1
Survival of a brown dwarf after engulfment by a red
giant star
P.F.L. Maxted
1
, R. Napiwotzki
2
, P.D. Dobbie
3
, M.R. Burleigh
3
Many sub-stellar companions (usually planets but also some brown dwarfs) have
been identified orbiting solar-type stars. These stars can engulf their sub-stellar
companions when they become red giants. This interaction may explain several
outstanding problems in astrophysics
1-
5
but is poorly understood, e.g., it is unclear
under which conditions a low mass companion will evaporate, survive the
interaction unchanged or gain mass.
1,
4,
5
Observational tests of models for this
interaction have been hampered by a lack of positively identified remnants, i.e.,
white dwarf stars with close, sub-stellar companions. The companion to the pre-
white dwarf AA Doradus may be a brown dwarf, but the uncertain history of this
star and the extreme luminosity difference between the components make it
difficult to interpret the observations or to put strong constraints on the models.
6,
7
The magnetic white dwarf SDSS J121209.31+013627.7 may have a close brown
dwarf companion
8
but little is known about this binary at present. Here we report
the discovery of a brown dwarf in a short period orbit around a white dwarf. The
properties of both stars in this binary can be directly observed and show that the
brown dwarf was engulfed by a red giant but that this had little effect on it.
1
Astrophysics Group, Keele University, Keele, Staffordshire, ST5 5BG, UK.
2
Centre for Astrophysics Research, University of Hertfordshire, College Lane, Hatfield
AL10 9AB, UK.
3
Department of Physics and Astronomy, University of Leicester, University Road,
Leicester, LE1 7RH, UK
2
WD0137-349 (BPS CS 29504-0036) was first noted as an unremarkable faint,
blue star in a survey for metal poor stars.
9
Improved spectroscopy showed it to be a
white dwarf.
10
We obtained high resolution spectra of WD0137-349 as part of the SPY
programme to identify the progenitors of Type Ia supernovae.
11
We noticed features in
these spectra due to a low mass companion in a close orbit so we obtained further
observations (shown in Figure 1) to measure the orbital period and mass ratio (Table 1).
From the mass of the white dwarf (Table 2) and the mass ratio we derive a mass for the
companion of (0.053±0.006)M
(solar masses). This mass is well below the limit of
about 0.075M
commonly used to distinguish stars from brown dwarfs.
12
Brown
dwarfs, by definition, are not massive enough to support core hydrogen burning but do
undergo a brief phase of deuterium burning soon after they form. They then start to
cool, so the spectral type of a brown dwarf, which is a measure of its temperature, is
also a measure of its age. The observed infrared flux distribution of WD0137-349
(Figure 2) is consistent with a model of an old brown dwarf companion with a mass of
0.055M
but inconsistent with models for companions that are young brown dwarfs or
stars. The orbital period of WD0137-349 is approximately 116 minutes and the stars are
separated by only 0.65R
(solar radii).
Brown dwarf companions to white dwarfs are rare – less than 0.5% of white
dwarfs have a brown dwarf companion at any separation.
13
However, there are many
white dwarfs that are known to have a low mass star as a companion in a short period
orbit. In cataclysmic variable stars (CVs) mass transfer onto the white dwarf from the
companion through the inner Lagrangian point produces strong emission lines in the
optical spectrum. No such lines are seen in our data for WD0137-349 so we conclude
that it is not a CV. There are several good candidates for sub-stellar companions in CVs,
but the extra light due to accretion makes it difficult to confirm their masses.
14,
15
3
Short period white dwarfs binaries with low mass companions that do not transfer
mass are known as pre-CVs because the loss of orbital angular momentum by
gravitational wave radiation (GWR) and other mechanisms will result in the shrinkage
of the orbit, the initiation of mass transfer and the formation of a CV.
16
The timescale
for WD0137-349 to become a CV through the loss of GWR is about 1.4Ga, at which
time the orbit period will be 60 – 80 minutes. This is close to the minimum orbital
period seen in CVs. This raises the possibility that some fraction of CVs with very low
mass companions may have formed as the result of the evolution of binaries like
WD0137-349, rather than by extensive mass loss from the companion. A simulation of
the population of pre-CVs formed from binaries like WD0137-349 shows that most of
these binaries will have evolved from a solar-type star with a brown dwarf companion
separated by a few au.
17
Although a few such systems have been found
18
, such binaries
are known to be rare
19
so it is likely that the contribution of stars like WD0137-349 to
the total CV population is a few percent or less.
The simulation of the population of binaries like WD0137-349 assumes that white
dwarfs with close, low mass companions are the result of “common envelope
evolution”. In this scenario, the more massive star in a binary system becomes a red
giant once it has exhausted hydrogen in its core. The red giant will interact with its
companion when its radius becomes comparable to the separation of the binary. The
details of the interaction are uncertain but some low mass companions will be engulfed
by the red giant, i.e., the core of the red giant and the low mass companion share a
common envelope. If the companion is not sufficiently massive to force the envelope to
co-rotate with its orbit, the drag on the companion will cause it to quickly spiral in
towards the core of the red giant. Some fraction of the orbital energy released, α
CE
, will
be deposited as kinetic energy in the envelope, which is ejected from the binary system.
The radius of a red giant is determined principally by the mass of its core. This is
effectively the mass of the resulting white dwarf, so we can calculate the value of α
CE
4
required to explain the formation of WD0137-349 assuming a range of red giant masses,
M
g
. The lowest possible value of M
g
is 0.8 M
because stars less massive than this do
not evolve to the red giant stage within the lifetime of the Galaxy. This
provides a lower
limit of α
CE
≈0.6. This value of α
CE
is similar to that derived for pre-CVs with more
massive companions.
20
If the red giant was more massive than 1.25M
the value of α
CE
required exceeds 1. These values can be compared directly to the results of simulations
of the common envelope phase, although no such simulations for systems resembling
WD0137-349 are available to us at present. Simple physical arguments suggest that low
mass companions to red giants will be evaporated during the common envelope phase if
they are less massive than some limit m
crit
. The value of m
crit
is uncertain, but is
expected to be about 0.02M
.
1,
4,
5
The properties of WD0137-349 show that m
crit
is at
most 0.05 – 0.06M
. The mass of WD0137-349 (0.4M
) is lower than the typical mass
of a single white dwarf (0.6M
), as is expected for systems in which a common
envelope phase has prematurely removed the envelope of a red giant.
21
Some models predict that planets may accrete a substantial fraction of the mass in
the red giant envelope prior to a common envelope phase, resulting in the formation of a
binary with similar properties to WD0137-349.
4
Since most of the current mass of the
brown dwarf was accreted from the red giant in this scenario, its spectral type would
imply an age similar to that of the white dwarf, i.e. about 250 Ma in the case of
WD0137-349. The spectral type in this case is expected to be ≈L1-2 (T
eff
≈ 2200K). In
contrast, if the brown dwarf has always been close to its current mass, by the time the
common-envelope phase occurs the brown dwarf will have been cooling for the lifetime
of the solar-type star (giga-years). The common envelope phase proceeds on a
dynamical timescale of a few years, which is negligible when compared to this thermal
timescale so very little mass or heat can be gained by the brown dwarf in this phase. The
spectral type of the brown dwarf is then expected to be in the much cooler T-dwarf
range (T
eff
<1500K). The companion may appear to be slightly hotter than this because
5
it intercepts about 1% of the light from the white dwarf. The asymmetric heating and
rotation of the brown dwarf will produce a small but detectable modulation of the
brightness at infrared wavelengths of WD0137-349 on the orbital period (“reflection
effect”). An accurate measurement of the intrinsic spectral type and luminosity of the
brown dwarf will therefore require infrared spectroscopy and photometry at a range of
orbital phases to determine and account for the irradiation from the white dwarf.
Despite this complication, the existing infrared photometry (Figure 2) is more consistent
with a spectral type for the brown dwarf slightly earlier (hotter) than T5, rather than
with a spectral type of L1-2. Therefore, the existing data favour the scenario in which
WD0137-349 formed by a common envelope phase which had little effect on the brown
dwarf, rather than by accretion onto a planet.
1. Nelemans, G., Tauris, T. M., Formation of undermassive single white dwarfs
and the influence of planets on late stellar evolution. Astron. Astrophys. 335,
L85 – L88 (1998).
2. Soker, N., Harpaz, A., Rotation, planets, and the `second parameter' of the
horizontal branch. Mon. Not. R. Astron. Soc. 317, 861 – 866 (2000).
3. Harpaz, A., Soker, N., Evaporation of brown dwarfs in AGB envelopes. Mon.
Not. R. Astron. Soc. 270, 734 – 742 (1994).
4. Livio, M., Soker, N., Star-planet systems as progenitors of cataclysmic binaries:
tidal effects. Astron. Astrophys. 125, L12 – L15 (1983).
5. Siess, L., Livio, M., The accretion of brown dwarfs and planets by giant stars –
II. Solar-mass stars on the red giant branch. Mon. Not. R. Astron. Soc. 308, 1133
– 1149 (1999).
6. Rauch, T., AA Dor – An Eclipsing Subdwarf - Brown Dwarf Binary. Rev. Mex.
Astron. Astrophys. (Serie de Conferencias) 20, 246 – 246 (2004).
