Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~inr/post/sdss.ps Äàòà èçìåíåíèÿ: Fri Jun 8 23:31:48 2007 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 03:26:30 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 103

Accepted to Ap.J. Letters
Preprint typeset using L A T E X style emulateapj
AN IMPROVED RED SPECTRUM OF THE METHANE OR T­DWARF SDSS 1624+0029: ROLE
OF THE ALKALI METALS
James Liebert 1 , I. Neill Reid 2 , Adam Burrows 1 , Adam J. Burgasser 3 , J. Davy
Kirkpatrick 4 , and John E. Gizis 4
Accepted to Ap.J. Letters
ABSTRACT
A Keck II low resolution spectrum shortward of one micron is presented for SDSS 1624+0029, the
first field methane or T dwarf discovered in the Sloan Digital Sky Survey. Significant flux is detected
down to the spectrum's short wavelength limit of 6200 š A. The spectrum exhibits a broad absorption
feature centered at 7700 š A, which we interpret as the K I 7665/7699 resonance doublet. The observed
flux declines shortward of 7000 š A, due most likely to the red wing of the Na I doublet. Both Cs I doublet
lines are detected more strongly than in an earlier red spectrum. Neither Li I absorption nor Hff emission
are detected.
An exploratory model fit to the spectrum suggests that the shape of the red spectrum can be primarily
accounted for by the broad wings of the K I and Na I doublets. This behavior is consistent with the
argument proffered by Burrows, Marley and Sharp that strong alkali absorption is principally responsible
for depressing T dwarf spectra shortward of 1¯m. In particular, there seems no compelling reason at this
time to introduce dust or an additional opacity source in the atmosphere of the SDSS object. The width
of the K I and strengths of the Cs I lines also indicate that the Sloan object is warmer than Gl 229B.
Subject headings: stars: individual (SDSS1624+0029 -- stars: brown dwarfs -- stars: atmospheres
1. INTRODUCTION
Methane or T dwarfs are substellar objects cooler
than L and M dwarfs, and have near­infrared (1­2¯m)
spectra dominated by molecular absorption due to wa­
ter, methane, and pressure­induced molecular hydrogen.
Methane is expected to remain an important atmospheric
constituent down to the temperature of Jupiter (¸125 K),
where it also is prominent in the infrared spectrum. The
prototype of the class is the companion to the nearby M
dwarf star Gl 229 (Nakajima et al. 1995, Oppenheimer
et al. 1995). Model atmosphere analyses fitting synthetic
spectra to detailed spectrophotometric and photometric
observations indicate a temperature for this object near
or slightly below 1,000 K (Marley et al. 1996, Allard et
al. 1996). This past year has seen the discovery of several,
similar field T dwarfs, found first in the Sloan Digital Sky
Survey (Strauss et al. 1999, S99; Tsvetanov et al. 2000),
shortly thereafter in the Two Micron All Sky Survey (Bur­
gasser et al. 1999) data sets, and also in an ESO survey
(Cuby et al. 1999).
All of the new objects also have 1--2¯m spectra char­
acterized by the very strong molecular absorbers listed
above. Unfortunately, at least at low spectral resolution,
the differences among their spectra appear somewhat sub­
tle. The current field surveys by SDSS and 2MASS are
magnitude limited, and therefore likely to identify the
warmest (highest luminosity) T dwarfs. Still, the known
parallax for Gl 570D shows that this object must be sub­
stantially cooler than the prototype, yet its infrared spec­
trum is similar to the others (Burgasser et al. 2000). The
relative strengths of the molecular bands are not strongly
dependent on the effective temperature. The molecular
absorbers also are effective in hiding the weaker atomic line
transitions which might be useful discriminants of temper­
ature. At least initially, it is proving difficult to establish
spectral types and a temperature sequence for the T dwarfs
at the wavelengths where they are easiest to observe.
It is possible to observe the brightest of the new T dwarfs
at wavelengths significantly shortward of 1¯m, where the
atmospheres may prove to be more transparent. Model
calculations indicate that there may be few molecular and
atomic opacity sources, and those that are present may be
more sensitive to temperature. In particular, the behavior
of the alkali resonance doublet features which reside gener­
ally in this red part of the spectrum (0.5­0.9¯m) could be
particularly useful in diagnosing the temperature and test­
ing for the formation of dust grains (Burrows and Sharp
1999; Lodders 1999; Tsuji, Ohnaka & Aoki 1999; Burrows,
Marley & Sharp 2000; hereafter, BMS). These papers gen­
erally predict that different alkalis should precipitate out
as sulfides, salts or other condensates over a range of T eff
below about 1,500 K, in the order (with decreasing T eff )
according to BMS of Li first, then Cs, K and Na. Indeed,
the latter two alkali features are the most prominent fea­
tures in the red spectra of the somewhat­warmer late L
dwarfs (Kirkpatrick et al. 1999, Mart'in et al. 1999, Reid
1 Department of Astronomy and Steward Observatory, The University of Arizona, Tucson AZ 85721, liebert@as.arizona.edu,
aburrows@as.arizona.edu
2 Department of Physics and Astronomy, University of Pennsylvania, 209 South 33rd St., Philadelphia, PA 19104­6396,
inr@morales.physics.upenn.edu
3 Division of Physics, M/S 103­33, California Institute of Technology, Pasadena CA 91125, diver@cco.caltech.edu
4 Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA 91125, davy@ipac.caltech.edu,
gizis@ipac.caltech.edu
1

2
et al. 2000).
The red spectra of field T dwarfs can be relevant to an­
other controversy regarding the spectrum of Gl 229B: The
companion object shows too little red flux relative to the
model predictions of Marley et al. (1996) and Allard et
al. (1996). These authors concluded that an additional
opacity source exists shortward of 1¯m. Golimowski et
al. (1998) suggested as the solution that TiO returned
to gaseous form (it precipitates out in M­L dwarfs above
2,000 K) in Gl 229B. However, this hypothesis predicts
TiO band absorption at the wavelengths seen in M dwarfs,
but these are seen neither in late L dwarfs nor in Gl 229B.
Griffith et al. (1998) turned to solar system physics for
an intriguing answer: they hypothesized a population of
small photochemical haze particles analogous to the red
Titan Tholins (Khare and Sagan 1984), heated by ultra­
violet radiation from the primary M dwarf to tempera­
tures at least 50% higher than the T eff . Dust is invoked
in two other proposed solutions to the Gl 229B spectral
slope: Tsuji, Ohnaka & Aoki (1999) describe a hybrid at­
mospheric model with a warm dust layer that effectively
blocks short­wavelength flux. Pavlenko, Zapaterio Osorio
& Rebolo (2000) attempt to fit the red spectrum with a
scattering dust opacity which increases sharply to shorter
wavelengths. Finally, BMS suggest that the alkali opacity
alone -- in particular the broad wings of K and Na -- is the
agent which depresses the emergent flux out to 1¯m. Pro­
vided K and Na still exist in atomic form at the relevant
temperature, they argued that there is no need to invoke
dust or some additional absorber at short wavelengths.
SDSS 1624+0029 (hereafter, SDSS 1624), the first field
T dwarf, was found in preliminary Sloan Digital Sky Sur­
vey data (S99). The discovery paper includes a red spec­
trum with the Apache Point 3.5­m telescope, with detected
flux down to 8000 š A. The accessible Cs I features at 8521 š A
and 8943 š A appeared weak or absent, while both are dis­
tinct features in the Gl 229B red spectrum (Oppenheimer
et al. 1998). The apparent weakness of the Cs I features
and shallower red spectral slope of SDSS 1624 led BMS
to the preferred conclusion that this object ``is tied to a
lower core entropy,'' which would normally imply a lower
T eff than that of Gl 229B. In contrast, the Sloan source
showed somewhat shallower methane absorption in the in­
frared spectrum, suggesting to Nakajima et al. (2000) that
it is warmer than the prototype. The red spectrum also
shows that a field object can show a similar excess below
1¯m as the companion object Gl 229B, thus demonstrat­
ing that the excess does not depend upon the presence
of a nearby source of potential ultraviolet photons (ie., to
produce the ``Titan Tholins'').
We discuss here a red spectrum of SDSS 1624 obtained
with the Keck II LRIS spectrograph that extends the de­
tection of flux down to 6200 š A, and has improved signal­to­
noise ratio at longer red wavelengths. We believe this ob­
servation allows us to test the roles of the alkali metals and
the need for dust and/or an additional short­wavelength
absorber.
2. THE SPECTRUM AND ITS FEATURES
Two consecutive spectra of SDSS 1624 were obtained on
1999 July 16 with the Keck II telescope and LRIS using
the configuration described for most observations in Kirk­
patrick et al. (1999). Each had an exposure time of 1800
seconds. Reduction was done with standard IRAF tools.
The averaged spectrum spanning the entire wavelength in­
terval of 6,300--10,100 š A at 9 š A resolution is shown in Fig. 1.
Significant flux is detected over this entire range, as shown
in an inset, which details the 6,300--8,200 š A spectrum on an
appropriate vertical scale. The variance spectrum from the
standard IRAF task is also shown in the short wavelength
inset where the signal­to­noise ratio (SNR) is smallest.
Longward of 8200 š A the variance spectrum (not shown)
rises slowly to about 4\Theta10 \Gamma18 near one micron, except for
the noisier intervals affected by the strongest atmospheric
OH bands. Some significant conclusions may be drawn
just from inspection of these figures.
(1) The spectrum at the shorter wavelengths, not acces­
sible in previous observations, shows a strong, broad dip
to zero flux centered precisely on the 7700 š A blended dou­
blet (see upper inset). Note that the variance plot remains
flat over this interval, a strong indication that the feature
is not due to any change in the noise. A ``pseudo''­EW
(pEW) of 390 š A for the line cores was measured over the
interval 7300­8100 š A. This was estimated from the IRAF
``splot'' routine which simply fits a linear continuum across
the stated interval. It is recognized that the flux levels at
the interval boundaries are not the true continuum. More­
over, the procedure ignores the the considerable absorp­
tion in the extended wings of this doublet. Nonetheless,
the pEW estimate may serve as a useful benchmark of
the K I strength for quick comparison with any similar
spectra obtained for other objects. The strength of the
feature confirms the suggestion of Tsuji et al. (1999) and
BMS that the red wing of this feature is a substantial ab­
sorber shortward of one micron in this T dwarf; presum­
ably the same feature contributes to the flux deficiency in
Gl 229B. Indeed, it was already recognized that this fea­
ture increases in strength with later types among the L
dwarfs (Kirkpatrick et al. 1999; Mart'in et al. 1999).
(2) The detected flux rises shortward of 7700 š A, reveal­
ing the blue wing of the K I feature. A broad maximum of
the flux level may be reached near 7000 š A, but significant
flux is detected down to 6200 š A, after moderate decline in
the flux level shortward from 7000 š A. This last decline is
likely due to the red wing of the Na I resonance doublet
centered near 5892 š A. Na is normally the most abundant
of the five alkalis in a Popuation I mix. Both Na and K
are expected to survive to lower temperatures than Li, Cs
and Rb, so their presence here is not surprising. Indeed,
subordinate lines of K I were reported in the S99 (see also
Nakajima et al. 2000) infrared spectrum. The SNR dips
below unity at the last few hundred š A, but the conclu­
sions about the shape of the continuum are robust. No
significant absorption features are claimed.
(3) Significant Cs I absorption lines, both members of
the well­separated doublet, are easily detected. The pEWs
of 6.5 š A and 6.1 š A, for the respective 8521 š A and 8943 š A
transitions, were measured over full width intervals of 25 š A
each. The Cs lines appear much stronger than was appar­
ent in the Arc 3.5­m spectrum of S99. We note that the
apparent strength of the 8943 š A line could be enhanced by
contribution from an overlapping weak CH 4 band. For the
same intervals, pEWs of 6.5 š A and 5.4 š A were measured for
the web­posted Oppenheimer et al. (1998) spectrum of
Gl 229B. While detailed modeling of the alkali line pro­

3
files and red spectrum needs to be done, the conclusion
of BMS that SDSS 1624 may have a lower temperature
than Gl 229B needs to be reconsidered. It is also possible,
however, that the Cs line strength varies with time and/or
location on the surface.
(4) No significant Hff emission or Li I absorption is de­
tected, to limits we estimate as pEWs of 15 š A very approx­
imately, since the SNR is near unity.
(5) The ¸9300 š A H 2 O band reported by S99 is strong
in this spectrum. A possible absorption which appears to
be strongest near 9955 š A, could be due at least in part to
FeH, if the wavelength calibration is poor near the edge of
the spectrum. A possible absorption feature near 8343 š A
may be H 2 O; one near 8624 š A (see lower inset of Fig. 1)
coincides with both CH 4 and CrH bands, more likely the
former.
3. AN EXPLORATORY MODEL FIT AND POSSIBLE
IMPLICATIONS
Several authors have explored quantitative fits to the
alkali resonance lines as a tool for estimating T eff values
for L and T dwarfs. Unfortunately, the detailed treatment
of the line broadening poses a complicated problem -- see
BMS Section 3. Only a few key points are mentioned here.
Available empirical data (cf. Nefedov, Sinel'shchikov, and
Usachev 1999) provide valuable clues to what might be
the best functional form for the profile. However, broad­
ening parameters remain uncertain. In any case, simple
treatments (ie. Lorentzian profiles) used previously in the
literature appear inappropriate.
The detailed spectral fit depends upon several parame­
ters, including the T eff , gravity, and abundances, but also
the line profile shapes, and the degree of rainout (Burrows
and Sharp 1999; BMS). We emphasize that the physical
treatment of these last two remain uncertain. A detailed
exploration of these many parameters would be required,
at minimum, to achieve quantitative estimates of the T eff ,
surface gravity and abundances. It is questionable whether
a satisfactory, unique solution will be found until a trigono­
metric parallax can help fix the luminosity and radius.
What is shown below is an exploratory fit which we ar­
gue nonetheless yields useful qualitative conclusions. The
techniques employed are described more generally in BMS.
Figure 2 shows a comparison between a representative
model spectrum (dashed line) and a smoothed version
of the SDSS 1624 spectrum (solid). Superposed is the
Leggett et al. Gliese 229B spectrum of Oppenheimer et
al. (1998). However, the observed flux has been trans­
formed to an absolute flux for an assumed distance to the
Sloan dwarf of 10 parsecs (S99), and is in milliJanskys,
while the wavelength is in microns.
It may be seen clearly that the K I and Na I doublets
and their broad wings dominate the spectrum. For this
illustrative model fit, we assumed T eff = 1100 K, gravity
= 10 5 cm s \Gamma2 , abundances of one half solar, alkali line
wing cutoff parameters defined in BMS of 0.2 (Na I) and
0.5 (K I), and an intermediate degree of rainout for the
alkalis (BMS). The SDSS 1624 data were smoothed with a
10­ š A boxcar function, which, among other things, muted
the depth of the Cs lines relative to the model, but are of
similar strength. No strong Li I absorption is predicted.
To obtain a reasonable fit, it is not clear to us that a
dust component or additional source of red opacity is re­
quired. Tsuji's need for additional red opacity may be ex­
plainable by (1) underestimation of the alkali wing opacity
due to the assumption of a Lorentzian, and (2) the I­band
broad band flux was plotted at the wrong mean wave­
length (see BMS). Although the presence of dust in the
atmosphere certainly cannot be precluded, the alkalis ap­
pear to be the dominant cause of the unique shape of the
red energy distribution. The detection of flux to the blue
boundary of the spectrum also has consequences. In par­
ticular, a Rayleigh scattering dust opacity, as suggested
by Pavlenko et al (2000), would have more than double
the opacity at 7000 š A than at 8400 š A. Finally, the observed
narrowness of the 7700 š A feature (relative to our models
at the Gl 229B temperature near 950 K) and the presence
of strong cesium features together argue that the effective
temperature of SDSS 1624 is above that of Gliese 229B
(BMS), in concurrence with Nakajima et al. (2000).
We emphasize that no concerted attempt was made to
find a rigorous fit, that other combinations of parameters
are still viable, and that, given the SNR of the data at the
shorter wavelengths, there are indeed parameter degenera­
cies.
This research is supported by a NASA JPL grant
(961040NSF) permitting us to undertake a core science
project on very low mass objects discovered in the 2MASS
survey. AB acknowledges support from NASA grants
NAG5­7499 and NAG5­7073. The model curve was com­
puted based upon a temperature/pressure profile gener­
ated by M. Marley (private communication) and the mod­
els in Burrows et al. (1997). We wish to acknowledge
helpful suggestions from an anonymous referee.
REFERENCES
Burgasser, A., 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, 522, L65
Burgasser, A., Kirkpatrick, J.D., Cutri, R.C., McCallon, H., Kopan,
G., Gizis, J.E., Liebert, J., Reid, I.N., Brown, M.E., Monet, D.G.,
Dahn, C.C., Beichman, C.A., & Skrutskie. M.F. 2000, ApJ, in
press.
Burrows, A., Marley, M., Hubbard, W.B., Lunine, J.I., Guillot, T.,
Saumon, D., Freedman, R., Sudarsky, D., & Sharp, C.M. 1997,
ApJ, 491, 856
Burrows, A., Marley, M., & Sharp, C. 2000, Ap.J., 531, in press (1
Mar 00 issue)
Burrows, A. & Sharp, C.M. 1999. ApJ, 512, 843
Cuby, J.G., Saracco, P., Moorwood, A.F.M., D'Odorico, S., Lidman,
C., Comeron, F., & Spyromilio, J. 1999, A&A, 349 41
Golimowski, D.A. et al. 1998, AJ, 115, 2579
Griffith, C.A., Yelle, R.V., & Marley, M.S. 1998 Science, 282, 2063
Khare, B.N. & Sagan, C. 1984, Icarus, 60, 127
Kirkpatrick, J.D., Henry, T., & McCarthy, D.W. 1991, ApJS, 77, 419
Kirkpatrick, J.D., Reid, I.N., Liebert, J., Cutri, R.M., Nelson, B.,
Beichman, C.A., Dahn, C.C., Monet, D.G., Skrutskie, M.F., and
Gizis, J. 1999, ApJ, 519, 802
Leggett, S.K., Toomey, D.W., Geballe, T.R., & Brown, R.H. 1999,
ApJ, 517, L139
Lodders, K. 1999, ApJ, 519, 793
Mart'in, E.L., Delfosse, X., Basri, G., Goldman, B., Forveille, T., &
Zapatero Osorio, M.R. 1999, AJ, 118, 2466
Nakajima, T., Oppenheimer, B.R., Kulkarni, S.R., Golimbowski,
D.A., Matthews, K. & Durrance, S.T. 1995, Nature, 378, 463
Nakajima, T., Tsuji, T., Maihara, T., Iwamuro, F., Motohara, K.,
Taguchi, T., Hata, R. Tamura, M., & Yamashita, T. 2000, PASJ,
in press.

4
Nefedov, A.P., Sinel'shchikov, V.A., & Usachev, A.D. 1999, Physica
Scripta, 59, 432
Oppenheimer, B.R., Kulkarni, S.R., Matthews, K. & Nakajima, T.,
1995, Science, 270, 1478
Oppenheimer, B.R., Kulkarni, S.R., Matthews, K. & van Kerkwijk,
M.H. 1998, ApJ, 502, 932
Pavlenko, Ya., Zapatero Osorio, M.R., & Rebolo, R. 2000, Astr. Ap.,
in press
Reid, I.N., Kirkpatrick, J.D., Williams, R., Liebert, J., & Burgasser,
2000, Astronomical Journal, in press
Schultz, A.B. et al. 1998, ApJ, 492, L181
Strauss, M.A., Fan, X., Gunn, J.E., Leggett, S.K., Geballe, T.R.,
Pier, J.R., Lupton, R.H., Knapp, G.R., et al. 1999, ApJ, 522, L61
(S99)
Tsuji, T., Ohnaka, K., & Aoki, W. 1999, ApJ, 520, L119
Tsvetanov, Z.I. et al. 2000, submitted to ApJ
6500
7000
7500
8000
8500
9000
9500
10000
0
SDSS
1624
6500
7000
7500
8000
0
8300
8400
8500
8600
8700
Fig. 1.--- The Keck II spectrum of 2M1624+0029, F – vs. –( š A), boxcar­smoothed by 5 pixels (10 š A). Two inset boxes are discussed in the
text. The noise (variance) spectrum is shown in the top inset (shifted upward by 10 \Gamma18 ) and is flat over this interval.

5
.6 .65 .7 .75 .8 .85 .9 .95 1
­4
­3
­2
­1
0
Model
SDSS 1624
Gliese 229B
Fig. 2.--- Comparison between a representative model spectrum (dashed) and a smoothed version of the SDSS1624 spectrum (solid), along
with the published spectrum of Gl 229B (see text for details). The flux is in milliJanskys, and the wavelength in microns. The normalizations
of the observations are based on an arbitrary distance assumed for SDSS1624 and the trigonometric parallax of Gl 229B.