3 The spectrum of HD149404

A FEROS spectrum of HD149404 is shown in Fig.1. Besides the lines of the upper Balmer series (above H$\gamma$), the spectrum reveals many typical O-star absorption lines of HeI, HeII, NIII, SiIII, SiIV, CIV and OIII. The less massive star exhibits strong NIII absorptions whereas NIII is seen as a weak absorption or emission (NIII $\lambda \lambda$4634-41) in the primary's spectrum. The spectrum reveals also a number of diffuse interstellar bands (DIBs), as well as interstellar absorptions due to NaI, Ca II, CH and CH⁺. However, the most remarkable feature of HD149404 is the important number of emission lines in its spectrum: H$\alpha$, H$\beta$, HeII $\lambda$4686, CIII $\lambda$5696, as well as several NII and SiIV lines. We will come back to some of these features in the next sections. Throughout this paper we will refer to the more massive star as the primary.

Figure 3: Variation of the HeII $\lambda$4686, HeI $\lambda$5876 and H$\alpha$ lines with phase

3.1 Spectral types

Only the least blended spectra are suitable for the spectral classification. Following Conti & Alschuler (1971), Conti & Frost (1977) and Mathys (1988, 1989), we adopt the $\log{W'} = \log{(EW(4471)/EW(4542))}$ classification criteria. We caution that the equivalent width (EW) of the HeII $\lambda$4542 line of the individual components is rather difficult to measure in the case of HD149404 because of the severe blending.

For the secondary, we find $\log{W'}$ between 0.51 and 0.83, corresponding to a spectral type O9.7 (using the Mathys criteria). The O9.7 classification is also supported by the fact that the HeII $\lambda$4542 and SiIII $\lambda$4552 absorptions have about the same strength (Walborn & Fitzpatrick 1990).

For the primary, we obtain $\log{W'}$ between -0.26 and 0.13. More specifically, we find an O6.5 spectral type for orbital phases near 0.75, i.e. when the primary is moving towards us (see Fig. 2), and an O7.5 type half a period later. In fact, Fig.2 indicates that the orbital variation of $\log{W'}$ is essentially due to the variation of the equivalent width of the HeI $\lambda$4471 line. It is interesting to note that the HeI $\lambda$4471 absorption of the secondary displays the opposite behaviour, i.e. it appears stronger around $\phi = 0.75$ and weaker near $\phi = 0.25$ (Nazé et al. 2000). The behaviour of both absorption components at HeI $\lambda$4471 could therefore at least partially reflect the blending with a weak, slightly red-shifted emission component that does not follow the orbital motion of neither star. Such an emission feature is indeed seen in the HeI $\lambda$5876 line (see Fig.3) and we actually detect a weak emission on some of the HeI $\lambda$4471 spectra taken near conjunction. At phases near 0.75, the emission is blended with the primary's absorption, hence reducing its equivalent width. The reverse situation occurs half a cycle later, when the secondary's absorption is blended with the emission. As the primary's EW is most probably less affected by the emission feature around $\phi = 0.25$, we adopt the O7.5 classification for this star (see also Nazé et al. 2000).

For the luminosity classification, we rely on the $\log{W''} = \log{(EW({\rm Si\,IV}\ \lambda}$4089) $/EW({\rm He\,I}\ \lambda$4143)) criterion introduced by Conti & Alschuler (1971). This turned out to be even more difficult than the spectral classification. In fact, HeI $\lambda$4143 could be deblended on a few spectra only and we could only set an upper limit of $\leq 0.07$Å on the EW of the primary's HeI $\lambda$4143 line. For the secondary, we find $\log{W''}$ between 0.36 and 0.65 corresponding to a supergiant classification. In the same way, our upper limit on the primary's EW(HeI $\lambda$4143) yields $\log{W''} \geq 0.4$ (at all orbital phases), also corresponding to luminosity class I. We caution that a similar effect as for the HeI $\lambda$4471 line could alter the EWs of the HeI $\lambda$4143 lines, although we do not expect this effect to significantly affect our conclusions.

On the other hand, HeII $\lambda$4686 could be in absorption in the spectrum of both stars (see below), an unlikely feature for luminosity class I stars, especially for Ia stars (Walborn 1971). Finally, the NIII $\lambda \lambda$4634-41 emission lines appear to follow more or less the motion of the primary. Therefore we add an (f) tag to the spectral classification of the primary.

3.2 Interstellar lines

We have measured the radial velocities (RVs) of the most prominent interstellar lines (Table1). Both CaII lines exhibit a blueshifted much fainter component which is only visible on the FEROS spectra.

We emphasize the very good night-to-night stability of the FEROS and Coralie spectrographs as revealed by the small 1-$\sigma$ dispersions of the RVs in Table 1. There is a slight difference of the order of 0.5kms^-1 between the RVs derived from the FEROS and the Coralie data. Moreover, we notice that there is a slight RV-difference between the CaII and NaI lines. Stickland & Koch (1996) adopted a radial velocity of -7.5 kms^-1 for the interstellar features in their IUE spectra, which is consistent with our results for the NaI lines.

 

 

Table 1: Radial velocities (in kms^-1) of the most prominent interstellar lines in the spectrum of HD149404. The quoted uncertainties correspond to the 1-$\sigma$ dispersion of the RVs measured during the same observing run

Line	FEROS 1999	Coralie	FEROS 2000
CaII $\lambda$3933	$-3.9 \pm 0.2$	$-4.5 \pm 0.3$	$-2.7 \pm 0.1$
CaII $\lambda$3968	$-3.8 \pm 0.3$	$-5.3 \pm 0.5$	$-3.8 \pm 0.1$
CH+ $\lambda$3958	$-2.4 \pm 0.6$	$-3.6 \pm 2.9$	$-2.3 \pm 0.8$
CH+ $\lambda$4233	$-1.6 \pm 0.6$	$-2.7 \pm 0.4$	$-1.9 \pm 0.4$
CH $\lambda$4300	$-0.4 \pm 0.2$	$-1.0 \pm 0.3$	$-0.4 \pm 0.4$
NaI $\lambda$5890	$-7.8 \pm 0.3$	$-8.4 \pm 0.4$	$-8.2 \pm 0.0$
NaI $\lambda$5896	$-7.9 \pm 0.3$	$-8.2 \pm 0.3$	$-7.9 \pm 0.0$

 

 

Table 2: Barycentric radial velocities obtained for HeI $\lambda$4471 and by a correlation with a synthetic mask. The dates of the observations are given in the format HJD-2450000 and the orbital phases are those corresponding to the circular orbital solution derived from the cross-correlation RVs (see Table3). The method used to derive the RVs of the HeI $\lambda$4471 line is specified in the sixth column: d(nbr of Gaussians) = Gaussian fit, c1 (c2) = 1st (2nd) technique (see text), m = mean of techniques listed in brackets

Instr.	Date	$\phi$	HeI $\lambda$4471			cross-correlation
		RV1	RV2	Method	RV1	RV2
			kms-1	kms-1		kms-1	kms-1
LC	622.588	0.23	-107.0	71.7	d(2)
LC	623.595	0.34	-107.1	35.5	m(c1,c2)
LC	624.552	0.43	-86.9	-14.2	m(c1,c2)
LC	625.593	0.54	-58.2	-58.2	d(1)
LC	626.556	0.64	-12.3	-109.4	c1
VLC	925.944	0.14	-103.4	26.8	m(d(2),c1)
VLC	926.944	0.24	-124.2	47.6	d(2)
VLC	930.884	0.65	0.2	-136.9	m(d(2),c1)
VLC	932.946	0.86	-21.2	-124.6	m(d(2),c1)
VLC	937.914	0.36	-111.9	13.7	m(d(2),c1)
VLC	995.635	0.24	-116.6	72.5	d(2)
VLC	996.565	0.34	-96.9	31.3	m(c1,c2)
VLC	997.581	0.44	-66.3	4.1	m(c1,c2)
VLC	998.572	0.54	-73.9	-73.9	d(1)
VLC	999.572	0.64	-48.5	-147.4	m(c1,c2)
VLC	1000.544	0.74	36.7	-160.6	d(2)
FER.	1299.800	0.23	-117.8	52.9	d(2)	-108.0	55.5
FER.	1300.793	0.34	-118.8	43.5	d(2)	-109.0	49.7
FER.	1301.797	0.44	-79.6	-23.6	c2	-67.1	3.0
FER.	1302.788	0.54	-67.5	-67.5	d(1)	-39.2	-70.5
FER.	1304.796	0.74	28.5	-127.6	d(2)	12.4	-138.5
FER.	1323.741	0.67	43.8	-119.7	d(2)	15.4	-124.0
FER.	1327.766	0.08	-77.6	4.6	c2	-75.0	2.5
BME	1328.700	0.18	-135.5	-0.5	c2
BME	1329.705	0.28	-131.3	46.0	d(2)
BME	1330.707	0.38	-92.9	0.6	c2
BME	1331.717	0.49	-52.5	-52.5	d(1)
BME	1332.728	0.59	-32.8	-120.5	c2
COR.	1578.866	0.67	38.8	-115.1	d(2)	6.8	-123.1
COR.	1579.879	0.77	38.0	-139.7	d(2)	18.8	-146.6
COR.	1580.872	0.87	-45.5	-122.1	c2	-9.6	-111.9
COR.	1581.864	0.97	-58.8	-58.8	d(1)	-34.2	-64.0
COR.	1582.869	0.08	-70.5	11.2	d(2)	-80.6	-3.7
COR.	1583.877	0.18	-92.8	63.5	d(2)	-98.0	54.7
COR.	1584.857	0.28	-101.0	65.9	d(2)	-99.7	58.1
COR.	1585.845	0.38	-84.4	8.9	d(2)	-85.9	8.2
COR.	1586.870	0.48	-46.6	-46.6	d(1)	-65.0	-30.9
COR.	1587.861	0.58	-48.2	-104.7	c2	-10.0	-89.8
COR.	1588.886	0.69	31.8	-133.8	d(2)	8.1	-132.9
COR.	1590.886	0.89	-49.8	-108.3	c2	-9.1	-98.5
FER.	1668.892	0.84	-44.2	-150.0	c2	-4.4	-127.2
FER.	1669.769	0.93	-73.5	-73.5	d(1)	-18.1	-82.2
FER.	1670.770	0.03	-46.1	-46.1	d(1)	-61.0	-15.0
FER.	1671.775	0.13	-95.3	27.8	d(2)	-85.5	27.9
FER.	1672.766	0.23	-104.6	67.4	d(2)	-102.5	64.8
FER.	1673.895	0.35	-106.7	18.8	d(2)	-89.6	33.7