Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.eso.org/~fcomeron/yipmo.ps Дата изменения: Fri Oct 31 11:21:48 2003 Дата индексирования: Mon Oct 1 20:54:02 2012 Кодировка: Поисковые слова: photosphere

The unusual infrared colors of a faint object in the Chamaeleon I
star forming region 1
F. Comeron
European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748 Garching bei
Munchen, Germany
fcomeron@eso.org
and
P. Claes
European Space Agency, ESTEC/APP-NSG, NL-2200 AG Noordwijk, The Netherlands
pclaes@estec.esa.int
Received ; accepted

{ 2 {
ABSTRACT
We present deep near-infrared (J S HK S ) imaging observations carried out
with the ESO Very Large Telescope (VLT) of a eld in the Chamaeleon I star
forming region in an attempt to identify possible members with masses compa-
rable to, or below, the mass of Jupiter. We focus on an object, ChaI J110814.2-
773649, which stands out as an outlier in color-color and color-magnitude dia-
grams of the eld, with H = 22:16 +0:21
0:17 , H K S = 0:01 +0:26
0:24 , J S H = 2:00 +1
0:62 .
H-band spectroscopy of this object shows that the unusual colors are not due to
emission lines in that region, and even provide a clear detection of its continuum.
Assuming membership in Chamaeleon I and an age of 2 Myr like for the bulk of
the members of that region, the blue H K S color and the absolute magnitude
are consistent with model predictions for a cool, sub-Jupiter mass object with
strong dust depletion in the atmosphere. However, the very red J S H color
implied by the marginal detection in the J S band is unexpected in an object with
such atmospheric properties. We speculate that this might be due to dierences
in the properties of dust and its depletion under the photosphere with respect
to eld objects (T dwarfs) with a similar temperature, resulting from both the
youth and the low surface gravity of a low mass member of a star forming re-
gion. We also consider the alternative possibility that ChaI J110814.2-773649
might actually be a high redshift object, whose red J S H color could result
from absorption of the ux bluewards of Lyman  in the J band. We nd that
such possibility would be marginally compatible with the J S HK S photometry of
ChaI J110814.2-773649 if it were an unreddened starburst at 8:5 < z < 11.
Subject headings: stars: low-mass, brown dwarfs | stars: formation

{ 3 {
1. Introduction
The search for the bottom of the stellar mass function, one of the main goals of surveys
of star forming regions just less than one decade ago (Tinney 1995), has been replaced
now by the search for objects beyond the bottom of the brown dwarf mass function and the
ultimate limit below which compact objects cannot form in isolation. The realization that
freely- oating objects keep being found as surveys probe lower and lower masses (Zapatero
Osorio et al. 1999, 2000, 2002; Martn & Zapatero-Osorio 2003; Lucas & Roche 2000;
Lucas et al. 2001) has forced to rethink object formation theories (Reipurth and Clarke
2001; Boss 2001) and even the basic denitions and nomenclature that apply to objects
not orbiting a larger body and whose mass is near or below the deuterium-burning limit
of ' 12 M J (Saumon et al. 1996). Despite recent progress, the direct detection of truly
Jupiter-mass and even lighter objects outside our Solar System remains an observational
challenge pushing current instrumental capabilities to their limits.
The complex molecular and dust chemistry in the atmospheres of objects with
temperatures near T  1000 K causes major ux redistributions that increase by orders of
magnitude the near-infrared brightness with respect to black bodies, and produce highly
peculiar colors that are diфcult to mimic by any other astrophysical objects (Burrows et al.
1997). These two properties suggest that such cool objects may be relatively easy to detect
thanks to their brightness, and to recognize thanks to their colors. Actual observations of
objects with spectral type T (Kirkpatrick et al. 1999), brown dwarfs much cooler than
the coolest stars, conrm the main results of theoretical modelling (Burgasser et al. 1999;
Leggett et al. 2000; Chabrier et al. 2000; Burrows et al. 2001; Dahn et al. 2002). The
large age dierence between million-years-old star forming regions and the gigayears-old
1 Based on observations obtained at the European Southern Observatory using the the
Very Large Telescope, Cerro Paranal, Chile (programmes 64.L-0049(A) and 67.C-0109(B))

{ 4 {
eld population implies dramatic dierences between the masses of newly formed and of
evolved T-dwarf-like objects: while a brown dwarf of 40 Jupiter masses (M J ) reaches the
temperature of ' 1200 K marking the approximate boundary between L and T dwarfs
(Burgasser et al. 2002a) at the age of 1 Gyr, the same temperature corresponds to a mass
around 5 M J at the age of 10 Myr, and to less than 2 M J at 1 Myr (Barae et al. 2003).
The near-infrared luminosities predicted by models indicate that T-dwarf-like objects in
the nearest star forming regions should be within the reach of current instrumentation in
4- to 8-meter class telescopes, as conrmed by the recent discovery of a possible member
of the  Ori aggregate with T-dwarf-like spectral features (Zapatero Osorio et al. 2002).
However, the faint magnitudes that such observations need to reach also extend the horizon
to high-z objects with appearances very dierent from those of objects in the local Universe,
thus requiring extra care in the interpretation of objects with unusual colors.
This paper presents observations carried out with the ESO Very Large Telescope
(VLT) of a very faint object in the direction of the Chamaeleon I star forming region with
highly unusual infrared colors, characterized by a blue H K S index combined with a
very red J S H. Spectroscopy demonstrates that these colors are due to a continuum
spectrum, and not to the presence of emission lines in the H band. We hypothesize that
the object might be a very low mass member of Chamaeleon I, whose faintness would then
suggest an extremely low, possibly sub-Jupiter mass that would make it the least massive
compact object detected so far outside the Solar System. However, we also contemplate the
possibility that the object might be an extremely high-z object, nding marginal consistency
between its colors and those expected from an unobscured starburst at a redshift z > 8:5
that would make it a J-band dropout.

{ 5 {
2. Observations and data reduction
The Chamaeleon I cloud, at 160 pc from the Sun (Wichmann et al. 1998), is one of
the nearest star forming regions. Besides its proximity, other factors make it an attractive
target for the study of its lowest mass contents: the low internal extinction over most of
its subtended area yields a mostly unobscured view of its contents, and its moderately
high galactic latitude (18 ф South of the galactic plane) decreases the density of background
sources. Moreover, several very low mass stellar and substellar members have been already
detected there with masses down to 20 30 M J (Comeron et al. 2000, 2003), thus
encouraging the search for even lower mass objects.
2.1. Imaging
We obtained deep imaging in the J S , H, and K S bands of a eld covering
2:5  2:5 arcmin 2 on the sky using ISAAC, the infrared imager and spectrograph at the
VLT, on the nights of 2000 March 27 and 28. The eld is centered at (2000) = 11 h 08 m 16 s 5,
ф(2000) = 77 ф 35 0 36 00 , and was chosen because of the existence of previously identied
brown dwarf candidates in its proximity (Comeron et al. 2000), while avoiding regions of
high extinction and bright nearby stars that might produce ghost images on the detector.
The images were obtained by integrating on the eld for approximately 30 minutes
in a given lter, splitting the integration into several individual exposures obtained with
telescope osets of a few tens of arcseconds in between. We then switched to a dierent lter
and repeated the procedure. The observations in each of the three lters were distributed
along our two observing nights to average over time any possible source variability with
timescales shorter than the span of our observing run. Each block of integrations in each
lter was combined into a single image, in which the individual frames were reduced by

{ 6 {
subtracting a sky frame (constructed by median averaging of the eld frames uncorrected
for telescope osets, so as to remove the source images and cosmic ray hits) and dividing by
a master at eld based on sequences of twilight sky exposures with dierent illumination
levels. All the blocks in each lter were then combined into a deep image of the eld, by
means of an average in which the contribution of each individual block was weighted with
the inverse square of the dispersion of the counts of its background. In this way, "noisy"
frames obtained with relatively high background levels had a reduced contribution to the
average. Total exposure times were 2.7 hours in J S , 2.5 in H, and 7.9 in K S . The longer
exposure times in K S are due to the blue H K S indices expected for the objects of interest
in this work.
The targeted region does not show any obvious nebulosity in the visible and, given the
progressively lower eфciency of scattering by small dust grains towards longer wavelengths,
the visibility of any possible remaining faint nebulosity at infrared wavelengths is expected
to be even lower. Indeed, tenuous lamentary nebulosity across the eld is perceptible only
in the nal combined images, most clearly in the J S image. The sky frame constructed
according to the procedure described above eectively lters out extended, uniform
nebulosity as well as point sources and small extended sources. However, it would leave
unltered any nebulosity with brightness variations over angular scales comparable to the
amplitude of the oset pattern, which would then be apparent in the sky-subtracted frames
as zones with local negative backgrounds, and may bias in this way the ux estimates of the
detected sources. No such eect is at all seen in our images, thus giving us condence that
the sky subtraction procedure followed is well suited for the reduction of our target eld.
Photometric calibration was performed by observing at the beginning of each night the
standard star S064-F, lying near the Chamaeleon I region, followed by an observation of
the target eld at a nearly identical airmass. This standard is a part of the set that denes

{ 7 {
the Las Campanas/Palomar NICMOS system (Persson et al. 1998), which the broad-band
lter plus detector system of ISAAC match well. This allowed us to establish a network
of local secondary standards, which were used to determine the photometric zeropoints of
the deep images. We then stacked together the three deep frames in each lter, and used
the DAOFIND task under the DAOPHOT package (Stetson 1987), available in IRAF 2 , for
automated source detection. Photometry was performed by point-spread function (PSF)
tting using as a reference for the PSF determination the brightest point sources in the
eld, using the DAOPHOT PEAK task. After experimenting with dierent PSF functional
forms available under the DAOPARS package a PSF consisting of a Gaussian analytical
part plus a lookup table of residuals, both constant across the eld, was chosen. Comparison
with the results obtained using other forms for the analytical component showed no major
dierences in the results. The measured FWHM of the PSF in the combined images is 0 00 81
(J S ), 0 00 71 (H), and 0 00 66 (K S ). Since the photometry was performed by tting of a stellar
PSF it is accurate only for point sources, which are the objects of interest in this work.
Photometric errors and detection limits for point sources were derived by adding articial
stars to the frames, constructed by scaling the derived PSF according to the desired
magnitude, and then recovering their photometry in the same way as for the real stars. The
errors quoted thus correspond to the 1- scatter in the input magnitudes of the articial
stars that produce a given recovered magnitude. We obtain average 5 detection limits for
point sources of J S = 23:20, H = 22:30, K S = 22:35, with slight regional variations below
the 0.1 mag level due to the presence of faint lamentary nebulosity across the eld.
Figure 1 shows the (H K S ; H) color-magnitude diagram of all the sources detected in
our images having a detection above the 3 level in at least one of the bands. At the bright
2 IRAF is distributed by NOAO, which is operated by the Association of Universities for
Research in Astronomy, Inc., under contract to the National Science Foundation

{ 8 {
end (H < 18) late-type stars, with a narrow range of colors and a well dened blue boundary
at H K S ' 0:3, dominate. Both the narrowness of the range and the low value of the
(H K S ) index show that the extinction is low and fairly uniform throughout the region.
This is conrmed by the color-color magnitude (Figure 2), where most objects cluster at the
colors expected for mid-to-late type stellar photospheres reddened by extinctions between
0.2 and 0.6 mag at K S .
At fainter magnitudes extragalactic sources, generally redder and often resolved,
progressively take over. However, there is an obvious outlier at H K S = 0:01 +0:26
0:24 ,
H = 22:16 +0:21
0:17 . The object (hereafter referred to as ChaI J110814.2-773649 after its
J2000 equatorial coordinates) is at most barely detected at J S ' 24:16 +1
0:59 , implying
J S H = 2:00 +1
0:62 .
Even though measurement uncertainties are marginally compatible with a (H K S )
color close to the blue boundary dened by the other sources in the eld, its combination
with the high upper limit in (J S H) implies very unusual near-infrared colors for
ChaI J110814.2-773649. This is more clearly visible in Figure 2, where the separation of
ChaI J110814.2-773649 with respect to all the other objects in the eld is very clear. No
other objects with comparable photometric properties are detected in the eld. Figure 3
shows a small section of the eld centered on ChaI J110814.2-773649 in the three bands.
The fact that it is clearly seen at H but hardly at J S would anticipate a considerable
brightness at K S but, as can be seen by comparing to the objects surrounding it, this does
not happen. A comparison with the radial plots of the PSF reference stars does not show
any evidence for ChaI J110814.2-773649 being spatially extended.

{ 9 {
2.2. Spectroscopy
The photometric properties of ChaI J110814.2-773649 are highly unusual for an object
with a continuum-dominated spectrum, but they might be reproduced by a AGN or QSO
with intense emission lines conveniently redshifted (vanden Berk et al. 2001): one or a few
strong lines in the H band would make the object appear red in J S H and blue in H K S .
Obvious candidates are H redshifted to z ' 1:5, H+[OIII] to z ' 2:3, or [OII] to z ' 3:4.
We obtained a 10-hours-long H band spectrum of ChaI J110814.2-773649, also using
ISAAC at the VLT, on the nights of 2001 April 29 and May 1. The goal was to discern
between the possibilities of emission lines dominating the H-band ux of the object (which
should be relatively straightforward to detect in such a deep spectrum) and of an unusual
continuum spectral energy distribution (which should result in a non-detection of the
object, or a marginal detection at most, due to the spread of the ux over the spectral range
covered). We were fortunate to enjoy excellent transparency conditions and subarcsecond
seeing during virtually the entire time devoted to these observations.
Individual spectra were obtained with the 1 00 0 slit oriented so as to include in it a
brighter, H = 16:83 reference star located 16 00 from ChaI J110814.2-773649. A total of 200
individual spectra of 180 seconds of exposure time each were obtained with nodding along
the slit combined with a small random oset around each of the two nod positions. The
presence of the reference star provides an easy way to monitor the object centering during
the integrations and to correct for telescope nodding at the reduction stage, and it also
includes in each frame a secondary calibrator for telluric absorption correction. Consecutive
frames taken at two dierent nod positions were subtracted to cancel sky emission lines,
and then divided by a spectroscopic at eld frame constructed from calibration frames
taken with an internal continuum lamp respectively switched on and o. We refer in what
follows to these sky-subtracted, at-elded frames as "pairwise-subtracted frames".

{ 10 {
At the time when our observations were obtained, ISAAC suered from a pickup
noise problem that imposed a wave pattern along detector columns, corresponding to the
spatial direction. The amplitude of the pattern is far too low to be seen in the individual
pairwise-subtracted frames, where it is hidden by the background and readout noise.
However, it becomes readily apparent at the very high sensitivity levels obtained when
coadding the 200 pairwise-subtracted frames after correcting for the telescope nodding
using as reference the trace of the spectrum of the bright star. We removed this pattern
by adopting an additional step in the data reduction, which consisted of subtracting from
each pairwise-subtracted frame another frame containing almost uniquely the wave pattern.
To do this, we took advantage of the fact that the wavelength of the pickup noise pattern
was almost exactly 6.5 pixels along the spatial direction. We thus produced a wave pattern
frame by combining and median-ltering the pairwise-subtracted frame with copies of itself
shifted respectively by 13, 26, 39, 52, 65, and 78 pixels along the detector columns, and
trimming it to the same dimensions of the original image. Then, we subtracted this wave
pattern frame from the pairwise-subtracted frame from which it was produced. In this
way, the deep spectroscopic frame obtained by combining the pairwise-subtracted, wave
pattern-subtracted frames after correcting for the telescope nodding contains no detectable
trace of the pickup noise, thus greatly increasing the visibility of possible spectral traces of
extremely faint objects.
3. Discussion
3.1. The spectrum of ChaI J110814.2-773649
Examination of the deep spectroscopic frame that combines our 10 hours of integration
shows no hints of emission lines at the expected position of the spectrum of ChaI J110814.2-
773649, while it is possible to show that the sensitivity of this frame should be suфcient

{ 11 {
to easily reveal any emission lines dominating the H-band ux. The background noise in
the spectroscopic frame greatly varies depending on whether or not strong OH airglow
lines are present at the corresponding wavelength. On the other hand, the spectrum of
the H = 16:83 reference star allows us to perform an approximate ux calibration and
thus determine the number of counts that should be expected in a single line dominating
the emission. In this way, we nd that if the H-band ux of ChaI J110814.2-773649 were
contributed by a single line with a Gaussian prole peaking at the wavelength most aected
by airglow emission, its full-width at half maximum should exceed 360  A so that its peak
intensity integrated over the spatial prole would fall below 3 times the local background
noise at that detector column. Such width would correspond to a rest frame line width of
145  A for H at z = 1:5, 110  A for H or [OIII] at z = 2:3, or 82  A for [OII] at z = 3:4.
For H or H these values are about three times greater than those found in typical QSO
spectra (vanden Berk et al. 2001). The discrepancy would be smaller for Ly, whose
corresponding rest frame width would be only 27  A, although the redshift of that line
into the H band would then imply 10:6 < z < 13:8. We should keep in mind nevertheless
that these are are deliberately conservative limits that are made even more stringent (i.e.,
involving greater intrinsic widths) if we assume that the line peaks in a region less aected
by skyglow. Similarly, such large widths greatly exceed those of the telluric OH line blends,
thus implying intensities comparable to the peak one also at wavelengths where airglow
lines are absent and the background noise is thus much lower. We thus consider that our
spectrum eectively excludes the possibility that the H band ux of ChaI J110814.2-773649
may be dominated by line emission.
In view of the non-detection of suфciently strong emission lines to account for the
H band detection, we decided to boost up the sensitivity to attempt the detection of
the continuum by strongly binning the spectrum along the dispersion direction. The
1024 detector columns were thus collapsed into 20 bins. Each column contributed to the

{ 12 {
bin with a weight inversely proportional to the squared dispersion of the counts along it
(excluding the regions aected by the trace of the reference star or by the artifacts that
it produced when building the wave pattern frame), so as to reduce the contribution of
those wavelengths aected by the presence of sky airglow lines. The binned spectrum is
shown in Figure 4, where we have indicated the position where the trace of the spectrum of
ChaI 110814.2-773649 should be expected. Indeed, a slight increase of the number of counts
over the background can be appreciated at precisely that position, especially on the left
half of the frame corresponding to shorter wavelengths. The excess of counts is detectable
over several bins and its peak along the spatial direction seems to be parallel to the trace
of the spectrum of the reference star. We thus consider it as the actual detection of the
spectrum of ChaI 110814.2-773649, demonstrating that the H band is indeed dominated by
continuum emission and that the unusual colors of this object are due to an exotic spectral
energy distribution rather than to the presence of emission lines.
3.2. The nature of ChaI J110814.2-773649
3.2.1. An extremely high-z galaxy?
The existence of an object with the colors of ChaI J110814.2-773649 in exposures like
the ones presented here must be interpreted with caution. On the one hand, such deep
images have the potential of detecting objects spanning a wide range of redshifts, sampling
the rest-frame ultraviolet spectrum of objects in early evolutionary stages that are absent
from the local Universe, and thus being sensitive to sources with colors widely dierent from
those characteristic of brighter, and typically more nearby, objects. On the other hand,
observations of elds in the JHK S bands to a depth comparable to that of our observations
are still scarce, although recent and ongoing eorts such as the deep JHK S imaging of the
Hubble Deep Field South (HDFS) (Labbe et al. 2003), the MS1054-03 cluster (Franx et al.

{ 13 {
2000) or the multiband imaging of the Chandra Deep Field South (Renzini et al. 2003) are
providing valuable information on the near-infrared colors found at such faint magnitudes.
Other deep near-infrared surveys recently carried out with 8m-class telescopes, such as
the Subaru Deep Field (Maihara et al. 2001) have tended to use the J and K S bands
alone, due to their focus on the detection of very red objects. As a result, the framework
for comparison of our results, especially concerning objects with unusual colors such as
ChaI J110814.2-773649, is still rather limited. In any case, a comparison to published
JHK S data for the HDFS over a similar eld of view (Labbe et al. 2003) does conrm that
the colors of ChaI J110814.2-773649 are quite unique, despite of the fact that the HDFS
survey reaches down to limiting magnitudes (3) that are respectively 1.6 (J S ), 2.7 (H),
and 2.2 (K S ) fainter than those of ChaI J110814.2-773649.
No objects with similar colors are detected either in deep infrared images of the Hubble
Deep Field North (HDFN), although the presence of an object with very atypical colors,
HDFN-JD1, has been reported by Dickinson et al. (2000) in near-infrared images of that
eld. The object in the HDFN is outstanding because of its very red H K color and
for being undetected at J , thus being by far the reddest object in the eld in J H.
Although HDFN-JD1 may be considered as the opposite of ChaI J110814.2-773649 as far
as H K is concerned (HDFN-JD1 being extremely red, while ChaI J110814.2-773649
is unusually blue), the large J H values are puzzling in both cases. It is interesting in
this respect to consider Figure 7 of that paper, which considers the variation of the J H
and H K S colors with redshift for a number of galaxy types modelled after the spectral
energy distributions of Bruzual and Charlot (1993). The curves presented there range
from a passively evolving elliptical galaxy formed at z = 15 to a unreddened starburst
galaxy, and can be considered as bracketting the entire range of colors encountered among
extragalactic objects. The bluest (H K S ) colors predicted by such models are produced

{ 14 {
by unreddened starbursts, reaching (H K S ) ' +0:2 3 near z = 2:5 and z = 5:5. Such
redshifts are however ruled out by the very blue (J S H) predicted in both cases, which
are incompatible with the extremely red (J S H) of ChaI J110814.2-773649. At larger
redshifts the Ly forest enters the near infrared, progressively decreasing the ux in the
J band. At z = 8:5 and greater the ux depletion on the blue side of the J band moves
the (J S H) color above 1.4, thus becoming compatible with our lower limit, while still
maintaining a relatively blue (H K S ) at 0.35 mag. Similarly blue (H K S ) colors and
ever redder (J S H) colors are predicted as the starburst is shifted up to z ' 11, where
Ly enters the H band and starts reddening (H K S ) as well. While the error bars of our
observations do not completely exclude (H K S ) = 0:35 for ChaI J110814.2-773649 it must
be recalled that, as noted in Section 2.1, a foreground extinction of at least 0.2 mag caused
by dust in the Chamaeleon I cloud pervades the eld, thus reddening any background
objects by at least 0.1 mag in (H K S ). The bluest high-redshift starbursts should thus
have (H K S ) > 0:45, which we consider hardly compatible with our measured uxes. In
summary, we consider an unreddened starburst at 8:5 < z < 11 as a marginally consistent
explanation for the unusual colors of ChaI J110814.2-773649, based on arguments similar
to the ones that lead Dickinson et al. (2000) to propose a z = 12:5 object as a possible
interpretation of HDFN-JD1.
3.2.2. A young planetary-mass object?
Although an extremely high redshift object may indeed account for the spectral energy
distribution of ChaI J110814.2-773649, the serendipity of our nding and its location in the
3 The AB magnitudes used by Dickinson et al. (2000) have been transformed into the
Vega magnitudes used throughout this paper for comparison.

{ 15 {
direction of a nearby star forming region leads us to consider also the possibility that it
may be a young, isolated planetary mass member of Chamaeleon I. Near-infrared spectra
of T-dwarf-like objects are dominated by strong molecular absorption caused by CH 4 , H 2 O,
and H 2 , producing a broad-band spectral energy distribution characterized by blue J H,
H K colors. At the somewhat higher temperatures of L-type dwarfs dust is present in
the atmosphere in the form of clouds containing a complex mixture of metals and minerals
(Burrows et al. 2001), producing instead extremely red colors. The disruption of clouds
and precipitation of its condensates below the photosphere is thought to happen over a
relatively narrow margin of temperatures around T ' 1200 K (Burgasser et al. 2002a),
marking the transition between the L and T spectral types.
The tracks overplotted in Figure 1 represent the colors and magnitudes expected for
young L- and T-dwarf-like objects at the distance of Chamaeleon I, after the evolutionary
tracks of Chabrier et al. (2000) (DUSTY models) and Barae et al. (2003) (COND
models). The former (dashed lines in the gure) are expected to represent L-dwarf-like,
dusty atmospheres, appropriate for objects hotter than T ' 1200 K, while the latter
reproduce the characteristics of cooler objects in which dust has precipitated below the
photosphere. The transition temperature between the DUSTY and the COND tracks is
attained at a H magnitude between 18.5 and 19.2 on the COND tracks, and between
20.5 and 21.3 on the DUSTY tracks. The magnitude intervals quoted re ect the mild age
dependence of the temperature-magnitude relationship over the 1 10 Myr range. As seen
in Figure 1, both the faint H magnitude of ChaI J110814.2-773649 and its location oset
bluewards from the general distribution of objects in the color-magnitude diagram can be
well reproduced by an extremely low mass member of the star forming region. This is fairly
independent of the assumed age, due to the broad overlap of the tracks. However, the
assigned mass does depend strongly on the age, as the symbols indicating the location of
objects with a given mass show. Remarkably, adopting the age that best reproduces the

{ 16 {
locations of very low mass stars and massive brown dwarfs in Chamaeleon I, 2 Myr, yields
a mass of only  0:7 M J for ChaI J110814.2-773649, or about two Saturn masses. The
spectrum of the object (Figure 5) displays hints of a possible peak around 1.55  A followed
by a drop to near zero emission levels at longer wavelengths, consistent with the overall
spectral energy distribution of T dwarfs that also peaks at the same position. However, the
extremely low signal in the spectrum makes the actual existence of the peak doubtful. It
should be pointed out too that the rise near 1.8 m, if real, would have no correspondence
in T dwarf spectra, thus casting doubts on this interpretation.
The agreement of the H and K S photometry with models of very young T-dwarf-like
objects does not hold when we consider the J S H color as well, which is generally
close to zero in T dwarf spectra. As seen in the color-color diagram of Figure 2, while
H K S is fully within the range expected for a T-dwarf-like object, the J S K S > 1:4
of ChaI J110814.2-773649 seems instead characteristic of a cool L dwarf. In this respect,
the photometric properties of ChaI J110814.2-773649 qualitatively resemble those of
transition objects between the L and T types observed by Leggett et al. (2000), which
are attributed to the persistence of dust in the atmosphere coexisting with the onset of
strong CH 4 absorption in the K band. The sequence of colors along this transition is not
well known, as it depends on factors such as the partial disruption of clouds (Burgasser
et al. 2002a), the composition of grains (Lodders 1999), and their size distribution,
which in turn are sensitive functions of the surface gravity (Cooper et al. 2003). The low
gravity of cool young planetary mass objects, over two orders of magnitude lower than that
of the eld, evolved objects with which models have been mainly confronted so far, also
imply important structural dierences in both the atmosphere and the layers beneath the
photosphere. The lower gravity brings convection much closer to the photosphere than in
evolved T dwarfs (Burrows et al. 1997), thus facilitating the presence of small grains in
the atmosphere due to convective overshooting or turbulent diusion (Chabrier & Barae

{ 17 {
2000) at temperatures under which they would be absent in more evolved objects. All
these factors might account for the maintenance of hybrid features such as L-dwarf-like
reddening at J S H and T-dwarf-like bluening at H K S at even lower temperatures
and luminosities in young planetary-mass objects, accounting for the observed colors of
ChaI J110814.2-773649 despite a brightness that is well below that predicted for a young
early-T object. At this point this is a very speculative explanation unsupported by proper
modelling, but the identication of such hybrid features as being common among other
young very low mass objects could lend support to it.
It is interesting in this respect to compare ChaI J110814.2-773649 with S Ori 70, a
conrmed T dwarf-like object recently discovered by Zapatero Osorio et al. (2002) in the
direction of the  Ori association. The membership, and therefore the youth, is considered
as likely by these authors based on both probabilistic arguments derived from the actual
density of similar objects in the eld, and on ts to theoretical spectra with dierent surface
gravities, which tend to give a better match for low surface gravity values. This has been
recently conrmed by Martn & Zapatero-Osorio (2003) based on mid-resolution infrared
spectroscopy. This object is considerably brighter than ChaI J110814.2-773649 despite of
the distance of  Ori, over twice as large as that of Chamaeleon I: if both are members
of their respective star forming regions, the dierence in absolute H magnitudes is about
3.5 mag. Interestingly, S Ori 70 also has unusual colors for a T dwarf-like object, albeit in
a dierent sense as ChaI J110814.2-773649: while the J H colors of S Ori 70 is very blue
and consistent with those of eld T dwarfs, its H K is unusually red when compared to
an object of this class (Dahn et al. 2002; Leggett et al. 2002) as well as to the best-tting
theoretical models. Moreover, its brightness places it between 1 and 2 magnitudes above
the sequence predicted by models in color-magnitude diagrams for objects of the same
color. The spectroscopic conrmation of S Ori 70 as a T dwarf-like object and its likely
membership in the  Ori thus cautions about the diversity found among otherwise similar

{ 18 {
objects of this kind, and the limited reliance that must be placed on the models.
4. Conclusions
In this paper we have reported the discovery with the ESO Very Large Telescope of an
object in the direction of the Chamaeleon I star forming region, ChaI J110814.2-773649,
characterized by its unusual near-infrared colors. The object is faint (H = 22:16 +0:21
0:17 ),
very red in J S H (2:00 +1
0:62 ), and very blue in H K S ( 0:01 +0:26
0:24 ). A 10h long H-band
spectrum demonstrates that the colors are not due to the presence of strong emission lines
in the H band, and actually reveals the continuum of the object.
We consider two possible interpretations for ChaI J110814.2-773649. In the rst one,
corresponding to a very high redshift object, a marginally acceptable match to the colors
is found by assuming that ChaI J110814.2-773649 is an unreddened starburst turned into
a J-band dropout by its extremely high redshift (8:5 < z < 11) that makes the Ly forest
penetrate well into the J band. Other spectral energy distributions at dierent, lower
redshifts cannot simultaneously reproduce both the red J S H and the blue H K S colors.
The very high redshift implied by this explanation, which greatly exceeds that of any
spectroscopically conrmed source known so far, and the important quantitative dierence
between the bluest (H K S ) color allowed by the models and the observed one makes us
consider it quite unlikely.
The second explanation, motivated by the location of ChaI J110814.2-773649 in the
general direction of the Chamaeleon I cloud, is that it may be an actual very low mass
member of that region. Taken together, the H K S color and the H magnitude are
consistent with the model predictions for a planetary mass object of roughly 0:7 Jupiter
masses with an age of 2 Myr, or slightly above one Jupiter mass if it is 5 Myr old. However,

{ 19 {
this explanation does not seem consistent with the very red J S H implied by the marginal
detection at J S , which models predict only for the coolest dusty atmospheres. We speculate
that ChaI J110814.2-773649 might be an object with spectral features hybrid between those
of L and T dwarfs, displaying both a dusty atmosphere that reddens J S H while at the
same time having strong CH 4 bands that make H K S blue, with the coexistence of both
features being possible due to the low surface gravity of such a young object.
Both possible explanations are quite exciting: the rst one would imply that
ChaI J110814.2-773649 is a record-breaking high redshift object, while the second one would
make it the least massive compact object so far identied outside the solar system. It is thus
tantalizing not being able to distinguish between both radically dierent possibilities with
the data currently at hand, and it must be kept in mind that the subsisting photometric
uncertainties do not completely exclude other less exotic possibilities. Further ground-based
spectroscopy will demand great amounts of integration time at the largest telescopes, and
detection at longer wavelengths must wait for the next generation of spaceborne infrared
observatories (Burrows et al. 2001). However, new observations of ChaI 110814.2-773649
and other possible new young, isolated planetary-mass objects with currently available
means are important. A better determination of the unusual colors of ChaI 110814.2-773649
and monitoring for variability are obvious follow-up tasks, while astrometry, which could
conrm kinematical membership over a timescale of 5-10 years, would address the origin
of ChaI 110814.2-773649 as either an isolated object or as a former planet expelled from
its parent stellar system (de la Fuente Marcos & de la Fuente Marcos 1998; Reipurth and
Clarke 2001). Finally, new deep searches in star forming regions will tell us if young,
isolated planetary mass objects are a common outcome of stellar and planetary system
formation, or rather a rarity among astrophysical objects.
We are very pleased to thank Dr. Adam Burrows for providing details on his model

{ 20 {
calculations and for a critical reading of an earlier version of this manuscript. Drs. France
Allard and Isabelle Barae are thanked for useful remarks on the validity of the Barae et
al. (2003) model predictions at the very young ages of relevance here. Insightful comments
by the referee, Dr. Phil Lucas, as well as a second anonymous referee, and the editor, Dr.
James Liebert, helped improving this paper in both contents and presentation. Last but
not least, the excellent support received from the Paranal Science Operations sta during
our VLT observations in 2000 and 2001 is warmly appreciated.

{ 21 {
REFERENCES
vanden Berk, D.E., Richards, G.T., Bauer, A., et al., 2001, AJ, 122, 549.
Barae, I., Chabrier, G., Barman, T.S., Allard, F., Hauschildt, P.H., 2003, A&A, 402, 701.
Bessell, M.S., Brett, J.M., PASP, 100, 1134.
Boss, A.P., 2001, ApJ, 551, L167.
Bruzual, G., Charlot, S., 1993, ApJ, 405, 538.
Burgasser, A.J., Kirkpatrick, J.D., Brown, M.E., Reid, I.N., Gizis, J.E., Dahn, C.C., Monet,
D.G., Beichman, C.A., Liebert, J., Cutri, R.M., Skrutskie, M.F., 1999, ApJ, 552,
L65.
Burgasser, A.J., Marley, M.S., Ackerman, A.S., Saumon, D., Lodders, K., Dahn, C.C.,
Harris, H.C., Kirkpatrick, J.D., 2002a, ApJ, 571, L151.
Burgasser, A.J., Kirkpatrick, J.D., Brown, M.E., Reid, I.N., Burrows, A., Liebert, J.,
Matthews, K., Gizis, J.E., Dahn, C.C., Monet, D.G., Cutri, R.M., Skrutskie, M.F.,
2002b, ApJ, 564, 421.
Burrows, A., Marley, M., Hubbard, W.B., Lunine, J.I., Guillot, T., Saumon, D., Freedman,
R., Sudarsky, D., Sharp, C., 1997, ApJ, 491, 856.
Burrows, A., Hubbard, W.B., Lunine, J.I., Liebert, J. 2001, Rev. Mod. Phys., 73, 719.
Chabrier, G., Barae, I., 2000, ARA&A, 38, 337.
Chabrier, G., Barae, I., Allard, F., Hauschildt, P., 2000, ApJ, 542, 464.
Comeron, F., Neuhauser, R., Kaas, A.A., 2000, A&A, 359, 269.
Comeron, F., Reipurth, B., Henry, A., Fernandez, M., 2003, A&A, submitted.

{ 22 {
Cooper, C.S., Sudarsky, D., Milsom, J.A., Lunine, J.I., Burrows, A., 2003, ApJ, 586, 1320.
Dahn, C.C., Harris, H.C., Vrba, F.J., Guetter, H.H., Canzian, B., Henden, A.A., Levine,
S.E., Luginbuhl, C.B., Monet, A.K.B., Monet, D.G., Pier, J.R., Stone, R.C., Walker,
R.L., 2002, AJ, 124, 1170.
Dickinson, M., Hanley, C., Elston, E., Eisenhardt, P.R., Stanford, S.A., Adelberger, K.L.,
Shapley, A., Steidel, C.C., Papovich, C., Szalay, A.S., Bershady, M.A., Conselice,
C.J., Ferguson, H.C., Fruchter, A.S., 2000, ApJ, 531, 624.
Franx, M., Moorwood, A., Rix, H.-W., Kuijken, K., Rottgering, H., van der Werf, P., van
Dokkum, P., Labbe, I., Rudnick, G., 2000, The Messenger, 99, 20.
de la Fuente Marcos, C., de la Fuente Marcos, R., New Astron., 4, 21.
Geballe, T.R., Saumon, D., Leggett, S.K., Knapp, G.R., Marley, M.S., Lodders, K., 2001,
ApJ, 556, 373.
Kirkpatrick, J.D., Reid, I.N., Liebert, J., Cutri, R.M., Nelson, B., Beichman, C.A., Dahn,
C.C., Monet, D.G., Gizis, J.E., Skrutskie, M.F., 1999, ApJ, 519, 802.
Labbe, I., Franx, M., Rudnick, G., Forster Schreiber, N.M., Rix, H.-W., Moorwood, A.F.M.,
van Dokkum, P.G., van der Werf, P., Rottgering, H., van Starkenburg, L., can de
Wel, A., Kuijken, K., Daddi, E., 2003, AJ, 125, 1107.
Leggett, S.K., Geballe, T.R., Fan, X., et al., 2000, ApJ, 536, L35.
Leggett, S.K., Golimowski, D.A., Fan, X., et al., 2002, ApJ, 564, 452.
Lodders, K., 1999, ApJ, 519, 793.
Lucas, P.W., Roche, P.F., 2000, MNRAS, 314, 858

{ 23 {
Lucas, P.W., Roche, P.F., Allard, F., Hauschildt, P.H., 2001, MNRAS, 326, 695.
Maihara, T., Iwamuro, F., Tanabe, H., 2001, PASJ, 53, 25.
Martn, E.L., Zapatero-Osorio, M.R., 2003, ApJ, 593, L113.
Persson, S.E., Murphy, D.C., Krzeminsky, W., Roth, M., Rieke, M.J., 1998, AJ, 116, 2475.
Reipurth, B., Clarke, C.J., 2001, AJ, 122, 432.
Renzini, A., Cesarsky, C., Cristiani, S., da Costa, L., Fosbury, R., Hook, R., Leibundgut,
B., Rosati, P., Vandame, B., 2003, in The mass of galaxies at low and high redshifts,
eds. R. Bender and A. Renzini, Springer-Verlag.
Saumon, D., Hubbard, W.B., Burrows, A., Guillot, T., Lunine, J.I., Chabrier, G., 1996,
ApJ, 460, 993.
Smail, I., Owen, F.N., Morrison, G.E., Keel, W.C., Ivison, R.J., Ledlow, M.J., 2002, ApJ,
581, 844.
Stetson, P.B., 1987, PASP, 99, 191
Tinney, C.G. (ed.), 1995, The Bottom of the Main Sequence - and Beyond, Springer-Verlag.
Wichmann, R., Bastian, U., Krautter, J., Jankovics, I., Rucinski, S.M., 1998, MNRAS, 301,
L39.
Zapatero Osorio, M.R., Bejar, V.J.S., Rebolo, R., Martn, E.L., Basri, G., 1999, ApJ, 524,
L115.
Zapatero Osorio, M.R., Bejar, V.J.S., Martn, E.L., Rebolo, R., Barrado y Navascues, D.,
Bailer-Jones, C.A.L., Mundt, R., 2000, Science, 290, 103.

{ 24 {
Zapatero Osorio, M.R., Bejar, V.J.S., Martn , E.L., Rebolo, R., Barrado y Navascues, D.,
Mundt, R., Eisloel, J., Caballero, J.A., 2002, ApJ, 578, 536.
This manuscript was prepared with the AAS L A T E X macros v5.0.

{ 25 {
Fig. 1.| (H K S ; H) color-magnitude diagram of the objects detected in our images.
Only objects exceeding a detection threshold of 5 in our H band images are plotted.
ChaI J110814.2-773649 is the point at the bottom left with error bars plotted on it. The
curves are theoretical isochrones for ages 1 Myr, 5 Myr, and 10 Myr according to two
dierent sets of models (see Section 3.2.2) at the distance of Chamaeleon I. The dashed lines
correspond to the DUSTY models of Chabrier et al. (2000), in which dust grains form
and stay in the atmosphere. The solid lines correspond to the COND models (Barae et al.
2003), in which dust grains precipitate under the photosphere as soon as they are formed.
The latter set of models is expected to better represent the cool (T  1000 K) objects of
interest in this work. The symbols plotted on the curves denote the position of objects of
dierent masses at the age of each isochrone: triangles correspond to 5 M Jup , squares to
1 M Jup , and circles to 0:5 M Jup . The large overlap among the isochrones corresponding to
each model implies a large degeneracy between age and mass. The red, bright object near
the dashed curves at H = 17:74, H K S = 1:01 is not a candidate L dwarf, but a well
resolved elliptical galaxy

{ 26 {
Fig. 2.| (H K S ; J S H) diagram of all objects having H-band detections at a level higher
than 5 in our images. ChaI J110814.2-773649 is the point at the upper left with error bars
plotted on it. Solid and dashed lines, and the empty symbols along them, represent theoret-
ical isochrones as in Figure 1. The dotted line is the locus of dwarf stars between spectral
types B and M6 from Bessell and Brett (1988). The displacement between the highest
density of points in the diagram, corresponding to stars background to the Chamaeleon I
cloud, and this locus is due to the moderate extinction caused by the dust in the star forming
region that pervades the imaged area.

{ 27 {
Js
Ks
H
Fig. 3.| Images of ChaI J110814.2-773649 in the J S , H, and K S bands. This is a small
portion (23 00  15 00 ) of the eld imaged in our observations. ChaI J110814.2-773649 is the
faint object in the center of the images.

{ 28 {
Fig. 4.| Combined frame containing the VLT/ISAAC spectrum of ChaI J110814.2-773649.
The white trace in the upper part corresponds to the bright reference star, and the arrow
marks the expected position of the trace of the spectrum of ChaI J110814.2-773649. The
black stripes are artifacts of the data reduction. The spectrum ranges from 1.4 m (left) to
1.8 m (right) and is strongly binned along the dispersion direction to enable detection of
the H = 22:16 object. See Section 2.2 for a detailed explanation of the reduction process.

{ 29 {
Fig. 5.| Extracted spectrum of ChaI J110814.2-773649. Relative ux calibration was
achieved by ratioing its binned spectrum by that of the bright reference star included in
the slit, multiplying by the ratio between a spectrum of this star and that of the B4V star
HD 94414 observed one after another at similar mass degraded to the same resolution, and
then multiplying by a T = 17; 200 K blackbody that approximates the spectral energy
distribution of HD 94414. The lower dotted line marks the zero ux level.