Two new hot subdwarf binaries in the GALEX survey

We report the discovery of two new hot, hydrogen-rich subdwarfs (sdB) in close binary systems. The hot subdwarfs, GALEX J0321+4727 and GALEX J2349+3844, were selected from a joint optical-ultraviolet catalogue of hot sub-luminous stars based on GSC2.3.2 and the Galaxy Evolution Explorer all-sky survey. Using high-dispersion spectra of the Halpha core obtained using the 2m telescope at Ondrejov Observatory we measured the radial velocities of the sdB primaries and determined orbital periods of 0.26584+/-0.00004 days and 0.46249+/-0.00007 days for GALEX J0321+4727 and GALEX J2349+3844, respectively. The time series obtained from the Northern Sky Variability Survey with an effective wavelength near the R band show that GALEX J0321+4727 is a variable star (Delta m=0.12 mag) while no significant variations are observed in GALEX J2349+3844. The period of variations in GALEX J0321+4727 coincides with the orbital period and the variability is probably caused by a reflection effect on a late-type secondary star. Lack of photometric variations in GALEX J2349+3844 probably indicates that the companion is a white dwarf star. Using all available photometry and spectroscopy, we measured the atmospheric properties of the two sdB stars and placed limits on the mass and luminosity of the companion stars.


INTRODUCTION
Subdwarf B (sdB) stars are helium core burning stars with very thin hydrogen envelopes that lie at the blue end of the horizontal branch and hence are identified with the extreme horizontal branch (EHB) stars (see a recent review by Heber 2009). D' Cruz et al. (1996) showed that a high massloss rate on the red giant branch produces a thin hydrogen envelope and prevents the star from ascending the asymptotic giant branch. The evolution of single EHB stars from zero-age to helium exhaustion may be followed on a series of tracks narrowly centred on 0.475 M⊙ (Dorman et al. 1993). After helium exhaustion these objects evolve directly onto the white dwarf sequence.
On the other hand, following the original proposal of Mengel et al. (1976) for the formation of sdB stars through binary evolution, it has been found that a significant fraction of these stars reside in close binary systems (e.g., ⋆ E-mail: kawka@sunstel.asu.cas.cz (AK); vennes@sunstel.asu. cas.cz (SV); pnemeth@fit.edu (PN); kraus@sunstel.asu.cas.cz (MK); kubat@sunstel.asu.cas.cz (JK) Maxted et al. 2001;Morales-Rueda et al. 2003). The onset of a common envelope phase or Roche lobe overflow contributes to the removal of the hydrogen envelope and directs the star toward the EHB. Han et al. (2002Han et al. ( , 2003 propose three formation channels for the formation of sdB stars through binary interaction, either involving common envelope (CE) phases, episodes of Roche lobe overflow (RLOF), or the merger of two helium white dwarfs. The CE scenario involving primary stars that experience a helium flash accompanied by a low-mass or white dwarf secondary star is expected to create short-period binaries (log P (d) ≈ −1 to 1) and a final primary mass distribution narrowly centred on 0.46M⊙. The CE scenario with primary stars massive enough to avoid a helium flash is expected to achieve a much lower final mass for the primary (0.33-0.35 M⊙). On the other hand, the RLOF scenario creates longer period binaries and a wider distribution of primary final masses. Studies of the binary components, and an estimate of the frequency of such systems are required to constrain these models and determine the relative contribution of these formation channels to the sdB population.
In this context, we have initiated a program to identify c 2010 RAS 2), to build a list of blue stellar candidates, while follow-up spectroscopic measurements further constrained the properties of the candidates. In Section 2, we present available spectroscopic and photometric measurements of two new, hot sdB stars. Section 3 presents our analysis of the radial velocity measurements, and in Section 4 we constrain the properties of the binary components. We summarise and conclude in Section 5.

OBSERVATIONS
The ultraviolet sources GALEX J234947.7+384440 (hereafter GALEX J2349+3844) and GALEX J032139.8+472716 (hereafter GALEX J0321+4727) were originally selected based on the colour index N U V − V < 0.5 and the brightness limit N U V < 14, where N U V is the GALEX near ultraviolet bandpass, and V is the GSC photographic magnitude. The sources were also identified with optical counterparts in the Tycho-2 catalogue (Høg et al. 2000) and the Third U.S. Naval Observatory CCD Astrograph Catalog (UCAC3, Zacharias et al. 2010

Photometry and astrometry
We extracted the ultraviolet photometry from the GALEX all-sky survey using CasJobs at the Multimission Archive at STScI (MAST). GALEX obtained photometric measurements in the FUV and NUV bands with effective wavelengths of 1528Å and 2271Å, respectively. We corrected the photometry for non-linearity (Morrissey et al. 2007). Although the statistical errors are small (< 0.02 mag) we estimate the total errors to be 0.2 mag because of large uncertainties in the linearity corrections for bright sources (see Morrissey et al. 2007). In the present case, the N U V measurements are more reliable than the F U V measurements. We also obtained infrared photometry from the Two Micron All Sky Survey (2MASS, Skrutskie et al. 2006) and optical photometry from the Tycho catalogue 1 . We transformed the Tycho BT and VT photometric magnitudes to the Johnson B and V magnitudes using the recommended transformation equations (Perryman 1997). Table 1 lists the available photometry for the two sources and astrometric measurements from Tycho-2 and UCAC3. The tabulated coordinates (epoch and equinox 2000) are the averages of the Tycho-2 and UCAC3 coordinates. Both optical counterparts lie within ∼ 3 ′′ of the ultraviolet sources.
Finally, we have extracted photometry from the Northern Sky Variability Survey (NSVS). The photometric bandpass is very broad, ranging from 4500 to 10000Å, with an effective wavelength close to the Johnson R band (Woźniak 2004). The modified julian dates supplied by NRVS were converted to the barycentric julian dates. The time series comprise 173 and 240 good measurements for GALEX J0321+4727 and GALEX J2349+3844, respectively, and allow the examination of our objects for variability. We also obtained a photometric series (167 good measurements) of the known eclipsing and variable sdB+dM binary 2M 1533+3759 ) to test our methodology.

Spectroscopy
We observed GALEX J2349+3844 and GALEX J0321+4727 using the spectrograph at the coudé focus of the 2m telescope at Ondřejov Observatory (Šlechta &Škoda 2002). We obtained the spectroscopic series using the 830.77 lines per mm grating with a SITe 2030 × 800 CCD that delivered a spectral resolution R = 13 000 and a spectral range from 6254Å to 6763Å. The exposure time for both targets is 30 minutes, with each exposure immediately followed by a ThAr comparison arc. The fast rotating B star HR 7880 was observed each night to help remove telluric features from the spectra. We verified the stability of the wavelength scale by measuring the wavelength centroids of Oi sky lines. The velocity scale remained stable within 1 km s −1 .
We also obtained two low dispersion spectra of GALEX J0321+4727 using the R-C spectrograph attached to the 4m telescope at Kitt Peak National Observatory (KPNO) on UT 2010 March 23. We used the KPC-10A grating (316 lines/mm) with the WG360 order blocking filter. The slitwidth was set to 1.5 arcseconds to provide a resolution of FWHM = 5.5Å. A HeNeAr comparison spectrum was obtained following the target spectrum. We exposed GALEX J0321+4727 for 60 and 180 s and we co-added the spectra weighted by the exposure times. All spectra were reduced using standard procedures within IRAF.

Radial velocity variations
We measured the radial velocities by fitting a Gaussian function to the Hα core and by measuring the shifts relative to the rest wavelength. The shifts were then converted into radial velocities and adjusted to the solar system barycentre. Tables 2 and 3 lists the barycentric julian dates, radial velocities and the spectra signal-to-noise ratios for GALEX J0321+4727 and GALEX J2349+3844, respectively. The accuracy of individual measurements varied from 1 km s −1 in high signal-to-noise spectra to 10 km s −1 in lower quality spectra. The orbital parameters were determined by fitting to the velocity series a sinusoidal function of the form  where P is the period, γ is the systemic velocity, K is the velocity semi-amplitude, and T0 is the initial epoch. The initial epoch T0 corresponds to the inferior conjunction of the sdB (Φ = 0). We applied a χ 2 minimisation technique with each velocity measurement weighted proportionally to the signal-to-noise ratio achieved in the corresponding spectrum. Figures 1 and 2 show the periodograms and best-fit radial velocity curves for GALEX J0321+4727 and GALEX J2349+3844, respectively. The residual to the best-fit radial curve is ≈ 6 km s −1 for both data sets. Table 4 lists the corresponding binary parameters and the calculated mass functions. The measured radial velocities were also employed to apply doppler corrections to individual spectra and build phase-averaged spectra for each star (Section 4).

Photometric variations
We investigated possible variability in GALEX J0321+4727 and GALEX J2349+3844 using the NSVS light curves. We analyse the light curves using a Lomb periodogram for unevenly sampled time series (Press et al. 1992). The power spectra (Fig. 3) show a peak signal at a frequency close to the orbital period in GALEX J0321+4727 (Table 4) and 2M 1533+3759 ), but not in GALEX J2349+3844 (Table 4). We estimated the probability of a given frequency relative to the probability of the peak frequency following Press et al. (1992) and determined the 1σ (66%) error bars on the period of the photometric variations for GALEX J0321+4727: P = 0.26586 ± 0.00003 d, and 2M 1533+3759: For both stars, the period of photometric variations is equal, within error bars, to the measured orbital period, thereby validating the method. Figure 4 shows the NSVS light curves for GALEX J0321+4727 and GALEX J2349+3844 folded on the orbital period with an arbitrary phase adjustment so that Φ = 0 corresponds to the inferior conjunction of the primary star (sdB). The distant NSVS epoch ( 1999) precluded phasing with the current orbital ephemeris (Table 4). The light curve of GALEX J0321+4727 is fitted with the sine curve: m = (12.034 ± 0.003) + (0.061 ± 0.004) sin 2πΦ, that we interpret as a reflection effect on a latetype secondary star (Section 4). The variations in GALEX J2349+3844 are not significant with a mean magnitude of < m >= 12.281 ± 0.003 and a semi-amplitude of ∆m/2 = 0.009 ± 0.004.

PROPERTIES OF THE COMPONENTS
Our analysis of the spectroscopic observations of the primary stars is based on a grid of non-LTE models and  synthetic spectra calculated using TLUSTY/SYNSPEC . The grid covers the effective temperature from T eff = 21000 to 35000 K (in steps of 1000 K), the surface gravity from log g = 4.75 to 6.25 (in steps of 0.25), and the helium abundance from log (nHe/nH) = −4.0 to −1.0 (in steps of 0.5). The neutral hydrogen and helium atoms include 9 (n 9) and 24 (n 8) energy levels, respectively, and the ionized helium atom includes 20 (n 20) energy levels. Figure 5 shows a preliminary analysis of the atmospheric properties of the two sdB stars using their observed spectral energy distribution (SED) from the infrared to the ultraviolet, and using representative sdB models at T eff = 29000 K, log g = 5.5, and log (nHe/nH) = −2.65 (GALEX J0321+4727) and −3.25 (GALEX J2349+3844). The model corresponds to a hot sdB star with absolute magnitude MV = 4.1 and a canonical mass of 0.47 M⊙ (see Dorman et al. 1993). We corrected the model spectra for interstellar extinction using a variable extinction coefficient E(B − V ) and a parametrised extinction law (R = 3.2, Cardelli, Clayton & Mathis 1989). GALEX J0321+4727 lies close to the plane of the Galaxy (l = 147.5, b = −8.1) and the total extinction in the line of sight is E(B − V ) = 0.61 (Schlegel, Finkbeiner, & Davis 1998). Excluding the FUV band, the observed SED is well matched by the model assuming E(B − V ) = 0.23. A much higher coefficient (≈ 0.4) is required to match the FUV band although the accuracy of the FUV photometry is most probably affected by nonlinearity (see Section 2.1). GALEX J2349+3844 is located lower below the plane (l = 110.0, b = −22.6) and the total extinction is lower E(B − V ) = 0.17. In this case, the SED is well matched assuming E(B − V ) = 0.13. These estimates appear reasonable if we locate both stars at a distance of ∼ 330 pc by assuming MV ∼ 4.1 and V = 11.7 for both stars. Taking the scale height for dust in the Galactic plane as h ≈ 150 pc, the total distance accross the dust layer is ≈ 1060 and 390 pc at |b| = 8.1 and 22.6 • , respectively, so that the path toward GALEX J0321+4727 covers ∼ 31% of the total distance and ∼ 85% for GALEX J2349+3844. According to this simple calculation, the scaled E(B − V ) indices are predicted to be 0.19 and 0.15 for GALEX J0321+4727 and GALEX J2349+3844, respectively, and are similar to our estimates based on the SED.

Spectral energy distribution
Taking into account the effect of interstellar reddening, the entire spectral energy distribution of both systems is dominated by the sdB stars.

Line profile analysis
We measured the sdB effective temperature (T eff ), surface gravity (log g), and helium abundance (log (nHe/nH)) by fitting the observed line profiles with our grid of model spectra. We employed χ 2 minimisation techniques to find the best-fit parameters and draw contours at 66, 90, and 99% significance. The model spectra were convolved with Gaussian profiles with a F W HM = 5.5Å for the analysis of the KPNO spectra, while we adopted a F W HM = 0.66Å that includes the effect of orbital smearing for the Ondřejov coudé spectra.
First we analysed the KPNO spectrum of GALEX J0321+4727 (Fig. 6). We included in the analysis the Balmer line spectrum from Hα to H11 and five blue Hei lines normally dominant in sdB stars. We repeated the analysis with Hα excluded and obtained the same atmospheric parameters as the analysis that included Hα. The mid-exposure time of the co-added KPNO spectrum is BJD 2455278.59501 corresponding to an orbital phase Φ = 0.51 ± 0.03 or close to the inferior conjunction of the secondary star. This phase also corresponds to minimum contamination to the sdB spectrum due to the reflection effect.
Next, we analysed phase-resolved Hα spectra of GALEX J0321+4727 to take into account the light contamination due to the reflection effect on the temperature measurements. This effect was notable in the analysis of the similar systems such as HS2333+3927 (Heber et al. 2004) and 2M 1533+3759 For et al. (2010. Variations of ≈ 6000 K were observed in HS2333+3927, while weaker variations of ≈ 1000 K were observed in 2M 1533+3759. To investigate this effect in GALEX J0321+4727 we built three spectra inclusive of phases 0.0 − 1.0 (average), 0.35 − 0.65 (minimum reflection effect), and 0.85 − 0.15 (maximum reflection effect). Figure 7 shows our analysis of the Hα and He Iλ6678 A line profiles in the co-added (0.0 − 1.0) coudé spectra of GALEX J0321+4727 and GALEX J2349+3844. Table 5 summarises our measurements of the hot subdwarf atmospheric parameters. Our measurements show that variability in GALEX J0321+4727 is affecting the temperature and abundance measurements but that the surface gravity does not vary significantly. In our limited phase resolution, the peak-to-peak temperature variation reaches ≈ 4000 K. The Ondřejov coudé and KPNO temperature measurements at phase 0.5 agree but the surface gravity measurements differ by 0.5 dex. This is most likely due to a systematic effect in the model spectra themselves with the Hα line profile analysis overestimating the surface gravity relative to an analysis involving the complete series. The shape and strength of the upper Balmer lines are very sensitive to surface gravity and offer a more reliable surface gravity diagnostics. In summary we estimate the parameters representative of the sdB GALEX J0321+4727 at phase 0.5. The effective temperature and helium abundance were es-  timated by taking the weighted average of the results from the Hα-H11 and Hα (0.35 -0.65) fits and for the surface gravity we adopted the result from the Hα-H11 analysis: T eff = 29200 ± 300, log g = 5.5 ± 0.1, and log nHe/nH = −2.6 ± 0.1. Applying a correction of −0.5 ± 0.2 to the surface gravity measurement of GALEX J2349+3844 that is based on Hα alone we conservatively estimate the sdB parameters: T eff = 28400 ± 400, log g = 5.4 ± 0.3, and log nHe/nH = −3.2 ± 0.1.
Our analysis assumes a hydrogen/helium composition and the inclusion of heavy elements in the model atmospheres is likely to affect our results (see Heber et al. 2000;Edelmann et al. 2003). A self consistent analysis including abundance measurements (e.g., O'Toole & Heber 2006;Ahmad et al. 2007) awaits high signal-to-noise and resolution ultraviolet and optical spectroscopy.

Nature of the companions
Using the mass function (Table 4) and assuming a mass of 0.5 M⊙ for the hot subdwarf GALEX J0321+472, we calculate a minimum secondary mass of 0.13 M⊙ that corresponds to a spectral type of M5 (Kirkpatrick & McCarthy 1994). Assuming an absolute J magnitude of 3.9 for the hot subdwarf, a M5 star with MJ = 8.8 would be outshone by the hot subdwarf.
The NSVS lightcurve shows GALEX J0321+4727 to be variable and the search for a period in the photometric data resulted in a best-period corresponding to the orbital period. The observed variations are most probably caused by irradiation of the atmosphere of the cool companion by the hot subdwarf. Using the observed semi-amplitude of 0.061 mag we may constrain the binary parameters further. We estimated the system inclination and secondary mass using the reflection model of Maxted et al. (2002) and assuming two different masses for the sdB star. For a sdB mass of 0.4 M⊙, the inclination is predicted to be between 63 • and 71 • and the secondary mass between 0.124 M⊙ and 0.133 M⊙. For a sdB mass of 0.5 M⊙, we obtain an inclination ranging from 65 • to 70 • , and a secondary mass between 0.143 M⊙ and 0.149 M⊙.
Again, using the mass function and assuming a mass of 0.5 M⊙ for the hot subdwarf GALEX J2349+3844, the minimum secondary mass is 0.27 M⊙ that corresponds to a spectral type of M4 (Kirkpatrick & McCarthy 1994). Assuming an absolute J magnitude of 5.6 for the hot subdwarf, a M4 spectral type star with MJ = 8.6 would also be outshone by the hot subdwarf.
However, the NSVS time series do not show variations down to a limit of ∆m = 0.009. Illumination of a 0.3 M⊙ star, which is the suggested mass at a high inclination, would cause a variation of ∆m ∼ 0.4 magnitudes. Lower inclinations would require larger companions causing even larger variations that are incompatible with the observations. The lack of variability suggests that the companion is most likely a white dwarf (see Maxted, Morales-Rueda, & Marsh 2004).

We
show that GALEX J0321+4727 and GALEX J2349+3844 are hot hydrogen-rich subdwarfs in close binaries. Based on a preliminary analysis of periodic light variations in GALEX J0321+4727 we infer that its companion is a low-mass star (M ∼ 0.13 M⊙). The secondary star in GALEX J2349+3844 is probably a white dwarf with M 0.3 M⊙. The two new systems are post-CE systems with a hot subdwarf primary. Their orbital periods locate them close to the peak of the period distribution for such systems (see Heber 2009). A future study of GALEX J0321+4727 will involve phase-resolved high signal-to-noise ratio spectroscopic and photometric observations aimed at resolving the nature of the companion. Searches for close binaries in the sdB population have a relatively high yield (69%, see Maxted et al. 2001), and, therefore, we expect that many new systems remain to be discovered in our GALEX/GSC catalogue of EHB stars.

ACKNOWLEDGEMENTS
S.V. and A.K. are supported by GA AV grant numbers IAA300030908 and IAA301630901, respectively, and by GAČR grant number P209/10/0967. A.K. also acknowledges support from the Centre for Theoretical Astrophysics (LC06014). Some of the data presented in this paper were obtained from the Multimission Archive at the Space Telescope Science Institute (MAST). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX09AF08G and by other grants and contracts.