The first pre-supersoft X-ray binary
S. G. Parsons, M. R. Schreiber, B. T. Gansicke, A. Rebassa-Mansergas, R. Brahm, M. Zorotovic, O. Toloza, A. F. Pala, C. Tappert, A. Bayo, A. Jordan
aa r X i v : . [ a s t r o - ph . S R ] J un Mon. Not. R. Astron. Soc. , 1–10 (2015) Printed 9 October 2018 (MN L A TEX style file v2.2)
The first pre-supersoft X-ray binary
S. G. Parsons ⋆ , M. R. Schreiber , , B. T. G¨ansicke , A. Rebassa-Mansergas ,R. Brahm , , M. Zorotovic , O. Toloza , A. F. Pala , C. Tappert , A. Bayo , and A. Jord´an , Instituto de F´ısica y Astronom´ıa, Universidad de Valpara´ıso, Avenida Gran Bretana 1111, Valpara´ıso, Chile Millenium Nucleus ”Protoplanetary Disks in ALMA Early Science”, Universidad de Valparaiso, Valparaiso 2360102, Chile Department of Physics, University of Warwick, Coventry CV4 7AL, UK Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China Instituto de Astrofsica, Facultad de Fsica, Pontificia Universidad Cat´olica de Chile, Av. Vicu˜na Mackenna 4860, 7820436 Macul, Santiago, Chile Millennium Institute of Astrophysics, Av. Vicu˜na Mackenna 4860, 7820436 Macul, Santiago, Chile
Accepted 2015 June 22. Received 2015 June 15; in original form 2015 March 25
ABSTRACT
We report the discovery of an extremely close white dwarf plus F dwarf main-sequencestar in a 12 hour binary identified by combining data from the RAdial Velocity Experi-ment (RAVE) survey and the Galaxy Evolution Explorer (GALEX) survey. A combination ofspectral energy distribution fitting and optical and Hubble Space Telescope ultraviolet spec-troscopy allowed us to place fairly precise constraints on the physical parameters of the bi-nary. The system, TYC 6760-497-1, consists of a hot T eff ∼ , K , M WD ∼ . ⊙ whitedwarf and an F8 star ( M MS ∼ . ⊙ , R MS ∼ . ⊙ ) seen at a low inclination ( i ∼ ◦ ).The system is likely the descendent of a binary that contained the F star and a ∼ M ⊙ A-typestar that filled its Roche-lobe on the thermally pulsating asymptotic giant branch, initiating acommon envelope phase. The F star is extremely close to Roche-lobe filling and there is likelyto be a short phase of thermal timescale mass-transfer onto the white dwarf during which sta-ble hydrogen burning occurs. During this phase it will grow in mass by up to 20 per cent, untilthe mass ratio reaches close to unity, at which point it will appear as a standard cataclysmicvariable star. Therefore, TYC 6760-497-1 is the first known progenitor of a super-soft sourcesystem, but will not undergo a supernova Ia explosion. Once an accurate distance to the sys-tem is determined by Gaia, we will be able to place very tight constraints on the stellar andbinary parameters.
Key words: binaries: close – stars: white dwarfs – stars: early-type – stars: evolution
The unique properties of Type Ia Supernovae (SN Ia) as distanceindicators, sufficiently bright to serve as yardsticks on cosmolog-ical scales (e.g. Branch & Tammann 1992), has resulted in thembecoming some of the most important objects in the Universe, hav-ing led to the discovery of its accelerating expansion (Riess et al.1998; Perlmutter et al. 1999).Although it is generally accepted that SN Ia are related to thethermonuclear ignition of a white dwarf that surpassed the Chan-drasekhar mass limit, there is no consensus yet on the pathwaysleading to the explosion. The two main formation channels arethought to be the “single-degenerate” channel, where a white dwarfaccretes material from a non-degenerate companion via Roche-lobe overflow (Whelan & Iben 1973) during a phase of thermal ⋆ [email protected] time-scale mass transfer known also as the super soft source (SSS)phase, and the “double-degenerate” channel in which the explosionis the result of the merger of two white dwarfs (Webbink 1984).In recent years the idea of a “double-detonation” scenario has alsobeen advanced. In this case a layer of helium on the surface of thewhite dwarf detonates, triggering the explosion of the underlyingcore either directly, or via a compressional shock wave (Fink et al.2007, 2010; Shen & Moore 2014), hence it is possiblevia this chan-nel to detonate the white dwarf while its mass is still below theChandrasekhar limit. In theory these “sub-Chandrasekhar SN Ia”are possible via both main channels. In the single degenerate chan-nel the white dwarf accretes helium-rich material from a donorstar, building up a large surface layer (Tutukov & Yungelson 1996),whilst the merger of a carbon-oxygen core white dwarf with a low-mass helium core white dwarf can lead to helium ignition and hencea supernova via the double degenerate channel (Fink et al. 2007).The search for Galactic SN Ia progenitors from either of the c (cid:13) S. G. Parsons et al. channels is a difficult task. In the single degenerate case very fewSSS systems sufficiently nearby for detailed parameter studies areknown (Greiner 2000). This is due to a combination of the shortduration of the SSS phase and because their very soft X-ray emis-sion is easily absorbed by neutral hydrogen in the galactic plane.Whilst double white dwarf binaries are intrinsically faint objectsand difficult to distinguish from single white dwarfs without dedi-cated spectroscopic monitoring (see Napiwotzki et al. 2003, for ex-ample). However, in both cases the progenitor systems are the de-scendents of detached white dwarf plus F, G or early K type main-sequence star companions. With no ongoing accretion these sys-tems are easy to characterise. Furthermore, they are expected to benumerous (Holberg et al. 2013) and so are optimal for populationstudies and hence testing both SN Ia progenitor channels.However, identifying white dwarfs with early type compan-ions has been extremely difficult until recently. This is due to thefact that the main sequence star completely outshines the whitedwarf at optical wavelengths (Rebassa-Mansergas et al. 2012a),hence a combination of optical and ultraviolet (UV) coverage isrequired to find these systems. Maxted et al. (2009) attempted todetect systems of this type by combining data from the GalaxyEvolution Explorer (GALEX) survey and the Sloan Digital SkySurvey (SDSS), but were limited by the optical colour selection oftheir main-sequence stars. Whilst Burleigh et al. (1997) found onlyfour unresolved white dwarf plus FGK main-sequence star systemssearching for extreme-UV and soft X-ray excesses.We have recently begun a project combining the largedataset of the RAdial Velocity Experiment (RAVE) survey(Kordopatis et al. 2013) in the southern hemisphere and the LAM-OST (Large Sky Area Multi-Object Fiber Spectroscopic Telescope)survey in the northern hemisphere, with UV data from the GALEXsurvey in order to identify main-sequence F, G and early K starswith significant UV excesses.The RAVE survey spectroscopically observed more than400,000 bright (8 < I <
12 mag) southern hemisphere stars in thespectral region 8410-8794 ˚A (the infrared calcium triplet) with aresolution of R ∼ Table 1.
Radial velocity measurements for the main-sequence star inTYC 6760-497-1.BJD Velocity Uncertainty Telescope/(mid-exposure) ( km s − ) ( km s − ) instrument2456811.5672390 30.33 0.50 Du Pont/Echelle2456827.6293689 -48.79 0.48 MPG2.2/FEROS2456828.4841811 53.36 0.28 MPG2.2/FEROS2456828.5729529 -13.12 0.28 MPG2.2/FEROS2456828.5904822 -26.22 0.28 MPG2.2/FEROS2456828.6012976 -34.38 0.27 MPG2.2/FEROS2456828.6663049 -61.48 0.30 MPG2.2/FEROS2456828.6775301 -63.33 0.30 MPG2.2/FEROS2456828.7213267 -55.85 0.28 MPG2.2/FEROS2456829.5163187 32.99 0.29 MPG2.2/FEROS2456829.5491571 7.39 0.27 MPG2.2/FEROS2456829.5939247 -29.35 0.30 MPG2.2/FEROS2456829.6737526 -62.99 0.31 MPG2.2/FEROS2456829.6855582 -61.90 0.29 MPG2.2/FEROS2456829.6963120 -60.80 0.29 MPG2.2/FEROS2456829.7123239 -58.27 0.29 MPG2.2/FEROS2456829.7603942 -33.58 0.30 MPG2.2/FEROS2456829.7718605 -24.86 0.29 MPG2.2/FEROS2456829.7818813 -15.99 0.28 MPG2.2/FEROS2456834.6122883 -49.91 0.30 MPG2.2/FEROS2456834.6226973 -52.15 0.29 MPG2.2/FEROS2456835.7338779 -41.03 0.29 MPG2.2/FEROS and binary parameters of the systems, reconstruct its evolutionaryhistory, predict its future and discuss the implications for our un-derstanding of close compact binary star evolution. We obtained high resolution spectroscopy of TYC 6760-497-1with the echelle spectrograph (R ∼ ∼ ∼ ∼ c (cid:13) , 1–10 he close binary TYC 6760-497-1 Figure 1.
Observed FEROS spectrum (black line) of TYC 6760-497-1 withthe best fit model overplotted (red line, v rot sin i = 75km s − ). We alsoplot the model spectra without any rotational broadening (grey line) to high-light the rapid rotation of the star. A medium resolution spectrum (R ∼ µ m tothe K -band at 2.4 µ m. Our data consisted of three 60 second ex-posures, which were reduced using the standard pipeline releaseof the X-shooter Common Pipeline Library (CPL) recipes (version2.5.2). The instrumental response was removed and the spectrumflux calibrated by observing the spectrophotometric standard starCD-38 10980 and dividing it by a flux table of the same star to pro-duce the response function. The spectrum was extinction correctedbut not corrected for telluric features. TYC 6760-497-1 was observed with the Hubble Space Telescope(HST) on the 9th of January 2015 with the Space Telescope Imag-ing Spectrograph (STIS) as part of program GO 13704. We usedthe G140L grating centered on 1425 ˚A, for one spacecraft orbit, re-sulting in a total exposure time of 2381 seconds. The data wereprocessed using
CALSTIS
V3.4. We de-reddened the STIS spec-trum using a value of E ( B − V ) = 0 . , determined from ourSED fit to the main-sequence star (see Section 4.1). Radial velocities were computed from our optical echelle spec-tra using the cross-correlation technique against a binary maskrepresentative of a G2-type star. The uncertainties in radial ve-locity were computed using scaling relations (for more detail seeJord´an et al. 2014) with the signal-to-noise ratio and width of thecross-correlation peak, which were calibrated with Monte Carlosimulations. The results are listed in Table 1. We note however,that these relations were calibrated using slowly rotating stars and
Figure 2. χ plotted against orbital period for the fit to the radial velocitymeasurements. There is a clear minimum at a period of 0.5 days hence the uncertainties may be underestimated for rapidly rotatingstars, such as the one in TYC 6760-497-1.We also estimated the projected rotational velocity ( v rot sin i )of the main-sequence star in TYC 6760-497-1 by comparing theobserved FEROS spectra against a synthetic grid of stellar spec-tra (Coelho et al. 2005). The synthetic spectra were degraded tothe resolution of FEROS by convolving them with a Gaussianwith R = λ/δλ = 53000 and then further degraded to differ-ent values of v rot sin i using the limb-darkening coefficients ofClaret (2004). The optimal fit (which also yields a set of stellarparameters) was found by chi-square minimisation in 3 echelleorder of the spectra which include the zone of the magnesiumtriplet (5000-5500 ˚A). The measured rotational broadening was v rot sin i = 75 ± − . We also obtained the following stel-lar parameters: T eff = 5750 ± K, log g = 3 . ± . (in cgsunits), [Fe/H]= − . ± . . Figure 1 shows the observed spectrawith the best fitted model, as well as the same model with no ro-tational broadening. This high rotation rate means that the fittedstellar parameters from the spectral fit are not necessarily reliablesince most of the weak lines that have a log g dependence are com-pletely smeared out. Moreover, the system is extremely close toRoche-lobe filling (see Section 4) and hence the effects of gravitydarkening and Roche-distortion likely have an effect on the spectralfit. We measured the binary period of TYC 6760-497-1 by fitting a con-stant plus sine wave to the velocity measurements over a range ofperiods and computing the χ of the resulting fit. The result of thisis shown in Figure 2 and shows a clear minimum in χ at a periodof 0.5 days. We determine the ephemeris of TYC 6760-497-1 as BJD = 2456823 . . E, (1)where E is the binary phase and phase 0 occurs at the conjunctionof the main-sequence star.The phase-folded radial velocity plot is shown in Figure 3. Wefind a radial velocity semi-amplitude for the main-sequence starof . ± . − and a systemic velocity of . ± . − .The lower panel of Figure 3 shows the residuals to the fit, whichshow variations larger than their uncertainties. This is likely due to c (cid:13) , 1–10 S. G. Parsons et al.
Figure 3.
Phase-folded radial velocity plot for the main-sequence star inTYC 6760-497-1. The lower panel shows the residuals to the best fit. the main-sequence star’s large rotational broadening causing smallsystematic errors during the cross correlation process.
In this section we detail how we constrained the physical parame-ters of the binary.
In order to get an initial estimate of the physical parameters of themain-sequence star in TYC 6760-497-1 and the reddening towardsthe system we fitted its spectral energy distribution (SED) usingthe virtual observatory SED analyzer (VOSA; Bayo et al. 2008).We convolved our flux calibrated X-shooter spectrum with a num-ber of generic narrow and broad band filters including Str¨omgren uvby , Bessell
BV RI
Johnson
BV R and Cousins RI filters. Wecombined these with archival data from the GALEX (Martin et al.2005), 2MASS (Skrutskie et al. 2006) and WISE (Wright et al.2010) catalogues. We excluded the GALEX data when fittingthe SED since we expected the white dwarf to contribute a non-negligible amount of flux at these wavelengths.We used a large grid of BTSettl (Allard et al. 2012) mod-els to determine the main-sequence star’s parameters and allowedthe extinction to vary from zero to three times the maximum ex-pected value of A V = 0 . (Schlafly & Finkbeiner 2011). As ex-pected the fit was insensitive to the metallicity of the system andonly mildly sensitive to the surface gravity, preferring a value of log g ∼ . . However, we found that the temperature was wellconstrained to T eff = 6400 ± K and the extinction to beA V = 0 . , implying E ( B − V ) = 0 . , in good agreementwith what is expected from the Schlafly & Finkbeiner (2011) maps.The uncertainties were determined via a Bayes analysis (for moredetails see Bayo et al. 2008). These measurements are consistentwith those from the RAVE survey, in which TYC 6760-497-1 wasdetermined to have T eff = 6150 ± K and log g = 3 . ± . (Kordopatis et al. 2013), although these measurements are likelyaffected by the large rotational broadening, similar to our FEROS Figure 4.
The spectral energy distribution of TYC 6760-497-1 in the UVand optical wavelength ranges (the SED fit also included infrared data, butwe limit the plot to this range to demonstrate the relative contributions ofthe two stars). The black lines are the observed UV HST/STIS spectrum andoptical X-shooter spectrum (no telluric correction was applied). Broad bandphotometry is shown in red (their errors are too small to see on this scale),the best fit BTSettl main-sequence star model is shown in grey and wasrotationally broadened to match the measured value for the main-sequencestar in TYC 6760-497-1. We also plot in dark blue a TLUSTY/SYNSPECmodel for a T eff = 20400 K, log g = 8 white dwarf. The fit to the UV partof the spectrum is better illustrated in Figure 7. fit. In Figure 4 we show part of the SED in the UV and optical wave-length ranges with the best fit model and observed spectra overplot-ted. While the GALEX NUV measurement shows little evidence ofan excess, the FUV measurement is clearly dominated by the whitedwarf. Assuming that the main-sequence star is tidally locked to the whitedwarf, which it should be since the tidal synchronisation timescalefor this system is less than 0.5 Myr (Zahn 1977), much shorterthan the white dwarf’s cooling age (see Section 5.1), the rotationalbroadening measurement can be used to place some constraints onthe binary parameters via v rot sin i = K MS (1 + q ) R MS a , (2)where K MS is the radial velocity semi-amplitude of the main-sequence star, q = M MS /M WD , the mass ratio of the binary and R MS /a is the radius of the main-sequence star scaled by the orbitalseparation. Combining this with Kepler’s third law, a = P GM MS π (cid:18) q (cid:19) , (3)where P is the orbital period and G is the gravitational constant,we can effectively solve for q (hence M WD ) by assuming a massand radius for the main-sequence star. Once this is known all theother binary parameters can then be determined. For example, theorbital inclination via sin i = P K (1 + q ) πGM WD . (4)Ideally the fit to the FEROS spectra would yield the main- c (cid:13) , 1–10 he close binary TYC 6760-497-1 Figure 5.
Limits on the physical parameters of the main-sequence star. Themeasured rotational broadening, period and radial velocity amplitude ex-clude the light grey areas as at low surface gravities the star fills its Rochelobe, whilst high surface gravities lead to unphysical results. The horizonalblue lines indicate the possible temperature range from the SED fit. Thedark grey hatched area indicates the parameter range that is consistent withthe fit to the white dwarf’s spectrum (see Section 4.3). The small region thatis consistent with all our constraints is highlighted by a red star symbol. sequence star’s effective temperature, surface gravity and metal-licity, which can then be used to estimate its mass and radius viathe Torres relation (Torres et al. 2010) and hence determine the bi-nary parameters. Unfortunately, due to its rapid rotation the FEROSfit does not give reliable stellar parameters. Therefore, keeping themetallicity fixed at the solar value, we used a grid of T eff , log g val-ues and the Torres relation to determine the range of possible binaryparameters. The result of this is shown in Figure 5. We found that atsurface gravities below ∼ > M ⊙ and 0.7 M ⊙ ,the main-sequence star mass between 1.08 M ⊙ and 1.26 M ⊙ andthe inclination larger than 32 degrees. Varying the metallicity doesnot have a large effect on these limits.We estimated the distance to TYC 6760-497-1 usingisochrones from the PAdova and tRieste Stellar Evolution Code(PARSEC) (Bressan et al. 2012). For a given temperature we usedthe isochrones to calculate the absolute magnitude of the main-sequence star over the full range of allowed surface gravities inthe BV RIJHK bands and compared these to the measured val-ues to calculate the distance. This results in a range of distancesfor a given main-sequence star temperature. We take into accountthe variations in the calculated distances from the different bandsand the effects of the unknown age of the star. The result of this isshown in Figure 6. While the unknown age of the star generally haslittle effect on the predicted distance, if it is very young ( <
30 Myr,which is in fact ruled out by the cooling age of the white dwarf) orvery old ( > Figure 6.
Distance to the main-sequence star in TYC 6760-497-1 as a func-tion of its effective temperature. The grey region shows the permitted range.
We fitted the HST/STIS spectrum of the white dwarf in TYC 6760-497-1 with a grid of white dwarf models computed usingTLUSTY/SYNSPEC (Hubeny & Lanz 1995), stepping in surfacegravity from 7.0 to 9.5 in steps of 0.1 (see Figure 7). At each stepthe best fit to the spectrum gives the temperature and hence themass and radius using the cooling models of Fontaine et al. (2001).Scaling the flux then also gives an estimate of the distance. Hencewe are able to determine the white dwarf mass, effective tempera-ture and distance as a function of the surface gravity. These rela-tions are illustrated in Figure 8. T eff and log g are correlated pa-rameters, so increasing log g results in a higher best-fit T eff . Atthe same time, increasing log g implies a higher mass, and, via themass-radius relation, a smaller radius. This drop in radius domi-nates over the increase in T eff in the UV flux emitted by the whitedwarf, i.e. increasing the surface gravity leads to a lower best-fitdistance.Several narrow absorption features are seen in the STIS spec-trum, primarily from Si and C as a result of the white dwarf ac-creting material from the wind of the main-sequence star as iscommonly seen in other close white dwarf binaries (Tappert et al.2011; Parsons et al. 2012; Pyrzas et al. 2012; Ribeiro et al. 2013).We found setting the metal abundances in the model to 0.1 timessolar for all elements fits these well. We can combine the constraints from fitting the white dwarf’s spec-trum with those from the main-sequence star to determine whichset of parameters are consistent with all our data. For a given main-sequence star effective temperature, we have a range of possibledistances from the isochrone fitting (see Figure 6) , which thenleads to a range of possible white dwarf masses from the relationsshown in Figure 8. The rotational broadening measurement alsoyields a white dwarf mass for a given main-sequence star temper-ature and gravity, allowing us to determine the range over whichthese two measurements are consistent. This range is shown in Fig-ure 5 as the dark grey hatched region. We can instantly exclude amain-sequence star with a temperature less than ∼ > M ⊙ , which is only possible if the c (cid:13) , 1–10 S. G. Parsons et al.
Figure 7.
HST/STIS spectrum of the white dwarf in TYC 6760-497-1 witha model fit overplotted (red line, log g = 8 . ). The dashed parts were notincluded in the fit, these include the core of the Ly α line, which is contam-inated by geocoronal emission, and the red end of the spectrum, which haslow signal-to-noise. There are also several emission lines including the quitestrong C IV system fills its Roche lobe. It also means that the system must be ata minimum distance of 250 pc.These constraints are also consistent with those from the main-sequence star temperature determined from the SED fitting ( T eff =6400 ± K), which is also illustrated in Figure 5. There is there-fore a small region in which all of our measurements and fits areconsistent, highlighted by a red star in Figure 5. The full set ofranges for all the parameters are given in Table 2.
Having determined the stellar and binary parameters of TYC 6760-497-1, we can now investigate the evolution of the system and itsimplications for models of compact binary star evolution and SN Iaformation channels. We start by reconstructing the systems evolu-tionary history.
Common envelope evolution is typically described with aparametrized energy equation: E bind = α CE ∆ E orb , (5)where α CE represents the fraction of orbital energy that is used tounbind the envelope usually called the common envelope efficiency(Paczynski 1976).The binding energy of the envelope is often assumed to beequal to the gravitational energy of the envelope: E bind = E gr = − GM M , e λR , (6)where M , M , e and R are the total mass, envelope mass andradius of the primary star, and λ is a binding energy parameter. Al-though very often ignored, the binding energy parameter stronglydepends on the mass and evolutionary state of the white dwarf pro-genitor when filling its Roche-lobe. This is especially true if the Figure 8.
Constraints on the white dwarf’s parameters as a function of itssurface gravity.
Table 2.
Physical and binary parameters of TYC 6760-497-1.Parameter ValueRA 15:02:22.4896Dec -29:41:15.666 B V R J H K . K MS ( km s − ) . ± . v rot sin i ( km s − ) . ± . Inclination ( ◦ ) 33–43Separation ( R ⊙ ) 3.20–3.28White dwarf mass ( M ⊙ ) 0.52–0.67White dwarf temperature (K) 19,500–21,000Main-sequence star spectral type F8Main-sequence star temperature (K) 6300–6500Main-sequence star surface gravity 4.31–4.48Main-sequence star mass ( M ⊙ ) 1.22–1.25Main-sequence star radius ( R ⊙ ) 1.18–1.40 recombination energy U rec available within the envelope supportsthe ejection process. Therefore, a more general form for the bindingenergy equation is: E bind = Z M M , c − Gmr ( m ) dm + α rec Z M M , c U rec ( m ) , (7)where α rec is the fraction of recombination energy that contributesto the ejection process. It is of outstanding importance for ourunderstanding of compact binary star evolution to observationallyconstrain the values of both common envelope efficiencies and toinvestigate possible dependencies on the binary parameters.With its period of less than half a day TYC 6760-497-1 is thefirst short orbital period ( P orb < day) post common envelope bi-nary (PCEB) with a massive secondary star (spectral type earlierthan K). The two systems that have been previously discovered areIK Peg and KOI-3278 with much longer orbital periods of 21.722and 88.18 days respectively (Wonnacott et al. 1993; Kruse & Agol2014). These two systems would have formed through commonenvelope evolution only if the common envelope efficiency α CE has been larger than it seems to be required to understand PCEBswith M dwarf secondaries, where α CE ∼ . − . seems towork best (Zorotovic et al. 2010; Rebassa-Mansergas et al. 2012b; c (cid:13) , 1–10 he close binary TYC 6760-497-1 Figure 9.
The mass of the white dwarf after the common envelope phaseas a function of its initial progenitor mass. The resultant white dwarf massdepends heavily upon the evolutionary stage at which the common envelopestarted. We show the results for a progenitor on the first giant branch (lightgrey), on the early AGB (dashed line) and on the TP-AGB (dark grey). Thehorizontal lines correspond to the range of white dwarf masses in agreementwith our observations.
Toonen & Nelemans 2013; Camacho et al. 2014) and/or if in addi-tion to a high fraction of the released orbital energy also recom-bination energy contributed to some degree to the ejection processexpelling the envelope (Zorotovic et al. 2010, 2014).To investigate whether TYC 6760-497-1 confirms or disprovesthe trend of larger efficiencies being required to understand PCEBswith massive secondaries we follow Zorotovic et al. (2010) andreconstruct the evolutionary history of the system using the bi-nary star evolution code BSE (Hurley et al. 2002) and the param-eter constraints derived in the previous sections, i.e. M WD =0 . − . ⊙ , M MS = 1 . − . ⊙ , P orb = 0 . daysand T eff , WD = 19500 − K. The current cooling age of thewhite dwarf is only ∼ . Gyr (Fontaine et al. 2001) hence the cur-rent orbital period is virtually identical to the period the system hadat the end of the common envelope (even if efficient magnetic brak-ing is assumed). Reconstructing the common envelope evolution asin Zorotovic et al. (2014) we find a large range of possible valuesfor the common envelope efficiencies for TYC 6760-497-1 and dif-ferent possible evolutionary stages at which the progenitor of thewhite dwarf could have filled its Roche-lobe.Figure 9 shows the solutions with α CE < in a final ver-sus initial mass plot for the primary if contributions from re-combination energy are ignored (i.e. α rec = 0 . ). Solutions ex-ist for massive progenitors on the first giant branch (FGB), onthe early asymptotic giant branch (AGB) and on the thermallypulsating-AGB (TP-AGB). The more massive the progenitor, thelarger the value of the common envelope efficiency must havebeen and the younger is the system today. The possible rangesare α = 0 . − . for a . − . ⊙ progenitor that fillsits Roche-lobe on the FGB (becoming first a naked helium starthen eventually a carbon-oxygen white dwarf, Hurley et al. 2002)at an age of . − . Gyr; α = 0 . − . if the progenitorreached the early AGB, after . − . Gyr and with an ini-tial mass of . − . ⊙ ; and finally α CE = 0 . − . fora . − . ⊙ progenitor that filled its Roche-lobe on the TP-AGB at an age of . − . Gyr, making it the oldest option. Inthe latter case we get the largest ranges as the core mass on the TP-AGB reaches the current white dwarf mass for a large rangeof initial masses. We therefore consider this the most likely sce-nario. This is further supported by our measurement of the radiusof the secondary. Assuming solar metallicity a . ⊙ main se-quence star needs ∼ . Gyr to expand from its ZAMS radius tothe current radius of the secondary of . ⊙ . We therefore con-clude that the most consistent scenario for the evolutionary historyof TYC 6760-497-1 is that the progenitor of the white dwarf wasof a relatively low mass ( ∼ ⊙ ) and filled its Roche-lobe onthe TP-AGB, which implies that the common envelope efficiencymust have been small α CE = 0 . − . . This is much smallerthan values obtained for the long orbital period systems with sec-ondary stars of similar mass (IK Peg and KOI-3278). In addition,contributions from recombination energy are not required to un-derstand the existence of TYC 6760-497-1. If recombination en-ergy contributed to expelling the envelope of the progenitor of thewhite dwarf, the fraction of orbital energy lost during the commonenvelope is reduced even further while the possible initial masseswould remain virtually identical. It therefore seems that a simpledependence of the total common envelope efficiency on the sec-ondary mass as speculated by Zorotovic et al. (2014) remains anincomplete prescription for common envelope evolution and otherparameters such as perhaps the evolutionary state of the progenitorwhen it fills its Roche-lobe may play an important role. Character-izing more PCEBs with massive secondaries is therefore urgentlyrequired to progress with our understanding of compact binary evo-lution which is directly related to our understanding of SN Ia pro-genitor systems. While it is clear that PCEBs with M dwarf secondaries evolve intocataclysmic variables (CVs) as long as the mass transfer will bedynamically stable, the future of PCEBs with more massive sec-ondary stars is more uncertain and potentially very interesting asthe total mass of the binary usually exceeds the Chandrasehkarlimit. Indeed, PCEBs with massive (G or F type) secondaries arethe progenitors of SN Ia explosions for both the double and the sin-gle degenerate channel. If the separation of a given PCEB is large itpotentially survives a second common envelope forming a doubledegenerate system which may then merge and produce a SN Ia. Ifon the other hand, the system is close it might start thermal timescale mass transfer and reach the rates required for stable nuclearburning on the surface of the white dwarf allowing its mass to in-crease and potentially reach the Chandrasehkar limit.With its short orbital period and given that the system is ratheryoung, TYC 6760-497-1 will clearly start mass transfer before thesecondary evolves off the main-sequence. It is also clear that thesystem will not undergo dynamical time scale mass transfer asthe critical mass ratio for dynamically unstable mass transfer is q cr > ∼ . (Ge et al. 2013) for donor stars with masses and radii sim-ilar to the Sun. It is less clear, however, if the mass transfer will bethermally stable (i.e. stable against thermal time scale mass trans-fer). The critical mass ratio for marginal stability against thermaltime scale mass transfer in a semi-detached binary star can be ob-tained by equating the mass-radius exponent of the Roche-lobe andthe star, i.e. ζ th = d ln( R MS ) d ln( M MS ) = d ln( R L ) d ln( M MS ) , (8)(see also e.g. de Kool 1992), where the Roche-lobe radius ( R L ) isa function of the mass ratio and the binary separation. Despite the c (cid:13) , 1–10 S. G. Parsons et al. . . . M sec [M ⊙ ] . . . . q = M s ec / M W D M WD > . ⊙ Figure 10.
Critical mass ratio for thermal time scale mass transfer as afunction of the main-sequence star’s mass for conservative (grey) and non-conservative (black) mass transfer. The position of TYC 6760-497-1 is in-dicated in red. For completeness we also provide the limits for dynamicaltime scale mass transfer (dashed lines) where we used the fits to detailedcalculations of the adiabatic mass-radius exponent (Hjellming 1989). secondary in TYC 6760-497-1 being slightly evolved we can geta first hint about the future of the system by calculating the limitimplied by Eq. 8 using the ZAMS M-R relation from Tout et al.(1996). For conservative mass transfer we get a critical mass ra-tio for thermal time scale mass transfer of q cr = 1 . − . forTYC 6760-497-1. Thus in the case of conservative mass transferthe system will certainly experience thermal time scale mass trans-fer. However, the assumption of conservative mass transfer is notnecessarily correct. As long as the mass transfer rate stays belowthe critical value for stable hydrogen burning on the white dwarf,mass transfer will most likely not be conservative. Instead, novaeruptions will occur, the white dwarf mass will remain nearly con-stant and angular momentum will be taken away from the systemby the expelled material. In Figure 10 we show the critical valuesfor the mass ratio as a function of secondary mass for both con-servative and non-conservative mass transfer. In the latter case weassume the mass expelled during nova eruptions to carry the spe-cific angular momentum of the white dwarf. The white dwarf massis assumed to be constant and the mass (and angular momentum)loss to be continuous which has been shown to be equivalent to adiscontinuous sequence of nova cycles in terms of the secular evo-lution of CVs (Schenker et al. 1998). These assumptions are typicalassumptions for CVs (e.g. Ritter 1988). The position of TYC 6760-497-1 is very close to the limit for thermal time scale mass transferin the case of non-conservative mass transfer so the system mayindeed be a progenitor of a super soft source.In order to investigate the future of TYC 6760-497-1 inmore detail we performed dedicated simulations using MESA(Paxton et al. 2011). We assume the initial masses of the two starsto be M MS = 1 . ⊙ and M WD = 0 . ⊙ and secondaryradius of . ⊙ , which corresponds to the mean values of theranges in agreement with our observations. The starting model wasobtained by evolving the secondary until it reaches the required ra-dius (after . Gyr) and the white dwarf is approximated as a pointmass. We assume mass transfer to be nearly conservative (90 percent of the transferred mass remains on the white dwarf) if the mass P orb (hr) -12.0-10.0-8.0-6.0 Log ˙ M ( M ⊙ / y r) . . P orb (hr) M W D , M ( M ⊙ ) Figure 11.
The predicted future evolution of the mass transfer rate (toppanel), the white dwarf mass (dotted line, bottom panel), and the secondarystar mass (solid line, bottom panel) as a function of orbital period. The hor-izontal red line in the top panel represents the mass transfer rate requiredfor stable hydrogen burning for the respective mass of the white dwarf (i.e.when the mass transfer rate is above this line stable hydrogen burning oc-curs). At P orb ∼ . h TYC 6760-497-1 reaches this critical mass transferrate. As a consequence mass transfer becomes conservative and the whitedwarf grows in mass. At P orb ∼ h the white dwarf mass has reached . ⊙ and the mass transfer rate drops below the rate required for stableburning. At this moment TYC 6760-497-1 becomes a “normal” CV withnon-conservative stable mass transfer driven by angular momentum lossonly and a constant white dwarf mass. transfer is large enough to generate stable hydrogen burning on thesurface of the white dwarf as we run into numerical problems inthe fully conservative case. For mass transfer rates below the limitfor stable burning we assume that all the accreted mass is expelledduring nova eruptions leaving the system with the specific angularmomentum of the white dwarf. Finally, for angular momentum lossdue to magnetic braking we assume the standard prescription fromRappaport et al. (1983).Based on these assumptions, the future of TYC 6760-497-1can be estimated. The black line in the top panel of Figure 11 showsthe predicted mass transfer rate of TYC 6760-497-1 as a functionof orbital period. At an orbital period of 11.9 h (in less than 5 Myr)mass transfer will start. The mass transfer rate will quickly increaseas the system is thermally unstable, but does not reach the limit forstable hydrogen burning until the system has an orbital period of6.2 h (13.1 Myr from now). At this point, mass transfer becomesconservative (we assume that 90 per cent of the burned materialremains on the white dwarf). As a consequence, the mass transferrate increases significantly and the white dwarf mass grows (bot-tom panel). When the white dwarf mass reaches . ⊙ (14.1 Myrfrom now) the mass ratio will be close to unity (secondary mass of . ⊙ at an orbital period of 5.1 h) and the system becomes stableagainst (conservative) thermal time scale mass transfer. Thereforethe accretion rate drops below the value required for stable hydro-gen burning. From this moment on, TYC 6760-497-1 will behaveas expected for normal CVs: after 105 Myr it will enter the periodgap and restart mass transfer at a much lower rate at an orbital pe-riod of 2.1 h (1.5 Gyr from now). Finally, the system will reach theorbital period minimum ( ∼ c (cid:13) , 1–10 he close binary TYC 6760-497-1 P orb (hr) -2.0-1.00.01.02.0 Log ( C / N , C / C ) Figure 12.
Surface abundance ratios (C/N, dashed line and C / C , solidline) for the main-sequence star in TYC 6760-497-1 as a function of orbitalperiod. Once a deep convective envelope forms (P orb ∼ / C , black diamond). scending from PCEBs with less massive secondaries, its past as aPCEB containing a more massive secondary star will be imprintedin the relative abundances of carbon and nitrogen of the accretedmaterial as soon as the outer convection zone reaches regions con-taining CNO processed material. Decreased values for C/N andC / C in CVs descending from thermal time scale mass transfersystems have been predicted by Schenker et al. (2002) and foundin CVs based on the presence of C/N line ratios measured withHST by G¨ansicke et al. (2003). In Figure 12 we predict the C/Nand C / C abundance ratios for TYC 6760-497-1 as a functionof orbital period. As soon as the secondary star starts to developa deep convective envelope both surface abundance ratios decreasesignificantly. Most dramatically, C/N decreases by several orders ofmagnitude. TYC 6760-497-1 is therefore likely a progenitor of theCVs with decreased C/N line ratios found by G¨ansicke et al. (2003)which are frequently called failed SN Ia.A final note concerning the future of TYC 6760-497-1 con-cerns the expected white dwarf mass when the systems becomes aCV. The mean white dwarf mass in CVs derived from observations( ∼ . ⊙ , Zorotovic et al. 2011) is significantly higher than pre-dicted by binary population models of CVs and significantly largerthan the mean value observed in CV progenitors containing lowmass secondaries (see Zorotovic et al. 2011, for details). The valuepredicted by our simulations for TYC 6760-497-1, however, is veryclose to the mean white dwarf mass of the observed CV sample.One could therefore speculate that if the number of CVs descend-ing from PCEBs with massive secondaries has been underestimatedpreviously, the problem with the white dwarf masses in CVs couldbe solved. Assuming that a large number of CVs are descendantsfrom PCEBS with massive secondary stars such as TYC 6760-497-1 indeed predicts an increased number of CVs with white dwarfmasses ∼ . ⊙ . However, the large number of CVs with evolvedsecondaries predicted in this scenario violates the general expla-nation for the orbital period gap (see Wijnen et al. 2015, for moredetails). We therefore believe that the fraction of CVs descending from systems with initially massive secondaries does not signifi-cantly exceed the ∼ per cent of CVs with decreased C/N lineratios as measured by G¨ansicke et al. (2003) and the white dwarfmass problem remains to be solved.We emphasize that the above evolutionary scenario forTYC 6760-497-1 is based on several assumptions such as thestrength of magnetic braking, the angular momentum taken awayfrom the system during nova eruptions, and the fraction of massthat is lost during thermal time scale mass transfer. However, oursimulation shows that it is likely that TYC 6760-497-1 is the firstknown progenitor of a super-soft source. We have identified TYC 6760-497-1 as a close binary consistingof a white dwarf and an F8 star that is extremely close to fillingits Roche lobe. Assuming that the F star is tidally locked, we con-strain the masses of the two stars to M WD = 0 . − . ⊙ , M MS = 1 . − . ⊙ and the radius of the F star to R MS =1 . − . ⊙ . The white dwarf is still hot ( , − , K)and so the system only emerged from the common envelope 0.1 Gyrago. Its progenitor was likely on the TP-AGB when the commonenvelope began. However, this means that the common envelopeefficiency was quite low, in contrast to other systems containingwhite dwarfs with early type companions, implying that the effi-ciency depends upon more than just the mass of the companionstar. TYC 6760-497-1 will become a semi-detached system in lessthan 5 Myr and will undergo a short phase ( ∼ Myr) of thermaltimescale mass-transfer in 13.1 Myr resulting in a roughly 20 percent increase in the white dwarf’s mass, still well short of the Chan-drasekhar limit, after which it will become a standard cataclysmicvariable system.Although we have placed some constraints on the binary pa-rameters, these will be greatly improved with an accurate distancemeasurement from Gaia which should reach a precision of ∼ . at the expected distance of TYC 6760-497-1 . This will dramati-cally reduce the reliance on evolutionary models in our measure-ments allowing us to more rigorously test the evolutionary scenar-ios of this (and similar) systems. ACKNOWLEDGMENTS
We thank the referee for useful comments and suggestions. SGP,MZ and CT acknowledge financial support from FONDECYTin the form of grant numbers 3140585, 3130559 and 1120338.AB acknowledges financial support from Proyecto FONDECYTde Iniciaci´on 11140572. The research leading to these resultshas received funding from the European Research Council underthe European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. 320964 (WDTracer). MRSthanks for support from FONDECYT (1141269) and Millen-nium Science Initiative, Chilean ministry of Economy: NucleusRC130007. ARM acknowledges financial support from the Post-doctoral Science Foundation of China (grants 2013M530470and 2014T70010) and from the Research Fund for InternationalYoung Scientists by the National Natural Science Foundation (cid:13) , 1–10 S. G. Parsons et al. of China (grant 11350110496). RB is supported by CONICYT-PCHA/Doctorado Nacional. RB acknowledge additional supportfrom project IC120009 ”Millenium Institute of Astrophysics(MAS)” of the Millennium Science Initiative, Chilean Ministryof Economy. AJ acknowledges support from the Ministry forthe Economy, Development, and Tourism’s Programa IniciativaCient´ıfica Milenio through grant IC 120009, awarded to the Mil-lennium Institute of Astrophysics (MAS), FONDECYT project1130857 and from BASAL CATA PFB-06. Based on observationsmade with the NASA/ESA Hubble Space Telescope, obtained atthe Space Telescope Science Institute, which is operated by the As-sociation of Universities for Research in Astronomy, Inc., underNASA contract NAS 5-26555. These observations are associatedwith program
REFERENCES
Allard, F., Homeier, D., Freytag, B., 2012, Royal Society of Lon-don Philosophical Transactions Series A, 370, 2765Bayo, A., Rodrigo, C., Barrado Y Navascu´es, D., Solano, E.,Guti´errez, R., Morales-Calder´on, M., Allard, F., 2008, A&A,492, 277Branch, D., Tammann, G. A., 1992, ARA&A, 30, 359Bressan, A., Marigo, P., Girardi, L., Salasnich, B., Dal Cero, C.,Rubele, S., Nanni, A., 2012, MNRAS, 427, 127Burleigh, M. R., Barstow, M. A., Fleming, T. A., 1997, MNRAS,287, 381Camacho, J., Torres, S., Garc´ıa-Berro, E., Zorotovic, M.,Schreiber, M. R., Rebassa-Mansergas, A., Nebot G´omez-Mor´an,A., G¨ansicke, B. T., 2014, A&A, 566, A86Claret, A., 2004, A&A, 428, 1001Coelho, P., Barbuy, B., Mel´endez, J., Schiavon, R. P., Castilho,B. V., 2005, A&A, 443, 735de Kool, M., 1992, A&A, 261, 188D’Odorico, S., et al., 2006, in Proc. SPIE, vol. 6269, p. 98Fink, M., Hillebrandt, W., R¨opke, F. K., 2007, A&A, 476, 1133Fink, M., R¨opke, F. K., Hillebrandt, W., Seitenzahl, I. R., Sim,S. A., Kromer, M., 2010, A&A, 514, A53Fontaine, G., Brassard, P., Bergeron, P., 2001, PASP, 113, 409G¨ansicke, B. T., et al., 2003, ApJ, 594, 443Ge, H., Webbink, R. F., Chen, X., Han, Z., 2013, in Zhang, C. M.,Belloni, T., M´endez, M., Zhang, S. N., eds., IAU Symposium,vol. 290 of
IAU Symposium , p. 213Greiner, J., 2000, NewA, 5, 137Hjellming, M. S., 1989, Rapid mass transfer in binary systems,Ph.D. thesis, Illinois Univ. at Urbana-Champaign, Savoy.Holberg, J. B., Oswalt, T. D., Sion, E. M., Barstow, M. A.,Burleigh, M. R., 2013, MNRAS, 435, 2077Hubeny, I., Lanz, T., 1995, ApJ, 439, 875Hurley, J. R., Tout, C. A., Pols, O. R., 2002, MNRAS, 329, 897Husser, T.-O., Wende-von Berg, S., Dreizler, S., Homeier, D.,Reiners, A., Barman, T., Hauschildt, P. H., 2013, A&A, 553, A6Jord´an, A., et al., 2014, AJ, 148, 29Kordopatis, G., et al., 2013, AJ, 146, 134Kruse, E., Agol, E., 2014, Science, 344, 275 Marsh, T. R., 1989, PASP, 101, 1032Martin, D. C., et al., 2005, ApJ, 619, L1Maxted, P. F. L., G¨ansicke, B. T., Burleigh, M. R., Southworth, J.,Marsh, T. R., Napiwotzki, R., Nelemans, G., Wood, P. L., 2009,MNRAS, 400, 2012Napiwotzki, R., et al., 2003, The Messenger, 112, 25Paczynski, B., 1976, in Eggleton, P., Mitton, S., Whelan, J., eds.,Structure and Evolution of Close Binary Systems, vol. 73 of
IAUSymposium , p. 75Parsons, S. G., et al., 2012, MNRAS, 420, 3281Paxton, B., Bildsten, L., Dotter, A., Herwig, F., Lesaffre, P.,Timmes, F., 2011, ApJS, 192, 3Perlmutter, S., et al., 1999, ApJ, 517, 565Pyrzas, S., et al., 2012, MNRAS, 419, 817Rappaport, S., Verbunt, F., Joss, P. C., 1983, ApJ, 275, 713Rebassa-Mansergas, A., Nebot G´omez-Mor´an, A., Schreiber,M. R., G¨ansicke, B. T., Schwope, A., Gallardo, J., Koester, D.,2012a, MNRAS, 419, 806Rebassa-Mansergas, A., et al., 2012b, MNRAS, 423, 320Ribeiro, T., Baptista, R., Kafka, S., Dufour, P., Gianninas, A.,Fontaine, G., 2013, A&A, 556, A34Riess, A. G., et al., 1998, AJ, 116, 1009Ritter, H., 1988, A&A, 202, 93Schenker, K., Kolb, U., Ritter, H., 1998, MNRAS, 297, 633Schenker, K., King, A. R., Kolb, U., Wynn, G. A., Zhang, Z.,2002, MNRAS, 337, 1105Schlafly, E. F., Finkbeiner, D. P., 2011, ApJ, 737, 103Shen, K. J., Moore, K., 2014, ApJ, 797, 46Skrutskie, M. F., et al., 2006, AJ, 131, 1163Tappert, C., G¨ansicke, B. T., Schmidtobreick, L., Ribeiro, T.,2011, A&A, 532, A129Toonen, S., Nelemans, G., 2013, A&A, 557, A87Torres, G., Andersen, J., Gim´enez, A., 2010, A&A Rev., 18, 67Tout, C. A., Pols, O. R., Eggleton, P. P., Han, Z., 1996, MNRAS,281, 257Tutukov, A., Yungelson, L., 1996, MNRAS, 280, 1035Webbink, R. F., 1984, ApJ, 277, 355Whelan, J., Iben, Jr., I., 1973, ApJ, 186, 1007Wijnen, T. P. G., Zorotovic, M., Schreiber, M. R., 2015, ArXive-printsWonnacott, D., Kellett, B. J., Stickland, D. J., 1993, MNRAS, 262,277Wright, E. L., et al., 2010, AJ, 140, 1868Zahn, J.-P., 1977, A&A, 57, 383Zorotovic, M., Schreiber, M. R., G¨ansicke, B. T., Nebot G´omez-Mor´an, A., 2010, A&A, 520, A86Zorotovic, M., Schreiber, M. R., G¨ansicke, B. T., 2011, A&A,536, A42Zorotovic, M., Schreiber, M. R., Parsons, S. G., 2014, A&A, 568,L9 c (cid:13)000