ebook img

The warm gas atmosphere of the HD 100546 disk seen by Herschel (Evidence of a gas-rich, carbon-poor atmosphere?) PDF

3 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview The warm gas atmosphere of the HD 100546 disk seen by Herschel (Evidence of a gas-rich, carbon-poor atmosphere?)

Astronomy&Astrophysicsmanuscriptno.hd100546 c ESO2012 (cid:13) January25,2012 The warm gas atmosphere of the HD 100546 disk seen by ⋆ Herschel Evidence of a gas-rich, carbon-poor atmosphere? SimonBruderer1,2,EwineF.vanDishoeck3,1,StevenD.Doty4,andGregoryJ.Herczeg1 1 MaxPlanckInstitutfu¨rExtraterrestrischePhysik,Giessenbachstrasse1,85748Garching,Germany 2 InstituteofAstronomy,ETHZurich,8093Zurich,Switzerland 2 3 LeidenObservatory,LeidenUniversity,POBox9513,2300RALeiden,TheNetherlands 1 4 DepartmentofPhysicsandAstronomy,DenisonUniversity,Granville,OH43023,USA 0 2 Submittedon7October2011/AcceptedtoA&Aon5January2012 n ABSTRACT a J Context.With the Herschel Space Observatory, lines of simple molecules (C+, O, and high-J lines of CO, Jup & 14) have been 3 observedintheatmosphereofprotoplanetarydisks.Whencombinedwithground-baseddataon[CI],allprincipleformsofcarbon 2 canbestudied.Thesedataallowustotestmodelpredictionsforthemaincarbon-bearingspeciesandverifythepresenceofawarm surfacelayer.Theabsenceofneutralcarbon[CI],whichispredictedbymodelstobestrong,canthenbeinterpretedtogetherwith ] ionizedcarbon[CII]andcarbonmonoxide. R Aims.Westudy thegastemperature, excitation, and chemical abundance of thesimplecarbon-bearing species C,C+,and CO,as S well asO by themethod of chemical-physical modeling. Using themodels, weexplore the sensitivityof the linestothe entering . parametersandconstraintheregionfromwhichthelineradiationemerges. h Methods.Numericalmodelsoftheradiativetransferinthelinesanddustareusedtogetherwithachemicalnetworksimulationand p acalculationofthegasenergeticstoobtainthegastemperature.Wepresentournewmodel,whichisbasedonourpreviousmodels - o butincludesseveralimprovementsthatwereportindetail,togetherwiththeresultsofbenchmarktests. r Results.AmodelofthediskaroundtheHerbigBestarHD100546isabletoreproducetheCOladdertogetherwiththeatomicfine- t structurelinesof[OI]andeither[CI]or[CII].Wefindthatthehigh-J linesofCOcanonlybereproducedbyawarmatmosphere s a withTgas ≫ Tdust.Thelow-JlinesofCO,observablefromtheground,aredominatedbytheouterdiskwitharadiusofseveral100 [ AU,whilethehigh-JCOobservablewithHerschel-PACSaredominatedfromregionswithinsometensofAU.Thespectralprofiles ofhigh-J linesofCOarepredictedtobebroaderthanthoseofthelow-J lines.Westudytheeffectofseveralparametersincluding 1 thesizeofthedisk,thegasmassofthedisk,thePAHabundanceanddistribution,andtheamountofcarboninthegasphase. v Conclusions.Themain conclusions of our work are(i) only awarmatmosphere withT T can reproduce theCO ladder. 0 (ii)TheCOladdertogetherwith[OI]andtheupperlimitto[CI]canbereproducedbymgoadse≫lswidtuhstahighgas/dustratioandalow 6 abundanceofvolatilecarbon.Thesemodelshoweverproducetoosmallamountsof[CII].Modelswithalowgas/dustratioandmore 8 volatilecarbonalsoreproduceCOand[OI],areincloseragreementwithobservationsof[CII],butoverproduce[CI].Owingtothe 4 uncertainoriginofthe[CII]emission,wepreferthehighgas/dustratiomodels,indicatingalowabundanceofvolatilecarbon. . 1 Keywords.Protoplanetarydisks–Stars:formation–Astrochemistry–Methods:numerical 0 2 1 : 1. Introduction violet(UV)radiationandX-rays.Throughheatingofthedisk’s v upperatmosphere,adisk-windisdriven,whichmayeventually i At an important phase of a young low-mass star’s life, the na- X disperse the disk (Alexanderetal. 2006; Ercolanoetal. 2008; talenvelopehasdispersed,butthestarisstillsurroundedbyan Gorti&Hollenbach 2009). The most abundant species tracing r optically thick protoplanetary disk, which is thought to be the a theionizingradiationandwarmtemperatureofafew100Kare location of planet formation (see Dullemond&Monnier 2010 (ionized)atomiccarbon(C,C+),atomicoxygen(O),andhigh-J and Armitage 2010 for recent reviews). Our own solar system linesofCO (with J & 14).Apartfromneutralcarbon,which up containsbothgaseousgiantsandrockyplanets.Itisthusevident isaccessibletoground-basedtelescopes,thesespeciescannow thatboththedustandthegascontentofthediskneedtobestud- be observed using the PACS instrument (Poglitschetal. 2010) iedinordertounderstandplanetformation.Thefeedbackofthe onboardtheHerschelSpaceObservatory(Pilbrattetal.2010). forming star onto the disk can help decide the disk’s fate. The Themainformsofcarbon(C,C+,andCO)canthusbeob- starirradiatesthedisk,heatingboththegasanddustandchang- served for the first time and through high-J CO lines also in ing the chemical composition of the gas by high-energy ultra- warm regions. This allows us to probe the carbon chemistry and the physicalstructureof planet- and comet-formingzones, Sendoffprintrequeststo:SimonBruderer, e-mail:[email protected] which were previously inaccessible to us. The carbon budget ⋆ Herschel is an ESA space observatory with science instruments is important for planet formation, since the terrestrial planets providedbyEuropean-ledPrincipalInvestigatorconsortiaandwithim- in the inner Solar System are known to have a deficit in car- portantparticipationfromNASA. bonofthreeordersofmagnituderelativetothesolarabundance 1 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk (Alle`greetal. 2001). Indications of a carbon deficit are also lyzed scattered-light images in the optical using ACS/Hubble foundtowardsotherplanetarysystems(Jura2008).Inthediffuse finding evidence of minimal grain sizes larger than those of ISM,about40 60%ofthecarbonislockedintoamorphouscar- typicalISMgrains.Thescatteredlightobservationsalso reveal − bon grains (Savage&Sembach 1996; Sofiaetal. 2004, 2011). structuresresemblingspiral armspossibly caused by either the TobeconsistentwiththeEarth’scomposition,thesegrainsneed perturbationsof a companion(Quillenetal. 2005) or a warped tohavebeendestroyedbeforeplanetformation.Incomets,how- disk structure (Quillen 2006). Features in the near-infrared ever,carbongrainshavebeendetected,suggestingthatgrainde- (NIR) reveal a high crystalline silicate dust fraction in the in- structionmechanismsplaynorole(Leeetal.2010). nerdisksurfacelayers(Bouwmanetal.2003)andanabundant What region of the disk does the far-infrared (FIR) C, C+, amountofPAHswhoseemissionisdetectedonextendedscales CO and O lines trace? How does the partitioning between (Geersetal. 2007). A wealth of continuum images in the near volatile and refractory carbon vary from clouds to disks? The andmid-IR(e.g.Pantinetal.2000;Gradyetal.2001;Liuetal. simultaneous measurement of the species tracing a significant 2003; Gradyetal. 2005) have helped to constrain the structure fractionofthe disk mayprovideanswersto these questions.In of the inner disk using SED fitting. They reveal a large inner addition, besides the hard to observe H , these simple species hole of radius 10 AU (Malfaitetal. 1998; Bouwmanetal. 2 ∼ formthe mostabundantconstituentsof the gas. Their lines are 2003), which possibly has a Jupiter-sized planet orbiting in- thus importanttracers of the physical structure of the disk and side (Acke&vandenAncker 2006). Interferometric observa- allow us to finally test the predictions made by many exist- tions(VLT-AMBERandMIDI,Leinertetal.2004;Petrovetal. ingphysical-chemicalmodelsofdisks.Neutralcarbon[CI],for 2007) confirm the presence of the inner hole directly and set example, is predicted to be strong by models with a range of r 13 AU (Benistyetal. 2010). Some CO ro-vibrational hole ∼ gas masses and dust settling (Jonkheidetal. 2007) but is not emission at 4.7 µm was reported by vanderPlasetal. (2009) detected in observationsof HD 100546 (Panic´etal. 2010) and and Brittainetal. (2009). The latter paper inferredthat the CO CQTau(Chapillonetal.2010).Inadditiontothesubmillimeter emissionoriginatesfromradiiof13-100AU,whichisconsistent lines discussed in our work, neutral carbon also has forbidden withtheinnerholeseeninthedustcontinuum.Theyfoundahot lines in the near-infrared (at 8729, 9826, 9852 Å), which are rotational temperature of 1000 K and indications that both ∼ predicted to have detectable strengths (Ercolanoetal. 2009b). UVfluorescenceandcollisionalexcitationareimportantforthe Todate,onlyafewdetectionstowardspre-main-sequencestars excitation.Hotgasinthedisksurfacelayersisalsodetectedby havebeenreported(Looperetal.2010). observingtheH v=1 0S(1)lineat2.12µmbyCarmonaetal. 2 − The physics and chemistry of gas in protoplanetary disks, (2011). which is irradiated by the central star, is similar to the in- Thebulkofthediskmassisatcoldtemperatures,astraced terface of general molecular clouds, where nearby massive bysubmillimeterlinesandcontinuum.Submillimetercontinuum stars heat and photoionize the gas. Modeling the infrared wasreportedbyHenningetal.(1994,1998),whoderivedadust (IR) response of gas irradiated by UV and X-rays has been massofa fewtimes10−4 M . Thegasupto 100Kis traced developed in the context of these photo-dominated or disso- byCOrotationallinesuptoJ⊙=7 6,asrepor∼tedbyPanic´etal. ciation regions (PDRs) and X-ray dominated regions (XDRs) (2010). They derived a gas mass −of > 10−3 M and concluded (e.g. Tielens&Hollenbach 1985; Sternberg&Dalgarno 1989; fromthehigh12COJ =6 5/12COJ =3 2ratio⊙thatthediskat- − − Kaufmanetal.1999;Meijerink&Spaans2005).Thesemodels mosphereneedstobewarmerthaninTTauristars,probablyow- simulate the infrared line emission by solving the coupled ingtophotoelectricheatingofthegasbythestrongUVfieldof system of chemical evolution and thermal balance. The theformingB9star.Stronghigher-JlinesuptoCO J =30 29 − coupling is the result of the cooling rates for the thermal were detected by Sturmetal. 2010 using the PACS instrument balance depending on the abundance of the coolants, which onboardHerschel.Thedetectedlineshaveupperlevelenergies is determined by the chemical network that itself also de- ofupto2500K. pends on the temperature. These physical-chemical models Inthiswork,we studythe chemistryandexcitationofsim- were applied in the context of protostars by Spaansetal. ple species (C, C+, CO, and O) in the disk arounda pre-main- (1995) and Brudereretal. (2009a, 2010) to the walls along sequenceHerbig Ae/Be star. Our approachis twofold.We first outflows, irradiated by UV radiation arising from either want to interpret the PACS observationsof Sturmetal. (2010) accretion hot-spots or the protostar’s photosphere. For pro- and the ground-basedobservations of Panic´etal. (2010) taken toplanetary disks, physical-chemical models have been towardsHD100546andsearchforevidenceofawarmdiskat- applied by e.g. Kamp&vanZadelhoff (2001), Glassgoldetal. mosphere from the CO ladder, which probes a wide range of (2004), Kamp&Dullemond (2004), Gorti&Hollenbach temperatures.Wehoweveralsowishtostudythedependenceof (2004), Nomura&Millar (2005), Aikawa&Nomura (2006), the lines on certain parameters such as the carbon abundance Jonkheidetal. (2004), Jonkheidetal. (2006), Jonkheidetal. in the gas phase, the gas/dust ratio, and others. In Section 2, (2007), Gorti&Hollenbach (2008), Woitkeetal. (2009a), we briefly summarize the modeling approach and describe the Woods&Willacy(2009),andErcolanoetal.(2009a). observations. In Section 3, we first present the results for one Thewell-studiedintermediate-masspre-main-sequencestar particular model that agrees reasonably well with the observa- HD 100546 is the focus of our present study. The B9.5Vne tions in Section 3 and then discuss the parameter dependences type star of 2.2 M , L 27 L , and T = 10500 K inSection4and5.InSection6,wesummarizeourresults. bol eff (vandenAnckereta⊙l. 1997)∼is at its⊙distance of 103 6 pc ± (Hipparcos) one of the closest Herbig Ae/Be sources. The 2. Modelandobservations star and disk have been extensively observed at all wave- lengths in lines and the continuum. X-ray observations are re- For this work, we use a new model based on Brudereretal. ported in Stelzeretal. (2006). Both FUSE and IUE observed (2009b,a,2010),withseveralchangesandimprovementsimple- HD100546 in the UV, detecting absorption in electronic tran- mented. The section provides a short summary of the model. sitions of H probingthe warm gas with a temperature of sev- DetailsandresultsofbenchmarktestsarereportedinAppendix 2 eral 100 K (Martin-Za¨ıdietal. 2008). Ardilaetal. (2007) ana- A. 2 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk Ourphysical-chemicalmodelstartswith(i)adustradiative Muldersetal.(2011).Thesurfacedensitypower-lawistakento transfercalculationforagivendustandgasdistributiontoobtain be Σ r 1, whichfits theVISIRimages. Anotherdustdensity − ∝ thelocalUVradiationfield andthedusttemperature.Usingan structurefortheHD100546diskwasproposedbyBenistyetal. initialguessofthegastemperature,(ii)thechemicalabundances (2010), basedon a analyticaldiskstructure andfitted to the IR arecalculated,whichserveasinputforamolecular/atomicexci- interferometricdata.Theirmainconclusionsabouttheinnerdisk tationcalculationtoobtainthecoolingrates.Next,(iii)thether- were incorporatedintothe Muldersetal. (2011) modelthatwe malbalanceiscalculatedtoobtainanimprovedguessforthegas use. It is well-known that SED fitting yields degenerate solu- temperature.Steps(ii)and(iii)arerepeateduntilconvergenceis tionsandtheobtainedduststructuredoesnotnecessarilyreflect achieved.Raytracingandconvolutiontothetelescopebeamthen thegasstructure,e.g.owingtodustsettlingeffects.Inaddition, yieldslinefluxesthatcanbeobservedwithobservations.These thestructureofthediskislikelynon-axisymmetric(Panic´etal. modelingstepsaresummarizedinFigureA.1. 2010; Quanzetal. 2011) and with substructures (Ardilaetal. DifferencesfromthestudyofBrudereretal.(2009b,a,2010) 2007).However,thisapproximateapproachisareasonableone are that we solve the chemical network for the steady-state forstudyingthemaincharacteristicsofthegasemission,given solution instead of interpolating from a pre-calculated grid of thelargeuncertaintiesinthegaslinemodeling(Section5.5). abundances (Brudereretal. 2009b). This new approach allows We adopted a Hipparcos distance of 103 pc to HD 100546 us to include detailed photodissociation and ionization rates (vandenAnckeretal.1998).Theinclinationandpositionangle based on the local wavelength-dependentUV field and the ac- weretakentobe42 and145 ,respectively(Pantinetal.2000). ◦ ◦ tual column densities for the self-shielding factors. The ther- WeusedaKeplerianvelocityfieldarounda2.5 M star,which malbalanceisrefinedbyobtainingatomicandmolecularcool- was found by Panic´etal. (2010) to closely reprod⊙uce the ob- ing rates from the excitation calculation used in Brudereretal. served spectra. The outer radius is uncertain but estimated to (2010). This (approximate)method, which is similar to that of bebetween200AUand400AU(Pantinetal.2000;Panic´etal. Poelman&Spaans(2005),allowsustocalculatethemolecular 2010). excitation quickly and with reasonable accuracy. We can now Inside the large inner hole of the HD 100546 disk, a includethedetailedvelocitystructureintothe excitationcalcu- smaller innerdiskextendingfrom0.25AU to 4AU was found lationandaccountforheatingof thegasbythe IRpumpingof (Benistyetal. 2010). We run models including the inner disk, molecules, which is then followed by collisional deexcitation. butfoundthatmorethan99%ofthefluxofthelinesdiscussed Themethodcouplescellsatdifferentradiiofthedisk,unlikethe in this work emerges from the outer disk. We thus considered “verticaldirectiononly”escapeprobabilityapproachemployed only the outer disk starting at 13 AU for this work. The dust inmostpreviousdiskmodeling.Thedisadvantageofourmethod opacityoftheinnerdiskwashoweverincludedtocalculatethe is that we need to iterate over the entire chemical structure to localUVfield. derive the local gas temperature, which increases the calcula- Asinputstellarradiationfield(Figure1),weanalyzeddered- tion time. However, the method allows us to calculate models withanarbitrarygeometryincludinge.g.disksembeddedinen- dened FUSE observationsfor wavelengths 911-1200Å, which velopes. Since our code contains all geometrical properties in had been previouslyanalyzed by Deleuiletal. (2004) and IUE onemodule,itcanbeextendedtothree-dimensionalstructuresin observations(Valentietal.2003)forlongerwavelengths(1200- thefuture.Thechemicalnetworkinthenewcodehasalsobeen 3200 Å). The spectra were dereddened using the extinction of extendedbyincorporatinghydrogenationreactionsongrainsur- AV = 0.36(Ardilaetal.2007)andtheextinctionlawofDraine faces,photodesorptionofspeciesfrozen-outontograins,andvi- (2003) for R = 3.1. The FUSE/IUE spectra were extendedto V brationallyexcitedmolecularhydrogen(H ),whichcanbeused longer wavelengths using the B9V template of Pickles (1998). ∗2 toovercomeactivationbarriers. TheX-rayluminosityhadbeenfoundtobelog(LX) = 28.9erg Inthisstudy,weconcentrateontherotationalladderofCO s−1 (Stelzeretal. 2006). Since the plasma temperature had not and do not consider vibration-rotationlines detected in the M- beenwell-constrainedbyobservations,weassumedaratherhigh bandat4.7µm,sincefluorescentexcitationbyUVphotonsplays temperatureofkTX = 6keV.ThishardspectraallowsX-raysto asignificantroleintheirpopulation.Wenotethattheexcitation penetratedeepintotheatmosphere.Nevertheless,thelinesstud- of the rotationalladderis determinedby collisions and the rel- ied in this work change by less than 5 % by switching on ∼ ative population of CO in the ν = 1 state is small at 4 % for X-rays.Thus,theassumptionontheshape(plasmatemperature) athermalizedpopulationat1000K(Brittainetal.2009).Thus, of the X-ray spectra does not affect our results (Section 3.1). the results presented in this work would not changeby includ- Inadditiontothestellarradiationfield,wealsoaccountforthe ingvibrationallyexcitedlevelsofCO.Forthesamereasons,we interstellarradiationfield(Draine1978).Anothersourceofion- also do notmodelthe H2 2.12 µm line. In addition,this line is izationarecosmicrays,forwhicharateofζc.r. = 5×10−17 s−1 verysensitive to the depthatwhich the dustbecomesoptically isadopted. thick and a quantitative comparison to the far-infrared lines is Thedustopacitiescontrolthepenetrationofthephotonsand thusdifficult. thus both the photodissociation and ionization of molecules as well as the dust temperature. The dust grains used in the SED fittingbyMuldersetal.(2011)aresilicateswitha5%(inmass) 2.1.ParametersfortheHD100546model compositionof carbonaceousmaterial. For gas/dust= 100 and To model HD 100546,we chose to use the dust density distri- apresent-daysolarphotospherecarbonabundanceof2.7 10 4 − × bution obtained by Muldersetal. (2011). They fit the SED us- relativetohydrogen(Asplundetal.2009),thisamountstoabout ingadustradiativetransfercalculationwithaverticalstructure 15 % carbon bound in this population of grains. The grains that obeys hydrostatic equilibrium assuming that T = T . have a size of 0.1 - 1.5 µm with a distribution proportional to gas dust Dropping this assumption results in a density structure with a a 3.5, employing the Mathis, Rumpl and Nordsieck size distri- − puffed-up inner rim with about 10 AU (Kampetal. 2010), but bution (Mathisetal. 1977). The opticalconstants are for irreg- this regiondoesnotcontributesignificantlyto the emission we ularly shaped distributions of hollow spheres (DHS, Minetal. probehere.Forconsistency,wealsousethedusttemperatureof 2005). The DHS dust modelhas been adoptedto be consistent 3 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk of recombinationreactions. We adopta 5 % PAH-to-dust ratio 104 FUV(Photoelectricheating) ) CHp.d. within 100 AU and 1% outside 100 AU, following a simulta- −1 CH2p.d. neous fitting of Spitzer and VISIR N-band observations using ˚A H2p.d.,COp.d.,Cp.i. −2 103 the NC = 100 opacities of Draine&Li (2007) (Gijs Mulders, m pers. comm.). The PAH-to-dustmass ratio of 5 % corresponds c −1 102 approximately to the average Galactic PAH-to-dust mass ratio s of 4.6 %, which in turn represents about 50 ppm of C per H g er Blackbody8000K nucleon bound in PAHs (Draine&Li 2007) and 15 % of the 13 101 Blackbody10500K present-day solar photosphere carbon abundance (for gas/dust − 10 Blackbody12000K = 100). The PAHs are only excited by UV in the uppermost x( 100 FUSE layer of the disk (Visseretal. 2007) and the observed features Flu IUE thustracethisregion.Inadditiontothelargeuncertaintiesinthe B9Vtemplate molecular properties of the PAHs, the PAH abundance is only 10−1 poorlydeterminedintheregionfromwhichthelinesdiscussed 1000 2000 5000 in this study emerge. We thus choose to scale the PAH abun- Wavelength(˚A) dance with the gas/dustratio, such that gaswith gas/dust= 20 contains five times smaller amount of PAHs than for gas/dust Fig.1.UV spectraofHD 100546atthe sourcedistance of103 = 100 (Jonkheidetal. 2007). We examine the influence of the pc.We indicatethefarultraviolet(FUV)wavelengthrangeim- PAH abundance separately in Section 4.3. We assume PAHs portantforgasheatingandthewavelengthrangeofphotodisso- to be photodestroyed in regions with integrated far ultraviolet ciation(p.d.)andphotoionization(p.i.)cross-sectionsfordiffer- (FUV)fluxes> 106 ISRF,followingtheresultsofamorecom- entmolecules. plete PAH chemistry by Visseretal. (2007). This affects how- ever only the upperlayer of the disk and we find that the lines discussed in this work are unaffected by this assumption. We with the SED fitting of Muldersetal. (2011). This dust popu- lation has a mass of 10 4 M , which agrees with observations notethattheemissionofalllinesstudiedhereemergesfromz/r − ofsmallgrainsbyDominike⊙tal.(2003).However,muchofthe lowerthantheregionofPAHdestruction(Section3.1,3.4). masscanbeinlargergrains,andthegas/dustratiosgivenrela- tivetothispopulationofgrainsarethusupperlimits.Thispop- ulationofgrainsdoesnotcontainthe UV/visualattenuationby 2.2.Observations verysmallgrains(VSGs)andpolycyclicaromatichydrocarbons (PAHs).WethusalsouseR =3.1and5.5opacitiesafterDraine V (2003)at wavelengths. 1 µm (Figure2) to includeVSG/PAH Spectrally resolvedobservationsof the low-J 12CO J = 7 6, absorption.Using these opacities at short wavelengthsonly af- − J = 6 5, and J = 3 2 lines were obtained with the 12m fects(i.e.increases)thedusttemperatureintheuppermostlayer, − − APEX telescopebyPanic´etal. (2010) anddisplayeda double- abovetheregionfromwhichthelineemissiondiscussedinthis peakedlineshape,thatisconsistentwitharotatingdisk.Wenote work emerges, as we checked. For simplicity, we thus use the that the uncertainty in the J = 7 6 line is considerable and samedusttemperatureforallmodelstakenfromMuldersetal. − the J = 6 5 line is detectedat a much highersignal-to-noise (2011). − ratio. Theyhavealso searchedforthe 370 µm (809GHz) [CI] 3P 3P line, but were unable to detect it. We note that they 2 1 − 105 havebeenusingpositionswitchingwithanoffposition16′from ) the source. Thus, it is unlikely that the emission at the source 1 − g hadbeencanceledoutbyextendedemission.The609µm(492 2m 104 GHz)3P1 3P0lineofneutralcarbonwasnotobserved. c − ( t fficien 103 andT1h4e5[µCmII3]P158µ3mP 2lPin3e/2s−an2dPm1/2id,-thJ/eh[iOghI-]J6l3inµemso3fPC1O−w3Pit2h, 0 1 e − co 0.1−1.5µm(κabs) Jup = 14 to 30 were observed by Sturmetal. (2010) with the ction 102 0R.V1−=13..51µ(mκab(sσ)scat) PanAdCgSasinisntrtuimmee”nt(DonIGboITar)dkeHyerpsrcohgerlamas(PpaI:rtNo.fEtvhaens“)D.Tushte, iCcOe, Massextin 101 RRRVVV===355...155(((σκσssaccbaastt))) l4nJi1on4t=ee−at3hn31ad0t3−Ct.hOT3e0haeJplsipne=aetrwew2no3atsjl−uibnm2lee2spndaiwnreeadCstOwhpuioltsihtnetnetahtkflieaeulnOlxyaHasbtl3Jue/pn>2pd,ee27rd5/l2wimm+iat−ihyts.5Hh/aW22vO−ee been introduced by uncertainties in the PACS flux calibration 0.1 0.2 0.5 1 2 5 10 20 50 100 when the first PACS results were published,includingthose of Wavelength(µm) Sturmetal. (2010). The larger uncertainty in these lines does Fig.2. Adopted dust opacities. The solid lines give absorption however not affect any of our conclusions. The PACS data are mass extinction coefficients, while dashed lines indicated scat- spectrally unresolved.Except for [CII], all lines are consistent teringmassextinctioncoefficients. withtheiremergencefromwithinthecentral9′.′4pixel.Tocor- rect for extended emission in [CII], we subtracted the pixels neighboring the central pixel. We also accounted for emission ThePAHabundanceaffectsboththeheatingbymeansofthe leakingfromthecenterpixelowingtotheextendedpointspread photoelectriceffect,aswellastheionizationbalance,bymeans function.Theoriginofthe[CII]lineisdiscussedinSection5.3. 4 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk Table1.LinefluxesmeasuredtowardsHD100546. the disk, X-rays can however be neglected, since the luminos- ity in theFUV band(6-13.6eV) isat 4 1034 ergs 1 much − Species Line Wavelengtha Flux/Errorb higherthantheX-rayluminosity.Neve∼rthel×ess,theconsiderable [µm] 10−17[Wm−2] ionization rate may affect the chemistry and for example lead APEX(Panic´etal.2010) to a destruction of water (e.g. Sta¨uberetal. 2006). The simple CO J=3 2 866.96 0.0149 0.003 speciesconsideredinthisworkarehoweverlittleaffectedbyX- CCOO JJ==67−−56 433731..5665 0.103.211±±0.00.0072 raysandwefindthattheCO,C,C+,andOfluxeschangebyless [CI] 3P2−−3P1 370.42 <±0.009 thanT∼he5(%intweghreanteXd)-rFaUysVarfleusxwiintcuhneditsoffof. the interstellar radia- HerschelPACS(Sturmetal.2010) CO J=14 13 186.00 7.4 0.9 tion field (Draine 1978; 1 ISRF 2.7 10−3 erg s−1 cm−2) is CO J=15−14 173.63 11.5±0.8 showninFigure3d.VeryhighFU∼Vflux×esofupto108ISRFare CO J=16−15 162.81 8.3±0.9 reachedinthesurfaceoftheinneredgeofthedisk.Intheupper CO J=17−16 153.27 10.5±0.8 atmosphereof the outerdisk at r = 400 AU and z/r = 0.3 the CO J=18−17 144.78 9.9±1.0 FUVfluxstillreaches105 ISRFatadensityofafewtimes105 CO J=19−18 137.20 8.9±0.7 cm 3.WenotethattheseareroughlytheconditionsoftheOrion CO J=20−19 130.37 7.1±0.7 − bar PDR (Hogerheijdeetal. 1995; Jansenetal. 1995). Similar CO J=21−20 124.19 5.9±0.6 CO J=22−21 118.58 5.9±0.9 conditions are also predicted for the outflow walls, the “dense CO J=23−22 113.46 <4.8±1.2 PDR” along the outflow of a high-mass star forming region CO J=24−23 108.76 10.6±1.3 (Brudereretal.2009a). CO J=26−25 100.46 12.7±2.4 Figure3egivestheFUVextinctioninunitsof AV,obtained CO J=27−26 96.77 10.2±1.4 by defining τ = log(I /I ), with the attenuated FUV CO J=28−27 93.35 10.0±1.0 flux I and thFUeVFUV−flux caottrreucntaettd only for geometrical dilu- CO J=29−28 90.16 11.8±1.6 tionIatt (e.g.Brudereretal.2009a,Eq.2).ConversiontoA is CO J=30−29 87.19 8.5±1.0 unatt V CO J=31−30 84.41 <4.9±1.5 donebyscalingwiththedustopacitiesat2070Å and555Å and [[OCIII]] 32PP3/2−3P−2P1/2 15673..7148 13.55±54±1.5c5 dacucroinugnttihnegcfaolrcuAlVati∼on,1b.0e8c6aτuVse. TthheepUhVotoeixotninizcatitoionnisanndotdiussseod- [OI] 3P1−3P2 145.53 35.7 ±1.3 ciationratesarecalculatedfromthewavelength-dependentUV 0− 1 ± spectra.TheFUVextinctionhowevershowsthatowingtoscat- Notes.(a)Restfrequency(b)Formalerrors,notincludingsystematicer- tering, FUV photons can penetrate much further into the disk, rors(c)Extendedemissionsubtracted,seetext. particularlyintheouterparts(e.g.vanZadelhoffetal.2003). Figure 3f shows the adopted PAH abundance. The drop in the inner, upper disk is due to the photodissociation of PAHs. 3. Results:Arepresentativemodel Thedropat100AUisobservationallyconstrained(Section2.1). The dustand gastemperatureare presented in Figure 3g/h. We now present the details of a representative model of HD 100546. By representative, we mean that it reproduces suffi- Figure3ishowstheratioofthegastodusttemperature.Thedust temperaturedoesnotexceed 250K, butstays above 30K cientlywelltheobservationsenablingustostudythemainphys- ∼ ∼ icalandchemicaleffects.Itishowevernotmeanttobeabestfit acrossthe entiremodel.Thus,CO doesnotfreeze outontothe dust grains. The gas temperature exceeds the dust temperature model.Thedependenceontheparametersassumedhereisdis- byfactorsofup to 50in the innerupperdisk andreachestem- cussedinthenextsection.Theouterradiusoftherepresentative peratures of up to 6000 K. In the outer upper disk, the gas model is assumed to be 400 AU. We define the depletion fac- ∼ temperatureisstillT 1000Kandthusaboutafactoroften tor δ as the carbon abundancein the gas phase relative to the gas C ∼ present-daysolarphotosphereabundanceof2.7 10 4 relative higherthanTdust.Deeperwithinthedisk,belowz/r =0.15,col- − × lisions between gas and dust grains result in a coupling of the tohydrogen(Asplundetal.2009).Inthissection,weassumea carbongasphaseabundanceof1.3 10 4 relativetohydrogen gasanddusttemperature(Tgas =Tdust). − × (δ 0.5),whichischaracteristicofthediffuseISM.Adopting C gas/d∼ust= 20,wegetagasmassof 2 10 3 M .Weusethe R =5.5dustextinctionlawforthis∼mod×el. − ⊙ 3.2.Chemicalstructure V Fractional abundancesof C+, C, CO, O, H , and electrons rel- 2 ative to the total hydrogen density (n = n(H) + 2n(H )) are 3.1.Physicalstructure H 2 shown in Figure 4. As found by previous chemical models of Thephysicalconditionsoftherepresentativemodelareshownin protoplanetarydisksthechemicalstructureresemblesaclassical Figure3.Onlyonequadrantofthediskisgivenandonlyregions PDRintheverticaldirection(e.g.Jonkheidetal.2007).Below withgasdensity>105cm−3areshown.Wepresentthegasden- a toplayerofatomichydrogen,H2 forms,asself-shieldingbe- sityinFigure3a/bbothinlinear(r,z)aswellasin(log(r),z/r)- comes sufficient. In the outer disk, the inward column is also space.Thelatterhastheadvantagethattheinnerpartandupper large enough to prevent H2 from photodissociation and hydro- diskcanbeseenclearly.Directraysfromthestartothediskcor- genisinmolecularform. respondtoverticallines.Thedensityreachesuptoafewtimes Carbonintheupperdisk(z/r 0.2)ismainlyintheformof 1010cm 3inthismodelintheinnerpartofthediskmidplane. C+, followedbyneutralcarbonan∼dCO deeperwithinthedisk. − TheX-rayionizationrateisgiveninFigure3c. Despitethe Anexceptionisthe“warmfinger”ofCOintheouterdisk(r > low X-ray luminosity of this source of 8 1028 erg s 1, the 100 AU, z/r > 0.25).In this region,CO forms throughthe re- − ionizationrateduetoX-raysintheinner∼disk×isstillaboutthree actionofC+ with H toCH+,followedbyareactionwithO to 2 ordersofmagnitudehigherthanthestandardcosmicrayioniza- form CO+ which then reacts with H to form HCO+ and sub- 2 tionrateof5 10 17s 1.Fortheheatingandcoolingbudgetof sequentlyCO(seeJonkheidetal.2007).Theinitiatingreaction − − × 5 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk Fig.3.Physicalconditionsintherepresentativemodel.Thecontourlinesareshownforvaluesindicatedinthelabelbysmallarrows. Thethindashedlineindicatesz/r = 0.15,0.2,and0.25.Thepanelsshow:a.)Gasdensityb.)Gasdensityinlinearscale.c.)Total ionizationrate.d.)LocalFUVfluxe.)FUVextinctionf.)PAHabundanceinunitsoftheISMabundanceg.)Dusttemperatureh.) Gastemperaturei.)Ratioofthegastodusttemperature. isstronglyendothermicwithanenergybarriercorrespondingto 13AU<r<20AUandz/r<0.1,gasanddustareatatempera- 4640 K, thus requires a high temperature to proceed. The OH ture>100KandthegasisnotexposedtostrongFUVradiation. formation at high temperature, by means of the reaction of O Inthispartofthedisk,oxygenislockedintowater.Furtherout, andH ,leadstoadditionalCO+ forthischainofreactions.The waterfreezesoutandlocksmostoftheoxygenintowaterice.In 2 OHreactswithC+ toformCO+.Attheinneredgeofthediskat thisregion,thecarbonisintheformofgaseousCH . 4 13AUbelowz/r 0.1,COisshieldedfromphotodissociation ∼ owingtoshadowingbytheinnerdisk.Thehigherabundanceof COatr =13AUandz/r =0.2isduetotheadoptedH forma- 2 tionrate:since the stickingcoefficientdecreasesabove 3000 ∼ TheelectronfractionroughlyfollowstheabundanceofC+. K(Cuppenetal.2010),thelowergastemperatureathigherz/r ChargeexchangewithPAHsisanotherimportantreactiontype leadstoanincreaseinH andsubsequentlyinCO. 2 for the ionization balance. The PAHs are positively charged in Atomicoxygenhasaconsiderablyhighabundancethrough- theupperatmosphere,neutralatintermediateheights,andnega- outthedisk,exceptforthemidplane.Thisisbecausetheelemen- tivelychargedatgreaterdepthswithinthediskduetocharging talabundanceofoxygenishigherthanthatofcarbonandnotall byelectronattachment.Sincetheelectronabundanceisverylow oxygen can be locked into CO, which is almost unaffected by inthemidplane,thePAHscanagainbecomeneutral.Chargeex- photodissociation,bymeansofself-shielding.Inaddition,oxy- changeC++PAH0 C+PAH+andrecombinationC++PAH − gencannotbeionizeddirectlybyFUVphotonssinceitsioniza- C+PAH0 areth→etworeactionsthataffectthesimplecarbon → tionenergyisslightlyhigherthan13.6eV.Intheinnerdiskwith chemistrymost,leadingahigherabundanceofneutralcarbon. 6 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk Fig.4.FractionalabundancesofC+,C,CO,O,H ,andelectronsrelativetothetotalhydrogendensity(n = n(H)+2n(H )).The 2 H 2 thindashedlineindicatez/r=0.15,0.2,and0.25.AbruptchangesatR=100AUareduetoavaryingPAHabundance,seetext. 3.3.Warmatmosphereversusdusttemperature 2011) and OH (Sturmetal. 2010) towards this disk, which are bothformedinendothermicreactionswithH . 2 Linefluxespredictedbytherepresentativemodelarepresented in Figure 5. The figure givesthe CO ladder on the left and the 3.4.Originofthelineemission [CI], [CII], and [OI] atomic fine-structure lines on the right. Atomicfine-structurelinesofCandCOlinesuptoJ =7 6are Where does the observed line emission originate? The angular convolvedtotheAPEXbeamandallotherlinestotheHe−rschel resolutionofHerschel,correspondingtoabout1000AU for63 beam.Inadditiontothemodelwiththecalculatedgastempera- µmatthedistanceofHD100546,doesnotresolvethedisk.The ture,amodelthatassumesthatthegastemperatureisequaltothe model however, may help us to locate the region of emission. dusttemperatureisshown.Afactoroftwoaroundtheobserved Figure 6 gives the relative contribution to the observed fluxes. fluxesisshownbyagrayshadedregion.Weconsiderthisregion We show the integrated contribution function (ICF), which in- as a goodagreementto observationsgivenall the uncertainties corporates all the effects involved in the line formation: abun- indataandmodeling(seeSection5.5). dance of the molecule, excitation, and opacity. In addition, we providethetworadiidelineatingtheregionfromwhich75%of The modelwith the calculatedgastemperaturefits the data theemissionemerges,eitherinsideoroutside. well, with the exception of the high-J CO lines with J > 25 The ICF, discussed in Tafallaetal. (2006) and (see Sections 4.5 and 5.5). However,the cool atmosphere with Pavlyuchenkovetal. (2008), shows the relative contribu- T set to T fails to reproduce the high-J CO ladder. The tiontothelineemissionalongaray.Weobtainfromtheformal gas dust J =16 15transitionwithE 750K(Table2)isalreadyun- solutionoftheradiativetransferequation(Eq.A.14) up − ∼ derproducedby aboutan order of magnitude,and the higher-J lines are underpredictedevenmore.For example, J = 25 24 ICF= (Sν Sdust)e−τν 1 e−∆τν dν, (1) (Eup 1800 K) is underpredictedby more than two orde−rs of Zline − (cid:16) − (cid:17) ∼ magnitude.Theatomicfine-structurelineshaveupperlevelen- wherethesourcefunctionisS ,thedustsourcefunctionisS , ν dust ergiesthataremuchlowerthanthehigh-JlinesofCO.Boththe theopacitytotheobserverisτ andtheopacitywithinonecellis ν [CI]and[CII]lineshave E < 100K andthe line fluxescal- ∆τ .Theintegrationisperformedoverfrequency.Forsimplicity, up ν culatedwithacoolandwarmatmosphereareverysimilar.The we take the disk to be face on and allow the rays to propagate [CII] line is somewhatunderproducedand [CI] overproduced. parallel to the z axis (as e.g. in Panic´etal. 2010). This avoids However,[OI]hasanupperlevelenergyof327K(145µm)and complications with the adopted Keplerian velocity profile but 227K(63µm).Consequently,the[OI]linesforthecoolatmo- conservesthemainphysicalconclusions.Thecontributiontothe spheredifferbyan orderof magnitudefromthoseobtainedfor observedlinefluxfromanannuluslocatedataradialdistancer thewarmatmosphere.Weconcludethattheprominentemission isgivenby2πI(r)rdr,withthefluxI(r)atradiusr.Thus,therel- inhigh-J COand[OI]atomicfine-structurelinesisstrongevi- ative contributionto the observedflux is I(r)dr. In Figure6, ∝ denceofahighertemperatureofthegasintheupperatmosphere wethusshowthecontributionfunctiontimestheradius(r ICF), ofthedisk.ThisagreestotherecentdetectionofCH+(Thietal. scaledtoitspeakvalue. × 7 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk 10−13 Warmatmosphere(Tgas>Tdust) 10−14 Coldatmosphere(Tgas=Tdust) ObservedFluxes −2) 10−15 m W ( 10−16 x u Linefl 10−17 Ffriogm.5t.heInrteepgrreasteedntalitniveeflmuoxdeesl.oTbhtaeinleefdt part of the figure shows the CO ladder 10−18 and the right part atomic fine-structure lines. Observations are indicated by blackcrosses. Theredlineandsquares 10−19 showmodelfluxesforamodelwithcal- 0 5 10 C1O5-Jup 20 25 30 µ370m µ810m µ158m µ63m µ145m caunldatterdianggalsestewmitpherTagtausres,etthtoeTbdluuset. Tlihnee [CI] [CI] [CII] [OI] [OI] gtwraoytoshtahdeeodbrseegrvioendiflnudxiceast.esafactorof The low-J emission of CO J = 3 2 originates at radii1 Table2.Linepropertiesandorigins − between r 70 220 AU at z/r 0.15 0.25. The density ∼ − ∼ − above this region is too low to contribute significantly to the Line λ E A n a Origin up ul crit emission,sincetheregionsbelowareshieldedbythelineopac- [µm] [K] [s 1] [cm 3] [AU] − − ity. Thus, the far side of the disk does not contribute much to CO J=3 2 867 33 3(-6) 2(4) 70-220 the emission, except for some radiation in line wings. Owing CO J=16− 15 163 752 4(-4) 1(6) 35-80 to the increasing density and temperaturetowardsthe inner re- CO J=30−29 87 2565 2(-3) 4(6) 20-50 gionatagivenz/r,evenspatiallysmallerregionsatradii< 100 [[CCIII]]32PP1−3−P22P 135780 9612 23((--67)) 51((33)) 115500--330000 AUcontributetotheobservableemission.Mid-J lines,e.g.CO [OI] 3P3/23−P 1/2 63 228 9(-5) 5(5) 50-180 aJn=do1n6ly−m1o5lweciuthleEswupit∼hi7n5r0K3,5are8n0oAtUexccoitnetdriibnuttehetootuhteeremdiissk- [OI] 3P10−−3P21 145 327 2(-5) 5(4) 80-210 ∼ − sion.Theselinesreachlineopacitiesofaroundunityandthefar Notes. a(b) means a 10b. (a) Critical density (A / C ) for colli- sideofthediskdoesnotcontributetotheemissionintheinner- sion with H at the te×mperature of E . Atomic/muloleclulaurl data from 2 up P mostregions(radii< 30AU).Thehighest-J emission,e.g.CO Scho¨ieretal.(2005). J = 30 29with E 2800K, detectedtowardsHD100546 up − ∼ emerges from the inner r 20 50 AU in a thin layer at the ∼ − surfaceofthedisk.There,COisformedbutthegastemperature isstill sufficientlyhighto excitethelines.Theopacityofthese linesisopticallythinandthedustextinctionisnotlarge.Thus, illustration purposes, we however also show CO J = 30 29, emissionfromboththenearandfarsidesreachestheobserver. which might eventually be accessible to resolved observa−tions The atomic fine-structure lines of [CI] and [CII] emerge withtheGREATinstrumentonboardSOFIA. mainlyfromthe outerdisk atradii 150 300AU. Owingto ∼ − thehigherabundanceofthosespeciesintheupperatmosphere, combinedwiththelowcriticaldensity( 103cm 3)andlowup- perlevelenergyofthe[CI]and[CII]l∼ines,thes−elinesemerge J =3−2 fromregionshigherinthediskthanthelow-J linesofCO.The J =10−9 oxygen[OI]lineshaveconsiderablyhighercriticaldensitiesof y 1 J =16−15 t∼hu1s05n−ot1e0x6ccitmed−3inantdhealosuotehrigdhiesrkuapnpderthleevleilneeneemrgiisessi.oTnhceoymaeres ensit J =30−29 t n fromregionswithin 50 210AU.Sincethislineisoptically I ∼ − d thick,mostoftheemissionisfromthenearsideofthedisk. ze 0.5 ali m r 3.5.COlineprofile o N ThepredictedshapesoftheCOJ =3 2,J =10 9,J =16 15 0 − − − andJ =30 29linesareshowninFigure7.ForCO J =3 2, − − theprofileisgivenfortheAPEXbeam,whilethehigher-Jlines arecalculatedfortheHerschelbeam.TheHIFIheterodynespec- −15 −10 −5 0 5 10 15 trometer onboard Herschel (deGraauwetal. 2010) can spec- Velocity(kms−1) trally resolve lines down to < 0.1 km s 1 and covers the fre- − quency range of CO J = 5 4 up to CO J = 16 15. For Fig.7.PredictedlineshapesfortheCOJ =10 9,J =16 15, − − andJ =30 29inthebeamofHerschel.TheC−OJ =3 2−line − − 1 Measuredbythe75%contributionradiigiveninFigure6andTable predictedforAPEX isshowntogetherwith theobservationsof 2. Panic´etal.(2010). 8 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk Fig.6.Contributionfunctiontothelineformation.Thediskisviewedfromthetopandthecontributionfunctionisnormalizedto thepeak.Total(line+dust)opacitiesofτ =1atthelinecenterareindicatedbyablackline,anddustopacitiesofτ =0.1 linecenter dust bygraylines.Theredarrowsindicatetheradiifromwhich75%oftheemissionemerges,eitherinsideoroutside.Thethindashed lineindicates z/r =0.15,0.2,and0.25. | | The modeled CO J = 3 2 line agrees well in terms of the main emission found in Section 3.4. We note that the CO − widthwiththespectraobservedbyPanic´etal.(2010).Aslight J = 30 29 line is broader than expected from the projected − asymmetryin the observedspectra witha strongerblue-shifted Kepler velocity at the inner rim. This is due to thermal broad- peakishowevernotreproducedbythemodel.Panic´etal.(2010) ening of the line, correspondingto more than 1 km s 1 for the − attributed the asymmetry to an asymmetric temperature struc- gas temperatures of a few 1000 K reached in the inner, upper ture that might be due to a warped inner disk (Quillen 2006). atmosphere. Modeling these asymmetries is beyond the scope of this study WeconcludethatthewidthoftheCOlinesisaninteresting and would requiretighterobservationalconstraintson the den- indicatoroftheradialoriginoftheemissionandthusaprobe/test sitystructureofthedisk. of the temperaturestructure. With currentfacilities, lines up to The CO J = 16 15 and J = 30 29 lines are predicted CO J =16 15canbespectrallyresolved,allowingustostudy to be considerably b−roader than those−of CO J = 3 2 and regions wit−hin radii of a few tens of AU as suggested by our J = 10 9. While CO J = 3 2 and J = 10 9 bo−th have results. a width−of 5 km s 1, CO J−= 16 15 and−J = 30 29 − have widths∼of 9 km s 1 and 12 k−m s 1. This can be−un- ∼ − ∼ − 4. Results:Dependenceonparameters derstoodbased on the originof those lines from regionscloser to the star. For the assumed Keplerian velocity field around a This section presents the results of a parameterstudy. We vary 2.5 M star,theprojectedvelocityalongthesemi-majoraxisis different input parameters of the model, such as the size of vproj =⊙sin(i)√M G/r = 31.5rA−U1/2 kms−1.Foragivenvelocity the disk or the amount of gas in the disk in order to under- v intheobserv∗edspectra,r correspondstothemaximumra- stand the dependence of the line fluxes on those parameters. proj diusfromwhichthelinephotonscanemerge.Forthehalfwidth Understandingthesedependenceswillhelpustointerpretobser- of the CO lines discussed here (2.5, 4, and 6 km s 1) we get vationaltrendsandalsotoquantifytherobustnessoftheresults, − r = 160, 50,and 27 AU, roughlycorrespondingto the radiiof giventhatlargeuncertaintiesenterthemodeling(Section5.5). 9 Brudereretal.:ThewarmgasatmosphereoftheHD100546disk 4.1.Varyingsize,gas/dustratio,andcarbonabundance considerablepartoftheemissionoriginatesfromwithin220AU (Section3.4). Keyparametersforthegasmodelingarethetotalamountofgas As expected, the fine-structure lines of [OI] for gas/dust and the size of the disk. In this section, we vary the outer ra- = 20donotchangewiththeamountofcarboninthegas.They diusofthegasdisktobeeither200or400AUandassumethat howeverslightly scale with the outer radiusby aboutthe same gas/dust = 20 or 100. A disk outer radius of r = 400 AU out factorasthelow-JlinesofCO.Thiscanbeunderstoodbythese and gas/dust = 20 (= 100) yields a gas mass of 8 10 4 M − linesemergingfromwithinaboutthesameradialregion,asdis- (4 10 3 M ). For an outer radius of r = 200 A×U, the ga⊙s − out cussedinSection3.4. ma×ssis4 1⊙0 4 M (2 10 3 M ),thusafactoroftwolower − − The[CI]and[CII]linesdependonboththeouterradiusand thanforr× =400A⊙U. × ⊙ out theamountofcarboninthegasphasebysimilaramounts.This Theouterradiusofthediskisdifficulttodetermine,asdis- istheresultofbothlinesbeingopticallythinandemergingfrom cussedinMuldersetal.(2011).ForHD100546,scatteredlight approximatelythesameregion.Thelinefluxesindeedscalewith hasbeendetected outto 1000AU (Ardilaetal. 2007), butit the amountof carbonin the atmosphere,while they changeby ∼ remainsunclearwhetherthis emission originatesfromthe disk aboutafactoroffourwithradius. or a diffuse remnantenvelope.Thus, the disk might be signifi- Themodelswith gas/dust= 100,shown in the lower panel cantlysmaller.Modelingthelow-JCOlines,Panic´etal.(2010) ofFigure8,exhibitverysimilardependencesontheouterradius constrainedthesizeofthegasdisktobe 400AUandthegas andamountofcarboninthegasphase.Howdothelineschange mass to be 10−3 M . Muldersetal. (2011∼) employed an expo- withthe largeramountofgasintheatmosphere?Theoptically nentialcutoffinsur⊙facedensityoutside350AU.Inotherdisks, thinhigh-JlinesofCOapproximatelyscalewiththeamountof Hughesetal.(2008)foundthatCOJ =3 2canbemoreclearly gas. However,since the thermalbalanceis also affected by the − explainedwhensuchaexponentialcutoffisused.Here,weuse higherdensity,leadingtoslightlyhighertemperatures,thelines a sharpcutoffatrout = 400AU (Panic´etal. 2010),butwe also increase by more than a factor of five. As expected, the low-J consideradiskwitharadiusofonly200AU(Pantinetal.2000). lines scale by a smaller factor, owing to optical depth effects. The gas/dustratio is assumed to be either an ISM value of The[OI]fluxesapproximatelyscalewithgasmass.Thecarbon 100oralowervalue,aspreviouslysuggestedbyChapillonetal. fine-structurelinesscalebyasmallerfactor,asthehigherdensity (2010)toexplainthenon-detectionof[CI]370µmand609µm movestheC+/COtransitionupintheatmosphereresultingina linestowardsthe HerbigAe star CQ Tau. We adoptgas/dust= smallerincreaseinthecolumndensity. 20,whichissimilartotheirchoice.Asmentionedbefore,these Thelineratio[CI]/[CII]isaboutafactoroftwolowerforthe gas/dustratios only refer to the populationof small grainsand highergas/dustratio.Thisisduetodifferentbranchingratiosin arethusupperlimits. theC+destruction.TheradiativeassociationofC+withH leads 2 The carbonbudgetofthe gasdependsonthe elementalde- toCH+,while grainsurfacereactionsofC+ with hydrogenated 2 pletion on dust grains. The warm ISM shows signs of little or PAHs (PAH:H) leads to CH+. Following a chain of reactions nodepletionofoxygenandwethusassumethatalloxygennot withH andelectronrecombinations,CH+ andCH+ leadtoCH 2 2 locked up in silicates is in the gas phase. We note that oxygen andCH .Forhigherdensities,CHandCH areturnedintoCO 2 2 mayfreeze-outas moleculesontothe dustgrains,e.g.as water rather than being photodissociated (see Tielens&Hollenbach ice,butthishappensonlyinthecoldshieldedregionsofthedisk 1985). This only slightly shifts the CO transition, thus affects (seeSection3.1).Carbon,however,maybedepletedbyaconsid- theCOfluxesonlyslightly,butdoesaffectthe[CI]lines. erablefractionbylockingitupinrefractorymaterial.Wechose Thereissomedegeneracybetweenmodelshavingmoregas acarbonabundanceofδ 0.1,correspondingtoanabundance (gas/dust=100)butlesscarbon(δ =0.1)andmodelswithless C C of2.4 10 5relativetohy∼drogen2.Inthisway,wecanstudythe gas(gas/dust=20)andmorecarbon(δ =0.5),sincetheyhave − C × degeneracybetweenhavingahighergas/dustratioof100com- the same amount of carbon in the atmosphere. These models paredto20andalowercarbongasphaseabundanceofδ =0.1 withthesameouterradiusagreemorecloselythanaboutafac- C comparedtoδ 0.5.Forgas/dust= 100,wealsorunmodels toroftwointhelow-J tomid-JlinesofCOand[CII](seealso C ∼ withδ =0.05. Figure9,modelswithR =5.5).Thehigh-JlinesofCOexhibit C V InFigure8,weshowtheintegratedlinefluxesforthemod- largerchanges.Asdiscussedbefore,neutralcarbonchangesby elswithdifferentradii,gas/dustratios,andcarbonabundancesin aboutafactorofthree. thegas,asinFigure5.Wefirstdiscussthemodelswithgas/dust =20shownintheupperpanelofthefigure.Thelow-Jandhigh- 4.2.Varyingthedustopacity J linesofCOhavedifferentbehaviors.Asexpected,thehigh-J linesdonotdependontheouterradiusofthedisksincetheradi- The dustopacitiesat UV wavelengthsare anotherkeyparame- ationemergesfromtheinnerdisk.Theopticallythinhigh-Jlines terofthegas,astheycontrolthepenetrationofthehigh-energy howeverscalewiththeamountofcarboninthegas.Wenotethat photonsintothediskatmosphere.Thisaffectsboththechemistry thisisalreadytrueformid-J linesthataremarginallyoptically andthermalbalance,forexamplebymeansofthephotodissoci- thick.Theopticallythicklow-Jlinesdependlessontheamount ation/ionization of molecules or the photoelectric effect on the ofcarbon,withachangebetweenthemodelofthesameradius dust grains. The opacity at short wavelengths is dominated by ofaboutafactoroftwo.Wenotethattheseopticallythicklines VSGsandPAHs. maystilldependontheabundance,sincee.g.thelinewingsmay Observationally,thesizeofthesmallestgrainsisdifficultto remainthin.Thelow-J linesalsodependontheouterradiusof determine.ForHD100546,indicationsofgrainprocessingand thedisk.Changingtheradiusfrom200AUto400AUincreases largerminimumgrainsizes were foundbyArdilaetal. (2007). the area of the disk by about a factor of four. The low-J lines Bouwmanetal.(2003)suggestedthatthesizedistributionsvary howeveronlychangebyaboutafactoroftwo.Thisisbecausea withradii,employingapopulationofsmallergrainsintheinner 40 AU and larger grains in the outer disk. Using such a grain 2 δ isdefinedasthefractionofcarboninthegas-phase(SeeSection distribution,Benistyetal.(2010)fittheSEDusingsilicatedust C 3) grainsandasurfacelayerof0.05 1µmgrainsat13 50AUand − − 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.