A&A 420, 647-653 (2004)
DOI: 10.1051/0004-6361:20035713
R. Neuhäuser 1 - E. W. Guenther 2
1 - Astrophysikalisches Institut, Universität Jena,
Schillergäßchen 2-3, 07745 Jena, Germany
2 - Thüringer Landessternwarte Tautenburg, Sternwarte 5, 07778 Tautenburg, Germany
Received 19 November 2003 / Accepted 3 March 2004
Abstract
We present infrared H- and K-band spectra of
a companion candidate
north of the young star GSC 08047-00232,
a probable member of the nearby young Horologium association.
From previously obtained JHK-band colors and the magnitude difference between primary and
companion candidate, the latter could well be substellar (Neuhäuser et al. 2000)
with the spectral type being roughly M 7-L 9 from the JHK colors (Chauvin et al. 2003).
With the H- and K-band spectra now obtained with ISAAC at the VLT,
the spectral type of the companion candidate is found to be M 6-9.5.
Assuming the same age and distance as for the primary star (
Myr, 50 to 85 pc),
this yields a mass of
Jupiter masses for the companion,
hence indeed substellar. After TWA-5 B and HR 7329 B, this is the third
brown dwarf companion around a nearby (
100 pc) young (
100 Myr) star.
A total of three confirmed brown dwarf companions (any mass,
separation
50 AU)
around 79 stars surveyed in three young nearby associations
corresponds to a frequency of
(with a correction for missing companions which
are almost on the same line-of-sight as the primary star
instead of being separated well),
consistent with the expectation, if binaries have the same
mass function as field stars. Hence, it seems that
there is no brown dwarf desert at wide separations.
Key words: stars: late-type - stars: low-mass, brown dwarfs - stars: pre-main sequence
Brown dwarfs (BDs) are objects with mass below the hydrogen burning mass limit,
i.e. below
(e.g., Burrows et al. 1997).
It is not yet clear, how BDs form.
According to one of the suggested scenarios, accreting companions
get ejected as BD embryos by encounters inside a forming multiple system
(e.g., Reipurth & Clarke 2001).
In such a case, BDs may be different from stars regarding
multiplicity, kinematics, and disk properties,
so that such a scenario can be tested observationaly,
e.g. by comparing the secular evolution of
star-star binaries with star-BD binaries: If single BDs became single by the
ejection from a multiple system, then the fraction of star-BD binaries should
decrease faster with time than star-star binaries.
A possible secular evolution in stellar binaries has been studied by Bouvier et al. (2001) and Patience et al. (2002) with as yet inconclusive results. In order to compare such results with star-BD binary fractions, one would first need significant statistics. Hence, several groups search for BDs as companions to stars.
As far as young stars (100 Myr) in nearby associations (
100 pc)
are concerned, only two brown dwarfs have been confirmed so far as companions
by both common proper motion as well as spectroscopy,
namely TWA-5 B (Lowrance et al. 1999; Neuhäuser et al. 2000)
in the TW Hya association (TWA, Webb et al. 1999)
and HR 7329 B (Lowrance et al. 2000; Guenther et al. 2001)
in the Horologium association (HorA, Torres et al. 2000)
and/or
Pic moving group (Zuckerman & Webb 2000).
In Sect. 2, we present previous and new imaging observation and investigate the astrometry, i.e. whether the pair is co-moving. Our spectroscopic observations, the data reduction, and the final H- and K-band spectra are shown in Sect. 3. We obtain a mass estimate for the companion and interprete the results in Sect. 4. A discussion in terms of BD companion frequency is given in the last section.
Table 1: Astrometry and photometry.
Recently, both Neuhäuser et al. (2000, henceforth N03) and
Chauvin et al. (2003, henceforth C03) have searched for faint
companions to confirmed and probable
members of the young Horologium and Tucana associations,
which may form one common association (TucHorA). N03 used the normal
infrared (IR) imaging camera SOFI and the MPE speckle camera Sharp,
both at the ESO-3.5 m-NTT, while C03 used the ADONIS AO system at the ESO-3.6 m
telescope. Their samples are partly overlaping and partly disjunct,
e.g. N03 also included
Pic members.
Both groups detected a promising companion candidate
north of GSC 08047-00232, a K3 V star suggested to be a Horologium member
by Torres et al. (2000) with H
emission
and a lithium equivalent width of 0.35 Å,
located at
and
(J2000.0);
the star is also called TYC 8047-232-1 (Tycho, Hog et al. 2000),
ERX 14 (Torres et al. 2000),
and 1RXSJ015215.7-521939 (ROSAT, Voges et al. 1999).
The position in the sky, the radial velocity, the proper motion,
and the kinematical parallaxe (and its position in the H-R diagram
at that distance) are all consistent with membership to HorA
(Torres et al. 2000) and also with the possible common TucHor association.
Also, the youth indicators like H
emission, lithium absorption,
X-ray emission, and rotational velocity are also consistent qualitatively
and quantitatively with an age of roughly 35 Myr as HorA/TucHorA (Torres et al. 2000).
Given the JHK-band colors, C03 found that the companion candidate could have a spectral type between M 7 and L 9 by comparing the Leggett et al. (2002) photometry with the Geballe et al. (2002) spectral classifications. From the magnitude difference in K, also N03 concluded that the companion candidate could be a BD companion of late-M or L spectral type.
To confirm or reject its companion status,
one can either try to (i) take a spectrum,
or (ii) prove the pair to be co-moving,
or (iii) show the radial velocity or astrometric
wobble of the primary due to the companion (both very small),
or (iv) detect orbital motion (after a long time).
Only the first option is immediately possible to achieve,
given the fact that the proper motion of the primary star is very
small (
mas/yr), almost exactly to the east,
so that the separation to the companion candidate (
north)
would hardly change, even if it is a background object.
Before obtaining a spectrum, we took aquisition images, which can be used to measure the separation between primary star and companion candidate. This separation should be constant with time, if its a common proper motion pair; a change in separation due to orbital motion in case of a non-circular orbit would be negligible.
![]() |
Figure 1:
Aquisition image through a narrow band filter in the K-band
taken with ISAAC at the VLT before our spectroscopy,
showing the companion candidate B three arcsec north of GSC 08047-00232 A.
The
![]() |
Open with DEXTER |
We compile the previous and new astrometric results in Table 1.
There are images at three epochs available now, first detected
on 04 July 2001 (N03), then on 29 Oct. 2001 (C03), and finally
on 20 July 2003 (here).
We measure the position of the primary star and the companion
candidate in every co-added image with center/gauss in MIDAS,
which provides errors.
Pixel scales, detector orientation, and their errors are measured on
calibration fields with many stars of known fixed positions taken
in the same nights
.
With these values, we can obtain separations and position
angles (PA), see Table 1.
Our three measurements - separated by two years - of the separation
and the PA are consistent with eachother within .
The companion candidate is located
almost directly north of the primary star
(only 1 to
west of the north direction)
and the primary star is known to move towards the east (just a bit south of east).
Hence, the separation should increase with time, if the companion
is a non-moving background star. Given the proper motion of the
primary star, it should have increased to
by July 2003. As observed (Table 1), the separation
has not changed within the errors.
The astrometry is
deviant from a background hypothesis.
However, due to the small proper motion and the small epoch difference,
we cannot yet reliably distinguish between the companion candidate being
either a common proper motion companion and a background star.
We obtained H- and K-band spectra with the Infrared Spectrometer
And Array Camera (ISAAC)
at Antu, Unit Telescope 1 (UT1) of the Very Large Telescope (VLT) of the
European Southern Observatory (ESO) on Cerro Paranal, Chile.
ISAAC is equipped with a
HAWAII detector.
All H- and K-band spectra obtained have 60 s exposure each and are taken
through a
slit with a low resolution of
.
In the H-band, centered on
m, we took 28 spectra on 20 July 2003.
In the K-band, centered on
m, we took 28 spectra on 11 July 2003
and again 28 spectra on 20 July 2003. All data were obtained in service mode.
Calibration frames were taken in the same nights: darks, flats, and
arcs for wavelength calibration.
We measured the full width at half maximum (FWHM) on the aquisition images
as well as in the individual spectra obtaining
for
both nights. The S/N in the (unbinned) K-band spectrum taken on 20 July 2003
is
24, but only
in the data taken on 11 July 2003,
so that we use the K-band spectrum from 20 July 2003.
Data reduction was done in the normal way with IRAF:
dark subtraction, normalization, flat fielding, sky subtraction,
wavelength calibration, co-adding the spectra, then
correction for instrumental sensitivity and atmospheric response.
We use the spectrum of the bright primary star
(located off the slit) for calibration purposes.
The S/N in the binned spectra is
for both H and K.
For the new companion candidate as well as for the spectral templates
HR 7329 and TWA 5 with their known companions, both the primary and the
secondary are observed simultaneously, so that we can do almost perfect
calibration of the continuum shapes by using the known spectral type
of the primary for estimating the response of the detector and, hence,
calibrating the secondary. This is different for the
spectral template Cha H
10.
However, even with templates observed simultaneously,
some uncertainties in the calibration remain,
so that spectral types obtained from spectral indices should
be more accurate than those obtained from continuum slopes.
![]() |
Figure 2:
H-band spectra of the companion candidate GSC 08047-00232/cc ( top)
compared to the H-band spectrum of the M 7.5-type young BDs Cha H![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 3:
H-band spectra of GSC 08047-00232/cc (middle) compared
with HR 7329 B (M 7-8, top) and TWA-5 B (M 8.5-9, bottom).
Here, we have smoothed the spectral resolution by a blocking factor of 5
yielding a resolution of
![]() ![]() |
Open with DEXTER |
![]() |
Figure 4: Rebinned K-band spectra of the companion candidate GSC 08047-00232/cc ( top) compared to the averaged K-band spectrum of M 7.5- to M 8-type young BDs ( bottom) in Chamaeleon (from Comerón et al. 2000), obtained with SOFI, an IR spectrograph similar to ISAAC, but at a different spectral resolution. Both spectra show the same absorption lines Na, Ca, and Mg, very similar molecular CO bands, and the same continuum shape, so that the companion is similar in spectral type as the template, i.e. M 7-8.5. |
Open with DEXTER |
The final co-added H- and K-band spectra are shown in Figs. 2 to 4 respectively. We compare the spectra of GSC 08047-00232/cc with those of young late-M type BDs in Chamaeleon, Tucana, and TWA.
The H-band spectrum is compared with Cha H
10
observed with ISAAC by us with the same set-up (in June 2000).
Cha H
10 is a young (
2 Myr) M 7.5 BD in
the Cha I star forming region (Comerón et al. 2000).
Both spectra are relatively featureless with the same
continuum shape indicative of a very similar spectral type.
The slope in the red part of the H-band at
1.7 to
m
is very sensitive to the temperature at late M and L spectral
types due to water absorption. As seen in Fig. 2, the red slope
in GSC 08047-00232/cc is slightly steeper than in Cha H
10;
hence, GSC 08047-00232/cc appears slightly cooler, i.e. slightly later,
around M 8-8.5,
while the broad maximum of the flux in GSC 08047-00232/cc is at a
slightly bluer wavelength compared to Cha H
10,
which would indicate a slightly earlier spectral type,
around M 6.5-7.
Hence, we classify the companion candidate tentatively as
M 6.5-8.5 (from the H-band continuum slopes).
Then, we estimate the spectral index H2O-B following Reid et al. (2001),
the ratio of the fluxes at 1.48 and m.
We obtain
for GSC 08047-00232/cc, hence M 8.5-9.5.
Then, the spectral index H2O-C is sensitive for the slope
in the red part of the H-band from 1.7 to 1.8
m,
i.e. very sensitive to the spectral type;
we obtain 0.56 for GSC 08047-00232/cc,
0.55 for TWA-5 B (M 8.5-9),
and 0.58 for HR 7329 B (M 7-8).
Hence, GSC 08047-00232/cc is intermediate between TWA-5 B and
HR 7329 B and can be classified as M 8-8.5 (from H2O-C).
Together with the above comparison of GSC 08047-00232/cc and
the M 7.5-type Cha H
10 yielding
M 6.5-8.5 (Fig. 2),
we classify GSC 08047-00232/cc as M
,
from the H-band.
The K-band spectrum (Fig. 4) is compared with an averaged K-band spectrum of
M 7.5- to M 8-type BDs in Chamaeleon (Comerón et al. 1999, 2000) obtained with
SOFI, because we did not take a K-band spectrum of Cha H 10 or any
other M 6-8 type Chamaeleon BD with ISAAC.
Here, we see again a very similar continuum shape as well
as several lines (Na, Ca, and Mg) similar in both spectra.
Also, the CO molecular absorption bands are very similar
in both spectra, so that we can classify GSC 08047-00232/cc
as roughly M 8 from the K-band, consistent with the H-band.
The K-band spectrum of GSC 08047-00232/cc is clearly later than
the average M 6 and average M 8-type spectra from Comerón et al. (1999).
To summarize, the companion candidate can be classified as
M
(or M 6-M 9.5)
from the H- and K-band spectra.
This classification is tentative and rough, because of the
relatively low spectral resolution, low S/N ratio.
Obviously, it is very unlikely to find by chance an
object with spectral type M just
off one of our targets.
According to the Besancon galactic model, to find an object as faint as
GSC 08047-00232/cc within
of the primary star at
the given galactic latitude is only
(C03) with sample
sizes being 25 stars in N03 and 23 stars in C03 with 13 overlaps; it is even
less likely, to find such an object with (at least roughly) the same proper
motion (see Table 1) and the late M spectral type.
Hence, the companion candidate most certainly is a truely bound companion.
For the remainder of the paper, we regard this companion candidate
as a bound companion of the star and call it GSC 08047-00232 B.
The primary star GSC 08047-00232 A, as a probable
member of Myr young TucHorA,
has a distance of
pc (Torres et al. 2000; Zuckerman & Webb 2000)
to as much as
pc (C03). The larger distance is obtained,
when forcing the star to be co-eval with the rest of the TucHorA stars
at
Myr with the Baraffe et al. (1998) models using the
hydrostatic scale-height
as mixing length parameter.
However, with e.g.
,
the star is again older.
On D'Antona & Mazzitelli (1998) tracks, the star is
30 to 50 Myr young at 85 pc, see Fig. 5.
Hence, we assume 85 pc as distance
and
Myr as age for both the primary star and its companion.
![]() |
Figure 5: H-R diagram for GSC 08047-0232 A and B (at 85 pc) with theoretical tracks and isochrones from D'Antona & Mazzitelli (1998), ages in Myr, masses in solar masses. The objects are co-eval at 30 to 50 Myr. |
Open with DEXTER |
Table 2:
Mass determination for the companion (
for Jupiter mass).
Given the spectral type M (Sect. 3), we obtain
2100 K
as temperature following Jones & Tsuji (1997) and Kirkpatrick et al. (1999),
or
K from the Luhman (1999) extrapolation of
the Leggett et al. (1996) scale for dwarfs
,
so that we use
K as temperature.
The companion has an absolute magnitude of
at 85 pc.
According to
mag (C03) and the spectral type M 6-9.5,
we use a bolometric correction in the K-band of
mag
following Leggett et al. (2002). With
mag (C03, N03),
we then obtain a bolometric luminosity of
for 85 pc
(or
for 50 pc).
With these values for temperature, luminosity, and age, we can obtain a
mass estimate using theoretical model calculations, see Table 2.
Using Fig. 7 in Burrows et al. (1997) for
at 35 Myr, we obtain
to 30
,
from Fig. 9 for
K at 35 Myr, we derive
to 35
,
and from Fig. 11 for
and
K,
we get
20 to 40
.
Using Fig. 2 (35 Myr)
in Chabrier et al. (2000) with
K at 35 Myr,
we obtain
to 40
;
from Table 1 in Chabrier et al. (2000) (i.e. for 100 Myr as their youngest age)
and using
K,
,
and
mag, we obtain
to 45
.
From Fig. 2 in Baraffe et al. (2002),
using L and T as above at 35 Myr,
we obtain a mass of
.
On D'Antona & Mazzitelli (1998) tracks and isochrones,
the primary star and the companion are co-eval at
85 pc with an age of
to 50 Myr and
with a mass of
to 50
for the companion
(see Fig. 5).
C03 show a color-magnitude diagram (CMD), MK versus J-K (their Fig. 4),
for the primary star with its companion candidate; they compare their locations
in the CMD at 60 and 85 pc distances with model calculations made by
Baraffe et al. (1998) and Chabrier et al. (2000) for 20 to 50 Myr;
they find the two objects to be co-eval for 85 pc only (at Myr)
with the mass of the companion then being
to 40
.
Hence, with both the D'Antona & Mazzitelli (1998) model and the
Baraffe et al. (1998) model, we consistently obtain
Myr
and
for the companion.
We have to caution, however, as discussed in Wuchterl (2001) and Baraffe et al.
(2002), that theoretical models are more uncertain for young ages (even tens of Myr)
as compared to older (Gyr) ages, because of uncertain initial conditions.
Hence, there could easily be a systematic shift in our mass estimates.
Extrapolating the Wuchterl & Tscharnuter (2003) models by scaling the luminosity
dependance in mass and age (to 35 Myr), one would obtain a mass estimate of
only
,
below the
deuterium burning mass limit (G. Wuchterl, private communication),
which we will study and discuss in more detail elsewhere.
To summarize, we obtain a mass estimate of
to 50
with
as most likely value, from conventional models by the
D'Antona & Mazzitelli, Chabrier & Baraffe, as well as Burrows et al. groups,
but possibly a lower mass when formation scenarios are taken into account.
GSC 08047-00232 B is probably a low-mass brown dwarf.
To confirm companionship of this late-M type object with the primary star beyond any
doubt, common proper (or radial) motion (or reflex motion of the primary due to the
companion) or orbital motion (
year orbit with
AU semi-major axis)
would have to be shown with high significance.
If the primary star GSC 08047-00232 would not be a member of HorA or TucHorA,
the assumed distance for primary and companion could be wrong. However,
the strong lithium absorption line (together with the other youth indicators
like strong X-ray emission, H
emission, and fast rotation) clearly
indicate that the primary star is a pre-main sequence star,
so that the spectral type of the companion,
M
,
alone points to a mass near or below the H burning limit.
Most young stars in the nearby TWA, TucHorA, and
Pic associations
have been surveyed by deep IR imaging (with speckle and/or AO and/or HST)
for substellar companions, so that we can try to do some statistics:
in TWA, there are 19 member systems (see Torres et al. 2003 for a recent list)
including visual multiples, so that there is a total number of 27 stars.
All of them have been surveyed by now with one BD companion found and
confirmed, namely TWA-5 B (Lowrance et al. 1999; Neuhäuser et al. 2000).
The sensitivity of the surveys are deep enough to have detected all
BD companions down to the D burning limit outside of roughly 50 AU;
however, they are quite inhomogenous, done by several groups with
several different telescopes and instruments.
There is a total of one BD companion among 27 stars,
which is a frequency of .
With imaging, one can detect only those companions separated from the
primary in the two dimensions of the plane of the sky,
i.e. those separated well in right ascension
and/or declination
.
One cannot detect those companions directly (or almost directly)
in front of or behind the primary, i.e. located (almost) on the same
line-of-sight as the primary, instead of having an angular separation.
By direct imaging, one can detect companions only in the two dimensions
and
,
not in the 3rd dimension
.
To correct the number above (from two to three dimenions, 2D to 3D),
we have to add
of the 2D value; hence, we obtain
(for TWA).
In the
Pic association, Neuhäuser et al. (2000) have done a
homogenous survey for companions down to the D burning limit outside
of
AU. Before, one BD companion was known, namely HR 7329 B
(Lowrance et al. 2000; Guenther et al. 2001). Neuhäuser et al. (2000)
did not find any additional BD companion (or candidates) among
a total of 17 stars in 12 systems.
Hence, the frequency of BD companions in
Pic is
(or
after correcting from 2D to 3D).
In TucHorA, 25 stars were observed by Neuhäuser et al. (2000) and
ten more by Chauvin et al. (2003), both down to roughly the D burning limit
for companions outside of AU. The number of BD companions here
is again one, namely the one confirmed above.
Hence, here we have a frequency of BD companions of
(or
for 3D).
In all three associations, the frequency of BD companions (of any mass
down to the D burning limit of
)
outside of roughly
50 AU separation is very low, around a few per cent.
The spectral types of the primaries ranges from A0 (HR 7329 A)
over K3 (here, GSC 08047-00232 A) to M2 (TWA-5 A).
Its still low number statistics with one BD companion each,
but since 17 to 35 stars have been observed, one can clearly
say that the frequency of wide BD companions (
50 AU) is quite low.
Putting together the three associations, we have three BDs around
79 stars, i.e. a frequency of
(or
after correction of the projection effect from 2D to 3D),
which is the
best estimate so far for the frequency of BDs (of any mass) outside of 50 AU
around
to 35 Myr young stars born in loose associations.
This value is consistent with the expectation, if the mass function of binaries
is similar to the mass function of field stars, in which case
to
of
stars should have BD companions (see Gizis et al. 2001),
so that there seems to be
no evidence for a BD desert at wide separation.
Acknowledgements
We would like to thank the ESO Service Mode observing team on Cerro Paranal as well as the ESO User Support Group for perfect support in both observing periods 65 and 71. M. Ammler and T. Schöning prepared Fig. 5. We also acknowledge Wolfgang Brandner and Günther Wuchterl for many stimulating discussions.