MNRAS 512, 2625–2635 (2022) https://doi.org/10.

1093/mnras/stac673
Advance Access publication 2022 March 11

The white dwarf binary pathways survey – VII. Evidence for a bi-modal
distribution of post-mass transfer systems?
F. Lagos,1,2,3 ‹ M. R. Schreiber,2,4 S. G. Parsons ,5 O. Toloza,2,4 B. T. Gänsicke ,6 M. S. Hernandez ,4

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
L. Schmidtobreick 3 and D. Belloni 4
1 Instituto ˜ 1111, Valparaı́so, Chile
de Fı́sica y Astronomı́a de la Universidad de Valparaı́so, Avenida Gran Bretana
2 MillenniumNucleus for Planet Formation, NPF, Valparaı́so, Chile
3 European Southern Observatory (ESO), Alonso de Cordova 3107, Vitacura, Santiago, Chile
4 Departamento de Fı́sica, Universidad Técnica Federico Santa Marı́a, Avenida Espana
˜ 1680, Valparaı́so, Chile
5 Department of Physics and Astronomy, University of Sheffield, Sheffield S3 7RH, UK
6 Department of Physics, University of Warwick, Coventry CV4 7AL, UK

Accepted 2022 March 9. Received 2022 March 9; in original form 2021 December 27

ABSTRACT
Binary systems consisting of a white dwarf (WD) and a main-sequence companion with orbital periods up to ≈100 d are
often thought to be formed through common envelope evolution which is still poorly understood. To provide new observational
constraints on the physical processes involved in the formation of these objects, we are conducting a large-scale survey of close
binaries consisting of a WD and an A- to K-type companion. Here, we present three systems with eccentric orbits and orbital
periods between approximately 10 and 42 d discovered by our survey. Based on Hubble Space Telescope spectroscopy and
high-angular resolution images obtained with the Spectro-Polarimetric High-contrast Exoplanet REsearch (SPHERE), we find
that two of these systems are most likely triple systems while the remaining one could be either a binary or a hierarchical triple
but none of them is a post-common envelope binary (PCEB). The discovery of these systems shows that our survey is capable to
detect systems with orbital periods of the order of weeks, but all six PCEBs we have previously discovered have periods <2.5 d.
We suggest that the fact that all of the systems we identify with periods of the order of weeks are not PCEBs indicates a transition
between two different mechanisms responsible for the formation of very close (࣠10 d) and somewhat wider WD + AFGK
binaries: common envelope evolution and non-conservative stable mass transfer.
Key words: (stars:) binaries (including multiple): close – (stars:) white dwarfs – techniques: high angular resolution.

Despite its importance for close compact binary formation the-

1 . I N T RO D U C T I O N
ories, CE evolution is still not fully understood and its physical
Observational and theoretical population studies of close binaries modelling is frequently simplified by using energy and angular
with white dwarf (WD) components are fundamental to progress with momentum conservation equations involving free parameters that
our understanding on their formation and evolution, with potentially must be fitted through observations. Previous surveys have shown
deep implications for the predicted occurrence rates of systems as that close WD binaries with M dwarf companions form through CE
important as the progenitors of Type Ia supernovae (SNe Ia). evolution and led to important constraints on the CE efficiency α CE ,
Close WDs binaries are believed to form mainly through common which corresponds to the fraction of orbital energy that is used to
envelope (CE) evolution (e.g. Webbink 1984; Zorotovic et al. 2010; unbind the envelope of the WD progenitor. Evidence is accumulating
Ivanova et al. 2013) which occurs when the more massive (1 M ) that for PCEBs with M dwarf companions α CE is rather small, i.e.
star of an initial main-sequence binary becomes a giant that is filling between 0.2 and 0.3 (e.g. Nelemans et al. 2000; Zorotovic et al.
its Roche lobe, which triggers dynamically unstable mass transfer. 2010; Nebot Gómez-Morán et al. 2011; Toonen & Nelemans 2013).
As the mass-transfer time-scale is shorter than the thermal time-scale To progress with our understanding of CE evolution it is important
of the companion, a CE forms engulfing the core of the giant star and to investigate the evolutionary history not only of PCEBs with M
the companion. Drag forces between the companion and the envelope dwarfs but over a wide range of companion masses as this may either
drive transfer of orbital energy and angular momentum from the orbit allow to derive constraints on a universal value for the CE efficiency
of the core of the giant and its companion to the envelope until the or indicate the existence of processes that are not included in the
latter is fully ejected. The close binary emerging from CE evolution, simple energy equation.
i.e. the post-common envelope binary (PCEB), consists of a WD and To that end, we began a large scale survey of close WDs with
a main-sequence companion star also known as the secondary star main-sequence star companions of spectral type A, F, G, or K
(e.g. Parsons et al. 2015). (WD + AFGK binaries). This survey is using the following strategy.
As the WD is outshone by the AFGK stars at optical wavelengths,
we combined spectroscopic surveys with the Galaxy Evolution
 Explorer (GALEX) data base (Bianchi et al. 2011; Bianchi, Shiao &
E-mail: [email protected]

© 2022 The Author(s)

Published by Oxford University Press on behalf of Royal Astronomical Society
2626 F. Lagos et al.
Thilker 2017) to identify AFGK stars with ultraviolet (UV) excess Observatory, Chile to obtain spectra of our targets. Each science ob-
as indicative of a potential WD companion (Parsons et al. 2016; servation was bracketed by ThAr spectra to correct for instrumental
Rebassa-Mansergas et al. 2017; Ren et al. 2020). We then use radial drift. However, we place a lower limit on the velocity precision
velocity measurements to identify close binary systems among our in a single spectrum of 0.5 km s−1 due to the unstable nature of
candidates. If radial velocity variations indicate the close binary the spectrograph. Standard image reductions were performed and
nature of a given object, we then finally measure the orbital periods the spectra optimally extracted and wavelength calibrated using
of those WD + AFGK candidate systems. We rely on Hubble Space the Collection of Elemental Routines for Echelle Spectra (CERES)

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
Telescope (HST) spectroscopy to confirm that the UV excess is indeed package (Brahm, Jordán & Espinoza 2017).
coming from a WD and to measure the mass of the WD.
Following this strategy, we have so far identified the first pre-
2.2 FEROS
supersoft X-ray binary system (Parsons et al. 2015), confirmed that
our target selection is reliable and in virtually all cases the UV High-resolution spectra were obtained with the FEROS echelle
excess is indeed indicative of a WD (Parsons et al. 2016), published spectrograph (R  48 000) on the 2.2 m telescope at La Silla, Chile
the first results of our radial velocity campaign (Rebassa-Mansergas (Kaufer & Pasquini 1998). FEROS covers the wavelength range
et al. 2017), estimated the limited contamination from triple systems from  3500 to  9200 Å. Observations were performed in Object-
motivated by the discovery of the first hierarchical triple system in the Calibration mode where one fibre is placed on the target while the
survey (Lagos et al. 2020), and identified and characterized 26 new other feeds light from a ThAr + Ne calibration lamp permitting
close WD + AFGK systems (Ren et al. 2020) through radial velocity velocity measurements to extremely high precision ( 10 m s−1 ) and
variations. Most importantly for this work, we measured the periods allowed us to correct for instrumental drift throughout the night.
of six systems and found all of them to be PCEBs with periods below FEROS data were reduced using the CERES package.
2.5 d and circular orbits. We reconstructed their evolutionary history
and found all of them to be consistent with a small CE efficiency
2.3 CHIRON
(Parsons et al. 2015; Hernandez et al. 2021; Hernandez et al.
2022), in agreement with what was found for PCEBs with M dwarf Spectra were also obtained with the CHIRON echelle spectrometer
companions (Zorotovic et al. 2010; Nebot Gómez-Morán et al. 2011). (Tokovinin et al. 2013) on the 1.5 m SMARTS telescope at Cerro
Here, we show that our observing strategy does permit the Tololo, Chile. We used 3 × 1 binning resulting in R  40 000. CH-
identification of systems with longer orbital periods of the order IRON observations are automatically reduced by the CHIRON team
of several weeks. We present the period determination of three using standard reduction methods. Like the Du Pont observations,
systems, i.e. 2MASS J06281844−7621467, TYC 6996−449−1 we place a lower limit on the velocity precision in a single spectrum
and TYC 8097−337−1 (hereafter 2MASS J0628, TYC 6996, and of 0.5 km s−1 .
TYC 8097 respectively), and find their orbital periods to be between
10 and 42 d and their orbits to be eccentric. Using high-resolution
2.4 UVES
spectroscopy at optical wavelengths we constrain the stellar prop-
erties of the AFGK star, while with high-contrast imaging in the Additional spectra were obtained with UVES (Dekker et al. 2000), a
infrared we look for additional companions that could potentially high-resolution echelle spectrograph mounted on the 8.2 m European
cause eccentric orbits, for instance, through Von Zeipel–Lidov–Kozai Southern Observatory Very Large Telescope (VLT) at Cerro Paranal,
oscillations (e.g. Naoz 2016). We find that one system (TYC 6996) Chile. We used the dichroic 1 set-up (390 + 564) with a 0.7 arcsec
is most likely a hierarchical triple system with the WD being the slit, resulting in R  50 000. The data were reduced using the UVES
distant tertiary. In the case of 2MASS J0628, HST spectroscopy does data reduction pipeline (version 5.8.2). A 0.5 km s−1 lower limit was
not detect a WD and the UV excess is most likely caused by stellar placed on the velocity precision in a single spectrum.
activity of a lower mass main-sequence star. For TYC 8097, given
the current available data, we conclude that it could be a triple similar
2.5 HST spectroscopy
to TYC 6996, or a binary system with an M- or K-type companion.
In case of the latter, the UV excess comes either from stellar activity We spectroscopically observed 2MASS 0628 with the HST in order to
or an active background galaxy. confirm that the UV excess is due to a WD companion. We obtained
The fact that all systems from our survey where we measured four spectra with the Space Telescope Imaging Spectrograph (STIS)
a period of the order of weeks are not PCEBs, while all systems on 2021 April 23 under the programme GO 16224 using the G140L
with periods of <2.5 d turned out to be PCEBs, provides crucial grating (R ∼ 1000), covering the wavelength range from  1150 to
constraints on CE evolution with potentially deep implications for  1700 Å. 1D spectra were extracted with CALSTIS-6 – version
our understanding of SN Ia progenitor channels. 3.4.2 (Hodge et al. 1998) from the bias-subtracted and flat-corrected
FLT files, which were downloaded from the STScI archive. We forced
the search to be performed around pixel 400 in the spatial axis,
2 O B S E RVAT I O N S
allowing a search of only 30 pixels. The location of the spectra was
We performed high-resolution spectroscopic and high-contrast imag- found to be around the pixel 404. The spectrum was extracted from
ing observations of three targets that have been identified as a box with a width of 11 pixels, while background regions were
WD + AFGK candidates by correlating optical surveys and GALEX extracted from boxes above and below the spectrum with sizes of
data. In what follows we describe the details of our observational seven pixels. These regions were scaled to the size of the spectrum
set-ups and the data reduction. region.

2.1 Du Pont echelle 2.6 High-contrast imaging

We used the high-resolution echelle spectrograph (1 arcsec slit, The three eccentric systems were observed with the high-contrast
R  40 000) on the 2.5 m Du Pont telescope located at Las Campanas imager VLT/SPHERE (Beuzit et al. 2019) under the programme

MNRAS 512, 2625–2635 (2022)

 Evidence for a bi-modal period distribution? 2627
100.D-0399. Acquisition of direct imaging with the N-ALC-YJH- accurate and precise stellar parameters for the AFGK stars in our
S coronagraph was made in the IRDIFS observing mode, which systems we followed the method outlined in Hernandez et al. (2021),
includes IRDIS dual band imaging (Dohlen et al. 2008; Vigan et al. whereby the high-resolution echelle spectra were fitted using iSpec
2010) plus the integral field spectrograph working in Y-J mode (Blanco-Cuaresma et al. 2014) to determine the effective tempera-
(Claudi et al. 2008). For IRDIS dual-band filters H2 (λH2 = 1593 nm) tures Teff, AFGK , surface gravities (log gAFGK ), and metallicities [Fe/H]
and H3 (λH3 = 1667 nm) were used. Furthermore, the pupil tracking (ZAFGK ), which were then used as priors when fitting the optical and
mode was implemented in order to perform angular differential infrared spectral energy distribution of the star in conjunction with

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
imaging (ADI, Marois et al. 2006). the Gaia EDR3 parallaxes (Gaia Collaboration 2021) to determine
The IRDIS data were first pre-processed (dark background the radius (RAFGK ). See Hernandez et al. (2021) for a more detailed
subtraction, flat-fielding, and bad-pixels correction) with the description of this fitting process. Combining RAFGK and log gAFGK
VLT/SPHERE Python package1 version 1.4.3 (Vigan 2020). The then yielded the mass of the AFGK star (MAFGK ). The best-fitting
frames were recentred based on star centre exposures using stellar parameters are listed in Table 1.
the four satellite spots. After pre-processing, and without any
post-processing technique to remove speckle patterns produced
3.3 Differential photometry of 2MASS 0628
by the coronograph, we detected one potential companion around
2MASS 0628 at 0.31 arcsec from the binary. We used the principal Unlike TYC 6996 and TYC 8097, 2MASS 0628 is the only system
component analysis (PCA) algorithm available in the Vortex Image which has a point source detection well located in the IRDIS field of
Processing (VIP, Amara & Quanz 2012; Soummer, Pueyo & Larkin view to perform photometry.
2012; Gomez Gonzalez et al. 2017) Python package to look for We use nine SPHERE/IRDIS flux calibration exposures (from the
fainter companions, finding one companion around TYC 6996, fits file SPHER.2017-12-30T03:54:37.371) to obtain the differential
which is partially visible on the edge of the IRDIS field of view, and magnitudes H2, H3 of the companion relative to the instrumental
two extended sources in TYC 8097 (see Fig. 1). photometry of the central binary. Given that a small fraction of the
companion’s Airy disc slightly overlaps the first Airy diffraction
pattern of the central binary, we use an aperture radius of 4 pixels
3 S T E L L A R A N D B I N A RY PA R A M E T E R S
(i.e. the measured full width at half-maximum of IRDIS in H2 and
From the observations, we derived basic binary and stellar parameters H3 filters) to avoid as much as possible flux cross-contamination.
for all three systems. The orbital periods can be derived from Sky substraction was performed by calculating the mean value of
our radial velocity measurements, the spectra also provide tight the pixels located inside an annulus centred at the same angular
constraints on the nature of the AFGK star in all three systems, distance of the companion and width equal to the aperture diameter.
and the SPHERE high-contrast imaging provides information on the The region of the annulus centred at the companion and with length
potential triple nature of our targets. equal to twice the diameter of the aperture was not considered in the
calculation. The final differential magnitudes (i.e. the average of the
nine exposures) for filters H2 and H3 are H2 = 2.17 ± 0.03 and
3.1 Orbital characterization H3 = 2.11 ± 0.05.
Radial velocities were computed from all our echelle spectra using
cross-correlation against a binary mask representative of a G2-type 4 THE THREE ECCENTRIC SYSTEMS
star (see Brahm et al. 2017 for more information). Barycentric
correction was applied using ASTROPY (Astropy Collaboration et al. The three targets we present in this paper turned out to have eccentric
2018). A full list of radial velocity measurements is given in orbits, which together with TYC 7218 (Lagos et al. 2020) brings the
Tables A1, A2, and A3 for 2MASS J0628, TYC 8097, and TYC 6996, number of measured eccentric orbits with periods of the order of
respectively, in the Appendix. weeks to four. Given their eccentric nature, these systems are not
In order to determine the orbital components of each system (Porb probably formed though CE evolution (e.g. Ivanova et al. 2013). One
– orbital period, e – eccentricity, K2 – radial velocity semi-amplitude, possibility is that, like TYC 7218, they are triple systems. To elucidate
γ – systemic velocity, TP – time of periastron, and ω – argument of their nature, we have performed detailed follow-up observations to
periastron) we fitted the radial velocities using EXOFAST (Eastman, characterize all three systems. In what follows we describe each
Gaudi & Agol 2013). For all three targets a clear best fit was found. system individually.
Uncertainties on the orbital components were determined using the
Markov Chain Monte Carlo method (Press et al. 2007) implemented 4.1 2MASS J06281844−7621467
using the Python package EMCEE (Foreman-Mackey et al. 2013),
where the best-fitting parameters from EXOFAST were used as the Of all the PCEB candidates observed with HST, 2MASS J0628 has
starting values in the fit. The result of these fits and the radial velocity the faintest GALEX FUV magnitude (mFUV, GALEX = 21.37 ± 0.39).
curves are shown in Table 1 and Fig. 1 respectively, where from the Based on the orbital properties derived from the radial velocity curve,
latter it is clear that the shape of the three radial velocity curves the minimum mass for the companion to the G-type star is ≈ 0.3 M .
correspond to eccentric orbits. The STIS spectra show no evidence for a WD companion or any
contribution from a spectral continuum. Averaging the two spectra
(from files oe9x08010 x1d.fits and oe9x08020 x1d.fits) with the best
3.2 Characterization of the AFGK stars signal-to-noise ratio, we found two emission features corresponding
The CERES pipeline provides an initial estimate of the stellar to C II at 1335 Å and C IV at 1550 Å (Fig. 2). Although the integrated
parameters of the AFGK star, but only over a narrow parameter flux of this spectrum through the GALEX FUV filter is insufficient
range relevant for main-sequence stars. In order to determine more to explain the observed UV excess, these lines are used as stellar
activity tracers and can contribute to the enhanced UV flux produced
during transient stages of high activity (e.g. Findeisen, Hillenbrand &
1 https://github.com/avigan/SPHERE Soderblom 2011; Shkolnik & Barman 2014; Loyd et al. 2018),
MNRAS 512, 2625–2635 (2022)
2628 F. Lagos et al.

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023

Figure 1. SPHERE/IRDIS H2 images (left-hand panel) and radial velocity curves (right-hand panel) for 2MASS 0628 (top panel), TYC 8097 (middle panel),
and TYC 6996 (bottom panel). Black circles in each SPHERE image indicate the position of the coronagraph, while detections are accompanied by its projected
separation in arcsec (white numbers). The star dominating at visible light is clearly part of an eccentric binary which excludes these systems to be PCEBs.
2MASS J0628 is best explained as a triple system with at least one active star while TYC 6996 is also a hierarchical triple with a WD as a tertiary. In the case of
TYC 8097 the origin of the UV excess remains mysterious. The three main candidates to the UV source(s) are the observed background galaxies (in case they
are active), stellar activity from the M or K-type companion, or a WD being the third companion in a hierarchical configuration.

MNRAS 512, 2625–2635 (2022)

 Evidence for a bi-modal period distribution? 2629
Table 1. Measured mass, radius, effective temperature, and metallicity of the magnitude in the H2 and H3 filters we proceed as follows. First,
AFGK star for each eccentric system. Orbital parameters are obtained from we calculate the synthetic H2 and H3 magnitudes of 2MASS J0628
radial velocity measurements of the AFGK star. Photogeometric distances using the synthetic photometry tool available in the Spanish Virtual
from Bailer-Jones et al. (2021) are shown with errors based on the 16th Observatory (Rodrigo, Solano & Bayo 2012; Rodrigo & Solano
and 84th percentiles of the distance posterior distribution. The last two rows
2020) and a BT-NextGen (Allard, Homeier & Freytag 2012) spectral
represent the most likely source of the UV excess and whether they are triple
systems.
template with effective temperature 5500 K, surface gravity log g =
4.5, and metallicity [Fe/H] = 0, i.e. resembling the stellar properties

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
Parameter TYC 6996 TYC 8097 2MASS J0628 shown in Table 1. Then, synthetic magnitudes were reddened using
the 3D interstellar dust map from Lallement et al. (2019).2 The final
MAFGK (M ) 1.12 ± 0.08 1.43 ± 0.16 0.99 ± 0.11 synthetic magnitudes of the G star in the H2 and H3 filters are H2G =
RAFGK (R ) 1.266 ± 0.020 1.715 ± 0.011 0.995 ± 0.009 10.72, H3G = 10.58, in good agreement with the archival magnitude
TAFGK (K) 6270 ± 35 5730 ± 40 5540 ± 50 in the 2MASS H filter of 10.44. With these values, and using the
ZAFGK ([Fe/H]) −0.31 ± 0.15 −0.07 ± 0.10 −0.07 ± 0.10
differential photometry, we derived synthetic magnitudes H2comp =
Porb (d) 41.9950 20.9850 10.2121
±0.0017 ±0.0022 ±0.0013
12.75 and H3comp = 12.83 for the companion candidate. Using this
e 0.497 ± 0.006 0.451 ± 0.003 0.266 ± 0.002 information, we estimate the likelihood of the companion to be a
KAFGK (km s−1 ) 24.21 ± 0.21 31.05 ± 0.26 28.11 ± 0.10 background source aligned by chance within an angular distance 
γ (km s−1 ) 2.54 ± 0.12 42.17 ± 0.09 −14.62 ± 0.02 following Brandner et al. (2000) as
TP (BJD) 2 457 040.00 2 457 423.837 2 457 235.406
P (, mlim ) = 1 − e−π
2 ρ(m
lim )
±0.02 ±0.001 ±0.001 , (1)
ω (deg) 151.9 ± 0.8 329.0 ± 0.3 197.7 ± 0.4
where ρ(m) is the cumulative surface density of background sources
Distance (pc) 392.1+ 5.3
−5.4 350.8+ 1.8
−1.5 265.9+ 1.9
−1.4 down to a limiting magnitude mlim (i.e. the magnitude of the
Triple system Most likely Inconclusive Most likely
UV excess source WD WD Active stars
detection). In order to calculate ρ(mlim ), we used the Besanc¸on galaxy
– Active galaxy – model3 (Robin et al. 2004) to generate a synthetic 2MASS H photo-
– Active stars – metric catalogue of point sources within 1 square degree, centred on
the coordinates of 2MASS J0628. To avoid underestimating the value
of P(, mlim ) we set a limiting magnitude of 16 in the H band, i.e.
three magnitudes fainter than the synthetic H2 and H3 magnitudes
of the companion candidate. We found that the probability of the
companion to be a background source is 0.001 per cent. It is worth
mentioning that this result, based on a synthetic catalogue, is purely
statistical and does not confirm the detected object being part
of 2MASS J0628, but rather allows to strengthen the hypothesis
about is triple nature, which needs to be ultimately confirmed via
common proper motion through new high contrast observations, as
the companion candidate is not resolved by Gaia.
Figure 2. Average HST/STIS spectrum of 2MASS J0628. A detailed extrac- With the evidence presented above, we conclude that
tion (see text for details) shows clear emission of C II and C IV. These emission 2MASS J0628 is most likely a main-sequence triple system with
lines may suggest the presence of stellar activity. at least one active component which produced the UV excess. This
system represents the first of the 11 targets in our sample observed
suggesting that the UV excess may be caused by chromospheric with HST that clearly does not contain a WD. Furthermore, by
and/or flare activity in at least one star in 2MASS J0628. placing 2MASS J0628 in the UV colour–temperature diagram of
This idea is further supported by the following reasoning. The NUV Parsons et al. (2016, their fig. 1), from which the PCEB candidate
flux from the stellar models of G stars does not include chromospheric assignment is given, we might expect that some candidates with
emission. This explains why the models fail to explain the NUV flux GALEX F U V − N U V colour close or greater than the one of
of active stars (see the magenta circle in Fig. 5 for 2MASS J0628). A 2MASS J0628 (F U V − N U V ≈ 3.6) contain UV sources different
similar discrepancy between the GALEX photometry, the HST data, to WDs, in particular, active stars.
and the stellar model for the AFGK star components has also been
observed in one of the six confirmed WD + AFGK PCEBs in our
survey: TYC 110−755−1 (Hernandez et al. 2022). The WD model 4.2 TYC 6996−449−1
that fits best the HST spectrum is consistent with the GALEX FUV Previous HST observations of TYC 6996 using STIS were obtained
flux, but fails to reproduce the observed GALEX NUV flux (see their by Parsons et al. (2016) under program GO 13704. They confirmed
fig. 9). The authors conclude that chromospheric emission from the the UV excess is due to a WD. However, the flux of the HST spectrum
G star is the most likely explanation for the flux difference. Based for this object falls substantially below its measured GALEX magni-
on these two systems, it could be inferred that (1) if the GALEX tudes (see bottom panel of Fig. 5), giving inconsistencies between the
FUV/NUV and HST/AFGK model fluxes disagree then variability distance of the main-sequence star obtained from the RAdial Velocity
due to stellar activity might occur in the binary, and (2) if the GALEX Experiment (Kordopatis et al. 2013) survey and the WD distance they
FUV and HST fluxes agree but the GALEX NUV flux is larger than estimated. As initially suggested by Parsons et al. (2016), we show
the synthetic one, this indicates steady chromospheric emission from below that this flux difference is because TYC 6996 is most likely
the AFGK star. a spatially resolved hierarchical triple with the WD as the tertiary
Furthermore, our SPHERE observations reveal the presence of a
companion candidate at an angular separation of 0.13 arcsec, corre-
sponding to a projected separation of ≈36 au using the photogeomet- 2 https://astro.acri-st.fr/gaia dev/#extinction
ric distance from Bailer-Jones et al. (2021). To calculate its apparent 3 https://model.obs-besancon.fr/

MNRAS 512, 2625–2635 (2022)

2630 F. Lagos et al.
2022), in which case the WD could still retain a fraction of the
intrinsic proper motion of the former triple.
The presence of triple systems with WDs in the survey has
been previously discussed in Lagos et al. (2020), who found that
the WD detected in TYC 7218 is in fact a tertiary companion
to a main-sequence binary star. They derived an upper limit of
15 per cent for contaminants (i.e. non-PCEB systems) consisting

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
of hierarchical triples with WDs and M-dwarf companions, of which
in 80 per cent of the cases the WD is the third object. We therefore
conclude that TYC 6996 is (or at least was) most likely a triple
system, with the close companion to the F-type star being a low-
mass star with a minimum mass of ≈0.5 M based on its orbital
solution.

4.3 TYC 8097−337−1

Based on the orbital solution derived from the radial velocity curve,
the minimum mass of the companion to the G-type star is ≈0.6 M .
Since this system was not observed with HST, we cannot confirm
the presence of a WD, and like 2MASS J0628, the companion to
the G-type star could, in principle, be an active star producing
the observed UV excess. Additionally, IRDIS images reveal the
presence of two extended sources at ≈2.4 arcsec from the binary,
Figure 3. GALEX FUV image of TYC 6996 (black cross) and the WD (black which are most likely background galaxies that also might be the
star). The SPHERE/IRDIS field of view centred in TYC 6996 is shown as source of the UV excess. This is the first time in our survey
a green circle. For display purposes, the image has been smoothed with a
that we have detected not one, but two potential non-stellar UV
Gaussian kernel.
sources.
To quantify the FUV/NUV flux excess produced by the unknown
companion. This caused the STIS slit to not be properly centred on UV source, we perform (as in Section 4.1) synthetic FUV/NUV
the WD losing a substantial part of its UV flux. photometry using a BT-NextGen spectral template with effective
Using PCA in the IRDIS science data cube we detected the temperature Teff = 5700 K, surface gravity log g = 4, and metallicity
presence of a potential tertiary companion at ≈6.2 arcsec from [Fe/H] = 0, according to the stellar properties of the AFGK
TYC 6996, close to the edge of the IRDIS field of view (see the star given in Table 1. While in the NUV band the difference
bottom-left panel of Fig. 1). This detection is consistent with a UV between observed and synthetic magnitudes (mNUV, GALEX = 17.06,
source detected by GALEX at the same coordinates, at ≈ 6.1 arcsec mNUV, synt = 17.22) is ≈ 0.16 mag, the difference in the FUV band is
from TYC 6996 as shown in Fig. 3, confirming the idea that the WD ≈ 5.6 mag (mFUV, GALEX = 19.93, mFUV, synt = 25.52). This translates
is either a resolved companion of TYC 6996 or a background source. to UV magnitudes of the unknown UV excess source of mNUV ≈
According to the Gaia EDR3 parallax, the WD is located somewhere 19.5 and mFUV ≈ 19.9.
between approximately 330 and 400 pc (Bailer-Jones et al. 2021), in To investigate if the presence of background galaxies can explain
agreement with the estimated distance of TYC 6996 (392 pc, see the UV excess, we used the GALEX UV atlas of nearby galaxies
Table 1) which supports the triple hypothesis. of Gil de Paz et al. (2007), which mostly contains regular (i.e.
In their search of wide companions around stars hosting giant non-active galactic nuclei, star-forming, or UV-luminous) galaxies.
planets and brown dwarfs, Fontanive et al. (2019) found that relative Since the galaxies in this catalogue have a diameter >1 arcsec,
differences of ≤20 per cent in the Gaia DR2 parallaxes and in at most FUV and NUV magnitudes were scaled to such a distance that the
one of the proper motion coordinates between the wide companion projected angular area of the galaxy is roughly equal to that of the
candidates and the planet-host star may suggest that they are likely two galaxies found with SPHERE, i.e. ≈ 2.3 × 10−5 arcmin2 for a
part of a hierarchical system. Following this criterion, we calculate direct comparison with the catalogue. Given that the rescaled FUV
the relative differences in proper motion (α ∗ and δ) and parallax magnitude of the brightest galaxies in the catalogue is ≈26 mag, it
( π ) between the WD and TYC 6996 using their Gaia EDR3 seems very unlikely that two regular background galaxies are the
astrometric solutions, finding that α ∗ = 0.3 ± 1.1, δ = 22.9 ± 1, unknown UV source. However, if one or both extended sources
and  π = 8 ± 14 per cent. The calculated relative differences meet are active galactic nuclei galaxies, the UV excess can be perfectly
the criterion that favours the triple nature of TYC 6996. However, explained as these types of galaxies can easily reach mFUV  20
the large error of the WD parallax (π = 2.31 ± 0.35) translates into mag (e.g. Welsh, Wheatley & Neil 2011).
a large uncertainty of  π , and therefore we cannot discard that In an alternative scenario, the UV excess might be caused by
its real value could be slightly above the threshold of 20 per cent. stellar activity. We find that using a BT-NextGen template with
However, if the latter was true, and taking into account that the orbital Teff = 4000 K, log g = 4.5, and metallicity [Fe/H] = 0 for a com-
separation of the WD from the inner binary is at least ≈2431 au, panion with 0.6 M , its estimated quiescent FUV magnitude is ≈ 39.
the system may also be a triple system that was disrupted due to This result may suggest that it is extremely unlikely that the observed
triple evolution, dynamical instability, galactic tides, encounters with FUV excess is due to stellar activity. However, the spectral template
passing field stars, or interactions with giant molecular clouds (e.g. used does not include the chromospheric component, which may con-
Weinberg, Shapiro & Wasserman 1987; Correa-Otto & Gil-Hutton tribute  99 per cent of the total emission in the FUV band and reach
2017; Toonen, Boekholt & Portegies Zwart 2021; Hamers et al. flux densities similar to the one observed in TYC 8097 (e.g. Stelzer

MNRAS 512, 2625–2635 (2022)

 Evidence for a bi-modal period distribution? 2631

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
Figure 4. 5σ contrast curve in the H2 filter for a full-frame ADI-PCA using
10 principal components (red line) for TYC 8097. The solid and dashed blue
lines represent the magnitude difference between the AFGK star and the two
best-fitting models for a WD companion as the source of the UV excess (see
the text for more details). The intersection of each blue line with the red line
at ≈ 0.27 arcsec represents the limiting angular separation at which the WD
would be visible with SPHERE/IRDIS. The peak observed at ≈2.4 arcsec is
due to the presence of the background galaxies. Curve generated with VIP.

et al. 2013). We therefore conclude that stellar activity cannot be

excluded.
Finally, we test whether a WD may be responsible for the UV
excess as follows. Using the WD cooling models4 from Holberg &
Bergeron (2006), Kowalski & Saumon (2006), Tremblay, Bergeron &
Gianninas (2011), and Bédard et al. (2020), we made a chi-square
minimization to the archival NUV/FUV GALEX fluxes, leaving
as a free parameter the distance of the WD. To include in the
minimization the FUV/NUV flux contribution from the AFGK star we
perform synthetic photometry using the same BT-NextGen spectral
template described above. Synthetic magnitudes were reddened as Figure 5. Best-fitting model spectrum of the AFGK star (green lines) and
explained in Section 4.1. We found that the best fit for the UV excess GALEX photometry (red stars) for all the three systems. The HST spectrum
of 2MASS J0628 (black line) did not confirm the presence of a WD and
corresponds to a WD with Teff = 14 500 K, M = 0.926 M , log g =
its flux integrated over the GALEX FUV filter is insufficient to explain the
8.5, and RWD = 0.97 R⊕ , located 356 pc away, in agreement with observed UV excess. Furthermore, the NUV flux obtained from the model of
the distance given in Table 1. In addition, we repeat the chi-square the G star (magenta circle) is ≈2.5 times less than the observed by GALEX.
minimization, but this time leaving as a free parameter the WD We conclude that transient events of stellar activity are responsible for the
cooling models and keeping fixed the distance at 351 pc, finding that enhanced GALEX UV flux. For TYC 8097 we also plot the best WD model fit
the best-fitting model corresponds to a WD with Teff = 17 000 K, (Teff = 14 500 K, log g = 8.5, and M= 0.926 M , blue line) to the observed
M = 1.2 M , log g = 9, and RWD = 0.63 R⊕ , being consistent UV excess. While this is clearly just one possible solution, it illustrates that
with the WD model at 365 pc. Fig. 4 shows that a WD with similar the UV excess can be explained by the presence of a WD. For TYC 6996
characteristics to our best-fitting models would have been detected we show the HST spectrum (black line) presented also in Parsons et al. 2016
in our SPHERE/IRDIS images, at least at an angular separation from which the presence of the WD was confirmed together with the best
WD model fit (Teff = 16 500 K, log g = 8, and M= 0.616 M ) to the GALEX
from the AFGK star >0.27 arcsec (projected separation of ≈95 au).
photometry. Since the STIS slit was not properly centred on the WD, the
Given that the eccentricity of TYC 8097 excludes CE evolution as observed flux in the HST spectrum is smaller than the real one.
the formation channel of the binary, the WD might be a tertiary with
a close orbit completely hidden by the coronagraph, or we might just the fit highly degenerate).
have been unlucky and the current location of a wide WD tertiary is We conclude that in the absence of HST observations, the UV
just behind the coronagraph. Although this interpretation is also in source could be either an active galaxy, an active star, or a WD.
agreement with the expected fraction of triple systems in the survey
with the WD being the third companion, like TYC 7218 and most
likely TYC 6996, the above scenario must be taken with caution as 5 DISCUSSION
we only used two photometric points in the chi-square minimization In the WD binary pathways survey we have so far discovered six
(Fig. 5) and only the FUV flux constrains the WD model (making short orbital period systems (Porb ࣠ 2.5 d) with circular orbits. These
systems can easily be interpreted as PCEBs. At somewhat longer
orbital periods (weeks), we have so far not yet identified a single
4 http://www.astro.umontreal.ca/∼bergeron/CoolingModels/ PCEB but instead four systems, three in this work and one in Lagos

MNRAS 512, 2625–2635 (2022)

2632 F. Lagos et al.
et al. (2020), with eccentric orbits, which are most likely main-
sequence binaries or the inner binaries of hierarchical triple systems,
either with no WD at all or with the WD being the distant tertiary
companion candidate. In what follows we discuss the implications
of this finding for our survey as well as for our understanding of WD
binary star formation.

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
5.1 Expected contribution from different types of contaminants
Despite the effectiveness of the UV excess criterion used to identify
WDs with short orbital periods around AFGK-type stars (PCEB
candidates), the sample is not free of contaminants, and even if
the UV excess comes from a WD, the system may not be formed
through CE evolution. In this regard, the orbital characterization
of each candidate, and in particular its eccentricity has become an
additional indicator of potential contaminants in the sample. The
three eccentric PCEB candidates presented here reveal at least three
main types of contaminants: (1) hierarchical triples with the AFGK Figure 6. Observed relation between orbital period and secondary mass for
star being orbited by an M-type star, with a WD as the third object WD + MS binaries. Red dots are PCEBs with M-type secondaries from
(AFGK/M + WD configuration); (2) binary (triple) systems with Nebot Gómez-Morán et al. (2011); magenta stars are WD + AFGK binaries
the AFGK star being orbited by one (two) M-type star(s) (AFGK/M from Wonnacott et al. (1993), O’Brien et al. (2001), Kruse & Agol (2014),
configuration for binaries and AFGK/M + M for triples), of which Kawahara et al. (2018), Masuda et al. (2019), and Krushinsky et al. (2020);
one or both are the source of the UV excess due to chromospheric blue squares are PCEBs with AFGK secondaries from The White Dwarf
activity; and (3) back(fore)ground UV sources like WDs or non- Binary Pathways Survey (Parsons et al. 2015; Hernandez et al. 2021, 2022);
and black triangles are the contaminants found in our survey presented in this
stellar sources like galaxies.
work and in Lagos et al. (2020).
The fraction of AFGK/M + WD triples in the survey was derived
in Lagos et al. (2020), ranging from 1 up to 15 per cent of the
sample of PCEB candidates. To roughly assess the fraction of type- we expect their contribution to be negligible compared with type-(1)
(2) contaminants, we use an approach similar to that in section 5 of and (2) contaminants.
Lagos et al. (2020). For a population composed of 107 single, binary,
and triple systems, the number of PCEBs with AFGK secondaries
5.2 Possible transition between systems formed through CE
is ≈2 × 105 (also including a small fraction formed in hierarchical
evolution and stable non-conservative mass transfer?
triples). From the initial mass function of Kroupa, Tout & Gilmore
(1993), we estimate that ≈3.4 per cent of the population are AFGK On the one hand, our survey has successfully identified six PCEBs,
stars (for the case of single stars) or have an AFGK star as the all of them with orbital periods <2.5 d. On the other hand, the four
most massive component (for the case of binary and triple systems). contaminants identified so far have orbital periods in the range 10 ࣠
By assuming that 8 per cent of those population of AFGK stars are Porb ࣠ 42 d. This shows that our survey is capable to detect wider
triples with uniform mass ratio distributions for the inner and outer WD + AFGK binaries thanks to the high precision of our radial
companions (Tokovinin 2014), we found that ≈0.06 per cent of the velocity measurements. Therefore, if PCEBs with periods >2.5 d
population have triple systems with configuration AFGK/M + M would exist in significant numbers, we should have detected them
(assuming that M-type stars are those with masses between 0.085 and with our observing strategy.
0.6 M ). Compared with the number of PCEBs, the fraction of triples In order to find possible explanations for this lack of wider WD +
with configuration AFGK/M + M is ≈3 per cent. Taking into account AFGK PCEBs, it is illustrative to look at the secondary mass-period
the estimated activity lifetimes for M5-M0 and >M5-type dwarfs of relation of confirmed WD + AFGK binary stars. In Fig. 6 (similar to
up to 2 and 8 Gyr, respectively (West et al. 2011), and a constant the fig. 6 of Hernandez et al. 2021), we show white dwarf plus main-
star-formation rate, the final fraction of AFGK/M + M systems with sequence (WD+MS) binaries collected from the literature and those
at least one active star is reduced to ≈1 per cent. Using the same discovered in our survey. Apart from the six PCEBs identified by our
reasoning, but a binary fraction of 33 per cent (Tokovinin 2014), survey, two more WD + AFGK systems have similar short orbital
we found that the fraction of AFGK/M systems with stellar activity periods, V471 Tau (Porb ≈ 0.52 d; O’Brien, Bond & Sion 2001) and
relative to the number of PCEBs is  9 per cent. Although the total GPX-TF16E-48 (Porb ≈ 0.3 d; Krushinsky et al. 2020). In the period
fraction of type-(2) contaminants (≈10 per cent) is close to the upper range where we found four systems that are clearly not PCEBs, only
limit estimated for AFGK/M + WD contaminants (≈15 per cent), we one WD + AFGK binary is known, IK Peg (Porb ≈ 21.7 d, e ≈ 0.03;
must recall that the obtained value represents an upper limit, as we Wonnacott, Kellett & Stickland 1993). At orbital periods longer than
assume that all AFGK/M and AFGK/M + M contaminants bypass 80 d, a population of WD + AFGK binaries has been discovered
the UV excess criterion. through the self-lensing effect (Kawahara et al. 2018; Masuda et al.
The probability for spurious alignments with back(fore)ground 2019). We note that a number of WD + A star binaries with similarly
UV sources is very difficult to measure as the fraction of this long orbital periods are suspected to be present in the Kepler sample
type of contaminants depends on the type of UV source taken statistically analysed by Murphy et al. (2018).
into account and on how crowded the region of the sky around Fig. 6 indicates a paucity of WD + AFGK binaries in the period
the candidate is. However, since most of the PCEB candidates range between a few days and more than two months. The only
have galactic latitude and longitude in the range |b|  30 deg and system in this period range is IK Peg. Given that four contaminating
200  l  360 deg, i.e. avoiding the crowded regions of the galaxy, systems but not a single PCEB in this period range have been

MNRAS 512, 2625–2635 (2022)

 Evidence for a bi-modal period distribution? 2633
discovered by our survey, this paucity is clearly not caused by the mass ratio qi = Ma /Md between the accretor and the WD progenitor is
fact that longer orbital periods are more difficult to measure by our above a critical mass ratio qcrit . For the particular case of conservative
radial velocity survey. mass transfer and donors in the range 1 ࣠ Md (M ) ࣠ 6, qcrit ≈ 1–0.7
Another striking feature illustrated in Fig. 6 is the large difference at the termination of the red giant branch (Ge et al. 2020), meaning
between the results obtained in previous surveys on WD + M binaries that stars that are able to evolve off the main sequence within the
(Nebot Gómez-Morán et al. 2011) and the results of our current Hubble time and develop cores (WDs) with masses ≈0.5 M (as
survey of WD + AFGK. In case of the first, long orbital period those observed in SBL1, SBL2, and SBL3) can undergo stable mass

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
(between 80 and 1000 d) systems do not seem to exist, i.e. only transfer when the accretor is an AFKG-type star. If a fraction of the
PCEBs with periods shorter than a few days and resolved systems mass lost by the donor escapes from the binary (non-conservative
with separations exceeding 100 au are known (Farihi, Hoard & case), then the value of qcrit may be somewhat lower compared to the
Wachter 2010; Ashley et al. 2019). Our survey of WD + AFGK conservative mass transfer case.
binaries finds PCEBs with similar characteristics; however, there Indeed, for KIC 8145411, one of the long orbital period post mass
seems to exist a population of systems with orbital periods of several transfer WD + AFGK systems, Masuda et al. (2019) suggests a for-
months which is not seen in the M dwarf sample. In what follows we mation channel involving stable (and most likely non-conservative)
discuss possible implications of these findings. mass transfer. Furthermore, non-conservative stable mass transfer,
Binary evolution simulations of PCEBs carried out by Zorotovic unlike CE evolution, can also explain the long orbital periods of
et al. (2014) showed that, for systems with Porb ࣠ 10 d, the observed SBL1, SBL2, and SBL3 and KIC 8145411. As shown in Woods
relation between the orbital period and the mass of the secondary is et al. (2012), the final semimajor axis is positively correlated with
well reproduced for a wide range CE and recombination energy effi- mass-loss rate and initial mass ratio (note that they used qi = Md /Ma
ciencies (0.25 ≤ α CE ≤ 1.0 and 0 ≤ α rec ≤ 0.25, respectively, see their instead), being able to increase the orbital period by a factor of five
fig. 5). However, WD + AFGK binaries with Porb  100 d (such as for a binary with qi ≈ 1, Md ≈ 1.1 M , and initial orbital period
SBL1, SBL2, SBL3, and KIC 8145411) are only reproduced for mod- of ≈100 d. Finally, if the longer period WD + AFGK stars are
els with significant contributions from recombination energy (i.e. descendants from stable mass transfer an obvious explanation for the
α rec = 0.25). Such strong contributions from recombination energy absence of WD + M binaries in this period range is available: For
would lead to the prediction of a relatively large number of WD + M these systems the mass ratio of the progenitor must have been below
PCEBs with periods between 10 d and several months which is clearly 0.5, i.e. the mass of the accretor (the M dwarf) was always too small
not observed. If all the systems (except of the contaminants of course) for stable mass transfer and CE evolution was unavoidable.
shown in Fig. 6 were indeed PCEBs, one would need to explain why While the outlined scenario is entirely consistent with the currently
nature switches from inefficient envelope extraction for WD + M available observations, as a note of caution, we stress that radial
PCEBs to very efficient envelope extraction including contributions velocity surveys of WD + M binaries have not been sensitive to
from recombination energy for a fraction of WD + AFGK binaries periods exceeding 10 d (Nebot Gómez-Morán et al. 2011) and that
(i.e. from the panels of the left hand side in fig. 5 of Zorotovic et al. the discovery of self-lensing WD binaries is biased towards systems
(2014) to the right hand panels of the same figure). Given that the with larger and brighter companions. A dedicated search of longer
amount of recombination energy that is used to unbind the envelope period WD + M binaries with a negative result would further confirm
is neither theoretically nor observationally well constrained, (see e.g. our hypothesis.
Ivanova 2018; Soker, Grichener & Sabach 2018), we cannot exclude In addition, given that only 10 systems in our survey are well char-
such a dependence of the CE efficiencies on the secondary mass. acterized so far (including both PCEBs and contaminants), we admit
However, we propose an alternative scenario which we think can that despite being plausible, the above outlined interpretation of the
naturally explain the observations illustrated in Fig. 6. observational result remains uncertain. By increasing the number of
We suggest that a different mechanism than CE evolution is well-characterized post-mass transfer WD + AFGK binary stars, we
responsible for the formation of wide WD + AFKG binaries (with can further test our hypothesis and provide crucial constraints on the
periods of months), and that the period range between 10 and 100 d, two formation channels. This is particularly relevant in the context of
populated so far only by our contaminants and two WD + AFGK SNe Ia progenitors, as WD + AFGK binaries evolving through stable
binaries (IK Peg and KOI-3278), might indeed contain very few post non-conservative mass transfer might experience a subsequent phase
mass transfer WD binaries. Assuming a universal and low value for of non-stable mass transfer and CE evolution (Woods et al. 2012),
the CE efficiency, like the one obtained for PCEBs with M-type leaving as end product a double WD binary that may contribute to
secondaries (α CE ≈ 0.2–0.3; Zorotovic et al. 2010), PCEBs mostly the SNe Ia rate via the double degenerate scenario.
populate the period domain up to 10 d, predicting only a tail of very
few systems with longer periods up to periods of Porb ≈ 100 d (see
6 CONCLUSION
fig. 2 of Zorotovic et al. 2014). CE evolution can explain all the WD
+ M systems as well as the short orbital period WD + AFGK binaries We report the discovery of three WD + AFGK close binary
and probably even IK Peg and KOI-3278 as two of the rare longer star candidates whose radial velocity curves revealed unexpected
period PCEBs if a very small fraction of the available recombination eccentric orbits, suggesting that these systems cannot have formed
energy contributed (Zorotovic et al. 2014). Under these assumptions, through CE evolution. By combining high-resolution imaging
however, CE evolution does not predict WD + AFGK with periods and spectroscopy techniques we confirm the latter, and conclude
of several months as observed and we therefore propose that these that two of them are most likely hierarchical triple systems.
systems are not PCEBs. Instead, dynamically stable non-conservative 2MASS J06281844−7621467 has a main-sequence inner binary
mass transfer (e.g. Kawahara et al. 2018) represents a reasonable consisting of G-type primary star and a M- or K-type companion,
candidate to be the mechanism responsible for the formation of these which is in turn orbited by another M or K-type star. Since the HST
wider WD + AFGK binaries. spectrum only shows C II and C IV emission lines, the source of the
The condition for stable mass transfer, under the assumption of UV excess in this system is most likely due to one or more active stars.
adiabatic response of the donor, is generally met when the initial TYC 6996−449−1 has a main-sequence inner binary, consisting of

MNRAS 512, 2625–2635 (2022)

2634 F. Lagos et al.
an F-type primary star and a K- or M-type companion, being orbited Correa-Otto J. A., Gil-Hutton R. A., 2017, A&A, 608, A116
by an outer WD. The remaining system, TYC 8097−337−1, could Dekker H., D’Odorico S., Kaufer A., Delabre B., Kotzlowski H., 2000, in Iye
be either a binary consisting of a slightly evolved primary G-type M., Moorwood A. F., eds, Proc. SPIE Conf. Ser. Vol. 4008, Optical and
star orbited by a K- or M-type lower mass companion, or a triple IR Telescope Instrumentation and Detectors. SPIE, Bellingham, p. 534
Dohlen K. et al., 2008, Proc. SPIE Conf. Ser. Vol. 7014, Ground-based and
system with a WD as the third companion.
Airborne Instrumentation for Astronomy II. SPIE, Bellingham, p. 70143L
The fact that only contaminants have been found in the orbital
Eastman J., Gaudi B. S., Agol E., 2013, PASP, 125, 83
period range between 10 and 40 d, may suggests two different Farihi J., Hoard D. W., Wachter S., 2010, ApJS, 190, 275

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
mechanisms to form close WD + AFGK binaries: CE evolution for Findeisen K., Hillenbrand L., Soderblom D., 2011, AJ, 142, 23
orbital periods less than ≈ 10 d, and stable non-conservative mass Fontanive C., Rice K., Bonavita M., Lopez E., Mužić K., Biller B., 2019,
transfer for orbital periods above ≈ 100 d. Since binaries evolving MNRAS, 485, 4967
through the latter mechanism are expected to end as double WDs Foreman-Mackey D., Hogg D. W., Lang D., Goodman J., 2013, PASP, 125,
after a phase of unstable mass transfer and CE evolution, they 306
could play an important role in the formation of SNe Ia through Gaia Collaboration, 2021, A&A, 649, A1
the double degenerate scenario. Given the still low number of Ge H., Webbink R. F., Chen X., Han Z., 2020, ApJ, 899, 132
Gil de Paz A. et al., 2007, ApJS, 173, 185
known WD + AFGK binaries, further characterization of additional
Gomez Gonzalez C. A. et al., 2017, AJ, 154, 7
candidates is required to test this hypothesis.
Hamers A. S., Perets H. B., Thompson T. A., Neunteufel P., 2022, ApJ, 925,
178
Hernandez M. S. et al., 2021, MNRAS, 501, 1677
AC K N OW L E D G E M E N T S
Hernandez M. S. et al., 2022, MNRAS, 512, 1843
FL was supported by the ESO studentship and the National Agency Hodge P. E. et al., 1998, in Albrecht R., Hook R. N., Bushouse H. A., eds, ASP
for Research and Development (ANID) Doctorado Nacional, grant Conf. Ser. Vol. 145, Astronomical Data Analysis Software and Systems
number 21211306. MRS acknowledges financial support from Fondo VII. Astron. Soc. Pac., San Francisco, p. 316
Holberg J. B., Bergeron P., 2006, AJ, 132, 1221
Nacional de Desarrollo Cientı́fico y Tecnológico (FONDECYT, grant
Ivanova N. et al., 2013, A&AR, 21, 59
numbers 1181404 and 1221059). SGP acknowledges the support
Ivanova N., 2018, ApJ, 858, L24
of the Science and Technology Facilities Council (STFC) Ernest Kaufer A., Pasquini L., 1998, in D’Odorico S., ed., Proc. SPIE Conf. Ser. Vol.
Rutherford Fellowship. OT was supported by a FONDECYT project 3355, Optical Astronomical Instrumentation. SPIE, Bellingham, p. 844
321038. BTG was supported by the UK Science and Technology Kawahara H., Masuda K., MacLeod M., Latham D. W., Bieryla A., Benomar
Facilities Council (STFC) grant ST/T000406/1. MSH acknowledges O., 2018, AJ, 155, 144
support through a Fellowship for National PhD students from Kordopatis G. et al., 2013, AJ, 146, 134
ANID, grant number 21170070. This project has received funding Kowalski P. M., Saumon D., 2006, ApJ, 651, L137
from the European Research Council (ERC) under the European Kroupa P., Tout C. A., Gilmore G., 1993, MNRAS, 262, 545
Union’s Horizon 2020 research and innovation programme (grant Kruse E., Agol E., 2014, Science, 344, 275
Krushinsky V. et al., 2020, MNRAS, 493, 5208
agreement no. 101020057). We thank the anonymous referee for
Lagos F. et al., 2020, MNRAS, 494, 915
his/her very helpful comments and suggestions on the manuscript.
Lallement R., Babusiaux C., Vergely J. L., Katz D., Arenou F., Valette B.,
We furthermore thank Mónica Zorotovic, Robert de Rosa, Anna Pala, Hottier C., Capitanio L., 2019, A&A, 625, A135
and Patricia Arévalo for helpful discussions and suggestions on this Loyd R. O. P., Shkolnik E. L., Schneider A. C., Barman T. S., Meadows V.
work. S., Pagano I., Peacock S., 2018, ApJ, 867, 70
Marois C., Lafrenière D., Doyon R., Macintosh B., Nadeau D., 2006, ApJ,
641, 556
DATA AVA I L A B I L I T Y Masuda K., Kawahara H., Latham D. W., Bieryla A., Kunitomo M., MacLeod
M., Aoki W., 2019, ApJ, 881, L3
The data underlying this article will be shared on reasonable request
Murphy S. J., Moe M., Kurtz D. W., Bedding T. R., Shibahashi H., Boffin H.
to the corresponding author.
M. J., 2018, MNRAS, 474, 4322
Naoz S., 2016, ARA&A, 54, 441
Nebot Gómez-Morán A. et al., 2011, A&A, 536, A43
REFERENCES
Nelemans G., Verbunt F., Yungelson L. R., Portegies Zwart S. F., 2000, A&A,
Allard F., Homeier D., Freytag B., 2012, Phil. Trans. R. Soc. A, 370, 2765 360, 1011
Amara A., Quanz S. P., 2012, MNRAS, 427, 948 O’Brien M. S., Bond H. E., Sion E. M., 2001, ApJ, 563, 971
Ashley R. P., Farihi J., Marsh T. R., Wilson D. J., Gänsicke B. T., 2019, Parsons S. G. et al., 2015, MNRAS, 452, 1754
MNRAS, 484, 5362 Parsons S. G., Rebassa-Mansergas A., Schreiber M. R., Gänsicke B. T.,
Astropy Collaboration et al., 2018, AJ, 156, 123 Zorotovic M., Ren J. J., 2016, MNRAS, 463, 2125
Bailer-Jones C. A. L., Rybizki J., Fouesneau M., Demleitner M., Andrae R., Press W. H., Teukolsky A. A., Vetterling W. T., Flannery B. P., 2007,
2021, AJ, 161, 147 Numerical Recipes. The Art of Scientific Computing, 3rd edn. Cambridge
Bédard A., Bergeron P., Brassard P., Fontaine G., 2020, ApJ, 901, 93 Univ. Press, Cambridge
Beuzit J.-L. et al., 2019, A&A, 631, A155 Rebassa-Mansergas A. et al., 2017, MNRAS, 472, 4193
Bianchi L., Herald J., Efremova B., Girardi L., Zabot A., Marigo P., Conti Ren J. J. et al., 2020, ApJ, 905, 38
A., Shiao B., 2011, Ap&SS, 335, 161 Robin A. C., Reylé C., Derrière S., Picaud S., 2004, A&A, 416, 157
Bianchi L., Shiao B., Thilker D., 2017, ApJS, 230, 24 Rodrigo C., Solano E., 2020, Contributions to the XIV.0 Scientific Meeting
Blanco-Cuaresma S., Soubiran C., Heiter U., Jofré P., 2014, A&A, 569, A111 (virtual) of the Spanish Astronomical Society. The SVO Filter Profile
Brahm R., Jordán A., Espinoza N., 2017, PASP, 129, 034002 Service XIV.0, Barcelona, SEA, p. 182
Brandner W. et al., 2000, AJ, 120, 950 Rodrigo C., Solano E., Bayo A., 2012, SVO Filter Profile Service Version
Claudi R. U. et al., 2008, in McLean I. S., Casali M. M., eds, Proc. SPIE 1.0, IVOA Working Draft 15 October 2012
Conf. Ser. Vol. 7014, Ground-based and Airborne Instrumentation for Shkolnik E. L., Barman T. S., 2014, AJ, 148, 64
Astronomy II. SPIE, Bellingham, p. 70143E Soker N., Grichener A., Sabach E., 2018, ApJ, 863, L14

MNRAS 512, 2625–2635 (2022)

 Evidence for a bi-modal period distribution? 2635
Soummer R., Pueyo L., Larkin J., 2012, ApJ, 755, L28
Stelzer B., Marino A., Micela G., López-Santiago J., Liefke C., 2013, Table A1 – continued
MNRAS, 431, 2063
Tokovinin A., 2014, AJ, 147, 87 BJD (mid-exposure) RV (km s−1 ) Err (km s−1 ) Instrument
Tokovinin A., Fischer D. A., Bonati M., Giguere M. J., Moore P., Schwab C.,
Spronck J. F. P., Szymkowiak A., 2013, PASP, 125, 1336 2457753.736 − 17.528 0.500 UVES
Toonen S., Nelemans G., 2013, A&A, 557, A87 2457760.689 6.203 0.500 UVES
Toonen S., Boekholt T. C. N., Portegies Zwart S., 2021, Stellar triples on 2457787.586 − 34.963 0.500 UVES

Downloaded from https://academic.oup.com/mnras/article/512/2/2625/6547013 by Lund University Libraries, Head Office user on 21 November 2023
the edge; Comprehensive overview of the evolution of destabilised triples 2457797.547 − 40.193 0.500 UVES
leading to stellar and binary exotica, https://arxiv.org/abs/2108.04272
Tremblay P.-E., Bergeron P., Gianninas A., 2011, ApJ, 730, 128 Table A2. Secondary star radial velocity measurements for TYC
Vigan A., 2020, Astrophysics Source Code Library, record ascl:2009.002 8097−337−1.
Vigan A., Moutou C., Langlois M., Allard F., Boccaletti A., Carbillet M.,
Mouillet D., Smith I., 2010, MNRAS, 407, 71 BJD (mid-exposure) RV (km s−1 ) Err (km s−1 ) Instrument
Webbink R. F., 1984, ApJ, 277, 355
Weinberg M. D., Shapiro S. L., Wasserman I., 1987, ApJ, 312, 367 2457386.724 48.370 0.010 FEROS
Welsh B. Y., Wheatley J. M., Neil J. D., 2011, A&A, 527, A15 2457387.640 43.159 0.010 FEROS
West A. A. et al., 2011, AJ, 141, 97 2457387.718 42.778 0.010 FEROS
Wonnacott D., Kellett B. J., Stickland D. J., 1993, MNRAS, 262, 2457388.695 38.390 0.018 FEROS
277 2457389.603 34.955 0.010 FEROS
Woods T. E., Ivanova N., van der Sluys M. V., Chaichenets S., 2012, ApJ, 2457471.552 43.303 0.011 FEROS
744, 12 2457586.921 24.652 0.010 FEROS
Zorotovic M., Schreiber M. R., Gänsicke B. T., Nebot Gómez-Morán A., 2457643.893 28.042 0.500 UVES
2010, A&A, 520, A86 2457645.824 24.803 0.500 UVES
Zorotovic M., Schreiber M. R., Garcı́a-Berro E., Camacho J., Torres S., 2457655.836 82.594 0.500 UVES
Rebassa-Mansergas A., Gänsicke B. T., 2014, A&A, 568, A68 2457674.689 61.560 0.500 UVES
2457680.816 46.421 0.500 UVES
2457684.782 30.497 0.500 UVES
APPENDIX A: RADIAL VELOCITY 2457711.565 23.734 0.500 UVES
MEASUREMENTS 2457712.817 24.490 0.500 UVES
2457723.621 41.890 0.500 UVES
2457725.542 34.282 0.500 UVES
Table A1. Secondary star radial velocity (RV) measurements for
2MASS J06281844−7621467.
Table A3. Secondary star radial velocity measurements for TYC
BJD (mid-exposure) RV (km s−1 ) Err (km s−1 ) Instrument 6996−449−1.

2457002.698 − 5.709 0.010 FEROS BJD (mid-exposure) RV (km s−1 ) Err (km s−1 ) Instrument
2457003.719 3.619 0.010 FEROS
2457004.723 6.349 0.010 FEROS 2456833.851 − 20.966 0.022 FEROS
2457025.740 5.803 0.500 Du Pont echelle 2456835.781 − 12.243 0.025 FEROS
2457026.735 2.613 0.500 Du Pont echelle 2457001.563 − 22.285 0.024 FEROS
2457027.704 − 3.866 0.500 Du Pont echelle 2457002.645 − 17.019 0.026 FEROS
2457028.749 − 16.597 0.500 Du Pont echelle 2457003.648 − 12.729 0.021 FEROS
2457269.813 5.767 0.976 CHIRON 2457025.548 16.137 0.500 Du Pont echelle
2457277.905 − 14.416 0.500 CHIRON 2457026.569 15.903 0.500 Du Pont echelle
2457291.845 3.494 0.500 CHIRON 2457027.586 16.080 0.500 Du Pont echelle
2457292.847 − 2.470 0.500 CHIRON 2457188.839 14.403 0.027 FEROS
2457299.769 2.753 0.500 CHIRON 2457255.781 − 12.018 0.500 CHIRON
2457303.876 − 10.430 0.500 CHIRON 2457266.632 8.456 0.500 CHIRON
2457317.805 − 36.106 0.500 CHIRON 2457269.684 11.963 1.515 CHIRON
2457332.714 3.576 0.500 CHIRON 2457276.712 15.748 0.500 CHIRON
2457365.629 − 16.683 0.500 CHIRON 2457283.761 13.860 0.500 CHIRON
2457378.850 − 41.177 0.500 UVES 2457297.742 − 12.244 0.500 CHIRON
2457386.668 − 26.322 0.010 FEROS 2457311.706 12.334 0.500 CHIRON
2457389.630 − 27.349 0.010 FEROS 2457313.682 13.547 0.500 CHIRON
2457439.599 − 49.261 0.500 UVES 2457319.630 16.206 0.500 CHIRON
2457472.531 − 4.665 0.010 FEROS 2457333.623 − 27.295 0.527 CHIRON
2457674.735 − 45.979 0.500 UVES 2457340.552 − 9.372 0.500 UVES
2457712.834 − 16.520 0.500 UVES 2458600.893 − 7.305 0.055 FEROS
2457723.627 − 26.336 0.500 UVES
2457744.722 − 37.877 0.500 UVES
This paper has been typeset from a TEX/LATEX file prepared by the author.

MNRAS 512, 2625–2635 (2022)