Transition disk chemistry and future prospects with ALMA
L. Ilsedore Cleeves, Edwin A. Bergin, Thomas J. Bethell, Nuria Calvet, Jeffrey K. J. Fogel, Juergen Sauter, Sebastian Wolf
aa r X i v : . [ a s t r o - ph . S R ] O c t Draft version November 19, 2018
Preprint typeset using L A TEX style emulateapj v. 11/10/09
TRANSITION DISK CHEMISTRY AND FUTURE PROSPECTS WITH ALMA
L. Ilsedore Cleeves, Edwin A. Bergin, Thomas J. Bethell, Nuria Calvet, and Jeffrey K. J. Fogel
Department of Astronomy, University of Michigan, 825 Dennison Building, 500 Church St, Ann Arbor, MI 48109 andJ¨urgen Sauter and Sebastian Wolf
Christian-Albrechts-Universit¨at zu Kiel, Institut f¨ur Theoretische Physik und Astrophysik, Leibnizstr. 15, 24098 Kiel, Germany
Draft version November 19, 2018
ABSTRACTWe explore the chemical structure of a disk that contains a large central gap of R ∼
45 AU, as iscommonly seen in transitional disk systems. In our chemical model of a disk with a cleared innervoid, the midplane becomes revealed to the central star so that it is directly irradiated. The midplanematerial at the truncation radius is permissive to reprocessed optical heating radiation, but opaque tothe photo-dissociating ultraviolet, creating an environment abundant in gas-phase molecules. Thus thedisk midplane, which would otherwise for a full disk be dominated by near complete heavy elementfreeze-out, should become observable in molecular emission. If this prediction is correct this hasexciting prospects for observations with the Atacama Large Millimeter/Submillimeter Array (ALMA),as the inner transition region should thus be readily detected and resolved, especially using high-Jrotational transitions excited in the high density midplane gas. Therefore such observations willpotentially provide us with a direct probe of the physics and chemistry at this actively evolvinginterface.
Subject headings: accretion, accretion disks — astrochemistry — circumstellar matter — stars: pre-main sequence INTRODUCTION
Over the last thirty years, our knowledge of pre-main sequence evolution has undergone significant ad-vances. Observations of the full spectral energy dis-tribution (SED) from disks around young stars haveshown that not all are alike, with the striking discov-ery of a subset of disks with optically thin inner “holes”devoid of small grains, surrounded by optically thickouter disks (e.g. Strom et al. 1989; Calvet et al. 2005;Cieza et al. 2010; Espaillat et al. 2010). This inner voidwas later confirmed by resolved sub-mm interferometry(e.g. Pi´etu et al. 2006; Brown et al. 2008; Hughes et al.2009; Andrews et al. 2009, 2011) and was interpretedas an intermediate stage between primordial and debrisdisks, coined “transition disks” (Strom et al. 1989).In this work we adopt the physical definition of “tran-sition disk” as an optically thick disk truncated withinsome inner radius within which there has either beensubstantial grain growth or removal. Such objectsare of particular interest, as the presence of a gaphas been attributed to clearing by young protoplanets(e.g. Skrutskie et al. 1990; Bryden et al. 1999; Rice et al.2003) or tidal interactions with young stellar companions(e.g. Dutrey et al. 1994).Dullemond et al. (2001) investigated the physicalstructure of Herbig Ae/Be disks that possess large innerradii similar to those seen in low mass transition disks.In this work it was found that at the inner rim, the nor-mal incidence angle of the stellar irradiation causes thedisk rim to be much hotter than it would otherwise befor a classical flared disk, where the radiation arrives ata glancing angle. Further observational evidence for thedirectly irradiated wall was later seen both in studies ofthe SED (e.g. Espaillat et al. 2007) as well as through scattered light imaging (Brown et al. 2008).ALMA will readily resolve such inner gaps and voidsin the dust disk (e.g. Wolf et al. 2002; Wolf & D’Angelo2005); however, here we seek to make predictions regard-ing the significant gas reservoir in transition disks. Inparticular, one interesting aspect of disk chemistry thathas yet to be explored is the potential that this warm UVirradiated inner rim should have unique chemical proper-ties. This has important implications since it is possiblethat the physical and chemical conditions of the previ-ously hidden, but now exposed, midplane will be revealedand potentially detectable with high spatial resolutionobservations, such as those anticipated by ALMA. Herewe present a chemical model of a protoplanetary diskwith a large inner gap similar to those seen in classicaltransition disks (e.g. Hughes et al. 2009; Espaillat et al.2010) as a prospective study of observability with ALMA. CHEMICAL MODELING
Disk Framework
For our model we adopt the comprehensive disk struc-ture from Sauter et al. (2009), originally purposed forCB26 and constrained by a large set of observations: theSED from nanometer to millimeter wavelengths, near-infrared scattered light images, and resolved millimeterimages. For the central star, the model of Sauter et al.(2009) assumes standard T Tauri values of M = 0 . ⊙ and L = 0 . ⊙ . The disk density profile is given by: ρ = ρ rr ! − α exp − zh ! ! (1) h = h rr ! β (2)Using this model and temperatures derived with MC3D(Wolf et al. 1999), Sauter et al. (2009) successfully re-produced the full set of observations with the followingbest-fit model parameters: r = 100 AU, h = 10 AU, α = 2 . β = 1 . R outer = 200 AU and R inner = 45AU. The measured size of the hole is typical of thatseen in resolved millimeter observations and SED mod-eling of transition disks, which range from a few AU toupwards of 70 AU (Andrews et al. 2011; Espaillat et al.2007). The inner disk irradiation substantially increasesthe disk temperature at the wall, reaching ∼
50 K, ascompared to a typical midplane temperature of ∼
15K as seen at larger radii. We note that this model istaken as a “snapshot” and any subsequent disk physicalevolution is beyond the scope of this work, though thechemistry of a variety of disk geometries will be exploredin an upcoming paper.For the gas density we assume a standard ISM gas-to-dust mass ratio of f g = 100 and that the dust andgas are co-spatial. The gas temperature is taken to beequal to T dust , which holds true near the dense mid-plane (Jonkheid et al. 2004). We furthermore note T dust was originally derived assuming passive heating by thecentral star, with accretion heating treated as negligible(Sauter et al. 2009). The effects of inclusion of accretionheating are discussed in Section 4.1. The model physicalstructure is shown in Fig. 1.For the opacities we adopt Weingartner & Draine(2001) for a blend of astronomical silicates and carbona-ceous grains with R V = 5 .
5. We also assume an unset-tled disk and uniform dust composition for simplicity.Deviations from these assumptions, such as the affectof vertical settling of small grains and grain-growth, arediscussed in Section 4.2.
Radiation Field
Both the ultraviolet and X-ray radiation field are be-lieved to be dominant factors driving the chemistry indisks (Glassgold et al. 1997, 2004; van Zadelhoff et al.2003). While low mass young stars peak in the opti-cal regime with negligible chromospheric contribution tothe FUV, the accretion shock at the central star providesa significant source of both FUV and soft X-ray flux (e.g.G¨unther et al. 2007). We neglect external contributionfrom the ISRF since the stellar FUV will dominate ex-ternal sources by orders of magnitude in the inner disk(Bergin et al. 2003), though its import may lie in under-standing the outer disk chemistry ( ¨Oberg et al. 2010).Due to the geometry of the star-disk system it is thennecessary to properly treat the radiative transfer into thedisk (Willacy & Langer 2000; van Zadelhoff et al. 2003;Fogel et al. 2011). This is especially true in transitiondisks, where voids and gaps can allow photons to prop-agate more freely into the outer disk material. Possibleimplications of the presence of a small amount of unde-tected dust within the gap are discussed in Section 4.3.
Continuum Radiative Transfer
We assume the FUV continuum opacity is dominatedby the dust, and consequently the UV field is depen-dent on settling of small grains, opacities assumed, anddisk geometry. Using the measured FUV spectrum ofTW Hydra (Herczeg et al. 2002, 2004) binned down to
Fig. 1.—
Left:
Plot of the transition disk model with the cen-tral star at the origin. Panels: a) gas density b) temperature c)FUV radiation field at 1600 ˚A.
Right:
Vertical cuts of the samequantities taken at R = 45, 50 and 100 AU. nine discrete wavelengths between 950-2000 ˚A, we cal-culate the continuum radiative transfer into the disk us-ing the method of Bethell & Bergin (2011), implement-ing the opacities described in Section 2.1. The choiceof nine wavelengths was motivated by the need to suffi-ciently capture the shape of the FUV continuum, to beinsensitive to individual weak emission lines, and to alsoremain computationally efficient.
Lyman- α Radiative Transfer
While many individual lines in the observed TW Hy-dra spectrum are weak, the Lyman- α line alone car-ries ∼
85% of the total FUV flux (Herczeg et al. 2004;Bergin et al. 2003). This line is furthermore expected toplay a significant chemical role as a number of molec-ular species have photodissociation cross sections near1216˚A (Fogel et al. 2011). In addition to dust scatter-ing, Lyman- α undergoes isotropic scattering off hydro-gen atoms (Bethell & Bergin 2011), which in principlerequires an iterative and computationally expensive cal-culation between the radiation field and the chemistry.In this work, we incorporate the effects of Lyman- α using an approximate treatment motivated byBethell & Bergin (2011). The procedure is illustratedin Fig. 2. If one assumes a priori that there is an opti-cally thick layer of atomic hydrogen on the disk surfacelocated at τ FUV = 1 as defined by the dust (Fig. 2: Re-gion II), the Lyman- α photons will first encounter thisisotropic H-scattering layer. A fraction of the radiationwill be lost to space ( . α photons diffusively propa-gating through the H-layer. Below the atomic layer, thehydrogen is predominantly H and the Lyman- α pho-tons proceed as continuum photons scattering off onlythe dust grains (Fig. 2: Region III). Thus, at the base ofthe hydrogen scattering layer (II → III), the photons effec-tively form a layer of isotropically emitting point sources,which together create a planar source of radiation thatshines at a nearly normal angle to the midplane. There-fore even though a significant fraction of Lyman- α pho- I. II. III.
Fig. 2.— Ly α schematic illustration. (I.) “Free-streaming” re-gion above τ radial ∼ α radiation escapes thedisk. (II.) H-scattering layer with a vertical depth of τ vertical ∼ . Dust now be-comes the dominant source of opacity to Ly α photons, and the Ly α radiation now behaves as vertically attenuated continuum photons,which significantly enhances its penetrating power. The (I.) → (II.) interface illustrates the τ ∼ → (III.) interface is effectively the τ ∼ α photons. tons are lost to space, the remaining photons will havegreater vertical penetration power into the disk and canpropagate many AU deeper than the UV continuum ra-diation. On the front edge of the disk, at the dense innerrim, Lyman- α is similarly “stopped” as was seen for thecontinuum photons. (cf. Fig. 1 (c)). We discuss theimplications of an inner void that is not empty of gas onLyman- α transfer in Section 4.4. Reaction Network
Combining the model detailed in Section 2.1 and theradiation field in Section 2.2, the resultant chemistryis calculated using Fogel et al. (2011)’s comprehensivedisk chemical model, based on the Ohio State Univer-sity Astrophysical Chemistry Group’s gas-phase network(Smith et al. 2004). The reaction types include photo-desorption, photo-dissociation, freeze-out, grain surfacereactions, ion and electron reactions, cosmic-ray and stel-lar X-ray ionization, and radiative reactions. In total,the network encompasses 5910 reactions and 639 react-ing species, including some time-dependent reactions,encompassing the main species of astrochemical impor-tance, and described in detail in Fogel et al. (2011).The model initially assumes uniform molecular cloudchemical abundances Aikawa & Herbst (1999) and fol-lows the chemical evolution for 3 Myr. In this work wetake the abundances at 1 Myr, which is long enough suchthat the chemistry has “relaxed” but not so long that thedisk would likely have physically evolved away from thisstate.Furthermore, we assume a typical integrated T Tauristar X-ray luminosity of 10 erg s − and a thermal X-ray spectrum between 1-10 keV (Glassgold et al. 1997,and references therein). The model of Fogel et al. (2011)incorporates the method of Aikawa & Herbst (2001) forX-ray propagation. For cosmic rays we adopt a typ-ical cosmic ray ionization rate of 1.3 × − s − per H, with an attenuation column of 96 g cm − (Umebayashi & Nakano 1981). RESULTS
Disk Chemistry
The resultant chemical abundances (relative to n H = n HI + 2 n H ) for six observed gas-phase species are shownin Fig. 3. Molecules such as CO, H CO, and N H + showan enhanced gas-phase abundance at the wall. For com-parison, in an untruncated disk model, all neutral speciesplotted in Fig. 3 would be otherwise frozen onto grainsat the midplane at R wall = 45 AU. The two ions shown,N H + and HCO + , would also not be present in an un-truncated disk, as their chemical precursors, N and CO,would be frozen out, inhibiting the formation of these twospecies. Consequently the gas-phase enhancement at thetruncation radius shown for all species plotted here, be-sides H O, would not exist if it were not for the largeinner gap.Water, with a freeze-out temperature of ∼
100 K, isfrozen onto grains at the ∼
50 K transition region. Onecould however envision a disk with a wall closer to thecentral star and thus warmer, such that water would sub-limate from grains and be observable. Thus the specificspecies present are not necessarily the most importantresult, but that the star can efficiently heat the wall,warming it above the sublimation temperature of a vari-ety of species, potentially allowing us to observationallyprobe disk physics as well as evolutionary state. There-fore the existence of a cleared inner gap should produceunique chemical features in the outer disk not present infull classical disks.
Observables
If such an enhancement is present it is of interest todetermine observability. While previous observationsreveal a diverse chemistry (e.g. Dutrey et al. 1997;¨Oberg et al. 2010), the sensitivity and resolving powerof ALMA is required to fully understand the detailedstructure of these systems.We adopt a disk inclination of 60 ◦ and calculate theresulting emission for rotational transitions of CO ando-H CO as would be seen at a distance of 140 pc us-ing the non-LTE line radiation transfer code, LIME(Brinch & Hogerheijde 2010). These species have beenchosen as both are commonly observed towards disksaround low mass T Tauri stars (e.g. ¨Oberg et al. 2010,2011). The densities reached at the frontally illuminatedmidplane preferentially excite high-J rotational transi-tions, and therefore transition disks should uniquely ex-hibit high-J bright molecular rings at the wall when re-solved by sub-mm observations with ALMA. Using thiscalculated line emission, we then use the
SIMDATA pack-age in CASA to calculate the ALMA visibilities and re-construct the image from the UV coverage of the fullALMA array. Fig. 4 shows the simulated observationsfor these lines as seen by ALMA for an antenna con-figuration with ∼ . ′′ resolution at 672 GHz, chosento provide adequate resolution and sensitivity assumingtypical thermal noise. FURTHER CONSIDERATIONS
In this work we propose the presence of gas-phasemolecules at the inner edge of transition disks. This ef-
Fig. 3.—
Chemical model results, plotted as abundance relative to the total number of hydrogen atoms. Shown are common species ofastrophysical interest: (a) CO, (b) HCO+, (c) H CO, (d) H O, (e) N H + , (f) HCN. fect could however be erased if either the dust at the in-ner edge is not sufficiently heated or if the wall becomespermissive to molecule destroying UV radiation. In thefollowing we discuss potential caveats of the model andhow these would alter the results presented here. Accretion Heating
The original model was computed assuming passiveheating by the central star and heating due to disk accre-tion was treated as negligible. Accretion heating, how-ever, predominantly increases the midplane temperature(D’Alessio et al. 1998), and, if significant, will increasethe thermal desorption of molecules from grains, enhanc-ing the predicted effect. Details of the dynamical trans-port of gas and dust and their effects on the chemistry,however, are beyond the scope of this Letter.
Dust Settling and Grain Coagulation
For this model we have assumed an unsettled outer diskwith grains of ISM abundance. In our analysis we havealso explored a model that includes mixed grain growthof dust particles of up to 1 mm in size, without verticalsettling. This lowers the UV opacity (cm g − ) by ∼ ρ dust ∼ − g cm ), however, arestill of sufficient magnitude such that the wall remains optically thick to UV radiation for a model with mixedgrain growth alone.Strong settling or removal of small grains can also makethe disk more permeable to the UV radiation. We canask how much dust would need to be removed/settled toallow the UV photons to penetrate and photodissociatethe enhanced abundance of molecules present at the wall.To approximate this, the unsettled wall density is ρ dust ∼ − g cm and the UV dust opacity is ∼ × cm g − at 1000 ˚A (Weingartner & Draine 2001). Based on thethermal model, the thickness of the “warm” inner edgeis ∼ < .
1% of theoriginal amount of dust present to erase the effect seenhere. We note however that settling will also increase thedepth of the heating, which will in principle thicken theextent of the wall further.
Dust in the Inner Gap
An underlying feature of the disk model is the presenceof a large inner “void” which allows the outer disk to beheated directly by the star. One possible explanationfor such a void is grain growth into rocky planetesimals(e.g. Skrutskie et al. 1990). Thus one could postulatethe existence of a small amount of undetected dust in-side the gap. Indeed some models infer the presence ofsome moderate mass of silicates in the inner disk (e.g.Calvet et al. 2002, 2005; Espaillat et al. 2007). This ma-
Fig. 4.—
Simulated ALMA observations (computed with one hour integration time) of CO and H CO for a disk at 140 pc, at differentrotational transitions with increasing J. This demonstrates the increasing contrast between the inner transition region and outer diskemission for higher J, along with ALMA’s future capabilities at these frequencies. terial has the ability to shadow the outer disk from beingdirectly heated by the star, thus inhibiting the presenceof the warm molecular interface presented here.First, one can ask how much dust would be requiredto cause the gap to become opaque ( τ ∼
10) to opti-cal radiation peaking at λ = 0 . µ m as is the case for a T = 4000 K star. If we assume that the inner disk followsthe outer disk profile (Section 2.1), a total mass in dustof M dust = 10 − M Moon inside 45 AU would be sufficientfor the midplane to become optically thick to the opti-cal heating radiation. While this is an extremely smallamount of dust, one must also consider the scale heightof the material: h ∼ . Gas in the Inner Gap
By assuming gas and dust are co-spatial we intrinsi-cally assume the gap is empty of gas. Gas within thegap would not alter the original optical heating calcula-tion of the outer disk or the propagation of UV contin-uum photons, both of which have dust-dominated opac-ities. However, the presence of gas would strongly affectthe propagation of Lyman- α as gas within the gap wouldlikely be predominantly atomic. The presence of hydro-gen within the gap would cause a net reduction in theLyman- α flux reaching the outer disk. This would ei-ther not change or marginally enhance the abundance ofgas phase molecules at the truncation radius since theamount of photo-dissociating radiation is reduced. CONCLUSIONS