EDP Sciences
Free Access
Issue
A&A
Volume 556, August 2013
Article Number A45
Number of page(s) 4
Section Stellar structure and evolution
DOI https://doi.org/10.1051/0004-6361/201321699
Published online 23 July 2013

© ESO, 2013

1. Introduction

The source φ Phe (HIP 8882, HD 11753, HR 558) is a fifth magnitude late-type B star whose chemical peculiarity was first noticed by Hyland (Dworetsky 1969). Its MK classification is B9pHgMn (Buscombe 1984).

It was suggested very early on (Campbell & Moore 1928) that the radial velocity of φ Phe was varying. However, that claim was only confirmed by Dworetsky et al. (1982) on a dataset too small to yield any orbit, not even the period. Leone & Catanzaro (1999) completed the dataset and derived the first spectroscopic orbit (e = 0.32, P = 41.489 days). Grenier et al. (1999) also complemented the dataset with one more radial velocity, but did not derive any orbit. In the original reduction of the Hipparcos data (ESA 1997), a circular nearly edge-on 878-day orbit was derived from the sole astrometric observations. The system φ Phe thus made it to the triple-star class (Tokovinin 2008) despite the excellent astrometric fit (F2 = 0.10), leaving no room for any detectable wobble caused by the inner component.

Even though the spectroscopic orbit has lately been substantially revised (Korhonen et al. 2013) with a period now reaching sightly more than three years, the multiplicity of the system was not questioned. Using the Hipparcos data, we show that φ Phoenici is indeed a genuine binary (Sect. 3), thus allowing the system to be fully characterised (Sect. 4).

Table 1

BESO radial velocities.

Table 2

Astrometric orbits based on the original Hipparcos data with and without spectroscopic help.

2. Revised spectroscopic orbit

Using about 150 Coralie spectra taken in 2000, 2001, 2009, and 2010 together with a few FEROS and HARPS data points in addition to the older radial velocities, Korhonen et al. (2013) have recently drastically changed the orbit. Instead of a month-long period, they obtained a three-year orbit. Although these new data strongly constrain the top of the velocity curve, its bottom (and thus partly its semi-amplitude) is still essentially set by the sole six additional data from Leone & Catanzaro (1999).

The Bochum Echelle Spectroscopic Observer (BESO) is a fibre-fed spectrograph based on slightly improved blueprints of the European Southern Observatory FEROS instrument (located at La Silla). It belongs to the Astronomical Institute of the Ruhr-Universität Bochum (AIRUB) and is located at Cerro Armazones in the Atacama desert in Chile through a partnership between AIRUB and the Universidad Católica del Norte in Antofagasta. BESO is attached to the 1.5-m Hexapod Telescope. It operates in the 370–860 nm wavelength range with an average resolution of 48 000, varying from order to order between 43 000 and 60 000 (see Fuhrmann et al. 2011, for more details about BESO).

For purely astrometric reasons, φ Phe has been monitored by BESO since late 2011. The initial goal of that monitoring was to secure the spectroscopic orbit of the inner components (even the period was still uncertain after Leone & Catanzaro 1999), which could then be used to improve the Hipparcos solution. Seventeen spectra were obtained between November 2011 and October 2012, yielding the radial velocities listed in Table 1. The radial velocities were determined both by fitting some of the strongest lines and by cross-correlation using the xcsao task in IRAF. Depending on the quality of the spectrum – the signal-to-noise ratio (S/N) varies between 30 to 70 – the formal errors of the cross-correlation are between 0.3 and 1 km s-1. The real error is higher because it combines systematics from the wavelength calibration and intrinsic variability of the object, as reported in Korhonen et al. (2013). The resulting rms for these data alone is 1.92 km s-1, very likely caused by the BESO calibration itself, which was reported to have had some problems during our run. At the time our investigation on φ Phe was initiated, the results from Korhonen et al. had not been released and the adopted S/N was thought to be enough to assess the 41-day orbit. However, unlike all other existing datasets, this one covers four consecutive months (because only one measurement was taken in November ’11 before φ Phe disappeared for six months).

Despite the extended phase coverage, the precision of these BESO radial velocities is not good enough to substantially refine the orbit from Korhonen et al. (2013). However, given their continuity, our data points definitively exclude a 40-day period orbit as proposed by Leone & Catanzaro (1999). The orbit from Korhonen et al. is plotted in the left panel of Fig. 1. The scatter of both the BESO data and the old velocities from Campbell & Moore (1928) restrict the usefulness of these sets, especially that of the latter as far as constraining the period is concerned.

thumbnail Fig. 1

Left panel: spectroscopic orbit with data from Leone & Catanzaro (1999), Grenier et al. (1999), Korhonen et al. (2013) (all as filled disks), BESO (open triangles), and Campbell & Moore (1928) as filled diabolos. Right panel: new astrometric orbit based on the original Hipparcos observations.

Open with DEXTER

3. Astrometric counterpart

The original choice of adopting a circular orbit to fit the Hipparcos data of this object was made for the sake of simplification. Indeed, it means that not only the eccentricity, but also the argument of the periastron (ω) can be set to 0. That orbit is listed in the first column of Table 2. Strangely enough, the very same orbit was also adopted in the new reduction of the Hipparcos observations by van Leeuwen (2007) despite the seemingly poor fit resulting from that choice: 0.1 versus 12.34 for the goodness of fit (Wilson & Hilferty’s cube root transformation, which follows a N(0,1) distribution Stuart & Ord 1994). However, in this second reduction of HIP 8882, there was no new orbit fitting per se, the original orbit was simply adopted.

In their general attempt to reprocess the Hipparcos observations of the spectroscopic systems listed in (Pourbaix et al. 2004), Jancart et al. (2005) concluded that the orbit derived in Leone & Catanzaro (1999) was not present in the astrometric data. Neither the spectroscopic orbit nor the astrometric one were ever questioned although there was no hint of the latter in the radial velocities.

The astrometric orbits resulting from both the Campbell and the Thiele-Innes approach are listed in Table 2. The assessment of the spectro-astrometric combinations was extensively described and improved in several papers (Pourbaix & Arenou 2001; Pourbaix & Boffin 2003; Jancart et al. 2005). It essentially consists in fitting the astrometric data with two different approaches (freezing some parameters in one while leaving them free in the other) and in evaluating the consistency between these two solutions through several statistical tests. In both cases, the eccentricity, period, and periastron time are adopted from the spectroscopic orbit. In Campbell’s approach, the amplitude of the radial velocity curve and the argument of the periastron are also adopted from the spectroscopic side. In the alternative, the four Thiele-Innes constants are left free. In no case, the radial velocities were fitted, therefore the K1 listed in the last column in Table 2 was derived from the astrometric solution alone. If both solutions are consistent (which is the case here), Campbell’s solution should be favoured because it results from the larger number of degrees of freedom.

All assessment indicators are listed in Table 3 and confirm the excellent agreement between the spectroscopic orbit and its astrometric counterpart. The goodness of fit is significantly reduced with the eccentric orbit with respect to the original circular one. Despite the change of the eccentricity and period, the parallax, and the proper motion, the last three lines of Table 2 remain unchanged (within the error bars).

Table 3

Assessment of the spectroscopic orbit in the Hipparcos observations.

Although the Thiele-Innes solution fits the Hipparcos data very well, the resulting K1 is slightly higher than the spectroscopic value of Korhonen et al. (2013), i.e. 9.21 km s-1. An astrometric value lower than its spectroscopic counterpart is often a sign that the assumption about the absence of light from the secondary is questionable (very unlikely, as we will see at the end of Sect. 4). Could the spectroscopic value be underestimated in the case of HIP 8882? This cannot be ruled out because the bottom of the velocity curve was constrained by the six points over five consecutive days from Leone & Catanzaro (1999), with the observed velocity ranging from –4.4 to –3.0 km s-1. The next minimum of the radial velocity will take place on JD 2 457 130 (2015.290).

For the sake of completeness, we fitted the data from the second reduction of the Hipparcos observations (van Leeuwen 2007). The resulting a0 is much too small (4.20 ± 0.33 mas) and, even though the goodness of fit decreases from 12.34 to 5.39, all statistical indicators are made useless by the poor weighting scheme adopted in that reduction. It is therefore safer not to use it, at least for binaries.

Although a perfectly edge-on orbit would be consistent with the astrometric solution, no sign of any eclipse was ever reported. The Hipparcos photometry (ESA 1997) shows a constant brightness during the whole mission. The difference between the 5th and 95th percentiles is 0.02 Hp mag, and there is no sign of brightness decrease at the epoch of conjunction.

thumbnail Fig. 2

Left panel: four evolutionary tracks from Bertelli et al. (2009). Right panel: log age(yr) = 8.41-isochrone with same Y and Z as in the best evolutionary track.

Open with DEXTER

Finally, the revision of the orbit leads to a significant change in the proper motion with respect to the Hipparcos one, especially along the declination axis; it becomes much closer to the Tycho-2 value (Høg et al. 2000). Unfortunately, the right-ascension component moves in the opposite direction. However, the UCAC4 (Zacharias et al. 2013) proper motion (−33.9 ± 1.0; −28.2 ± 1.0) mas yr-1 confirms neither the Tycho-2 value nor ours, illustrating once more how poorly known this bright system is.

4. Astrophysical outcome

In addition to the genuine binary nature of φ Phe, our new combined astrometric-spectroscopic solution confirms the parallax of the system and the orbital inclination, which can now be combined with the mass function to derive the mass of the secondary.

Dolk et al. (2003) reported Teff = 10   612 ± 200 K and log g = 3.79 ± 0.10 together with [Fe/H]  = 0.15 from Smith & Dworetsky (1993). On the grid of stellar evolutionary tracks from Bertelli et al. (2009), the three-solar-mass track with Y = 0.3 and Z = 0.017 gives the best match as far as log g and Teff are concerned (left panel of Fig. 2). According to Hakkila et al. (1997), the interstellar extinction AV = 0.03 ± 0.17, which coupled to V = 5.112, yields the absolute magnitude MV = 0.26 ± 0.13. The point (Teff,MV) agrees excellently with the log age(yr) = 8.41-isochrone from Bertelli et al. (2009) as illustrated in the right panel of Fig. 2.

From the spectroscopy, f(m) = 0.048 ± 0.0015 M, which, combined with i = 93° (astrometry) and M1 = 3.0 ± 0.12 M (evolutionary track), leads to M2 = 0.91 ± 0.025 M. Using one solar mass as initial guess for M2, the dynamical parallax method (Binnendijk 1960) yields M2 = 0.9045 M after a few iterations, very consistent with our track estimate. On the log age(yr) = 8.41-isochrone (Fig. 2), such a mass corresponds to the big point on the right panel. The magnitudes of the two components differ by 5.7 in V and 3.9 in K. Such a large difference in V explains the absence of the signal in the Hipparcos data and in the spectra. The semi-major axis of the relative visual orbit is 36.3 mas only, making this system essentially inaccessible to VLT/NACO (Schöller et al. 2010). The two components are nevertheless bright enough to be resolved with the VLTI.

Even though with Δm = 5.7 in V some hints of the secondary peak might be detectable in the FEROS-based cross correlation function, a thorough inspection of the publicly available spectra has not revealed anything. Should eclipses occur (JD 2 457 078, i.e. 2015.148), they would yield a change in magnitude of 0.010 and 0.057 in V and K, respectively, so the constant brightness reported by Hipparcos cannot be used to rule out the possibility of eclipses. These eclipses would also constrain the orbit to be edge-on within 0.0001°.

5. Conclusion

It is very disappointing to note that although φ Phe belongs to the Bright Star Catalogue, in which it was noted for its chemical peculiarities, one had to wait until 2013 to have the first plausible characterisation of the components of this binary system.

Out of the 235 orbital solutions published in Hipparcos (ESA 1997), 115 were derived without using any ground-based solution, 70 of which had their eccentricity and periastron angle arbitrarily set to 0 as for φ Phe. How many other systems deserve some revision?

Acknowledgments

We thank the referee, A. Tokovinin, for his useful comments and J. F. González and M. Briquet for sending the radial velocities from Korhonen et al. (2013) prior to their upload to the CDS. This publication is supported as a project of the Nordrhein-Westfälische Akademie der Wissenschaften und der Künste in the framework of the academy program by the Federal Republic of Germany and the state Nordrhein-Westfalen. This research has made use of the Simbad data base, operating at CDS, Strasbourg, France.

References

All Tables

Table 1

BESO radial velocities.

Table 2

Astrometric orbits based on the original Hipparcos data with and without spectroscopic help.

Table 3

Assessment of the spectroscopic orbit in the Hipparcos observations.

All Figures

thumbnail Fig. 1

Left panel: spectroscopic orbit with data from Leone & Catanzaro (1999), Grenier et al. (1999), Korhonen et al. (2013) (all as filled disks), BESO (open triangles), and Campbell & Moore (1928) as filled diabolos. Right panel: new astrometric orbit based on the original Hipparcos observations.

Open with DEXTER
In the text
thumbnail Fig. 2

Left panel: four evolutionary tracks from Bertelli et al. (2009). Right panel: log age(yr) = 8.41-isochrone with same Y and Z as in the best evolutionary track.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.