Near-Infrared Multi-Band Photometry of the Substellar Companion GJ 758 B

Abstract

GJ 758 B is a cold (~600K) companion to a Sun-like star at 29 AU projected
separation, which was recently detected with high-contrast imaging. Here we
present photometry of the companion in seven photometric bands from
Subaru/HiCIAO, Gemini/NIRI and Keck/NIRC2, providing a rich sampling of the
spectral energy distribution in the 1-5 micron wavelength range. A clear
detection at 1.58 micron combined with an upper limit at 1.69 micron shows
methane absorption in the atmosphere of the companion. The mass of the
companion remains uncertain, but an updated age estimate indicates that the
most likely mass range is ~30-40 Mjup. In addition, we present an updated
astrometric analysis that imposes tighter constraints on GJ 758 B's orbit and
identifies the proposed second candidate companion, "GJ 758 C", as a background
star.

Full text

Page 1

arXiv:1011.5505v1 [astro-ph.SR] 24 Nov 2010
Near-Infrared Multi-Band Photometry of the Substellar Companion
GJ 758 B∗
M. Janson1, J. Carson2, C. Thalmann3, M. W. McElwain4, M. Goto3, J. Crepp5, J. Wisniewski6,
L. Abe7, W. Brandner3, A. Burrows4, S. Egner8, M. Feldt3, C. A. Grady9, T. Golota8, O. Guyon8,
J. Hashimoto10, Y. Hayano8, M. Hayashi8, S. Hayashi8, T. Henning3, K. W. Hodapp11, M. Ishii8,
M. Iye10, R. Kandori10, G. R. Knapp4, T. Kudo10, N. Kusakabe10, M. Kuzuhara10,17, T. Matsuo12,
S. Mayama19, S. Miyama10, J.-I. Morino10, A. Moro-Mart´ ın13, T. Nishimura8, T.-S. Pyo8, E.
Serabyn12, H. Suto10, R. Suzuki10, M. Takami14, N. Takato8, H. Terada8, B. Tofflemire6, D.
Tomono8, E. L. Turner4,18, M. Watanabe15, T. Yamada16, H. Takami8, T. Usuda8, M. Tamura10
ABSTRACT
GJ 758 B is a cold (∼600K) companion to a Sun-like star at 29 AU projected separation, which was
recently detected with high-contrast imaging. Here we present photometry of the companion in seven
photometric bands from Subaru/HiCIAO, Gemini/NIRI and Keck/NIRC2, providing a rich sampling of
the spectral energy distribution in the 1–5 µm wavelength range. A clear detection at 1.58µm combined
with an upper limit at 1.69µm shows methane absorption in the atmosphere of the companion. The mass
of the companionremains uncertain, but an updatedage estimate indicates that the most likely mass range
is ∼30–40 Mjup. In addition, we present an updated astrometric analysis that imposes tighter constraints
onGJ 758B’s orbitandidentifiestheproposedsecondcandidatecompanion,“GJ 758C”, as a background
star.
Subject headings: planetary systems — brown dwarfs — techniques: high angular resolution
*Based on data collected at Subaru Telescope, which is operated
by the National Astronomical Observatory of Japan, on Gemini data
under program GN-2010A-Q-23, and on Keck data under N068N2.
1Univ. of Toronto, Canada; janson@astro.utoronto.ca
2College of Charleston, Charleston, South Carolina, USA
3Max Planck Institute for Astronomy, Heidelberg, Germany
4Dep. of Astroph. Sciences, Princeton Univ., Princeton, USA
5California Institute of Technology, Pasadena, USA
6University of Washington, Seattle, Washington, USA.
7Laboratoire Hippolyte Fizeau, Nice, France
8Subaru Telescope, Hilo, Hawai‘i, USA
9Eureka Scientific & Goddard Space Flight Center, USA
10National Astronomical Observatory of Japan, Tokyo, Japan
11Inst. for Astron., University of Hawai‘i, Hilo, Hawai‘i, USA
12Jet Propulsion Laboratory, Caltech, Pasadena, USA
13Dep. of Astroph., CAB - CSIC/INTA, Madrid, Spain
14Inst. of Astron. and Astroph., Academia Sinica, Taipei, Taiwan
15Dep. of Cosmosciences, Hokkaido University, Sapporo, Japan
16Astronomical Institute, Tohoku University, Sendai, Japan
17University of Tokyo, Tokyo, Japan
18Inst. for Ph. and Math. of the Universe, Univ. of Tokyo, Japan
1.Introduction
In recent years, a number of high-contrast compan-
ions have been thermally imaged around nearby stars
(e.g. Marois et al. 2008; Lagrange et al. 2010). One
interesting companion is GJ 758 B (Thalmann et al.
2009, hereafter P1). Its combination of a Sun-like
parent star (spectral type G8V) only 15.5 pc away
(Perryman et al. 1997), the close proximity to the star
(projected separation 29AU), and low surface temper-
ature (∼600 K) make it one of the most “planet-like”
objectsavailablefordirectstudy, actingas a laboratory
for current planet formation and evolution theories.
P1 presented two epochs of H-band imaging of
GJ 758 B that allowed for proof of common proper
motion and a first estimation of physical and orbital
properties.Furthermore, a candidate second com-
panion, tentatively called “GJ 758 C”, was found in
19Grad. Univ. for Adv. Studies, Kanagawa, Japan
1

Page 2

one epoch. Recently, Currie et al. (2010) published
L′-band data that confirmed the estimated tempera-
ture range of B but did not detect “C”. Both publi-
cations deduced similar results for best-fit orbital pa-
rameters (a ∼ 50 AU, e ∼ 0.7). Both also noted
that the mass range is ∼10–40 Mjupif the full range
of main-sequence ages for a Sun-like star is consid-
ered. In this work, we present photometriccoverageof
GJ 758B in the nearinfrared(JHKcL′M bands)with
additional photometry in the methane-sensitive CH4S
and CH4L narrow-band filters (see Table 1 for fil-
ter specifications), collected with high-contrast imag-
ing techniques at Subaru/HiCIAO, Gemini/NIRI, and
Keck/NIRC2. Thisisusedforupdatingtheestimations
of physical and orbital parameters of GJ 758 B. We re-
port the re-detection of “GJ 758 C” and the conclusive
identification of it as a background star by proper mo-
tion.
2. Observations and Data Reduction
H-band observations of GJ 758 B were obtained
with Subaru on November 11, 2009.
of AO188 (Hayano et al. 2010) coupled with HiCIAO
(Hodapp et al. 2008) as part of the SEEDS survey
(Tamura2009). Thefieldofview(FOV)was 20′′×20′′
with a 0.′′010 pixel scale. The total integration time
was 906 s.
Observations in J-, H-, CH4S-, CH4L-, and
Kc-band were obtained at Gemini North with the
Altair AO system (Herriot et al. 2000) and NIRI
(Hodapp et al. 2003). The observations were carried
out on 5 different nights between April 27 and May
08, 2010, with a 0.′′022 pixel scale over a 22′′×22′′
FOV. The total integration times were 900 s, 1800 s,
3810 s, 2220 s, and 2400 s, for the J-, H-, CH4S-,
CH4L-, and Kc- band, respectively. The nearby star
HD 226294 was observed before or after each ADI
dataset in the same filter to be used as a photometric
reference.
We imaged GJ 758 in the L′- and Ms-bands
on August 6, 2010 using the Keck II AO system
(van Dam et al. 2004) and NIRC2.
ameter coronagraphic mask was used for the L′ob-
servations to prevent saturation. Images without the
mask were also taken, to calibrate the brightness of the
star. For all observations, we avoided the bad quadrant
in the NIRC2 detector and used the 768×776 subar-
ray mode with a 0.′′010 plate scale. The integration
times were 1800 s and 1995 s for L′and Ms, respec-
making use
The 0.′′300 di-
tively.Three photometric standard stars (HD 162208,
HD 161903, Gl 748) were observed in both bands.
All data were taken usingangulardifferentialimag-
ing (ADI, e.g. Marois et al. 2006) with typical field
rotations of ∼30o, and the images were reduced with
the LOCI procedure (Lafreni` ere et al. 2007), using an
IDL implementation adapted for each of the instru-
ments. For the HiCIAO data, the same procedure as
described in P1 was used, including correction for the
partial subtraction of companion flux during the LOCI
reduction. The only practical differences for NIRI and
NIRC2 were the image registration procedures, where
the NIRI registration was performed on the basis of
cross-correlationof the diffractionspiders, and NIRC2
registration was caclulated by centroiding on the stel-
lar PSF core, which was non-saturated in the M-band
data and non-saturated behind the semi-transparent
mask in the L′-band data. Also, for the NIRC2 data,
high-pass filtering was applied at the same time as the
subtractionofaradialprofilefromthestellarPSF inall
images, to remove low-frequency spatial variations in
the high thermal background of the L′- and Ms-band
(e.g. Janson et al. 2008). Fluxes of all observedtargets
and standard stars were extracted using aperture pho-
tometry. The reduced images are shown in Fig. 1 and
Fig. 2.
3. Photometric Analysis
3.1. Calibration
Photometric calibration is of central importance for
these data, due to the fact that the JHKs-photometry
of GJ 758 A has been flagged as unreliable in 2MASS
(Skrutskie et al. 2006), and furthermore,for the broad-
band filters it is always saturated in NIRI images, even
for the shortest available integration times in subarray
mode. Hence, we used the standard star for photo-
metric calibration in all filters, with the exception of
CH4S. Those data were taken on Apr 29 under photo-
metricallyunstableconditions,sothat thestandardstar
photometry couldn’t be trusted. Instead, we calibrated
the CH4S flux based on the fluxes of the background
stars in the GJ 758 field in CH4L and CH4S, under the
assumption that they are equally bright in both filters,
which is reasonable since they are stars and cannot ex-
hibit methane absorption. Four background stars were
used with a dispersion of 0.19 mag. The brightness of
the background stars in CH4L was calibrated against
the standard star. The resulting absolute magnitudes
of GJ 758 B are summarized in Table 1.
2

Page 3

Fig. 1.— High-contrast imaging of GJ 758 B with ADI/LOCI. In all panels, the star is located approximately in
the upper left corner, North is up, and East is left. The location of GJ 758 B is marked with a solid arrow, that of
“GJ 758 C” with a dashed arrow. (a) HiCIAO H-band image (Nov 2009). GJ 758 B is clearly visible, whereas “C” is
marginally detectable at ∼3σ. (b–f) Gemini/NIRI images in five filters (Apr/May 2010). (g–h) NIRC2 images in L′-
and Ms-band (Aug 2010).
3

Page 4

Fig. 2.— Combination of Gemini images. (a) Co-add of J, H, Kc, and CH4S, after normalizing the flux levels such
that GJ 758 B appears approximately equally bright in each image. The “C” source is detectable. (b) A map of the
pixels that exhibit a S/N ratio of ≥+1 in all four images used for (a). Only GJ 758 B and “C” pass the test, indicating
that they are both real on-sky objects.
4

Page 5

GJ 758 A has no published L′-band or Ms-band
magnitudes, thus we use the standard stars to calibrate
those values. All three standard stars give consistent
results within 0.1 mag, yielding ML′ = 3.7 ± 0.1
mag and MMs = 3.6 ± 0.1 mag. During the sci-
ence exposures, GJ 758 A was unsaturated in the Ms-
band and so could be used to evaluate the brightness
of GJ 758 B. In the case of L′, the star was behind
the semi-transparent mask, which makes it unreliable
for photometry, hence for this case, standard star pho-
tometry was used directly to calibrate the brightness
of GJ 758 B. The results are in Table 1. There are two
cases where special circumstances apply: For CH4L,
the flux at the position of the companion is visually
unconvincing as a point source, and more reminiscent
of residual noise from the stellar PSF. It is therefore
possible that GJ 758 B is systematically overestimated
at this wavelength, such that the quoted CH4L flux
should rather be regarded as an upper limit rather than
the true brightness. In the case of Ms-band, there is
no visually clear signature of the companion,but aper-
ture photometry was performed on the same location
as the L′-band detection. This yielded an excess flux
at the ∼2σ level, hence this is a plausible flux of the
companion, since its position is known a priori. Aside
from CH4L and Ms, all detections are ∼10σ confi-
dence. For NIRI, the photometric error that we cal-
culate is dominated by an uncertainty in the linearity
behaviour of the detector at low counts/short integra-
tions, which we estimate could impose an error of at
most 20%.
3.2. Interpretation
GJ 758 B exhibits clear methane absorption. This
is generally expected for objects in the late T-type
range,althoughto a decreasedextentif theatmosphere
is in chemical non-equilibrium (e.g. Burgasser et al.
2006; Fortney et al. 2008). GJ 758 A has a metal-
licity of approximately +0.2 dex (e.g. Holmberg et al.
2009; K´ osp´ al et al. 2009). Assuming that the compan-
ion has the same composition, this super-solar metal-
licity is likely at least partly responsible for the high
flux in Kc. Since the commonly used COND models
(Allard et al. 2001; Baraffe et al. 2003) do not contain
any non-solar abundances at the relevant temperature
range, we do not use them for model comparison. In-
stead, we use models based on Burrows et al. (2006)
extended to colder temperatures (Hubeny & Burrows
in prep.). With our new adopted age range discussed
below, the feasible values of logg for the companion
are all close to 5.0, hence we use this as a baseline
value in our fitting. The comparison was done by fit-
ting model spectra to the flux densities of the photo-
metric points, which were derived from the magnitude
values using the NIRI and NIRC2 filter transmission
curves. All fluxes were normalized to a distance of
10 pc. The best-fit value was acquired using the mini-
mized χ2in the photometric bands as the quality met-
ric. The models cover temperatures of 500–700 K in
steps of 50 K, logg of 4.0–5.0 in steps of 0.5 dex, and
metallicities of +0.0 and +0.5 dex (note that the full
grid is not covered– in particular, no super-solar abun-
dace models exist below 600 K). The best-fit temper-
ature of this procedure is 600 K, in agreement with
previous results. A metallicity of +0.5 dex fits better
than +0.0 dex, which is consistent with the metal en-
richment expected in the system. However, since sur-
face gravity and metallicity are largely degenerate for
the wavelengths covered by our study and the models
are very uncertain, we do not attempt to optimize the
fit with respect to either of these quantities. Indeed,
the models are known to be unable to reproduce vari-
ous features of other cool companionsto stars (e.g. the
HR 8799 planets, see Marois et al. 2008; Janson et al.
2010; Hinz et al. 2010; Bowler et al. 2010), so deriv-
ing absolute values from model comparison is of lim-
ited relevance. Four examples of model fits are shown
in Fig. 3.
We also make a comparison with field brown
dwarfsfromLeggett et al.(2010). Forthispurpose,we
use the J-band absolute brightness MJ= 17.58±0.20
mag, and the color J − H = 0.58 ± 0.28, mag, since
this should be relatively insensitive to metallicity and
gravity. The companion fits very well to the tempera-
turesequenceoftheLeggett et al.(2010)fielddwarsin
an HR-diagram, among the latest-type brown dwarfs
(T8–T9). The J − H color alone with uncertainties
places the companion firmly beyond T5. A T8–T9
spectral type fits well to the temperature of ∼600 K
derived from model comparison.
In P1, we adopted an age range of 0.7–8.7 Gyr, on
thebasisofavarietyofageindicators. The0.7Gyrage
wasbasedonisochronalfittingbyTakeda et al.(2007).
However, other isochronal fits have been performed
yielding substantially higher ages of several Gyr (e.g.
Valenti & Fischer 2005; Holmberg et al. 2009). This
indicatesthatisochronalfittingis inadequatefordating
GJ 758, hence we revise our age estimates by remov-
ing the isochrone method, which yields a residual age
range of ∼5–9 Gyr (keeping estimates based on activ-
5

Page 6

ity and rotation). As an additional age determination
test, we have analyzed a high-resolution (R∼31,500)
spectrum of GJ 758 from the Apache Point Observa-
tory 3.5m. We detect no clear evidence of the Li I
6708 ˚ A doublet at an upper limit of ∼4 m˚ A equiv-
alent width. If Li had been detected, it would have
indicated a young age, but the non-detection does not
provide a strong lower limit on the age, as it only sug-
gests that GJ 758 should be older than the ∼100 Myr
old Pleiades cluster (Maldonado et al. 2010). We re-
iterate that age determination of main-sequence Sun-
like stars is highly uncertain, and that the adopted age
rangeof∼5–9Gyrshouldnotbeconsidereddefinitive,
but merely represents the range of mean ages from
methods that are considered sufficiently reliable. The
correspondingrangeofcompanionmassesfromevolu-
tionary models is ∼30–40 Mjup. The most promising
avenue for getting a reliable age estimate of GJ 758,
and thus a better constraint on the mass of its compan-
ion, is likely asteroseismology.
4. Astrometric Analysis
4.1.Data points
In P1, we demonstratedthe common proper motion
of GJ 758 and GJ 758 B on the basis of two epochs
of observation, May 3, 2009 (E1) and August 6, 2009
(E2). In this work, we include three additional epochs
in our analysis: Our data from November 1, 2009
(E3) and April 29, 2010 (E4), as well as the May 27,
2010 data point from Currie et al. (2010) (E5). The
Keck/NIRC2 data from August 2010 (E6) have a nar-
rowfieldofviewexcludingthe5–7knownbackground
stars thatwere usedtofine-tunethe pixelscales andro-
tation angles in our previous observation. Thus, they
do not deliver sufficient astrometric accuracy.
Figure4ashowstheresultingpositionsofallnearby
point sources relative to GJ 758 at E1–E4, as well
as the data point for GJ 758 B in E5. The E4 data
points represent the mean positions of the sources in
the four datasets that reveal GJ 758 B, i.e. the J,
H, Kc, and CH4S data. The error bars for the Hi-
CIAO data (first three epochs) are based on the pixel
size of the HiCIAO camera, 9.5mas. For the Gem-
ini data, the pixel size is 22mas, yielding an error of
22mas/√4 = 11mas for the combined data points.
Theseerrorsare consistentwiththe scatter ofthe back-
ground star data points around the projected motion
path, given the expected contributions from the proper
motions of the individual background stars. For the
E5 data point, we assume an isotropic error of 10mas
to represent the anisotropic error bars of 5 and 15mas
shown in Currie et al. (2010).
We observe that GJ 758 B pursues a trajectory to
the northwest consistent with orbital motionrelative to
GJ 758, clearly setting it apart from the background
star trajectory, which is dominated by GJ 758’s known
parallactic and proper motion. The source tentatively
referred to as “GJ 758 C” in P1 is found to follow
the background star track; while this was still indis-
tinguishable from orbital motion at the 2σ level in
November 2009, the May 2010 data are unambiguous.
This also allows us to identify the candidate signal in
Currie et al. (2010) as spurious.
4.2. Orbital Monte Carlo simulation
With a total of N = 10 scalar parameters (2 coordi-
nates × 5 epochs), the astrometric data on GJ 758 B
is now extensive enough to fit synthetic orbital so-
lutions generated by a Monte Carlo simulation with
the least-squares method rather than with the simpli-
fied approach previously used in P1 and Currie et al.
(2010). However, since the curvature of the orbit is
not yet measurable, the benefit of this improvement is
limited.
We generate a large number (>106) of orbital tra-
jectories with random values for eccentricity e, in-
clination i, argument of periastron ω, and longitude
of the ascending node Ω. The distributions are pre-
sumed to be flat, except for the inclination, where
larger angles are favored proportionately to sini in
order to represent their higher geometric likelihood.
The two remaining orbital parameters, the semimajor
axis a and the mean anomaly at epoch M0, are im-
plicitly chosen by defining an anchor point (xA,yA)
in the projected image plane where the companion is
located a given epoch tA. In order to achieve a high
production rate of valid orbital solutions, we choose
tA to be the mean of the five observational epochs,
?tobs?, and generate (xA,yA) randomly in a box of
20mas×20mas centered on the mean astrometric co-
ordinates (?xobs?,?yobs?).
In order to evaluate an orbital solution for consis-
tency with the data, we determine the predicted po-
sition of the companion at the five epochs and calcu-
late the χ2deviation. The minimum best fit achieves
χ2= 1.58. We select theset oforbitswithχ2< 2.5 ≈
χ2
Figure 4b illustrates the distribution of semima-
min+ 1 to represent the “good fits”.
6

Page 7

Fig. 3.— Photometricanalysis of GJ 758B. The red plus signs show the measuredflux values normalizedto a distance
of10pc forthe sevenfilter bands,whose wavelengthdomainsaremarkedwith blackbars. Theorangecurvesrepresent
model spectra for different assumptions of Teff, logg, and metallicity, and the black diamonds are the resulting flux
levels in the filter bands.
7

Page 8

jor axis a, eccentricity e and inclination i for 2516
“good fit” orbits. The solutions appear to lie in a two-
dimensional manifoldin the three-dimensionalparam-
eter space. We note that more stringent fitting require-
ments (e.g. χ2< 1.8) do not relevantly reduce the
size of this manifold, showing that this spread is due
to model degeneracy rather than fitting errors. The
degeneracy represents the fact that only the projected
position of GJ 758 B can be tracked, leaving the line-
of-sight component of its position and velocity unde-
termined. In order to break this degeneracy, it is nec-
essary to measure the curvature of the projected orbit.
Given the typical predicted orbital periods of several
centuries, this requires monitoring on time scales of at
least a decade.
The numerical results are summarized in Table 1.
We note that the weighted median eccentricity has
dropped from 0.73 in Currie et al. (2010) to 0.46; this
is because the new epochs E3 and E4 suggest that the
previousepochs were overestimatingthe orbital veloc-
ity.
We thank David Lafreni` ere for providing us with
the source code for LOCI, Eric Mamajek for use-
ful discussion, the staff at Subaru, Gemini and Keck
for their support, and an anonymous referee for use-
ful suggestions. The Gemini time was allocated by
NOAO. The Keck telescope was funded by the W.M.
Keck foundation, and time was allocated by NASA
through partnership with Caltech and UC. We ac-
knowledge the cultural significance that the summit of
Mauna Kea has to the indigenous Hawaiian commu-
nity. Part of the research was supported by NSF grants
AST 1009203,AST 0802230,AST 1009314,the AAS
Chretien grant, a Grant-in-Aid for Specially Promoted
Research,theMitsubishiFoundation,andJPL,Caltech
under NASA contract.
Facilities:
Subaru (HiCIAO, AO188), Gemini
North (NIRI), Keck (NIRC2).
Table 1: Numerical Results
Photometry (mag)
MJ(1.15-1.33 µm)
MH(1.49-1.78 µm)
MKc(2.08-2.11µm)
ML′ (3.43-4.13µm)
MMs(4.55-4.79 µm)
MCH4S(1.53-1.63 µm)
MCH4L(1.64-1.74µm)
GJ 758 A
—
—
—
3.7±0.1
3.6±0.1
—
—
GJ 758 B
17.58±0.20
18.16±0.20
17.12±0.20
15.0±0.1
≥13.9+0.5
17.74±0.20
≥18.86±0.20
−0.8
GJ 758 B orbital parameters
Semimajor axis a (AU)
Eccentricity e
Inclination i (deg)
Period P (yr)
Weighted median
41.8
0.464
50.6
270
68% interval
28.4–79.1
0.204–0.670
31.6–62.2
151–703
GJ 758 B’s position relative to GJ 758
∆RA (′′)
∆Dec (′′)
−0.574
−0.579
−0.597
−0.609
−0.616
— insufficient astrometry —
Epoch
E1: May 3, 2009
E2: Aug 8, 2009
E3: Nov 1, 2009
E4: Apr 29, 2010
E5: May 27, 2010
E6: Aug 6, 2010
References. (1) P1, (2) this work, (3) Currie et al.
(2010). Note that the error bars for astrometry in E1
and E2 were mistakenly listed as 5mas in Table 1 in
(1); however, the text and plots use the correct value
of 9.5mas.
Error (′′)
0.010
0.010
0.010
0.011
0.010
Ref.
(1)
(1)
(2)
(2)
(3)
−1.789
−1.765
−1.751
−1.735
−1.716
8

Page 9

Fig. 4.— Astrometric analysis. (a) Motions of point-sources near GJ 758 across five epochs (E1–E5), measured
relative to GJ 758’s position. GJ 758 B exhibits common proper motion with its parent star as well as systematic
orbital motion towards the northwest, whereas all other point-sources follow the expected trajectory for background
stars (solid arrows). The object referred to as “GJ 758 C” in P1 is unambiguously identified as a background star
(motion highlightedby dashed blue arrows). The grey plus signs are 1σ error bars. The circle marked as “PSF” shows
the size of the resolution element in H-band on HiCIAO. (b) Plot of eccentricity e against semimajor axis a for 2516
orbital solutions with χ2≤ 2.5 generated by the Monte Carlo simulation. The orbit selection is biased according to
the statistical weight ?v?/vobsas in P1.The inclination i is shown by color coding. The weighted median values of a
and e are marked with a white plus sign.
9

Page 10

REFERENCES
Allard, F., Hauschildt, P., Alexander, D., Tamanai, A.,
& Scweitzer, A. 2001, ApJ, 556, 357
Baraffe, I., Chabrier, G., Barman, T., Allard, F., &
Hauschildt, P. 2003, A&A, 402, 701
Bowler, B., Liu, M., Dupuy, T., & Cushing, M. 2010,
ApJ, 723, 850
Burgasser,A.,Geballe,T., Leggett,S., Kirkpatrick,D.,
& Golimowski, D. 2006, ApJ, 637, 1067
Burrows, A., Sudarsky, D., & Hubeny, I. 2006, ApJ,
640, 1063
Currie, T., Bailey, V., Fabrycky, D., Murray-Clay, R.,
Rodigas, T., & Hinz, P. 2010, ApJ, 721, L177
Fortney, J., Marley, M., Saumon, D., & Lodders, K.
2008, ApJ, 2008, 678, 1419
Hayano, Y., et al. 2010, Proc. SPIE, 7736, 21
Herriot, G., et al. 2000, Proc. SPIE, 4007, 115
Hinz, P., Rodigas, T., Kenworthy, M., Sivanandam, S.,
Heinze, A., Mamajek, E., & Meyer, M. 2010, ApJ,
716, 417
Hodapp, K. W., et al. 2003, PASP, 115, 1388
Hodapp, K. W., et al. 2008, Proc. SPIE, 7014, 42
Holmberg, J., Nordstrom, B., & Andersen, J. 2009,
A&A, 501, 941
Janson, M., Reffert, S., Brandner, W., Henning, Th.,
Lenzen, R., & Hippler, S. 2008, 488, 771
Janson, M., Bergfors, C., Goto, M., Brandner, W., &
Lafreni` ere, D. 2010, ApJ, 710, L35
K´ osp´ al,´A, Ardila, D.R., Mo´ or, A., &´Abrah´ am, P.
2009, ApJ, 700, L73
Lafreni` ere, D., Marois, C., Doyon, R., Nadeau, D., &
Artigau, E. 2007, ApJ, 660, 770
Leggett, S.K., et al. 2010, ApJ, 710, 1627
Lagrange, A.-M., et al. 2010, Science, 329, 57
Maldonado, J., Martinez-Arnaiz, R.M., Eiroa, C.,
Montes, D., & Montesinos, B. 2010, A&A, 521, 12
Marois, C., Lafreni` ere, D., Doyon, R., Macintosh, B.,
& Nadeau, D. 2006, ApJ, 641, 556
Marois, C., Macintosh, B., Barman, T., Zuckerman,
B., Song, I., Patience, J., Lafreni` ere, D., & Doyon,
R. 2008, Science, 322, 1348
Perryman, M.A.C. et al. 1997, A&A, 323, 49
Skrutskie, M.F. et al. 2006, AJ, 131, 1163
Takeda, G., Ford, E.B., Sills, A., Rasio, F., Fischer,
D.A., & Valenti, J.A. 2007, ApJS, 168, 297
Tamura, M. 2009, American Institute of Physics Con-
ference Series, 1158, 11
Thalmann, C., et al. 2009, ApJ, 707, L123 (“P1”)
Valenti, J.A. & Fischer, D.A. 2005, ApJS, 159, 141
van Dam, M. A., Le Mignant, D., & Macintosh, B. A.
2004, Appl. Opt., 43, 5458
This 2-column preprint was prepared with the AAS LATEX macros
v5.2.
10