Online material

Appendix A: Observation logs

Appendix B: Medium-resolution AMBER/VLTI data

The medium-resolution data from AMBER are very rich and deserve to be shown in detail. The atmospheric conditions during the observing run were excellent and the standard deviations of the dispersed differential and closure phases are better than 2°. We show in Figs. B.1 and B.3 the differential visibilities from five individual observations in the CO and Brγ region, respectively. The dataset consists of three dispersed raw visibilities from a calibrator, and two sets of three dispersed visibilities from V382 Car and HR 5171 A (out of five recorded each). The differential and closure phases are shown in Figs. B.2 and B.4 in the CO and Brγ region, respectively.

The comparison of the visibilities from these two sources shows the strong impact of the extended environment of HR 5171 A on the observations. We note also that a signal is observed in the visibilities from V382 Car through the CO lines, but only at the longest baselines. This signal indicates a thin molecular environment at a close distance from the photosphere. This must be some CO emission filling an already existing CO absorption since there is no significant signature of this CO in the spectrum. The strong phase signal of the CO lines shows the spatial complexity of HR 5171 A. We recall that a UD surrounded by a centered GD should yield a zero phase signal. One can also see that a significant closure phase signal is observed only in data from HR 5171 A. Furthermore, the signal was observed to strongly increase from one observation to the next during the night, following the Earth rotation of the projected baselines, clear evidence of binarity, but also an indication that the envelope of molecular gas also has a complex shape, not accounted for our geometrical model. No signal is observed in the Brγ region (visibilities and phases). A weak phase signal close to the noise limit is observed in the NaI2.2 μm sodium line of HR 5171 A.

Fig. B.1 Medium-resolution (R = 1500) AMBER/VLTI data centered in the CO molecular band. Some uncalibrated visibilities from a calibrator are shown in the upper row as a visual noise estimate. The dataset consists of three dispersed differential visibilities scaled to the level of the absolute visibility obtained using a calibrator. The spectrum (in arbitrary units) is shown in color at the bottom of each panel. The decrease in the dispersed visibilities betray some extended emission.

Open with DEXTER

	Fig. B.2 Same dataset and spectral region as that in the previous figure. The different panels show three differential phases and one closure phase per observation.
Open with DEXTER

Fig. B.3 Medium-resolution (R = 1500) AMBER/VLTI data centered on the Brγ line. Some uncalibrated visibilities from a calibrator are shown in the upper row as a visual noise estimate. The dataset on the science consists of 5 consecutive observations providing for each, three dispersed visibilities (this figure), three differential phases and one closure phase (next figure). The decrease of the dispersed visibilities betray some extended emission, particularly in the NaI sodium line of HR 5171 A whereas no signal is observed in the Brγ line. No significant interferometric signal is observed in the data from V382 Car.

Open with DEXTER

	Fig. B.4 Same dataset and spectral region as that from the previous figure. The different panels show three differential phases and one closure phase per observation.
Open with DEXTER

Appendix C: Archival X-ray data

HR 5171 A and HR 5171 B were observed serendipitously by XMM-Newton during 40 ks in Aug. 2001 (Rev number = 0315, thick filter used for the EPIC cameras). This archival dataset (ObsID = 0087 940 201) was downloaded and processed using SAS v12.0.0 and calibration files available on July 1, 2012, following the recommendations of the XMM team¹¹. A background flare affects the beginning of the observation, which was discarded. HR 5171 appears as a single faint source near the top edge of the field-of-view of the MOS2 and pn cameras. It is also near a gap in the pn dataset. The source detection was performed using the task edetect_chain on the three EPIC datasets and in two energy bands (soft = S = 0.3 − 2.0 keV, hard = H = 2.0 − 10.0 keV energy band). The best-fit position for this source is 13:47:10.138, −62:35:16.11, i.e. at 8.6′′ to the NW of HR 5171 A, a position compatible with that of HR 5171 B. The X-ray detection is also compatible with current knowledge on the X-ray emission of massive stars : O- and early-B stars, such as HR 5171 B, are moderate X-ray emitters. We infer the equivalent on-axis count rates are 0.008 ct/s for MOS2 and 0.02 ct/s for pn, which correspond¹² to fluxes of 1.5 − 2.55 × 10^-13 erg cm^-2 s^-1 considering a plasma temperature of 0.6 keV. These values yield log (L_X/L_BOL) ~  − 6.5, a value typical of massive OB stars. In contrast, the evolved objects with strong winds, such as the LBVs or WCs, are generally not detected (Nazé 2009; Nazé et al. 2012; Oskinova et al. 2003). It is thus unsurprising that HR 5171 A is not detected. There is also no evidence of a companion or any evidence of energetic radiations from the interaction between the two components in this spectral range. In August 2001 the orbital phase was about 0.66, a potentially favorable configuration for observing any emission from the companion.

Appendix D: Infrared photometry

Fig. D.1 Near-infrared two colour diagrams. The full dataset is shown with the same color coding as in Fig. 7 and the location of Cepheids (Laney & Stobie 1992), after correction for interstellar reddening, (assuming the reddening law from Laney & Stobie 1993 and AL = 0.15 E(B − V)) is shown for comparison. The arrow shows the effect of correcting HR5171A for interstellar reddening of E(B − V) = 1.13 mag (van Genderen 1992), following the same reddening law. We note that there may well be additional reddening due to circumstellar extinction of uncertain characteristics, evidenced by the spread (0.1−0.3 mag) of the temporal measurements. Nevertheless, the corrected colors are similar to those of the Cepheids, with an excess of emission that increases at longer wavelengths.

Open with DEXTER

HR 5171 A shows dust excess (Humphreys et al. 1971; Apruzese 1975; Odenwald 1986; Stickland 1985). Since the early 70 s, HR 5171 A has regularly been observed by space-borne infrared instruments, and we report these measurements in Table D.1. The aperture of these instruments is larger than several arcseconds. Since HR 5171 A cleaned up a large cavity, one can reasonably be confident that no other infrared source contributes significantly to the infrared flux. The flux from the B0Ia supergiant HR 5171 B is also negligible at these wavelengths. Some significant variability of the source is observed by comparing the MSX and WISE measurements obtained ~15 yr apart in filters with relatively close characteristics. The recent WISE fluxes are systematically lower at wavelengths shorter than 12 μm. Nonetheless, the large variations observed between instruments with filters covering the N band can be explained by the variations of the transmission that include different contributions of the very strong silicate features at 10 μm.

The JHKL colors of HR 5171 A are compared to those of YSGs, i.e. Cepheid variables in Figs. D.1. They are very similar in JHK, but HR 5171 A has distinctly redder colors in K − L, probably from the influence of circumstellar dust.

© ESO, 2014