MBM 12: young protoplanetary discs at high galactic latitude

0.83 MB·English

by  G. Meeus

Astronomy&Astrophysicsmanuscriptno.mbm12˙archiv (cid:13)c ESO2009
January23,2009
MBM 12: young protoplanetary discs at high galactic latitude
G.Meeus1,A.Juha´sz2,Th.Henning2,J.Bouwman2,C.Chen3,W.Lawson4,D.Apai3,I.Pascucci5,andA.
Sicilia-Aguilar2
1 AstrophysicalInstitutePotsdam,AnderSternwarte16,D-14482Potsdam,Germany
9
e-mail:[email protected]
0
2 MaxPlanckInstituteforAstronomy,Ko¨nigstuhl17,D-69117Heidelberg,Germany
0
2
3 SpaceTelescopeScienceInstitute,3700SanMartinDr.,Baltimore,MD21218,USA

n
a 4 School of Physical, Environmental and Mathematical Sciences, University of New South Wales, Australian Defence Force
J Academy,ACT2600,Canberra,Australia

2 5 DepartmentofPhysicsandAstronomy,JohnHopkinsUniversity,Baltimore,MD21218,USA
1

] Received2008;accepted
R
S ABSTRACT
.
h WepresentSpitzerinfraredobservationstoconstraindiscanddustevolutioninyoungTTauristarsinMBM12,astar-formingcloud
p athighlatitudewithanageof2Myrandadistanceof275pc.Theregioncontains12TTaurisystems,withprimaryspectraltypes
- between K3and M6; 5areweak-line andtherest classical T Tauristars. Wefirstuse MIPSand literaturephotometry tocompile
o
spectralenergydistributionsforeachofthe12membersinMBM12,andderivetheirIRexcesses.Ofthe8starsthataredetected
r
t withMIPS(spectraltypesbetweenK3andM5),only1lacksanIRexcess-theother7allhaveanIRexcessthatcanbeattributed
s toadisc.Thismeans thatinMBM12, for thedetected spectraltypesK3-M5, wehaveaveryhighdiscfractionrate,about 90%.
a
Furthermore,3ofthose7excesssourcesarecandidatetransitionaldiscs.Thefourlowest-masssystemsinthecloud,withspectral
[
typesofM5-M6,wereundetectedbySpitzer.Theirupperlimitsindicatethattheyeitherhaveatransitionaldisc,ornodiscatall.The

1 IRSspectraareanalysedwiththenewlydevelopedtwo-layertemperaturedistribution(TLTD)spectraldecompositionmethod.For
v the7TTauristarswithadetectedIRexcess,weanalysetheirsolid-statefeaturestoderivedustpropertiessuchasmass-averagedgrain
8 size,compositionandcrystallinity.Themass-averagedgrainsizewedeterminefromthe10micronfeaturehasawiderange,between
6 0.4and6µm.Thisgrainsizeismuchsmallerinthelonger-wavelengthregion:between0.1and1.5µm.Wefindthatlater-typeobjects
6 havelargergrainsizes,aswasalreadyshownbyearlierstudies.Furthermore,wefindawiderangeinmassfractionofthecrystalline
1 grains,between3and(atleast)30%,withnorelationtothespectraltypenorgrainsize.Wedofindaspatialgradientintheforsterite
. toenstatiterange,withmoreenstatitepresentinthewarmerregions.Thefactthatweseearadialdependenceofthedustproperties
1 indicatesthatradialmixingisnotveryefficientinthediscsoftheseyoungTTauristars.Thesourcesthathavetheleastamountof
0
discflaringhavethelargestgrainsizes,pointingtodustsettling.Acomparisonbetweentheobjectswithcompanionscloserthan400
9 AU(’binaries’)andthosewithwiderornocompanions(’singles’),showsthatdiscevolutionalreadystartstodifferentiatebetween
0 bothgroupsatanageof2Myr:theexcessat30µmisafactor3largerforthe’single’group.TheSEDanalysisshowsthatthediscs
v: inMBM12,ingeneral,undergorapidinnerdiscclearing,whilethebinarysourceshavefasterdiscevolution.Thedustgrainsseem
i toevolveindependentlyfromthestellarproperties,butaremildlyrelatedtodiscpropertiessuchasflaringandaccretionrates.
X
Keywords.Stars:pre-mainsequence;(Stars:)planetarysystems:protoplanetarydisks;Infrared:stars;Stars:late-type
r
a
1. Introduction massandsize ofthose discsprovideconstraintson theprocess
ofplanetformationintheseyoungenvironments.
Ourunderstandingofthestar andplanetformationprocesshas
Tostudytheyoungclustermembersinmoredetail,itisim-
advanced through comprehensive observations of nearby star-
portanttocharacterisetheircircumstellarenvironment,whichis
forming regions, such as the Taurus-Aurigacloud or the much
best done at infrared and/or millimetre wavelengths. With the
denser and massive Orion Nebula Cluster. Because these sites
launchofSpitzer,thisfieldhasmadehugestepsforward,asfor
are relativelyclose andyoung,it isoftenpossible todetermine
thefirsttime,itwaspossibletoobservethediscanddustchar-
thecompletecensusofclustermembers,fromthehighestmass
acteristicsoflargesamplesofTTauristars(e.g.Kessler-Silacci
stars down to brown dwarfs with masses of just a few Jupiters
etal.2006,Sicilia-Aguilaretal.2007,Pascuccietal.2008)and
(e.g.Luhmanetal.2003,Muenchetal.2002).Furthermore,the
evenbrowndwarfs(Apaietal.2005,Scholzetal.2007).
detectionofcompanionsthroughadaptiveopticsandotherhigh-
resolutionimagingandthesubsequentderivationofmultiplicity Cloud 12 in the catalogue of Magnani, Blitz and Mundy
ratesisanimportantpieceofthestarformation(SF)puzzle(e.g. (1985), in short MBM 12, is a molecular cloud at high galac-
Petr et al. 1998, Ratzka et al. 2005). In addition,near-IR stud- ticlatitude(l,b=159.4◦,-34.3◦),withrelativelyhighextinction
iesofSFregionshaveshownthatyoungobjectsexhibitnear-IR (A >5mag).RadiomapsoftheregioninCOgiveamassesti-
V
excesses,commonlyattributedtothepresenceofprotoplanetary mateforthewholecomplexof30to200M (Poundetal.1990).
⊙
discs (e.g.Adams et al. 1987, Lada et al. 2000). The structure, The clouddoesnotappearto be gravitationallybound,andthe
2 G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude
moleculargasisexpectedtodissipatewithinthenextfewMyrs showedthatthisobjectflares,withafactorsixincreaseinX-
(Zimmermann & Ungerechts 1990). Hearty et al. (2000) used raycountsduringtheflare.Spectralfitsofthecountratesug-
ROSAT observations to detect X-ray sources in MBM 12, and gesta coronalorigin, notuntypicalfor other flaringWTTS
followed-up stellar candidates by optical spectroscopy, to con- (Heartyetal.2000).
firm8X-rayemittingTTauristarswithanupperageof10Myr. LkHα 262 is a CTTS, with a disc detected at 2.1 mm, from
It is one of the few clouds at high galactic latitude known to whichItohetal.(2003)estimatedadiscmassof0.05M .It
⊙
harboursuchyoungobjects.Fromitscontent,thecloudcanbe islocatedatadistanceofonly15.′′3fromLkHα263.
seenasaprecursorofaTWHydrae-likeassociation,wherethe LkHα263isatriplesystem,withLkHα262possiblebelong-
molecularmaterialhasnotyetdisappeared.AlthoughMBM12 ingto this triple to thenforma quadruplesystem (Chauvin
wasinitially thoughtto beoneofthe neareststar formationre- et al. 2002); however, it is not clear whether this system
gions,atadistanceofonly65pc,itisnowdeterminedtolieata is bound. The C component of LkHα 263 has spectral
distanceof275pc(Luhman2001). type M0, and was foundto harbouran optically thick disc,
A census of this association - complete down to masses of that spatially-resolved observations showed to be edge-on
0.03 M - was determined by Luhman (2001), based on sen- (Jayawardhana et al. 2002). These authors derived a disc
⊙
sitive near-IR and optical photometry. The candidate members massof0.0018M forthis0.7M Ccomponent,basedon
⊙ ⊙
wereconfirmedbyfollow-upspectroscopy,whichalso allowed theirnear-IRimages.Furthermore,forbiddenlinesintheop-
to study their Hα and Li 6707 Å line properties.In total, 12 T ticalspectrumofLkHα263Csuggestthepresenceofajet
Tauristarswerefoundto berealmembersofMBM 12,andan (Jayawardhanaetal.2002).
age of 2+3 Myr was derivedfrom boththe lithium line and the LkHα 264 is a wide binary with a separation of 9.′′160.
−1
location of the objects in the H-R diagram. The spectral types Millimetreobservationsat1.3and2.1mmgiveadiscmass
for those 12 TTS is between K3 and M6. An additional study of0.09M⊙ aroundtheprimary(Itohetal. 2003).Emission
byLuhman&Steeghs(2004)of7candidatememberscouldnot ofmolecularhydrogenat2.1218and2.2233µmwasalsode-
confirmanymoremembers.TheTTauristarsinMBM12have tectedaroundtheprimary.Thewidthoftheselinespointsto
ahighbinaryfrequency:near-infraredadaptiveopticsstudiesby adiscorigin,whilefurthermodellinglocatestheNIRemit-
Chauvinetal. (2002) andBrandekeretal.(2003) revealedthat tingH2 in the inner10 AU, andshowsthatthe disc is seen
MBM12containsatleast4binariesand2tripleswithprojected nearlypole-on(Carmona et al. 2008). The line strength ra-
separations between 20 and 4000 AU (of which one is a can- tioisconsistentwithatemperaturelowerthan1500K, and
didate quadruple: LkHα 262 has a projected distance of only pointstothermalexcitationbyUV photons;LkHα264has
15arcsectothetripleLkHα263). a strong UV excess, so it is indeed plausible that there are
Inthispaper,wepresentSpitzerIRSandMIPSobservations enoughUV photonsto excitethe H2. Thetotal massof the
ofallMBM12members,andcombineourobservationswithlit- opticallythin,hotH2 inthediscoftheprimaryisestimated
eraturephotometryandspectroscopytoanalysethederiveddust tobeafewlunarmasses(Carmonaetal.2008).
characteristicsbyrelatingthemtothestellarparametersanddisc MBM 12-5(E0255+2018)is abinaryTTSwitha separation
properties.In Sect. 2, we presentthe individualtargetsand the of 1.′′144. Following the classification by White & Basri
Spitzer observations. The analysis in Sect. 3 first discusses the (2003), its spectral type of K3 and Hα equivalentwidth of
spectral energy distributions, and then the spectral features are 3.1Åputsitontheborderbetweentheclassicalandweak-
interpretedintermsofdustpropertieswiththeaidofaspectral lineTTauristars.
decompositionmodel.InSect.4,wediscussgraingrowth,crys- MBM 12-6 is a classical T Tauri star for which Brandekeret
tallisation,discpropertiesandaccretionrates.Weroundoffwith al.(2003)didnotfindacompanionwithinaradiusof1.′′6.
conclusionsinSect.5,andgivemoredetailsintheAppendixon MBM 12-7, 9 and 11 are all weak-line T Tauri stars, with
thedustmodelusedandthemassfractionsderived. the latest spectral types of the whole MBM 12 sample:be-
tween M5 and M6, corresponding to masses between 0.15
and0.1M .Wecouldnotfindanymultiplicitydataforthese
⊙
2. SampleandObservations relatively faint objects. High spatial resolution images are
neededtoestablishthemultiplicitystatusofthesesources.
2.1.Targets
MBM 12-8 is a classical TTS, with spectral type M5.5, and
Oursampleisunbiased,asitincludesalltheconfirmedmembers thehighestHαequivalentwidth:120Å,suggestingitisthe
oftheMBM12cloud(Luhman2001,2004).Theirparameters, mostactively accreting objectof the sample. No additional
spectraltypeandTTauriClass(weak-lineorclassical)arelisted informationconcerningitsmultiplicitystatusisavailable.
in Table 1. We re-derivedtheir T Tauri Class, followingWhite MBM 12-10 is a binary weak-line TTS with a separation of
and Basri (2003): T Tauri stars are classical when, for spectral 0.′′390.
typesbetweenK0andK5,theequivalentwidthoftheirHαline, S 18 is a triple system, consisting of the primary A and, at
|EW(Hα)| ≥ 3 Å, for K7 to M2.5, |EW(Hα)| ≥ 10 Å and for a projected distance of 0.′′747, a tight binary companion
M3 to M5.5, |EW(Hα)| ≥ 20 Å. We also list their multiplicity Bab (with a separation of only 0.′′063 - Brandeker et al.
statusandprojectedseparationofcompanions(whenpresent),as 2003). Millimetre observationsat 2.1 mm give a disc mass
determinedbyBrandekeretal.(2003)andChauvinetal.(2002) of0.07M aroundtheprimary(Itohetal.2003).
⊙
for 8 of the 12 members. The objects under consideration are
allTTauristarswithspectraltypesbetweenK3andM5.75,and
at least 6 out of the 12 targetsare known to have companions. 2.2.SpitzerIRSspectroscopy
Below,welistmoreinformationontheindividualtargets:
The T Tauri stars in MBM 12 were observed with Spitzer as
MBM 12-1 (RX J0255.4+2005) is a weak-line binary, with part of a larger programme on young stellar clusters to study
the primary of spectral type K6 and a separationof 0.′′533 the evolution of protoplanetarydiscs (GO proposal 20691, P.I.
(Chauvin et al. 2002). X-ray observations with ROSAT Bouwman). We obtained 7.5 − 35 µm low-resolution (R =
G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude 3
Table1.Targetcoordinatesandparameters:SpectralType,T ,HαequivalentwidthoftheprimaryfromLuhman(2001),TTauri
eff
Class derivedby us (see Sect. 2.1), multiplicity (S: single, B: binary, T: triple and Q: quadruple)and projected separationsfrom
Chauvinetal.(2002)andBrandekeretal.(2003).Anegativeequivalentwidthmeansemission;thewidthisgiveninangstroms,Å.
Object AlternativeName α(2000.0) δ(2000.0) Spectral T Eq.Width Class Multiplicity Projected
eff
(hms) (◦′′′) Type (K) (Hα) Status Separation(′′)
MBM12-1 RXJ0255.4+2005 025525.78 200451.7 K6 4205 -1 WTTS B 0.533
MBM12-2 LkHα262 025607.99 200324.3 M0 3850 -40 CTTS (Q)a 15.3
MBM12-3 LkHα263(A-B/B-C) 025608.42 200338.6 M3 3415 -25 CTTS T 0.416/4.1
MBM12-4 LkHα264 025637.56 200537.1 K5 4350 -18 CTTS B 9.160
MBM12-5 E02553+2018 025811.23 203003.5 K3 4660 -3 W/CTTSb B 1.144
MBM12-6 RXJ0258.3+1947 025816.09 194719.6 M5 3200 -29 CTTS S –
MBM12-7 RXJ0256.3+2005 025617.98 200609.9 M5.75 3024 -14 WTTS – –
MBM12-8 – 025749.02 203607.8 M5.5 3058 -120 CTTS – –
MBM12-9 – 025813.37 200825.0 M5.75 3024 -10 WTTS – –
MBM12-10 – 025821.10 203252.7 M3.25 3379 -12 WTTS B 0.390
MBM12-11 – 025843.80 194038.3 M5.5 3058 -14 WTTS – –
MBM12-12 S18(A-B/Ba-Bb) 030221.05 171034.2 M3 3415 -69 CTTS T 0.747/0.063
a PossiblecompaniontoLkHα263ABC
b Ontheborderbetweenweak-lineandclassicalTTS,seetargetnotesinSect.2.1
60 − 120) spectra of the MBM 12 cluster members with the Table2.MIPSphotometryat24and70µm,andtheirstatistical
Infrared Spectrograph (IRS, Houck et al. 2004) on-board the errors (1σ flux uncertainties, derived as described in the text).
Spitzer Space Telescope. A high accuracy PCRS peak-up was For those objects that were not detected, 3σ upper limits were
executedpriorto thespectroscopicobservationstopositionthe derived(seeSect.2.3).Theabsolutecalibrationuncertaintiesare
target within the slit. All targets have been observed with a 4and7%for24and70µm,respectively.
minimum of three observing cycles for redundancy.Our spec-
tra are based on the droopresproductsprocessed throughthe
Object 24µm(1σ) 70µm(1σ)
S15.3.0 version of the Spitzer data pipeline. Partially based on
(mJy) (mJy)
theSMARTsoftwarepackage(Higdonetal.2004),ourdatawas
further processed using spectral extraction tools developed for MBM12-1 2.8(0.4) <34.4
MBM12-2 142(0.4) 216a(12)
the ”Formation and Evolution of Planetary Systems” (FEPS)
MBM12-3 50.8(0.4) 216a(12)
Spitzer science legacy team (see also Bouwman et al. 2008).
MBM12-4 282(0.5) 266(13)
Thespectrawereextractedusinga6.0pixeland5.0pixelfixed-
MBM12-5 308(0.5) 253(12)
widthapertureinthespatialdimensionfortheobservationswith
MBM12-6 25.7(0.4) <34.4
thefirstorderoftheshort-(7.5−14µm)andthelong-wavelength MBM12-7 <1.2 <32.9
(14−35 µm) modules, respectively.The backgroundwas sub- MBM12-8 <1.2 <41.4
tracted using associated pairs of imaged spectra from the two MBM12-9 <1.2 <37.5
noddedpositionsalongtheslit,alsoeliminatingstraylightcon- MBM12-10 21(0.4) <35.6
tamination and anomalousdark currents. Pixels flagged by the MBM12-11 <1.2 <34.1
data pipeline as being ”bad” were replaced with a value inter- MBM12-12 45.5(0.4) <32.2
polatedfrom an 8 pixelperimetersurroundingthe errantpixel.
The low-level fringing at wavelengths > 20 µm was removed a CombinedfluxofLkHα262and263
usingtheirsfringepackage(Lahuis&Boogert2003).Tore-
moveanyeffectofpointingoffsetsperpendiculartotheslit,we
matched orders based on the point spread function of the IRS The targets with the latest spectral types, MBM 12-7, 8, 9
instrument,correctingfor possible flux losses (see Swain et al. and11,werenotdetectedwithIRS,sowewillnotincludethem
2008forfurtherdetails).Toremoveanyeffectofpointingoffsets intherestofouranalysisofthedustproperties.Thismeansthat
paralleltotheslit,wemadeaflatfieldcorrectiontothenominal thespectralrangeforwhichwehaveIRspectroscopyislimited
rsrf which was derivedfromcalibrationmeasurementsof stan- toK3–M5.
dard stars where the calibrator was mapped along the slit. The
spectraarecalibratedusingaspectralresponsefunctionderived
frommultipleIRSspectraofthecalibrationstarη Doradusand 2.3.SpitzerMIPSphotometry
1
aMARCSstellarmodelprovidedbytheSpitzer ScienceCentre.
AdditionalinfrareddatawasobtainedusingMIPS(Riekeetal.
The spectra of the calibration targethave been extracted in the
2004)onSpitzer(Werneretal.2004)inphotometrymodeat24
same way as our science targets. The relative errors between
and 70 µm (default scale). All of our targets were observed in
spectralpointswithin one order are dominatedby the noise on
February2006,using1cycleof3secintegrationsat24µmand
each individualpointandnotby the calibration.We estimate a
2-5cyclesof10secintegrationsat70µm,correspondingtoon-
relative flux calibration acrossan orderof ≈ 1 % and an abso-
sourceintergationtimesof24.1secand251.6-629secat24and
lutecalibrationerrorbetweenorders/modulesof≈3%,whichis
70µm,respectively.Ourobservationswereprocessedusingver-
mainlyduetouncertaintiesinthescalingoftheMARCSmodel.
sionS16.0.1oftheSpitzer ScienceCenter(SSC)datapipeline.
Wecreatedtwo24µmmosaicsforeachofoursourcesfromthe
4 G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude
resultingbasiccalibrateddata(BCD)imagesofeachdatacollec-
tionevent(DCE)usingtheSSC’sMOPEXsoftware(Makovitz
& Marleau 2005), one with a pixelscale approximatelythatof
the native pixelscale (2.′′45 pixel−1) and anotherresampled to
approximatelyonehalfofthenativepixelscale(1.′′23pixel−1).
We,similarly,createdtwo70µmmosaicsforeachofoursources
usingthefilteredBCDs(whichhavespatialandtemporalfilters
applied to the data in order to remove instrumentalsignatures)
withoutlierrejection,onewithapixelapproximatelythatofthe
native pixel scale (9.′′9 pixel−1) and another resampled to ap-
proximatelyonequarterofthenativepixelscale(2.′′0pixel−1).
We usedtheAPEXportionofMOPEXtoperformaperture
photometryon ourmosaickedimages. APEX appliesa median
filter to the data to estimate the sky backgroundatany pixelin
the image and subtractsthe median-filteredimage before sum-
ming the flux in an aperture. Since the estimated backgrounds
in the majority ofour fields, extrapolatedfromCOBE (Cosmic
MicrowaveBackgroundExplorer),wasmediumtohigh,weused
amedium-sizedaperturewithradiusof6′′ at24µmand16′′ at Fig.1. IRS Spitzer spectra of the seven MBM 12 sources that
showexcessemissionintheinfrared.Thestrengthandshapeof
70µmtoreducetheamountofbackgroundcontaminationinthe
theemissionfeatures,aswelltheslopeofthespectravariesfrom
aperture.Theseaperturesare notlargeenoughto containallof
sourcetosource.
the photons from a diffraction-limited point source; therefore,
weappliedscalaraperturecorrectionsof1.697and1.771at24
µm and 70 µm, as published on the SSC website. The 70 µm
aperturecorrectionissomewhatdependentonthecolourofthe distributions (SED) of the 12 targets, together with a MARCS
source;theaperturecorrectionusedhereassumesthatthemea- stellarmodel(Gustafssonetal.2008)andforthosesourceswith
sured flux has a red power law shape, Fν ∝ ν−2, because any Teff> 4200 K with a Kurucz atmosphere model (Kurucz 1994)
flux detected at 70 µm is dominated by the emission from a fortheappropriateeffectivetemperatureofthecentralsource,as
cool, dusty disc (T ∼ 100 K). The MIPS data handbook (ver- listedinTable1.ThephotometrywasdereddenedusingtheAV
sion3.2.1)statesthatproductsprocessedwithversionS14.4of valuesderivedbyLuhman(2001), assuming a standardextinc-
thedatapipelinehaveabsolutecalibrationuncertaintiesof4and tionlaw(RV =3.1).
7%at24and70µm,respectively. Of the 8 sources detected with Spitzer, only one object,
The majority of our sources were detected with signal to MBM 12-1, a K6-type weak-line TTS with a companion at
noise ratios greater than 10 at 24 µm and were not detected at 0.′′533,showsnoexcessemissionatall:thespectrumispurely
70 µm (see Table 2). We estimate 1σ fluxuncertaintiesfor ob- atmospheric.The other 7 sourcesshow an infraredexcess, and
jectsthatweredetectedand3σupperlimitsonthefluxesofob- alsohaveasilicateemissionfeatureat10µm,withvaryingde-
jectsthatwerenotdetectedfromuncertaintymosaicsproduced grees of strengths and shapes (see Fig. 1). As we do not have
byMOPEX(atthenativepixelresolution).Todeterminethe1σ anydatabetween∼3.4and8µm,itisdifficulttodeterminethe
uncertainty in a measured flux, we take the square root of the propertiesof the inner disc, e.g. whether there is an inner hole
sumoftheuncertainties(inthefluxineachpixelintheaperture) ornot,aswitnessedbyalackofexcessshortwardsofthe10mi-
in quadrature,multiplied by the aperture correction(1.164and cronfeature.SourcesMBM12-3,6and10,appearnottohavea
1.197at24or70µm,respectively),centeredattheexpectedpo- near-IRexcess(seeFig.2);theyarecandidatetransitionaldiscs,
sitionofthesource.Similarly,wedeterminethe3σupperlimit where the inner disc has alreadybeen cleared out.MBM 12-2,
onthe24and70µmfluxesasthreetimesthesquarerootofthe 4, 5 and 12 have ’fulldiscs’, as they alreadyhave an excess in
sum of uncertainties (in the flux in each pixel in the aperture) theKband.Additionaldata,between3and8micron,isneeded
in quadrature,multiplied by the aperture correction(1.164and toconfirmthestatusofthetransitionalcandidates.Anoverview
1.197, respectively), in a relatively large aperture, with radius ofthemainfeaturesobservedisgiveninTable3;theproperties
35′′,centeredattheexpectedpositionofthesource. ofthedust,asderivedfromthesolid-statefeatures,willbedis-
In Table 2, we list the MIPS photometry for the MBM 12 cussedinthefollowingsection.Wediscusstherelationbetween
members.The4starswith thelatestspectratypes,MBM12-7, excessluminosityandbinarityinSect.4.3.
8,9and11werenotdetectedat24norat70µm.InFig.2,we
showthattheMIPSphotometryat24µmandthefluxesfromthe
3.2.Dustmineralogy
IRSspectraagreeverywell.
Themostfrequentlyanalysedregioninthecontextofdustprop-
ertiesisthe10micronregion:itisthroughthiswindowthatdust
3. Analysis characteristicslikecomposition(e.g.olivines,Mg Fe SiO
2x 2(1−x) 4
and pyroxenes, Mg Fe SiO , or Mg/Fe content), structure
3.1.Spectralenergydistributions x (1−x) 3
(crystallinevs.amorphous)andgrainsizescanbederivedfrom
TocomplementourSpitzerobservations,wecollectedphotom- their solid-state features. The 10 µm regionis only sensitive to
etryfromtheliterature:Jayawardhanaetal.(2001)andLuhman dust grainsup to sizes of a few microns, as the feature flattens
(2001) for RI, JHK and LN band photometry. The V band is outwhenthe dustgrainsattain sizessimilar to the wavelength,
s
fromBroegetal.(2006),while thesubmillimetreandmillime- making them essentially invisible. Furthermore, it is important
tre photometry are from Hogerheijde et al. (2003) and Itoh et to realise that the dust causing this feature traces only a small
al. (2003), respectively.In Fig. 2, we show the spectral energy fraction of the disc material, namely those dust grains that are
G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude 5
Fig.2.Thespectralenergydistributionsofthe12TTauristarsinMBM12.Thelowest-massstars,MBM12-7,8,9and11were
notdetectedwithSpitzer,thereforewecanonlyshowupperlimitsintheinfraredforthosesources.MBM12-1istheonlysource
inourdetectedsamplethatshowsnoexcessaboveitsatmosphere.MBM12-3,6and10appeartonothaveanear-IRexcess,soare
candidatetransitionaldiscs,whileMBM12-2,4,5and12have’normalaccretiondiscs’,withanexcessextendingintothenear-IR.
located in the opticallythin disc atmosphere,while the bulk of In order to analyse the composition of the dust in the disc
thedustmassislocatedinthediscmidplane.Inaddition,asthe atmosphere,theradiationofwhichdominatestheIRSspectrum,
dusttemperatureneedstobehighenough(150-450K)inorder we use the two-layer temperature distribution (TLTD) spectral
toradiateat10µm,theradiallocationofthedustobservedisalso decomposition routines described in Juha´sz et al. (2009). This
limited:foratypicalstarinoursample,thisisaround1AU.At methodusesamulti-componentcontinuum(star,innerrim,disc
longer wavelengths, features of crystalline dust (forsterite and midplane), assuming that the region where the observed radia-
enstatite) witness another important part of dust processing in tion originates(bothopticallythin andthick) hasa distribution
theprotoplanetarydisc. oftemperatures.Inthisfittingmethod,theobservedflux-density
atagivenfrequencyisgivenby
In Fig. 3, we zoom in on the three most interesting ranges
thatareobservedwithIRS,andalsoindicatethelocationofthe N M Ta,out 2π 2−qa
pmhoosutsimsipliocrattaentatfe9a.t7ur(epsr.eAsepnatrtinfraolml 7thsetabrrsowadithfeaatudreeteocfteadmeoxr-- Fν = Fν,cont + Xi=1 Xj=1Di,jκi,jZTa,in d2Bν(T)T qa dT (1)
cess) and 18 µm, features of enstatite (MgSiO ) and forsterite whereN and M arethenumberofdustspeciesandofgrain
3
(Mg SiO ) are observed. We do not see evidence for carbona- sizesused,respectively.κ isthemassabsorptioncoefficientof
2 4 i,j
ceousdust,suchasfeaturesfrompolycyclicaromatichydrocar- thedustspeciesiandgrainsize j. B (T)isthePlanck-function,
ν
bons(PAHs).Thisisperhapsnotsurprising,giventhatPAHsare qa is the powerexponentof the temperaturedistributionand d
transientlyexcitedbyUVphotonsandthecentralsourceshave isthedistancetothesource.Thesubscript’a’intheintegration
lowtemperatures,althoughsomesourcesdoshowaratherlarge boundariesreferstothedisc atmosphere.Thecontinuumemis-
UVexcess(e.g.MBM12-2,4and12). sion(F )isgivenby
ν,cont
6 G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude
Fig.3.Zoominonthe3mainrangesintheIRSspectra:7-14µm,17-25µmand26-36µm.Weoverplotthepositionsofthemain
crystallinefeatures:long-dashedlinesindicateenstatite,whiledot-dashedlinesindicateforsterite.
nent(discatmosphere,innerrim,midplane),thehighesttemper-
ature is fitted to obtain T , while the lowest temperature,
a/r/m,in
Fν,cont = D0πdR22⋆Bν(T⋆) + D1Z Tr,out 2dπ2Bν(T)T2−qrqrdT (2) Tpear/ra/tmu,roeutc,oinstcraiblcuutelastmedorreeqtuhiarnin0g.1th%attothtehaentnoutalluflsuwxi.ththattem-
Tr,in For our fit, we used five dust species that are commonly
+ D2ZTmT,min,out 2dπ2Bν(T)T2−qmqmdT. (3) 6foµumnd).inWteheadpipslcisedofthyeouthnegosrtyaros,finditshtrriebeugtiroaninosfizheosl(l0o.w1,s1p.5hearneds
for the crystalline dust (Min et al. 2005) and the classical Mie
Thefirsttermonthe righthandsidedescribestheemission theory for sphericalparticles for the amorphousdust, to derive
of the star, while the second and third term describe the radia- themassabsorptioncoefficientsfromtheopticalconstants.The
tionoftheinnerrim(subscript’r’)andthediscmidplane(sub- listofthedustspecies,theoriginoftheopticalconstantsandthe
script’m’),respectively.Themeaningofthedifferentparameters grain model used are presented in Table 4. Furthermore, since
are summarised in Table 7 in the Appendix. For each compo- the dust grains radiating at 8-13 µm and 20-30 µm are at dif-
G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude 7
Fig.4.TheTLTDfitsofthe10micronsilicatefeature(thick,solidline),andtheIRSspectrawiththeirnoise(blackpointswitherror
bars).Noticethedifferentshapesthatarepresentinthissmallsample:MBM12-5hasatriangularshape,pointingtosmall,amor-
phoussilicate grains,whileMBM12-6hasa muchbroaderfeature,indicatinglargerandcrystallinegrains.Thenarrowemission
featureat9.3µmiscausedbycrystallineenstatite.
Table3.Summaryofthespectralappearanceofthedetectedtar- Table 4. Overview of dust species used in our fitting routines.
gets.Welisttheshapeofthespectralenergydistribution,where For each component, we specify its lattice structure, chemical
we distinguish between a ’full’ and a ’transitional’ disc, based composition, shape and reference to the laboratory measure-
on the presence or absence of excess emission at wavelengths ments of the optical constants. For the homogeneous spheres,
shorter than 8 micron, and write ’transitional object (TO) can- weusedMietheorytocalculatetheopacities.Fortheinhomoge-
didate’ when this classification needs to be confirmedby addi- neousspheres,weusedthedistributionofhollowspheres(Min
tionalphotometry.Furthermore,welistthewavelengthsatwhich etal.2005),tosimulategrainshapesdeviatingfromperfectsym-
features of enstatite and forsterite are present. Colons indicate metry.
uncertaindetections.
Species State Chemical Shape
Object SED Enstatite Forsterite formula
MBM12-1 Photosphere – – Amorphoussilicate1 A MgFeSiO Homogeneous
4
MBM12-2 Full – 19,27.8,33µm (Olivinestoichiometry) Sphere
MBM12-3 TOcandidate 9.3: 11.3:,19:,27.8,33µm Amorphoussilicate1 A MgFeSi O Homogeneous
2 6
MBM12-4 Full – 11.3:,19:,27.8,33µm (Pyroxenestoichiometry) Sphere
MBM12-5 Full – 19,27.8,33µm Forsterite2 C Mg SiO HollowSphere
2 4
MBM12-6 TOcandidate 9.3µm 27.8,33µm ClinoEnstatite3 C MgSiO HollowSphere
3
MBM12-10 TOcandidate 9.3µm 23.4,27.8µm Silica4 A SiO HollowSphere
2
MBM12-12 Full 9.3µm 11.3,19,27.8,33µm
References:(1)Dorschneretal. 1995;(2)Servoinetal.1973;
(3)Ja¨geretal.1998;(4)Henningetal. 1997
ferenttemperatures,henceatdifferentradialdistancesfromthe
star,wesplittedthewavelengthregionintwopartsbeforefitting.
This way, we can take into account that different species can
contributein differentamountsto thespectrum,ifthereisara- As the IRS spectra do contain noise, we repeated the fit of
dialgradientintheirabundancesand/orproperties.Wechosethe eachobject100times,everytimeaddingrandomGaussiannoise
1) the 7-17µm range for the shorter wavelengthregion,which on top of the measured spectrum. The final composition is de-
Juha´szetal.(2009)showedtobetheoptimalwavelengthrange rivedastheaverageofthose100fits,whiletheerrorsarederived
to balancethe presenceof spectralfeaturesandradialchanges, fromthestandarddeviation,takingintoaccountthepositiveand
and 2) the 17-37 µm range for the longer wavelength region, negative directions of the variations. Therefore, the errors we
whichincludesimportantcrystallinefeatures;bothregionswere givedescribetheS/Nandmeasurementerrorsinthespectra.We
separatedlyfitted,andwewillrefertothemasthe’shorter’and emphasizethatwecannotgiveerrorestimatesthatarerelatedto
’longer’wavelengthregions. species thatare notincludedin ourfits, norof differentshapes
8 G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude
Table 5. Distilled fitting results: given are mass-averaged size
of the amorphous and the crystalline silicates, < a > and
am.sil.
<a> ,respectively,massfractionofthecrystallinegrains,
cryst.sil.
f (ratioof crystallinesilicate tototalsilicate mass), andthe
cryst.
massratioofforsteritetoenstatite,asderivedforboththe7-17
and17-37µmregions.
Object <a> <a> f forst/enst
am.sil. cryst.sil. cryst.
(µm) (µm) (%)
7-17µmregion
MBM12-2 4.1(+0.1) 5.2(+0.1) 10.7(+0.7) 0.1(+0.0)
−0.1 −0.1 −0.8 −0.0
MBM12-3 4.0(+0.4) 1.7(+1.7) 5.8(+2.1) 0.4(+0.1)
−0.6 −1.1 −1.6 −0.2
MBM12-4 0.4(+0.1) 4.1(+0.5) 16.3(+2.4) 0.1(+0.0)
−0.1 −0.6 −2.6 −0.0
MBM12-5 2.2(+0.1) 4.1(+1.0) 2.9(+1.2) 0.3(+0.3)
−0.0 −1.6 −1.1 −0.1
MBM12-6 3.9(+2.0) 5.3(+0.2) 46.5(+24.9) 0.0(+0.0)
−3.1 −0.2 −16.6 −0.0
MBM12-10 5.3(+0.1) 1.9(+2.1) 4.6(+4.6) 0.8(+2.2)
−0.1 −0.9 −1.7 −0.5
MBM12-12 6.0(+0.0) 5.5(+0.1) 32.8(+2.9) 0.1(+0.1) Fig.5. Relation between the shape of the 10 µm feature (flux
−0.1 −0.1 −2.8 −0.0
ratio11.3over9.8µm)andthefeaturestrength(peakovercon-
17-37µmregion
tinuum). Our objects (squared symbols, numbered as listed in
MBM12-2 0.1(−+00..01) 0.8(−+00..00) 1.8(−+00..11) 1.1(−+00..11) Table1),followthetrendobservedforTTSinTr37(4Myr)and
MBM12-3 1.2(+0.1) 0.4(+0.3) 4.8(+0.4) 1.2(+0.1)
−0.1 −0.2 −0.4 −0.1 NGC7160(12Myr),shownbydiamonds(Sicilia-Aguilaretal.
MBM12-4 0.2(+0.2) 1.6(+0.8) 2.8(+0.8) 1.0(+0.3)
−0.1 −0.6 −0.5 −0.2 2007)andTTSinTaurus(2Myr),shownbytriangles(Pascucci
MBM12-5 0.7(+0.2) 4.0(+0.5) 3.5(+1.0) 0.9(+0.7)
−0.1 −0.8 −0.9 −0.3 etal.2008):thefluxratioiscorrelatedwiththefeaturestrength.
MBM12-6 1.5(+0.4) 0.1(+0.0) 7.1(+1.5) –
−0.2 −0.0 −1.3
MBM12-10 0.4(+0.9) – 2.9(+3.1) –
−0.3 −1.5
MBM12-12 0.6(+0.3) 0.8(+0.2) 3.3(+1.1) 0.9(+0.2)
−0.2 −0.2 −0.7 −0.1
minosity sources than for higher luminosity sources: under the
assumptionofasimilardisc structureforallspectraltypesdis-
cussed here, an M0-type star reaches 300 K between 0.7 and
orcompositionofthegrains.Thoseerrorsaresubjectofanother
1.5 AU, while itis between1.2 and3 AU for a K5-typestar, a
study(A.Juha´szetal.2009,inpreparation).
factor 2 difference in distance. Furthermore, the density distri-
Withthefits,wedeterminethesourceofthesolid-statefea-
bution has a radialdependence,as it decreases with increasing
turesthatareobservedinthespectra.Sinceseveralspecies(e.g.
radius, and grain growth occurs faster in more dense environ-
carbon) are featureless within our wavelength range (but can
ments.Thisimpliesthatgraingrowthwillbefasterinthemore
contributetothecontinuum),theyremainundetectedinourfits;
inwardregions,sothatlowerluminositysourceswillappearto
soitisimportanttorealisethatwecanonlydiscussthe’visible
have larger grains, as their 10 micron feature originates from
dust’,thathasfeaturesintherange7to37micron.InFig4,we
adenserregion,inwhichgrowthnaturallyoccursmorerapidly.
showthebestfitofthe7to17micronregion,andasummaryof
Thisisalsoseeninourspectra:inFig.6,weplotthecontinuum-
thederiveddustpropertiesforboththeshorterandlongerwave-
normalised spectra, and also list their spectral types. With the
lengthregionsisgiveninTable5.Thedetailedresultsofthefits
exceptionofMBM12-4,thefainterobjectsindeedhaveweaker
inbothregions,intermsofmassfractionsoftherelevantspecies
features. This is further quantified through our fitting with the
andtheirsizes,aregiveninTable8intheAppendix.
TLTDmethod:inFig.7,weshowtherelationbetweenthemass-
averaged grains size in the 10 micron region and the effective
4. Discussion:2Myr-oldTTauridiscs temperatureofthecentralstar. Thereisa cleartrendforcooler
starstohavelargergrainsizes.
4.1.Graingrowth
Ofcourse,notalldiscsneedtohavethesamestructure,and
A first thing to notice, when looking at the 10 micron silicate thedegreeofflaringcanalsoplayanimportantroleinthiscon-
feature,isthatthestrengthoftheemissionvaries.VanBoekelet text. However,to study these effects, detailed radiativetransfer
al. (2003) showedthatthe shapeandthe strengthofthe 10µm modellingofeachsourceisneeded,whichisbeyondthescope
feature in Herbig Ae/Be stars are related, and demonstrated it ofthispaper.
to be evidenceforgrain growth:a strongand triangular10 µm
feature is typical of small (submicron-sized)grains, whereas a
4.2.Crystallisation
weakerandbroaderfeatureindicatesthepresenceoflargersized
grains(up to a few microns).In Fig. 5, we probedthis relation Thedustthatisinitiallyincorporatedintotheprotoplanetarydisc
foroursampleofTTauristars,andalsofoundevidenceforgrain is largely amorphous, as it comes from the ISM, for which an
growthintheprotoplanetarydiscsoftheMBM12members. upperlimitof∼2%in massofthe crystallinegrainswasdeter-
Kessler-Silaccietal.(2006,2007)andApaietal.(2005)no- mined using spherical grains (Kemper et al. 2004). Min et al.
ticed that later type TTS have weaker 10 micron features than (2007) further improved this number using grains with irregu-
earliertypeTTS,andrelatedthistothelocationofthedustcaus- larshapesandderivedacrystallinityofonly∼1%.However,the
ingthis feature.Thisrelationwasconfirmedby Sicilia-Aguilar amorphousdustmaybecomecrystallinethroughthermalanneal-
etal.(2007),whofoundthatthepresenceofaveryweakfeature ing(e.g.Fabianetal.2000)orshockheating(e.g.Scott&Krot
was3timesmorefrequentforM-typestarsthanforearlier-type 2005). We derived the crystalline mass fraction in the shorter
stars. The temperature of the silicates causing the feature is ∼ wavelength region, and found a large degree of variation: be-
300K,atemperaturethatisreachedmoreinwardsforlowerlu- tween 2.9+1.2 % for MBM 12-5 and 47+25 % for MBM 12-6.
−1.1 −17
G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude 9
Fig.8.MBM12-6,hostofalargeamountofcrystallinesilicates,
witnessedbytheagreementinpeakpositionsbetweenthespec-
trumandthefeaturesofthemassabsorptioncoefficientsofen-
statite(Ja¨geretal.1998;dottedline)andforsterite(Servoinetal.
1973;dashedline).Forcomparison,wealsoshowtheHerbigBe
starHD100546(Malfaitetal.1998),andthesolar-systemcomet
Hale-Bopp (Crovisier et al. 1997), both host of large amounts
ofcrystallinedust.ThefluxesofHD100546andHale-Boppare
scaledtomatchthefluxofMBM12-6.
correlatedwithamorphousgrainsize(seeFig.9).However,we
do note that the sources with the largest crystalline mass frac-
tions(MBM12-6andMBM12-12)alsohavethelargestgrains
Fig.6.Continuumnormalisedfluxinthe10µmregion.Theob- size,bothfortheamorphousasforthecrystallinegrains(seealso
jectswiththelatestspectraltypestendtohavetheweakestfea- Fig.10).Inthelongerwavelengthregion,wefindthevariationin
ture. crystallinitytobeless,andthecrystallinemassfractionsmaller:
between 1.8+0.1 % for MBM 12-2 and 7.1+1.5 % for MBM 12-
−0.1 −1.3
6. This difference in crystallinity between both regions for the
wholesamplecouldberelatedtothepresenceofalargeamount
ofsmallamorphoussilicategrainsinthecoolerdiscregion(av-
erage size for the sample is 0.7 µm), while in the warmer disc
region the amorphous silicate grains are much larger (average
size for the sample is 3.7 µm). Larger grains will produce less
detectablefeatures, so that the crystalline grainsthat still show
featureswillappearmoreabundant.
InFig.8,weshowthe15to37µmspectrumofMBM12-6,
the object with the largest fraction of crystalline grains. Its IR
spectral appearance is similar to that of HD 100546, a Herbig
B9e star with highly evolved dust, which is remarkable, given
thattheireffectivetemperaturesareverydifferent:11000K for
theB9star,andonly3200KfortheM5typestar.Still,thedust
aroundbothobjectsisdominatedbycrystallinegrains,indicat-
ingthatthetemperatureofthecentralstardoesnotplayanim-
portantroleinthecrystallisationprocess.ThespectrumofMBM
12-6isalsosimilartothatofsolar-systemcometHale-Bopp,so
Fig.7. Mass-averaged size of the amorphous silicate grains in thatsimilar dustprocessingmechanismsmustexistinbothour
the warmer region, versus effective temperature of the central solarsystemandthediscofMBM12-6.
star. The cooler the star (the later the spectral type), the larger InthedustmodelbyGail(1998),thecompositionofthedust
thedustgrainstendtobe. isderived,basedoncondensationsequencesandchemicalequi-
librium considerations. He predicts that - assuming crystalline
silicatesformashightemperaturegasphasecondensates-ofthe
The crystalline fraction of MBM 12-6 is not very well deter- crystalline silicates, enstatite will be the dominant constituent,
mined,thereforeweusethelowerlimitof30%thatwasfound whileforsteriteisonlypresentinasmallregionclosetothestar.
for this object. The mass fraction of the crystalline silicates is Morerecentsimulationsshowthatradialmixingwithinthedisc
not related to the spectral type. Furthermore, we do not find a iscapableoftransportingasubstantialamountofcrystallinema-
correlationbetweencrystallinityandthe size oftheamorphous terialfromtheinnerdisctowardsmorecolder,outwardregions
grains; also the size of the crystalline grains appears to be un- (Keller&Gail2004).
10 G.Meeusetal.:MBM12:youngprotoplanetarydiscsathighgalacticlatitude
Fig.9.Relationbetweenthepropertiesofthecrystallinesilicates Fig.10. Properties of the the crystalline silicates, derived from
andthesizeoftheamorphoussilicatesinthe7to17micronre- the 7 to 17 micron region. Upper panel: the forsterite over en-
gion.Upperpanel:mass-averagedsizeofthecrystallinesilicates statite ratio in function of the size of the crystalline silicates.
versusmass-averagedsize of the amorphoussilicates. The two Thereisatrendforlargergrainstohaveasmallerforst/enstra-
appearunrelated.Lowerpanel:crystallinemassfractionversus tio.Lowerpanel:crystallinityinfunctionofcrystallinegrainsize
mass-averagedsizeoftheamorphoussilicates.Alsoherewedo .Thecrystallinefractiononlyexceeds20%whenthecrystalline
notseeacorrelation. grainsarelargerthan5micron.
Tobetterunderstandthecrystallisationprocessoccurringin unlikelyatlargerradialdistances,whereboththedensityandthe
the T Tauri discs, we look at the forsterite to enstatite ratio. In temperaturearelower.
Table5,welisttheforsteritetoenstatitemassratioforbothre-
gions.Inthewarmerregion,thesmallestratioisfoundforthose
sources with the largest grains (see Fig. 10), implying that en- 4.3.Discproperties
statiteismoreabundant(relativetoforsterite)whenlargergrains
4.3.1. Discfraction
are present. We further compare the relative mass fractions of
forsteriteandenstatiteinbothregions:theaverageratioforthe We obtained Spitzer data for the complete sample of MBM 12
sampleis0.3inthewarmerregion,and0.9inthecoolerregion. members; 4 of them, however, could not be detected. The re-
Itisthusclearthatthereisaspatialgradientintheforsterite maining 8 objects (spectral type range K3 to M5) have a disc
to enstatite ratio, with forsterite dominatingthe coolerregions, fraction rate of 7 out of 8, or nearly 90%. This is very high,
andenstatitemoreabundantinthewarmer,innerregions.Such when compared to a similar spectral type range in other star
a gradient was already found in a few earlier studies, e.g. forming regions: Damjanov et al. (2007) derive a disc fraction
Bouwmanetal. (2008).Thisis incontrastwith thepredictions of 47% for M0-M4 and 55% for K3-K8 type stars in the 2
byGail(1998,2004), suggestingthatthechemicalequilibrium Myr-oldChamaeleonI and63%region;however,Flaherty and
conditions needed for the forsterite to enstatite conversion are Muzzerole(2008)deriveadiscfractionbetween65and81%for
notreachedinsidethesediscs,andthatthecrystallisationprocess the2MyroldclustersNGC2068andNGC2071.Thehighdisc
mustbedifferent,evenwhenincludingradialmixing.Itisinter- fractionratefoundforMBM12may(partly)beattributedtothe
estingtoconsiderthatenstatiteismoreabundantinthewarmer absenceofclosecompanionstothediscbearingstars:thereare
(inner)regions,andthatwealsofindalargerenstatiteabundance nocompanionsfoundataprojecteddistancesmallerthan0.′′39
inthosesourcesthathavelargergrains(seeFig.10).Theseob- (or70AUatthedistanceofMBM12).
servationssuggestthatenstatiteformseasierinregionsofhigher Ofthe7TTauristarswithadisc,3arecandidatetransitional
density,wherealsograingrowthismoreabundant.Bouwmanet objects,whichisahighrateforaregionwithsuchayoungage
al.(2008)discusstheformationofenstatiteandforsteriteinde- (2Myr),pointingtofastinnerdiscdispersal.However,duetothe
tail, andlink theformationofenstatite in the innerregionwith lackofdatabetween3.5and8µm,wecannotfurtherdiscussthe
the conditions that prevail there: due to the higher density and differencesininnerdiscstructure;additionalobservationscould
temperature,ittakesthedustgrainslongertocooldownsothat, shed more light on this topic. For the four non-detected mem-
potentially,equilibriumconditionscanbereached-whatismore bers, we derived upper limits, to put some constraints on their