4686 (see Fig.3) consists of a mixture of emission and absorption lines and the line morphology displays strong phase-locked variations. The emissions do not follow the orbital motion of the stars. Instead, both stars exhibit this line in weak absorption though the visibility of these absorption features varies strongly with phase and the absorption lines vanish near phases 0.2 and 0.7;
- A similar situation is observed for the H$\beta$ line. As for HeII $\lambda$4686, the stellar H$\beta$ lines are again in (rather weak) absorption and the composite line profile suffers strong variability with an emission feature that does not follow the motion of either star. We caution that the H$\beta$ profile is blended with the NIII $\lambda \lambda$4858-67 absorption lines and this blend probably affects the radial velocity measurements of the H$\beta$ emission components;
- Our echelle spectra reveal NII emission features at $\lambda \lambda$5001-12, 5667, 5932, 5940-42, 5952, 6482 and 6611, that are associated with the O9.7 secondary. Walborn (1980) reported weak NII $\lambda \lambda$5940-42, 6482 emission in the ON9.7Iae star HD105056. The presence of the NII emissions and the strength of the NIII absorption lines in the spectrum of the secondary of HD149404 could therefore indicate that nitrogen is enhanced in the atmosphere of this star. Further support for an ON classification of the secondary comes from the relative weakness of the CIII $\lambda \lambda$4647-50 absorption compared to the intensity of this line in "normal" O9.7 supergiants (see Walborn & Fitzpatrick 1990); - The CIII $\lambda$5696 emission is probably formed by two independent features. The strongest component follows the orbital motion of the primary. This trend is confirmed by the variations of the line peak position (see Nazé et al. 2000). At some orbital phases (e.g. $\phi = 0.67$), a second emission peak appears with a velocity indicating that it is most probably associated with the secondary;
- In addition to the absorption lines, HeI $\lambda$5876 displays a variable emission which is particularly strong at phases near 0.0 and 0.5 (see Fig.3). This emission is probably present in other HeI lines too, maybe causing part of the intensity and EW variations;
- The strongest emission feature in the visible spectrum of HD149404 is H$\alpha$ (see Fig.3). The line exhibits a double peaked structure at certain phases, more or less in quadrature with the orbital motion. Though the visibility of the emission peaks in the H$\beta$ profile is hampered by the blending with the stellar absorption lines, they seem nevertheless to follow more or less the behaviour of the H$\alpha$ peaks, pointing towards a similar origin. In Sect.7.2, we will discuss the variability of the H$\alpha$ emission line in terms of a wind interaction model.

To further analyse the emission features, we apply a so-called "S-wave analysis" (see e.g. Richards et al. 1996). We adopt two axes centred on the center of mass of the binary; the x axis points towards the secondary star while the direction of the y axis is given by the orbital motion of the secondary. We measured the RVs of the peaks of the most prominent emission lines and fitted a sine-wave expression

$$v(\phi) = -v_x\,\cos(2\pi\phi) + v_y \sin(2\pi\phi) + v_z$$

to these points (see e.g. Rauw et al. 1999). The results are summarized in Table6 and a projected Doppler map is presented in Fig.6. We caution that the quality of the fit to the RVs of the NIII $\lambda \lambda$4634-41 lines is rather poor, mainly because of the variable shape of these lines.

 

 

Table 6: Results of the S-wave analysis: Coralie observations are not included for the NIII emissions

	vx	vy	vz
? $\lambda$4486	-0.1	87.3	-39.1
? $\lambda$4504	2.7	80.6	6.5
NIII $\lambda$4634	-9.6	-73.6	-36.9
NIII $\lambda$4641	-16.9	-105.8	-52.8
CIII $\lambda$5696	22.3	-62.6	-42.5
NII $\lambda$5932	8.3	108.0	-81.7
NII $\lambda$5942	16.3	109.7	-44.8
HeII $\lambda$4686	-128.6	-82.4	-99.7
HeII $\lambda$4686	80.6	-65.3	16.4
H$\beta$	-123.8	-153.2	-47.1
H$\beta$	136.7	57.0	-47.8
H$\alpha$	-119.2	-75.1	-44.4
H$\alpha$	106.2	24.0	-53.4

Figure 6: Doppler map showing the results of our S-wave analysis projected on the (vx, vy) velocity plane. The crosses indicate the velocity of the center of mass of the binary components as derived from our orbital solution while the dashed line is the equivalent of the Roche lobe in velocity space. The different symbols stand for different emission lines: filled circles = H$\alpha$; open circles = H$\beta$; filled triangles = NII $\lambda \lambda$5932, 5942; open triangles = NIII $\lambda \lambda$4634-41; filled squares = $\lambda \lambda$4486, 4504; open square = CIII $\lambda$5696 and stars = HeII $\lambda$4686

The positions of the $\lambda \lambda$4486, 4504, CIII $\lambda$5696, NIII $\lambda \lambda$4634-41 and NII $\lambda \lambda$5932, 5942 emission regions in velocity space indicate that the bulk of these emissions arises most probably within the atmospheres of the stars. On the other side, our analysis reveals that this is not the case for the H$\alpha$, H$\beta$ and HeII $\lambda$4686 emissions. The latter lines exhibit a double peaked structure at orbital phases near conjunction and a single peak around quadrature. Assuming that the two peaks actually cross over, we have fitted their RVs with the above sine-wave expression. We caution here that only the H$\alpha$ peaks are free from blends with absorption lines. The H$\beta$ and HeII $\lambda$4686 lines are usually "polluted" by absorption lines and the results for these lines should therefore be regarded only as trends roughly confirming the behaviour of the H$\alpha$ peaks.