On the circum(sub)stellar environment of brown dwarfs in Taurus

0.5 MB·English

by  S. Guieu

Astronomy&Astrophysicsmanuscriptno.disks-nb6-print-corrected (cid:13)c ESO2008
February5,2008
On the circum(sub)stellar environment of brown dwarfs in Taurus
⋆
7 S.Guieu1,C.Pinte1,J.-L.Monin1,2,F.Me´nard1,M.Fukagawa3,D.L.Padgett3,A.Noriega-Crespo3,S.J.Carey3,
0
L.M.Rebull3,T.Huard4,andM.Guedel5
0
2
n 1 Laboratoired’AstrophysiquedeGrenoble,BP53,38041Grenoble,Francee-mail:[email protected]
a 2 InstitutUniversitairedeFrance
J 3 SpitzerScienceCenter,Caltech,Pasadena,CA91125,USA

9 4 Harvard-SmithsonianCenterforAstrophysics,60GardenStreet,Cambridge,MA02138,USA
5 PaulScherrerInstitut,Wu¨renlingenandVilligen,CH-5232VilligenPSI,Switzerland

1
v
Received31July2006;Accepted18December2006
1
5
ABSTRACT
2
1
0 Aims.Wewanttoinvestigatewhetherbrowndwarfs(BDs)formlikestarsorareejectedembryos.WestudythepresenceofdisksaroundBDs
7 intheTauruscloud,anddiscussimplicationsforsubstellarformationmodels.
0 Methods.Weusephotometricmeasurementsfromthevisibletothefarinfraredtodeterminethespectralenergydistributions(SEDs)ofTaurus
/ BDs.
h
Results.WeuseSpitzercolorindices,Hαasanaccretionindicator,andmodelsfittotheSEDsinordertoestimatephysicalparametersofthe
p
disksaroundtheseBDs.WestudythespatialdistributionofBDswithandwithoutdisksacrosstheTaurusaggregates,andwefindthatBDs
-
o withandwithoutdisksarenotdistributedregularlyacrosstheTauruscloud.
r Conclusions.Wefindthat48%±14%ofTaurusBDshaveacircumstellardisksignature,aratiosimilartorecentresultsfrompreviousauthors
t
s inotherregions.WefittheSEDsandfindthatnoneofthedisksaroundBDsinTauruscanbefittedconvincinglywithaflaringindexβ = 0,
a
indicatingthatheatingbythecentralobjectisefficientandthatthedisksweobserveretainasignificantamountofgas.WefindthatBDswith
:
v disksareproportionallymorenumerousinthenorthernTaurusfilament,possiblytheyoungestfilament.Wedonotfindsuchaclearsegregation
i forclassicalTTauristars(CTTS)andweak-linedTTauristars(WTTS),suggestingthat,inadditiontotheeffectsofevolution,anysegregation
X
effectscouldberelatedtothemassoftheobject.Aby-productofourstudyistoproposearecalibrationoftheBarradoyNavascue´s&Mart´ın
r (2003)accretionlimitinthesubstellardomain.Theglobalshapeofthelimitfitsourdatapointsifitisraisedbyafactor1.25-1.30.
a
Keywords.Stars:formation-Stars:lowmass,browndwarfs-Stars:pre-mainsequence-Accretiondisks
1. Introduction & Bouvier2003a).These twomodelsarenotmutuallyexclu-
sive;othermechanismsarealsodiscussedintheliterature(e.g.,
In recent years, a large number of brown dwarfs (BDs) have
Whitworth&Goodwin2005,Whitworthetal.2006).
beendetectedinstarformingregions,openingtheopportunity
TheTaurusregionhasbeenextensivelystudiedforstarfor-
to study the stellar formation process and the corresponding
mation.Itisyoung(1-5Myr),sothedynamicaleffectsremain
IMF deep into the substellar domain, even down to the plan-
limited.Itextendsovera largeregion,soitcanbestudiedfor
etarymassregime(Chauvinetal.2005;Luhmanetal.2005).
largespatialdistributioneffects;therearenobrightstarstoir-
Twomainclassesofmodelshavebeenproposedfortheforma-
radiateanddisturbthestellarsurroundings.Ourstudyisbased
tion of substellar objects. In the standardformationscenario,
on a sample of 33 BDs in the Taurus cloud as presented in
BDsformlikestars,through(turbulent)gravitationalcollapse
Guieuetal.(2006)andreferencestherein.Withsuchanumber
and fragmentation of very low mass cores, followed by sub-
of objects at hand, we can now begin statistical studies. With
sequent disk accretion. In theejectionmodel, BDs are stellar
the aim of studying the proportionof TaurusBDs that harbor
embryosejectedfromtheirparentcoreeitherearlyintheirevo-
an accretion disk, we have combined Guieu et al. (2006) op-
lutionfromdynamicallyunstablemultipleprotostellarsystems
tical photometry with JHK 2MASS data and recent Spitzer
(Reipurth&Clarke2001),orthroughseculardynamicaldecay s
3.6to70µmdatatodeterminethespectralenergydistributions
in denseembeddedclusters(Sterzik& Durisen2003;Kroupa
(SEDs)ofBDsintheTauruscloud.
Sendoffprintrequeststo:S.Guieu Numerous studies have attempted to distinguish between
⋆ BasedonobservationsmadeatESO,CFHT,2MASS,&Spitzer. the two principal models of BD formation by examining the
2 S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus
circumstellardisksofyoungbrowndwarfs.Forinstance,there inthe mosaicimages.Thefluxesweremeasuredateachindi-
is now ample evidence that, like their more massive counter- vidualepoch,thenaveragedtoobtainthefinalvalues.Wehave
parts,TaurusBDsexperienceaTTauriphase.Broadasymmet- computed the correspondingmagnitudesadopting IRAC zero
ric Hα emission profiles characteristic of accretion have been magnitudeflux densities of 280.9,179.7,115.0,and 64.13Jy
found;see,e.g.,Jayawardhanaetal.(2003)andMuzerolleetal. inthe3.6,4.5,5.8and8micronbands,respectively(Reachet
(2005).LbandexcesseshavebeendetectedinTaurussubstellar al.2005);forMIPS,weusedzeropointsof7.14Jyforthe24
sources,indicatinga diskfrequency≈ 50%(Liuetal.2003). micronbandand0.775Jyforthe70micronband,basedonthe
Recently,Whelanetal.(2005)havedetectedanoutflowfroma MIPSDataHandbook.Theuncertaintyontheabsolutecalibra-
BDbyspectroastrometry.Ithasoftenbeenarguedthatthepres- tionis10%fortheIRACbandsand24µm,and20%for70µm
enceofaccretionand/oroutflowactivityinBDsisevidencethat measurements.
brown dwarfs form like stars (i.e., are not ejected). However, TheIRACfluxesfor23TaurusBDsareshowninTable1.
currentstarformationmodels(seee.g.,Bateetal.2003)show CFHT-Tau 12andKPNO-Tau 9werenotobservedatallIRAC
thatthe majorityof remnantdisksin browndwarfshaveradii bands. Eleven sources were detected at 24µm, and just one
lessthan20AU;thesecalculationsdonotpossessaresolution source (J04442713+2512164)was detected at70µm, with an
sufficienttofollowthefateofthesedisksafterejection.Hence uncertaintyof0.2mag.Theupperlimitsofundetectedsources
the possibility remains that BDs can retain a disk even when are 0.9mJy at 24µm, and . 100mJyat 70µm, dependingon
theyareejected.WithanaccretionrateofM˙ =510−12M⊙.yr−1 thebackgroundlevelduetothecloudemission.Theerrorbars
(Muzerolleetal.2005),evena510−5M⊙ diskwouldsurvivea are within the size of the symbol in Figure 2, except for the
fewMyr,alifetimeconsistentwiththeageofTaurus. 70 µm measurement. Note that IRAC photometry for objects
Inthispaper,wepresentcompletephotometryavailableon KPNO-Tau 4 to 7 andGM Tau hasalreadybeenpublishedin
those 23 BDs in Taurus for which Spitzer data are available, Hartmannetal.(2005).
from the visible to 70µm (Section 2). We combine this large
We havealsoused the2MASScatalogtoobtain J, H and
range of photometry with other observations such as spectral
K bandphotometryforthewholesampleofTaurusBDs.The
s
types to further aid in interpretation. We sort these SEDs de-
transformationbetween2MASSphotometryandabsoluteflux
pendingonthepresenceofaninfraredexcessandwefitthese
has been performed using the zero-point fluxes from Cohen
excess-bearingsourceswithadiskmodel(Section3).Wedis-
(2003),specifically1594,1024,and666.7Jyfor J, H and K
s
cusstheimplicationsofourresultsforBDformationmodelsin
bands,respectively.
Section4.2.
The optical data come from several telescopes. We have
obtainedI photometryusingCFHT12kandMegacamcameras
2. Observationsandresults on the Canada-France-Hawaii telescope for 16 BDs. In addi-
tion, 8 BDs possess R photometryobtainedwith the same in-
2.1.Observations struments.Themaincharacteristicsofthisphotometricobser-
vations are presented in Guieu et al. (2006). Additional visi-
Table 1 lists the names, spectral types, temperatures, and the
ble photometric data have been obtained from Briceno et al.
photometricmeasurementsdescribedinthissectionforthe23
(2002),Luhmanetal.(2003),andLuhman(2004).
BDs for which Spitzer photometry is available. The observa-
All R and I data have been transformed to the CFHT12k
tions reported here have been collected with various instru-
camera’s Cousins system (see Guieu et al. 2006; Briceno et
ments from the visible to the infrared range. Not all objects
al.2002)beforeconversiontoabsolutefluxes.Themagnitudes
havebeenmeasuredineveryphotometricbandavailable.
arelistedinTable1whiletheabsolutefluxesareusedinSED
Mid-infraredphotometryhasbeenobtainedwiththeIRAC
fitting(seeSection3.3).
(ateffectivewavelengthsof3.6,4.5,5.8,and8µm;Fazioetal.
2004) and MIPS (24, 70, and 160µm, Rieke et al. 2004) in-
strumentsonboardtheSpitzerSpaceTelescope(Werneretal.
2.2.SEDsandinfraredexcesses
2004).The Spitzerfluxeswere extractedfromthe mosaicim-
ages obtained as partof General Observerprogram3584 (PI: WehaveplottedtheSEDsinFigures1and2forallTaurusBDs
D. Padgett) for wide area (≈ 30deg2) mapping of the Taurus withavailableIRACphotometry.TheSEDsaresorteddepend-
cloud(Guedeletal. 2006).Theobservationswere carriedout ingonthepresenceofanIRexcessdeterminedasexplainedbe-
in February and March 2005, with exposures at two epochs low.Onthesameplots,wehavesuperimposedaphotospheric
severalhoursaparttoprovideforasteroidrejectioninthislow- fit using the DUSTY models from Allard et al. (2000). Each
ecliptic-latitude region. Photometry was performed using the modelhas been fit using the BD’s temperatureand visual ab-
APPHOT and PHOTCAL packages in IRAF. We measured sorption previously published in the literature combined with
fluxeswithinanapertureradiusof3pixels(3.′′6)andapplied the Draine (2003a, 2003b, 2003c) extinction law. Following
aperturecorrectionsto 8 pixels(9.′′6) for IRACbands. At 24 NattaandTesti(2001),whoshowthatBDdiskexcessemission
microns, we used 3 pixel (7.′′4) apertures and corrected the becomesclearlydetectableonlylongwardof3µm,wehavefit
flux to 11 pixels(27.′′0); at 70 microns,we also used 3 pixel the DUSTY models to the available data points just in the R,
(29.′′5)aperturesandcorrectedthefluxto8pixels(78.′′7).The I ,J,HandK bands.WedecidethatagivensourcehasanIR
c s
aperture corrections from 3 pixels to 8 or 11 pixels were de- excess when the IRAC photometry exceeds the photospheric
rivedbasedonthephotometryofbrightisolatedpointsources SED bymorethanonesigma.Forsomesources(e.g.,CFHT-
S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus 3
Table1.Visible-infraredphotometryforthe26BDsstudiedinthispaper.ThesourceswithIRexcessarelabeledwith∗.The
dataaregivenasmagnitudes(seetextfordetails)exceptforthe24and70µmfluxes,giveninmJy.
Name SpT Av R I J H K [3.6] [4.5] [5.8] [8.0] 24 70
mJy
CFHT-Tau 9∗ M6.25 0.91 ... 15.35 12.88 12.19 11.76 11.14 10.86 10.45 9.80 12.00 <57.20
KPNO-Tau 4 M9.50 2.45 20.54 18.75 15.00 14.02 13.28 12.52 12.34 12.14 12.10 <0.89 <57.20
CFHT-Tau 15 M8.25 1.30 ... 17.94 14.93 14.24 13.69 13.19 13.16 13.04 13.04 <0.89 <57.20
KPNO-Tau 5 M7.50 0.00 19.10 15.08 12.64 11.92 11.54 11.03 10.99 10.90 10.84 <0.89 <57.20
KPNO-Tau 6∗ M9.00 0.88 20.56 17.90 14.99 14.20 13.69 13.08 12.79 12.47 11.68 <1.65 <75.80
CFHT-Tau 16 M8.50 1.51 ... 17.91 14.96 14.24 13.70 13.26 13.16 13.01 12.97 <0.89 <57.20
KPNO-Tau 7∗ M8.25 0.00 ... 17.16 14.52 13.83 13.27 12.59 12.27 11.90 11.22 2.29 <57.20
CFHT-Tau 13 M7.25 3.49 ... 17.90 14.83 13.97 13.45 12.74 12.73 12.84 12.60 <1.20 <72.20
CFHT-Tau 7 M6.50 0.00 16.63 14.12 11.52 10.79 10.40 9.83 9.91 9.70 9.69 <1.41 <134.00
CFHT-Tau 5 M7.50 9.22 ... 18.79 13.96 12.22 11.28 10.48 10.25 10.03 10.01 <0.94 <57.20
CFHT-Tau 12∗ M6.50 3.44 ... 16.26 13.15 12.14 11.55 ... ... ... ... 3.38 ...
CFHT-Tau 11 M6.75 0.00 ... 14.88 12.53 11.94 11.59 11.13 11.09 10.96 10.96 <0.89 ...
KPNO-Tau 9 M8.50 0.00 ... 18.76 15.48 14.66 14.19 13.51 ... ... ... <0.89 ...
CFHT-Tau 2 M7.50 0.00 ... 16.81 13.75 12.76 12.17 11.54 11.41 11.32 11.33 <0.89 <65.90
CFHT-Tau 3 M7.75 0.00 ... 16.88 13.72 12.86 12.37 11.78 11.69 11.61 11.59 <0.89 ...
J04380083+2558572 M7.25 0.64 20.21 14.71 11.54 10.62 10.10 9.56 9.45 9.31 9.28 1.22 <79.00
J04381486+2611399∗ M7.25 0.00 20.33 17.84 15.18 14.13 12.98 10.77 10.19 9.66 8.91 62.80 <70.70
GM Tau∗ M6.50 4.34 ... 15.04 12.80 11.59 10.63 9.25 8.76 8.38 7.80 46.30 <103.00
CFHT-Tau 6∗ M7.25 0.41 18.40 15.40 12.64 11.84 11.37 10.72 10.44 10.00 9.11 15.90 <81.50
CFHT-Tau 4∗ M7.00 3.00 ... 15.78 12.17 11.01 10.33 9.48 9.06 8.58 7.79 66.00 <77.80
CFHT-Tau 8∗ M6.50 1.77 19.28 16.43 13.17 12.12 11.45 10.83 10.31 9.86 9.18 16.80 <57.20
J04414825+2534304∗ M7.75 1.06 ... 17.03 13.73 12.80 12.22 11.37 10.88 10.43 9.53 18.40 <57.20
J04442713+2512164∗ M7.25 0.00 ... ... 12.20 11.36 10.76 9.51 8.99 8.33 7.40 124.00 157.00
Tau2),theIRACmeasurementsfallsomewhatabovetheSED dataareplottedasfull/emptytrianglesdependingonthepres-
buttheslopeofthepointsisthesameastheunderlyingphoto- ence / absence of an IR excess in the SED. We find that the
sphere.Insuchcases,weconcludethatitismorelikelythatthe TaurusBDsareessentiallyplottedintwodistinctregions,iden-
shiftcomesfromasmallfitmismatchratherthanfromarealIR ticaltotheonesfoundforTTauristars(TTS)byHartmannet
excess.TheonlyexceptiontothisruleisCFHT-Tau12,which al. (2005). We interpret the distinction between these two re-
wasnotobservedwithIRAC,butshowsaconfirmeddetection gionsas due to the presenceor absence of circumstellar dust,
at24µm.WehavethusplacedthisBDintheIRexcesslist. in a way consistent with the classification adopted to sort the
Asapreliminaryconclusion,wefindthat11objectsoutof SEDs in Figures1 and 2. On the same figure,we havesuper-
23 have a significant IR excess, suggesting that 48%± 14% imposed the data points from Hartmann et al. (2005)as open
ofTaurusBDspossessdisks.Wecomputedtheuncertaintyon /filledcirclesforTaurusclassII/IIIstars.Thereappearstobe
this percentageusing simple Poisson statistics on the number no clear segregationbetween BDs and TTS colors when they
of BDs with disks. This proportionis very similar to the one haveadisk.Incontrast,inthelowerleftcorneroftheplot,BDs
of classical T Tauris (CTTs) among T Tauri stars in Taurus withoutdiskstendtoberedderontheaveragethantheirWTTS
(Hartmann et al. 2005), and is fully consistent with the pro- equivalents as measured by the [3.6]−[4.5] index. Figure 3
portion of BDs with disks found by Luhman et al. (2005) in showsthatCTTsandBDswithinfraredexcessesappearindis-
IC348andChaI.Insection3,wewillpresentthemodelsused tinguishable, consistent with their emission being dominated
tofittheSEDswithinfraredexcesses.Thesefitsaresuperim- bytheirdisksinbothcases,whiletheWTTsandBDswithout
posedontheobjects’SEDsplottedinFigure2.Wecouldonly diskscanbedistinguishedbytheirphotosphericcolors.
fit9outof11SEDsforreasonsexplainedinSection3.Inthe InFigure4wehaveplottedtheIRAC[5.8]−[8]colorindex
followingparagraphs,weexploreotherphysicalparametersto versustheunderlyingeffectivetemperatureofthecorrespond-
studythepresenceofdisksaroundourobjects. ing photosphere.Again,there is a clear gapof about0.7 mag
betweentheBDswithandwithoutdisks,withnocleardepen-
denceofthecolorindexonthecentralobjecteffectivetemper-
2.3.Spitzercolor-colordiagrams
ature,henceitsmass,atagivenage.
In the previous section, the presence or absence of a disk
aroundaBDisinferredmorefromtheslopeoftheIRACdata
2.4.Accretionsignatures
than from the actual value of the IRAC data comparedto the
photospheric fit. To confirm our results, we have plotted in In Figure 5, we have plotted the Hα equivalent width (EW;
Figure 3 the Spitzer [3.6]−[4.5] vs. [5.8]−[8] color-color in- Guieu et al. 2006;Bricen˜o et al. 2002)from spectra obtained
dices for all the BDs where the photometry is available. The with moderate resolution (R ≈ 1000) versus spectral type
4 S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus
−2m) 10−15 10−15 10−14
W.
10−15
(λ 10−16 10−16
F KPNO−Tau 4 CFHT−Tau 15 KPNO−Tau 5
λ
10−16
10−17
10−17
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
10−13
)
−2m 10−15 10−15 10−14
W.
F (λ 10−16 CFHT−Tau 16 10−16 CFHT−Tau 13 10−15 CFHT−Tau 7
λ
10−17
10−17 10−16
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
) 10−15
−2 10−14 10−14
m
W.
( 10−15 10−16
λ 10−15
F CFHT−Tau 5 CFHT−Tau 11 KPNO−Tau 9
λ 10−16 10−17
10−16
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
10−14 10−14 10−13
)
2
−
m
W. 10−15 10−15 10−14
(

λ Fλ 10−16 CFHT−Tau 2 10−16 CFHT−Tau 3 10−15 J04380083+2558572
10−16
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
λ (m) λ (m) λ (m)
Fig.1. Mosaic of SEDs for BDs classified as havingno excess (λF in W m−2). Empty trianglesdenoteSpitzer upperlimits.
λ
Dot-dashedline:DUSTYmodelsfromAllardetal.(2000)fittedonthevisible-NIRrange.
for all BDs where Spitzerphotometryand EW(Hα) are avail- and our disk / no disk criterion. All but one of the BDs with
able (filled triangles: BDs with disks; empty triangles: BDs disks have an Hα emission level in excess of that expected
withoutdisks).TheempiricalCTTS/WTTSboundaryextended from chromospheric activity, suggesting that most BDs with
to substellar analogs, defined by Barrado y Navascue´s and disks are experiencingan accretion phase, or a jet. Moreover,
Mart´ın (2003), is delineated by the dashed line. This bound- although we are dealing with small numbers, there appears
aryisdefinedbythesaturationlimitforchromosphericactivity to be two groups of accreting BDs in our data: BDs with
(L /L =−3.3). EW(Hα) > 300Å are strong accretors, while objects with
Hα bol
EW(Hα) < 100Åmayhavestoppedsignificantaccretionand
There is a general agreement between the accretion / non
accretion limit from Barrado y Navascue´s and Mart´ın (2003)
S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus 5
10−14
−2m) 10−14 10−15 10−15
W.
10−15
(λ 10−16 10−16
F CFHT−Tau 9 KPNO−Tau 6 KPNO−Tau 7
λ
10−16
10−17 10−17
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
10−13
10−14
2) 10−14 10−14
−
m 10−15
W. 10−15
( 10−15
λ 10−16
F CFHT−Tau 12 J04381486+2611399 10−16 GM Tau
λ
10−16 10−17 10−17
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
10−13
2) 10−14 10−14
− 10−14
m
W.
( 10−15 10−15 10−15

λ
F CFHT−Tau 6 CFHT−Tau 4 CFHT−Tau 8
λ
10−16 10−16 10−16
10−6 10−5 10−4 10−6 10−5 10−4 10−6 10−5 10−4
10−13 λ (m)
10−14
)
2
− 10−14
m
W. 10−15
( 10−15
λ
F J04414825+2534304 J04442713+2512164
λ 10−16
10−16
10−6 10−5 10−4 10−6 10−5 10−4
λ (m) λ (m)
Fig.2.Sameasfigure1forBDswithinfraredexcesses.ThegreenlinetracesourSEDmodels;seetextandSection3fordetails.
be surrounded by more passive disks, analogs to the one de- accretionlimitisat200kms−1).Moreover,thislatterobjectis
scribedinMcCabeetal.(2006). CFHT-Tau4,whichwasfoundbyPascuccietal.(2003)tohar-
boradisk,asshownbyitsmmemission.InourHαmeasure-
As the Hα EW mightnot providean unambiguousaccre- ments, we find that CFHT-Tau 4 has an Hα equivalent width
tionstatusofagivenobject,wehavecomparedourresultswith of 300km s−1. Additionally, among the 12 sources we clas-
thosefromMohantyetal.(2005).Theystudytheaccretionin sify asBDs withoutdisks, 4 arefoundto be non-accretorsby
BDsusingtheHα10%width.Amongthe23sourcesstudiedin Mohantyetal.(2005),andoneisapossibleaccretor(KPNO-
thispaper,9wereobservedbyMohantyetal.(2005).Wefind Tau4).UsingthislimitedoverlapbetweentheMohantyetal.
thatamongour11sourcesclassified asBDs with disks, 3are (2005)dataandours,wecanconcludethatthereisacorrelation
foundtobeaccretingbyMohantyetal.,andoneisfoundtobe betweenouraccretionclassificationandtheirs.Ifweignorethe
apossibleaccretorwithanHα10%width=150kms−1 (their
6 S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus
1.0
0.6
0.4
5] 8]
[4. − [ 0.5
6]− 8]
3. 0.2 5.
[ [
0.0
0.0
0.0 0.5 1.0 2600 2800 3000
[5.8] − [8] Teff
Fig.3.[3.6]−[4.5]vs.[5.8]−[8]color-colordiagramoftheBDs Fig.4. [5.8]−[8] µm color index of the disk emission versus
studiedinthiswork,superimposedontheTTSfromHartman effectivetemperatureofthe correspondingcentralobject.The
et al. (2005). Filled / empty triangles: BDs with / without IR solidlinedelineatesthemodelcomputedusingtheAllardetal.
excesses; filled circles: class II (CTTS); and empty circles: (2000)photosphere.Open/filledtrianglesdenoteBDwithout
classIII(WTTS).ThetypicaluncertaintyonSpitzercolorin- / with IR excess. The typical uncertainty on the Spitzer color
dicesis0.05(seesec.2.1) indexesis0.05.
objectsthatappearasintermediate,orpassive,thepresenceof
a disk inferredfromthe IRexcessis thushighlycorrelatedto find that, although globally, the fraction of BDs with disks is
thepresenceofaccretion. about50%,BDswithandwithoutdisksarenotdistributedreg-
As a word of speculation,it couldbe notedthat there is a ularly.Forinstance,aggregateIIIhas7outof8BDswithdisks,
smallmismatchbetweentheaccretionlimitandourthreelow while aggregate V has 4 out of 5 BDs without disks, the re-
EW(Hα) objects;the differencefalls withinthe errorbars, al- mainingonebeingBDCFHT-Tau12,whichhasthelowestIR
though the shift is in the same direction for all three objects. excessfromallthesampleat24µm.Wehavealsocountedthe
Rather than speculating about possible remnant accretion in proportionsof class I+II and III (C/W)TTs in the aggregates,
BDs withoutstrong disks, it could be possible that the exten- inside the 3 star/deg2 contour. The proportions of CTTs and
sion in the BD domain of the boundary drawn by Barrado y BDs with disks in the variousaggregateswhere this informa-
Navascue´s and Mart´ın (2003)has to be offset by +20−30% tion is available are listed in Table 2. We have also listed the
tofitournewdatapoints.Indeed,ifweraisetheirlimitbythis (binomial)probabilitytogettheaggregateBDdiskproportion
amount, it nicely follows our BDs without disks (open trian- if a given BD has a 48% probability to get a disk. Given the
gles) leaving all the low-accreting BDs with disks below the uncertaintyonthe proportionofBDs with disks,the numbers
chromosphericactivitylimit. listed in Table 2 are not absolutely inconsistentwith an over-
all 50% disk frequency,although with rather low probability.
Indeed, if the probability for a BD to have a disk falls down
2.5.Browndwarfsspatialdistribution
to its lowest value (34%), there is a 20% chance to find the
Figure6showsthespatialdistributionofourobjects,superim- repartition(0/4)inaggregateIV.However,thisensembleofre-
posedontheaggregatesdefinedbyGomezetal.(1993)plotted sults could be evidenceof differentphysicalconditionsin the
as stellar isodensity peaks; we label them from III to V ac- variousaggregates.WecomebacktothispointinSection4.2.
cordingly (the aggregate in the upper right part of the figure Inordertocheckifthereisadifferenceofspatialposition
wasnotlabeledbyGomezetal.1993).Weplotisodensitycon- of the BDs with and without disks relative to the stellar ag-
toursat3,6and12starspersquaredegree.BDswithdisksare gregates, we have checked a series of estimators: i) We have
plottedasfilledtrianglesandBDswithoutdisksareplottedas computedtheaveragedistancetotheneareststarforBDswith
opentriangles.WealsoshowthepositionsoftheclassI+IIand andwithoutdisks,andfoundnosignificantdifferencebetween
III (C/W) T Tauristars of the Tauruspopulation,as compiled thetwo.ii)TheaveragedistancefromaBD withandwithout
from Kenyon and Hartmann (1995), White and Ghez (2001), disktoitsnearestaggregatecenteraresimilar.iii)Wehavealso
Hartmannetal.(2005),andAndrewsandWilliam(2005).We computedthe weightedstellar density whereeach BD stands,
S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus 7
100.0
)))
ÅÅÅ
(((
) ) )
ααα
HHH
(((
WWW
EEE
10.0
6 7 8 9
MMM SSSpppeeeccctttrrraaalll TTTyyypppeee
Fig.5.HαequivalentwidthversusM 6-9spectraltypeforall Fig.6. Spatial distribution of TaurusBDs. Filled triangles are
theBDs inoursamplewheredataare available.Fulltriangle: BDs with disks; open triangles are BDs without disks. Filled
BDwithadisk;emptytriangle:BDwithoutadisk. and opencircles denoteclass I+IIand III TTauristars (TTS).
The few crosses mark stars that could not be classified. The
solid linesdelineatethe 3,6, and12 star/deg2 isodensitycon-
usingtheKernelmethod(Silverman1986),andwefindnosig-
tours.
nificantdifferencebetweenBDswithandwithoutdisks.
Table 2. CTTS and BDs with disk proportions in aggregates theequatorialplane,h the scale heightattheradiusr = r .
0 0 in
where the information is available. The number listed in the
Thedisk extendsfromaninnerradiusr to anouterlimitra-
in
thirdlinearethebinomialprobabilitiestogettheobservedBD
dius r . The central star is representedby a sphere radiating
out
repartition,computedwith p (disk)=0.48.
BD uniformlywithphotosphereparametersextractedfromthelit-
erature(Guieuetal.2006andreferencestherein)andthecorre-
Aggregate III IV V
spondingsyntheticbrowndwarfspectraofAllardetal.(2000)
CTTS 15/18 15/19 8/13
showninFigure1and2.
BDdisks 7/8 0/4 0-1/5
Probability 2.4% 7% 3.8-17%
3.2.Dustproperties
Weconsiderhomogeneoussphericalgrainsandweusethedi-
electric constants described by Mathis & Whiffen (1989) in
3. Diskmodels
theirmodelA,withtypicalinterstellarmediumvalues.Thedif-
Inordertoestimatethedistributionandtheamountofmaterial ferential grain size distribution is given by dn(a) ∝ a−3.7da
presentinthedisksaroundtheBDsshowinganIRexcess,we with grain sizes between amin = 0.03µm and amax = 1µm.
haveuseda3DMonte-Carlocontinuumradiativetransfercode Themeangraindensityis0.5gcm−3 toaccountforfluffiness.
(MCFOST, see Pinte et al. 2006), to model the SEDs of the Extinctionandscatteringopacities, scatteringphase functions
BDs with IR excess. Our model includes multiple scattering andMuellermatricesarecalculatedusingMietheory.Dustand
with passive dustheating,assumingradiativeequilibriumand gasareassumedtobeperfectlymixedandgrainpropertiesare
continuumthermalre-emission. taken to be independentof position within the disk. The total
disk mass (gas+dust)is fixed at 1M , assuming a gas to dust
J
massratioof100.
3.1.Dustdistributioninthedisk
We use a density distribution with a Gaussian vertical profile
3.3.Modelfitting
ρ(r,z) = ρ (r) exp(−z2/2h2(r)),assumingaverticallyisother-
0
mal, hydrostatic,non self-gravitatingdisk. We use power-law The fits were performed using a grid of SED models built
distributionsfor the surface density Σ(r) = Σ (r/r )α and the by variation of 5 free parameters whose values are listed in
0 0
scaleheighth(r)=h (r/r )βwhereristheradialcoordinatein Table3.
0 0
8 S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus
Table3.Parameterrangeofthemodelscomputedinthispaper. orbyahighlyflaringmodel(β = 1.25).InTable4welistthe
valuesobtainedbyourfittingproceduredescribedabove.
Parameter rangevalues
r (AU) 0.015 0.032 0.067 0.14 0.3
in
Table 4. Best fit results for all the BDs with IR excesses; the
β 0.0 1.0 1.125 1.25
α -0.5 -1 -1.5 sources are listed in the same order as they are presented in
h /r 0.02 0.04 0.06 0.08 0.1 Figure2
0 in
cos(i) from0.05to0.95bystepsof0.1
Object h /r r (AU) β
0 in in
CFHT-Tau9 0.02 0.032-0.067 1.125
KPNO-Tau6 0.02 0.032 1.0-1.125
0.6 KPNO-Tau7 0.04 0.015 1.0-1.125
GMTau 0.06-0.1 0.015-0.067 1.0
.
CFHT-Tau6 0.04 0.067-0.14 1.0-1.125
CFHT-Tau4 0.04-0.06 0.14-0.3 1.0
CFHT-Tau8 0.04-0.06 0.015-0.067 1.0-1.125
0.4 J04414825+2534304 0.06 0.015-0.032 1.125
5] J04442713+2512164 0.1 0.032-0.067 1.125
4.
[
−
]
6 Ourdiskfittingproceduredoesnotaddresstheentiredisk
3.
[ 0.2 parameter space. Indeed, even the 24µm Spitzer photometry
onlyprobestheinnerpartofthediskonascale∼1AU,sothat
mostofthediskmassremainshiddenintheouterpartsofthe
disk. In our disk model, 75% of the disk energy comes from
theinner1AUand90%comesfromtheinner3AU.Westress
0.0
that our model is able to compute the SED up to λ ≈ 1mm,
where the disk is optically thin, so future mm measurements
of the disks around our BD sample will be very valuable to
0.0 0.5 1.0 1.5
estimatetheirmass.However,ourresultsshowthatforallthe
sources that we have fitted, we can rule out a disk structure
[5.8]−[8]
without flaring, showing that in all the BD disks sampled in
Fig.7. Dereddened BD + TTS superimposed on disk models this study, the disk retains a significant amount of gas. This
in a color-colordiagram, with h and h /r taking the values isconsistentwiththelevelofaccretionobservedinalmostall
o o in
listedinTable3.Startingfromthelowerleftpointofthegrid BDswithinfraredexcess.
(0.5,0.2),h /r variesfrom0.02to0.1upward,whiler varies In our modeling, the disk structure is described by para-
o in in
from0.015to0.3AUtotheright. metric laws. Nevertheless,the dust scale heightinferred from
thebestmodelswascomparedwiththehydrostaticscaleheight
(computedfromthedisktemperatureandcorrespondingtothe
Foreachobject,themodelsaresortedfollowingapseudo- gas scale height). The dust scale height is marginallysmaller
χ2 minimisation. We decide that a parameter value is accept- than,butcompatiblewith,thehydrostaticscaleheight.
ablewhenthecorrespondingχ2 valueislessthantwicetheχ2 The fitted SEDs are superimposed on the objects’ SEDs
ofthebestmodel.ForeachmodelSED,wecomputeitsSpitzer in Figure 2. Only 9 fits are displayed there because we could
colors in the 3.6, 4.5, 5.8 and 8 µm bands and we plot them notfind a satisfactory fitresulteither forCFHT-Tau 12or for
inthe[3.6]−[4.5]vs.[5.8]−[8]colorcolordiagram.Usingthis J0438+2611.Fortheformerobject,thisisbecauseofthelack
diagramas a diagnostictool,we find thatwhen we varyβ, α, ofSpitzerphotometry.Forthelatter,theSEDofthisobjectis
andcos(i),thecorrespondingcolorpointsarespreadrandomly the only one in the sample to show such a rise toward longer
across the diagram.In contrast, the variationof r and h /r wavelengths.Weinterpretthisbehaviorasbeingduetothepe-
in o in
definea coherentgridacrossthe diagram.Figure7showsthe culiarorientationofthe diskaroundtheBD, close toedge-on
deredened TTS and BD color points superimposed on such a (Luhman2004).Moreover,if it were notforthe higheremis-
model grid obtained for h /r varying in the range listed in sioninthe10−100µmdomain,thissourceistheweakestone
o in
Table 3 (5 × 5 values). Each point is computed for a given ofthesample,consistentwithathickdiskoccultingthecentral
(h ,r )couple,withalltheresultsduetootherparametersvari- object.
o in
ationsaveraged.Ofcourse,theresultofagivenmodeldepends
on the central object used to compute the surrounding disk
3.4.Comparisonwithothermodels
parameters. In Figure 7, we show all the objects’ color-color
pointsbutonlyonemodelgrid,computedusingthecentralob- Strictly speaking, our study concerns the proportion of BDs
jectwithcorrespondingcolor-colorpointencircled. with inner disks only. Here, we compare our results with the
In regardto the flaringexponent,we find thatnoneofour ones of Scholz et al. (2006) who have surveyed the 1.3mm
sourcescanbefittedconvincinglybyaflatdiskmodel(β= 0) emissionofTaurusBDstostudythepropertiesoftheircolder
S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus 9
(hence farther) disks. Out of 12 objects in common between boradisk,thenthefactthattheaggregatenumberIIIharbors7
their study and ours, only two show NIR IRAC excess emis- outof8BDswithdisksisquiteimprobable(3%).Ontheother
sion without outer cold mm emission. All the BD with cold hand,inaggregateV,4outof5BDsarefoundwithoutdisks.
outer mm emission have IRAC emission in excess over the Moreoverthe remainingBD in this latter aggregateis CFHT-
photosphere model, hence were classified as ”BD with disk” Tau 12,whichpossessesthelowestIRexcessofallthesample
in ourstudy.We have comparedourdisk parameterswith the withdisks.
ones derived by Scholz et al. (2006) from their model of the In order to checkif this differencecould be an age effect,
objectsCFHT-Tau4,CFHT-Tau6,andJ044427+2512.When wehaveusedanHRdiagramtocomputetheageoftheobjects
Scholzetal.(2006)detectadisk,theymeasureadiskmassbe- present in all the aggregates. The ages are spread between 1
tween0.4and1.2MJ,i.e.,arangeofvaluesconsistentwithour and10Myr,butontheaverage,theobjectsinaggregateIIIdo
choice to fix the disk mass at 1MJ (see section 3). Similarly, notappearsignificantlyyoungerthantheonesinaggregateIV.
theyuseanindexβ = 1.15whenwefindvaluesrangingfrom ThusthehigherproportionofBDswithdisksinaggregateIII
1.0to 1.125.Theyuse an externalradiusRD = 300AUwhen cannotbeexplainedbyantimeevolutioneffectalone.
weuseR = 100AU,againaconsistentvalue.Ifwecompute
D Ontheotherhand,aggregatesIIIandIVhavesimilarstel-
thediskscaleheightatthestellarradius,therelativedifference
lardensitiesandhaveoppositeBDsrepartitionswithrespectto
with theirs is less than 25%. Finally, if we add the Scholz et
thepresenceofadisk.Thisisanothercluethatsimilarstellar
al.mmpointtoourSEDwhenitisavailable,wefindthatour
densities can result in different outcomes with respect to the
modelSED matchesthis new measurementto within a factor
proportion of BDs with disks. Either there can be large fluc-
of 2. Given the remaining uncertainties, we estimate that our
tations in the ejection rate with such low stellar densities, or
modelsareconsistentwithScholzetal.results.
the ejection process itself can have large efficiency variations
inremovingBDdisks.
4. discussion Goodwinetal.(2005)havemodeledtheevolutionofcores
containingsmallnumbersofstarsandBDs,astheyejecttheir
4.1.BDdistributionrelativetostars
lowest-massmembers.Two of their findingsappearto be rel-
evanthere:i) thespatialdistributiondifferencesbetweenstars
Insection 2.5, we foundthatthereis nosignificantdifference
andBDscandisappeardependingontheinitialclusteringcon-
between BDs with and without disks relative to the stars. If
ditions; ii) for clusters of young age such as Taurus, a signif-
BDswithoutdisksareactuallyejectedobjects,ourdatasuggest
icantdifferenceinthespatialdistributionsofstarsandBDs is
thattheyhavenothadthetimetotravelawayfromtheirparent
seen in only 1 out of 5 simulations. Then we can speculate
aggregate, or the aggregate stars have also been ejected with
that numerical simulations of low density aggregates with a
a similar velocity dispersion, as described in Bate & Bonnel
few tens of stars can produce situations where BDs may lose
(2005). It is also possible that for low density agregatessuch
theirdisksinshorttimes,butalsocanretaintheirdiskforlong
as the one found in Taurus, the comparison of the BD spa-
times.Inordertobecomparedtoourresults,suchsimulations
tial distribution relative to the surrounding stellar population
shouldbeabletofollowthefateofverysmalldisks.
isnotalwaysadiscriminantdiagnostic.Goodwinetal.(2005)
havemodeledtheevolutionoflowdensitycoresandfindthatin
somecases,thedifferencesbetweenthetwodistributions(stars
4.3.BrowndwarfsandthemoleculargasinTaurus
andBDs)candisappear.
Inordertostudyapossiblymorediscriminantdiagnosticthan
thenearestneighbourdistanceforstarsandBDs,wehavecon-
4.2.SpatialdistributionofBDwithandwithoutdisk
sideredthespatialdistributionofBDs(withandwithoutdisks)
We find that the global proportion of BDs with and without relativetotheunderlyingclustergas.InFigure8,wehaveplot-
disks in Taurus is similar to the C/W TTS global proportion tedthe13COmapadaptedfromMizunoetal.(1995)superim-
inthesameregion.Ifdisksarearobusttracerofthemainstar posed on the BD and C/W-TTS populations in galactic coor-
formationroute,thisshowsthattheBDformationprocesshasa dinates. Three main filaments appear on this figure, showing
lotincommonwiththatofthestars,andthusejectionbyitself thattheoddBDrepartitionmightinfactberelatedtotheirpo-
cannotbeinvokedtoexplainapossibledifferencebetweenBD sition relative to the filaments. Lepine & Duvert (1994) have
andstellarformation.Thefactthatthisglobalproportionisalso proposed that the Taurus filaments result from a collision be-
verysimilartotheonefoundbyLuhmanetal.(2005)inIC348 tweenahighvelocitycloudandthegalacticplane.Ifthecorre-
andChaIstarformingregionswithdifferentstellardensityand spondingwaveproceedsfromb=0,thentheleadingsouthern
physicalparametersbut similar ages, shows that the resulting filamentsaresomewhatolderthanthenortherndenser,trailing
proportionofBDwithdisksinagivenregiondoesnotdepend one where stellar formationwould be more recent, consistent
directlyonthelocalstellardensity,henceonthelocalphysical with a higher proportion of CTTs and BDs with infrared ex-
parameters,orontheoutcomeofpossibleejections. cess in this filament. Note that all the BDs found to have a
Oneofthemostintriguingresultofourstudyisthestrong mm disk by Scholtz et al. (2006)are also found in the north-
variationoftheproportionofBDwithdisksamongtheTaurus ernfilament,withaconcentrationofsuchBDsinaggregateIII.
aggregates,withaglobalproportionverycloseto50%atlarge However, the projected distance between the filaments corre-
scale.Ifonaveragethereisa50%probabilityforaBDtohar- spondstolessthan106yrforawavetravelingatafewkm/s,a
10 S.Guieuetal.:Onthecircum(sub)stellarenvironmentofbrowndwarfsinTaurus
&Armitage2006).Clearlymoreobservationalandtheoretical
work is needed to conclude about the question of a different
−12 lifetimeinBDdisksrelativetoTTSdisks.
5. Conclusions
III
−14
Using 0.6-70µm photometry, we have studied the disk prop-
erties around23 youngBDs in Taurus. Using optical to 2µm
b II data, we fit a photosphere model from Allard et al. (2000) to
IV allourobjects.FromtheirSEDs,we distinguishBDs withIR
−16
I excesses(stronglysuspectedtohaveadisk)fromBDswithno
excess(BDswithoutdisks).FortheBDsshowinganIRexcess
V longwardof3µm,wehavefitadiskmodel(Pinteetal.2006)
−18 andderivedthemaindiskparameters.
We find that 11/23 = 48%±14% of Taurus BDs show a
circumstellardisksignature.Thisratioissimilartotheoneob-
servedamongCTTS/WTTSinTaurus(Hartmannetal.2005),
−20 and to recent results on BDs from Luhman et al. (2005) who
176 174 172 170 168
derivedadiskfractionof42%±13%and50%±17%around
BDsinIC348andChamaeleonIrespectively.Withourmodel,
l
we find that disks around BDs in Taurus are all significantly
Fig.8.BDs(withknownSpitzerfluxes)andTTSsuperimposed flaring,indicatingthatheatingbythecentralobjectisefficient
on13COemission,plottedingalacticcoordinates.Contoursare andthatthedisksweobserveretainasignificantamountofgas.
2,4,6&8Kkms−1 integratedarea. UsingHαEWmeasurements,wefindthat6outof7BDwith
disks(85%)arestillsignificantlyaccreting.
We find that BDs with disks appear statistically more nu-
time differencepossibly too shortfor significantevolutionef- merousinoneoftheTaurusfilaments,specificallythenorthern
fectstotakeplace.Thisevolutionproblemisevenmorecrucial one. As this filament also contains a very high proportion of
if one considers that the unnamed aggregate between II and CTTsrelativetoWTTs,thissegregationcouldbeduetoatime
III has a BD disk frequencyof 50%. Once again,a time evo- evolutioneffectifthenorthernfilamentistheyoungestone,a
lution effect by itself cannot be invoked to explain the repar- resultpointingtowardasimilarformation&evolutionprocess
tition of BDs with and without disks in the Taurus filaments. for stars and brown dwarfs + disks. However, the age differ-
We have comparedthe levelof 13CO emission at the position encebetweenthedifferentfilamentsappearstoosmalltofully
ofthe BDs with andwithoutdisksandwe findthatBDs with explainsuch a difference.Moreover,if the variousaggregates
disksareplacedatpositionswheretheaverage13COemission arenotexactlyinthesameevolutionarystage,ourresultcould
(4.8±2.8Kkms−1)isalmosttwotimeslargerthanfortheBDs imply that the BDs belongingto an older aggregatehave lost
without disks (2.5±2.4Kkms−1), although with large uncer- theircircumstellardiskbeforethestars,implyingthatthedisk
tainties. This result mainly comes from the fact that the BDs lifetime depends on the mass of the central object, with BD
withdisksarealmostallfoundinthenorthernfilamentwhere diskshavingashorterlifetime.Thisisincontradictiontoare-
the13CO emissionis thestrongest.As 13CO emissionis opti- cent result from Alexander & Armitage (2006) who propose
cally thin, this could suggest that aggregateswhere the gas is that BD disk lifetimes could be larger than the stellar ones.
denserhaveproducedobjectswithlargerdisks. Alternatively, if all the Taurus aggregates have similar ages,
Finally,ifthethreefilamentsyetcorrespondtoatimeevo- and if ejection is the only process left to eliminate BD disks,
lutionseriesfromthesupposedlyyoungestIIItotheoldestV, thenwearestilltounderstandwhyejectionwouldresultinso
thereisapossibilitythatwearewitnessingatimeevolutiondif- differentoutcomesinaggregateswithsimilarstellardensities.
ferencebetweenBDs with disksand T Tauristars with disks. We also comparedthe underlying13CO emission forBDs
TheC/W-TTSproportionishighinfilamentsIIIandIV(con- with and withoutdisks and foundthatBDs with disks appear
sistent with them being younger) and similar to the one over to be foundwith strongermolecularemission than BDs with-
theTauruscloud(≈ 50%)infilamentV.However,theBDsin out disk, showing that the presence and/or the size of a disk
filamentIVhavealreadylostalltheirdiskswhenTTSretaina aroundaBDcouldbelinkedtotheunderlyingparentcoregas
significantamountofdisksthere.Ifthiseffectisreal,thiscould density.ThereisnosucheffectwhenwecomparetheBDpo-
be an evidenceof a centralobjectmass effectonits disk life- sitions with the underlyingstellar density, but Goodwinet al.
time.AsimilareffecthasbeenreportedbyLadaetal.(2006)in (2005) concluded that a lack of difference between stars and
IC348wherethediskfractionappearstobeafunctionofspec- browndwarfsspatialdistributiondoesnotnecessarilyexclude
traltypeandstellarmass.Althoughappealing,strongissuesre- theejectionscenario.Ourcurrentstudyofthespatialdistribu-
main: i) we could not clearly find a significant age difference tion of BDs and their disks does not allow us to distinguish
between the objects foundin the three filament; ii) this result between the two main BD formationmodels,althoughit pro-
isincontradictionwithadisklifetime∝1/M (e.g.,Alexander videsanotherpieceofevidencethatejectioncannotbetheonly
∗