New Mid-Infrared Imaging Constraints on Companions and Protoplanetary Disks around six Young Stars
D. J. M. Petit dit de la Roche, N. Oberg, M. E. van den Ancker, I. Kamp, R. van Boekel, D. Fedele, V. D.Ivanov, M. Kasper, H. U. Käufl, M. Kissler-Patig, P. A. Miles-Páez, E. Pantin, S. P. Quanz, Ch. Rab, R.Siebenmorgen, L. B. F. M. Waters
AAstronomy & Astrophysics manuscript no. main © ESO 2021February 26, 2021
New mid-infrared imaging constraints on companions andprotoplanetary disks around six young stars (cid:63)
D. J. M. Petit dit de la Roche , N. Oberg , M. E. van den Ancker , I. Kamp , R. van Boekel , D. Fedele , V. D.Ivanov , M. Kasper , H. U. Käufl , M. Kissler-Patig , P. A. Miles-Páez , E. Pantin , S. P. Quanz , Ch. Rab , , R.Siebenmorgen , and L. B. F. M. Waters , European Southern Observatory, Karl-Schwarzschild-Strasse 2, 85748 Garching, Germanye-mail: [email protected] Kapteyn Astronomical Institute, University of Groningen, P.O. Box 800, 9700 AV Groningen, The Netherlands Max-Planck Institut für Astronomie, Königstuhl 17, 69117 Heidelberg, Germany INAF-Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 Firenze, Italy European Space Agency, Camino Bajo del Castillo, s / n., Urb. Villafranca del Castillo, 28692 Villanueva de la Cañada, Madrid,Spain CEA, IRFU, DAp, AIM, Université Paris-Saclay, Université Paris Diderot, Sorbonne Paris Cité, CNRS, F-91191 Gif-sur-Yvette,France Institute for Particle Physics and Astrophysics, ETH Zurich, Wolfgang-Pauli-Strasse 27, 8093 Zurich, Switzerland Max-Planck-Institut für extraterrestrische Physik, Giessenbachstrasse 1, 85748 Garching, Germany Department of Astrophysics / IMAPP, Radboud University Nijmegen, P.O.Box 9010, 6500 GL Nijmegen, The Netherlands SRON Netherlands Institute for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, The NetherlandsReceived XXXX; accepted XXXX
ABSTRACT
Context.
Mid-infrared (mid-IR) imaging traces the sub-micron and micron-sized dust grains in protoplanetary disks and it o ff ersconstraints on the geometrical properties of the disks and potential companions, particularly if those companions have circumplanetarydisks. Aims.
We use the VISIR instrument and its upgrade NEAR on the VLT to take new mid-IR images of five (pre-)transition disksand one circumstellar disk with proposed planets and obtain the deepest resolved mid-IR observations to date in order to put newconstraints on the sizes of the emitting regions of the disks and the presence of possible companions.
Methods.
We derotated and stacked the data to find the disk properties. Where available, we compare the data to P ro D i M o ( Pro toplanetary Di sk Mo del) radiation thermo-chemical models to achieve a deeper understanding of the underlying physical pro-cesses within the disks. We applied the circularised point spread function subtraction method to find upper limits on the fluxes ofpossible companions and model companions with circumplanetary disks. Results.
We resolved three of the six disks and calculated position angles, inclinations, and (upper limits to) sizes of emission regionsin the disks, improving upper limits on two of the unresolved disks. In all cases the majority of the mid-IR emission comes from smallinner disks or the hot inner rims of outer disks. We refined the existing P ro D i M o HD 100546 model spectral energy distribution (SED)fit in the mid-IR by increasing the PAH abundance relative to the ISM, adopting coronene as the representative PAH, and increasingthe outer cavity radius to 22.3 AU. We produced flux estimates for putative planetary-mass companions and circumplanetary disks,ruling out the presence of planetary-mass companions with L > . L (cid:12) for a >
180 AU in the HD 100546 system. Upper limits of0.5 mJy-30 mJy are obtained at 8 µ m-12 µ m for potential companions in the di ff erent disks. We rule out companions with L > − L (cid:12) for a >
60 AU in TW Hydra, a >
110 AU in HD 169142, a >
150 AU in HD 163296, and a >
160 AU in HD 36112.
Conclusions.
The mid-IR emission comes from the central regions and traces the inner areas of the disks, including inner disks andinner rims of outer disks. Planets with mid-IR luminosities corresponding to a runaway accretion phase can be excluded from the HD100546, HD 169142, TW Hydra, and HD 36112 systems at separations > (cid:48)(cid:48) . We calculated an upper limit to the occurrence rate ofwide-orbit massive planets with circumplanetary disks of 6.2% (68% confidence). Future observations with METIS on the ELT willbe able to achieve a factor of 10 better sensitivity with a factor of 5 better spatial resolution. MIRI on JWST will be able to achieve250 times better sensitivity. Both will possibly detect the known companions to all six targets.
Key words.
Methods: data analysis - Protoplanetary disks - Planets and satellites: detection - Infrared: planetary systems - Infrared:stars
1. Introduction
Transition disks are believed to represent an intermediate stageof planet formation between the protoplanetary disk and a gas- (cid:63)
Based on observations collected at the European Southern Obser-vatory under ESO programmes 0101.C-0580(A), 60.A-9107(G), and60.A-9107(N). less, fully formed planetary system. Scattered light imaging inthe near-infrared (near-IR) and thermal sub-millimetre observa-tions with ALMA have revealed detailed structures in many tran-sition disks, including rings, spirals, and warps (e.g. Francis &van der Marel 2020). These features can be a result of the ac-cretion of gas and dust onto a planet, although they can also beexplained by other processes in the disk such as shadowing from
Article number, page 1 of 18 a r X i v : . [ a s t r o - ph . E P ] F e b & A proofs: manuscript no. main
Table 1.
Stellar and disk properties of the target stars. Stellar masses, luminosities, and temperatures, where possible, have been taken from theDIANA models of the targets, which are fit to multiple data sets. (a)
Gaia Collaboration et al. (2020). (b)
Wichittanakom et al. (2020); Miley et al.(2019); Casassus & Pérez (2019); Jamialahmadi et al. (2018); Mendigutía et al. (2017); Pineda et al. (2014); Avenhaus et al. (2014); Walsh et al.(2014); Leinert et al. (2004). (c)
Garufi et al. (2014); Wichittanakom et al. (2020). (d)
Pérez et al. (2019); Pani´c et al. (2008); Raman et al. (2006);van Boekel et al. (2005). (e)
Nayakshin et al. (2020); Sokal et al. (2018). (f)
Wichittanakom et al. (2020); Rosotti et al. (2020); Yu et al. (2019);Long et al. (2017); Benisty et al. (2017); Wagner et al. (2015). (g)
Isella et al. (2010); Meeus et al. (2012).
Target Age (Myr) M ∗ (M (cid:12) ) T (K) L ∗ (L (cid:12) ) d (pc) (a) PA ( ◦ ) i ( ◦ ) Structures CompanionsdetectedHD 100546 ( b ) ( c ) ( d ) ( e ) ( f )
11 1.5 - 6 103.8 145 35 gap, spiral arms M dwarfHD 36112 ( g ) ff ects (e.g. Sieben-morgen & Heymann 2012; van der Marel et al. 2018). Study-ing transition disks is an important step in understanding planetformation. Mid-infrared (mid-IR) direct imaging traces dust of ∼
150 K in the disk. Additionally, the disk is expected to re-emita large fraction of the stellar flux in the infrared (e.g. Dullemond& Monnier 2010). Mid-IR imaging can thus further constraindisk properties, especially when combined with observations atother wavelengths. It also allows us to search for thermal emis-sion from (planetary) companions, especially if these compan-ions still have circumplanetary disks (CPDs), which are expectedto be bright in the mid-IR.We used the VLT Imager and Spectrograph for the mid-InfraRed (VISIR; Lagage et al. 2004) and its upgraded versionNear Earths in the AlphaCen Region (NEAR; Kasper et al. 2017)to obtain the deepest resolved mid-IR images of five HerbigAe / Be (pre-)transition disks and one other circumstellar disk todate. The instruments that we used are more sensitive and theobservation time is longer than in any previous studies (Liu et al.2003; van Boekel et al. 2004; Leinert et al. 2004; Verhoe ff ro D i M o ( Pro toplanetary Di sk Mo del; see Sect. 4). Our HD 100546 disk model is the result ofa multi-wavelength spectral energy distribution (SED) fit, whichwill allow us to compare the predicted and observed total fluxwithin the observed bands (Woitke et al. 2019). Our syntheticimages of the HD 100546 circumstellar disk enabled us to searchfor a non-axisymmetric disk structure. The radiative transfer re-sults allowed us to determine the mid-IR extinction along line-of-sights to the midplane and the resulting obscuration of pu-tative embedded companions. The disk modelling code can beapplied further to produce SEDs for planetary companions andcircumplanetary disks to compare theoretical fluxes with detec-tion limits (Rab et al. 2019).Section 2 describes the targets and in Sect. 3 we show theobservations and the data analysis. The P ro D i M o model is dis-cussed in Sect. 4 and compared to the data in Sect. 5. Limitson possible companions are discussed in Sect. 6 and for three of the targets planetary models with circumplanetary disks areanalysed. Finally, our discussion and conclusions are presentedin Sect. 7.
2. Targets
The following targets were observed: HD 100546, HD 163296,HD 169142, TW Hydra, HD 100453, and HD 36112 / MWC 758(see Table 1). These stars were selected to study the influenceof features such as spiral arms, circular gaps, and inner cavities,seen in near-IR scattered light images on the mid-IR morphologyof the disk which is dominated by thermal emission.All six targets are young disks with ages of 3-16 Myrand, with the exception of HD 163296, are classified as (pre-)transition disks with a central cavity (or large inner gap). WhileHD 163296 does not have the traditional (pre-)transition diskSED and only some evidence of possible inner clearing, itnonetheless has other structures in the disk and proposed com-panions, similar to the remaining targets in the sample and wastherefore included here (Espaillat et al. 2014; Isella et al. 2016).In addition to central cavities, sub-millimetre dust emission andnear-IR scattered light imaging have revealed features such asrings, clumps, and spirals in all the disks. At distances of 60-160 pc, the extended disks of the targets are expected to be largeenough to be resolved with the Very Large Telescope (VLT) atParanal in the 8-12 µ m wavelength range.Below, we provide an overview of the structure and possiblecompanions of the targets, specifically those inferred through di-rect imaging. This disk is divided into an inner disk and an outer disk, sepa-rated by a single gap from ~1-21 AU (e.g. Bouwman et al. 2003;Grady et al. 2005; Menu et al. 2015; Jamialahmadi et al. 2018;Pineda et al. 2019). It is possible that the inner and outer disksare misaligned (Pineda et al. 2019; Kluska et al. 2020). The outerdisk has spiral structures that have so far only been detected inthe near-IR (Follette et al. 2017; Quillen 2006) and there is a ten-tative detection of a bar-like structure across the gap which couldindicate small-scale inflow or be the base of a jet (Mendigutíaet al. 2017; Schneider et al. 2020). There have been some sug-gestions of warping in the inner and outer disk, but this has sofar remained inconclusive (e.g. Quillen 2006; Pani´c et al. 2014;Pineda et al. 2014; Walsh et al. 2017; Sissa et al. 2018; Kluskaet al. 2020).
Article number, page 2 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars
There has been much discussion about possible companions.One companion, HD 100546 b, was identified at a separationof 55 AU at a position angle of 9 ◦ (Quanz et al. 2013; Currieet al. 2014; Quanz et al. 2015). However, this has been calledinto question by Rameau et al. (2017), who failed to detect anyaccretion at the planet location in H α and posit the L (cid:48) band(3 . µ m) detection might be related to the chosen method of datareduction. The lack of detection in H α is supported by Cugnoet al. (2019). A di ff erent companion, HD 100546 c, may havebeen detected just inside the central cavity at ~13 AU (Brittainet al. 2014; Currie et al. 2015), although this too has been con-tested (Fedele et al. 2015; Follette et al. 2017; Sissa et al. 2018).ALMA observations at 1.3 mm have revealed a 6 σ point sourceof 92 ± µ Jy at a position angle of 37 ◦ and a projected separationof 7.8 AU, which could represent an additional planetary candi-date (herafter HD 100546 d; Pérez et al. 2020). A final planetcandidate has also been suggested by the presence of a Dopplerflip observed in the disk CO kinematics. Such a planet wouldbe embedded within the disk continuum emission region exte-rior to the gap, corresponding to a projected radial distance of20 . ± Near-IR and sub-millimeter wavelength observations show thatHD 163296 has four gaps. They are centred on 10 AU, 50 AU,81 AU, and 142 AU with bright rings in between (e.g. Garufiet al. 2014; Isella et al. 2016; Isella et al. 2018).Companions have been suggested based on their possiblerole in forming the ring structures in the disk. For example, Liuet al. (2018) fitted three half-Jovian-mass planets and Teagueet al. (2018) found the radial pressure gradients can be explainedby two Jupiter-mass planet at 83 and 137 AU (see also Teagueet al. 2019). Additionally, Pinte et al. (2018) found a Jupiter-mass companion at 223 AU based on deviations from Keplerianvelocity in the gas of the disk. So far, observations have not beenable to confirm or rule out such companions due to a lack ofsensitivity. Guidi et al. (2018) claim to have found a 5-6 M
Jup companion at a separation of 50 AU from the star in the L (cid:48) bandwith Keck / NIRC2, but neither this object nor the one proposedby Pinte et al. (2018) was found by Mesa et al. (2019), who setupper limits of 3-5 M
Jup on possible companions in the gaps ofthe disk with SPHERE H band (1.6 µ m) and K band (2.2 µ m)data. Due to extinction from the disk setting, these kinds of masslimits remain challenging, especially outside the gaps, as only afraction of the intrinsic, modelled flux of the companion may beobservable. The disk around HD 169142 has been imaged at near-IR andat sub-millimetre wavelengths. Various teams have imaged two(Fedele et al. 2017; Quanz et al. 2013; Momose et al. 2015; Pohlet al. 2017), three (Macías et al. 2017; Osorio et al. 2014), or four(Macías et al. 2017; Pérez et al. 2019) rings around the star. Theinner ring is located at 20 AU and is more than twice as brightas the outer rings. As a result, it was found in all the previouslymentioned works. The three outer rings (located between 45 AUand 80 AU) are faint and close together, leading to blending insome observations and resulting in the di ff erent numbers of ringsfound in di ff erent studies.Four disk features that could be associated with formingplanets have been found. The first was found between the 20 AU and 50 AU dust rings by Osorio et al. (2014) at 7 mm, the secondwas found in the L (cid:48) band just within the edge of the inner gap byReggiani et al. (2014) and Biller et al. (2014). However, the L (cid:48) band source was not recovered by either team in the J (1.3 µ m),H, or K bands and it is concluded by Biller et al. (2014) thatthe feature cannot be due to planet photospheric emission andmust be a disk feature heated by an unknown source, althoughReggiani et al. (2014) argue that the accretion of material in thegap enhances the L (cid:48) band flux, resulting in a lower mass planet,which is not as easily observed in other bands. The presenceof circumstellar material with entrained dust grains spreadingacross the gap or being accreted onto a planet could also subjectthe planet to further extinction in the J band. Biller et al. (2014)detected the third source in the H band, with no L (cid:48) band counter-part, but Ligi et al. (2018) show that this is actually part of the in-ner ring. They did find another H band structure close to the starthat is consistent with the detections by Biller et al. (2014) andReggiani et al. (2014), but it appears to be extended and they can-not rule out that it is not part of a marginally detected ring at thesame separation. Finally, Gratton et al. (2019) combined di ff er-ent SPHERE datasets and suggest that this source could actuallybe a combination of two extended blobs observed in the disk.They find a di ff erent, fourth, feature located between the innerand outer rings that does not correspond to any of the previousdetections and could indicate the presence of a 2 . ± . Jup planet.
TW Hydra is a 3-15 Myr old T Tauri star (Vacca & Sandell 2011;Weinberger et al. 2013; Herczeg & Hillenbrand 2014). At a dis-tance of 60 . ± .
06 pc (Gaia Collaboration et al. 2020), it isone of the nearest known hosts of a protoplanetary disk. Stud-ies in the near-IR and sub-millimetre wavelength regimes havefound between three and six di ff erent gaps in eight di ff erent lo-cations between 0.6 AU and 90 AU (Nomura et al. 2016; Tsuk-agoshi et al. 2016; Andrews et al. 2016; van Boekel et al. 2017;Huang et al. 2018).Tsukagoshi et al. (2016) suggest the presence of a (cid:46)
26 M ⊕ planet interacting gravitationally with the gap at 22 AU. Tsuk-agoshi et al. (2019) found an azimuthally elongated 1.3 mm con-tinuum source in the south-west of the disk at a radial separationof 54 AU that could be either dust that has accumulated into aclump in a vortex or a circumplanetary disk associated with anaccreting Neptune mass planet. Nayakshin et al. (2020) arguethe feature can be explained by a Neptune-mass planet disruptedin the process of accretion and expelling dust into the circum-stellar disk. Observations with SPHERE suggest from the gapprofiles that if planets are responsible for forming the gaps in thecircumstellar disk, they are at most several 10 M ⊕ (van Boekelet al. 2017). HD 100453 has been found to possess a misaligned inner disk, agap between 1 AU and 21 AU, and an outer disk with two shad-ows, two spiral arms around 30 AU, and a faint feature in thesouth-west of the disk (Benisty et al. 2017; Kluska et al. 2020).It also has an M dwarf companion at a separation of 125 AUwhose orbit is not aligned with the disk plane (van der Plas et al.2019).Dynamical modelling has shown that tidal interactions withthe M dwarf companion are responsible for at least some of the
Article number, page 3 of 18 & A proofs: manuscript no. main disk features, such as the spirals and the truncation of the outerdisk (Wagner et al. 2018; van der Plas et al. 2019; Gonzalez et al.2020). However, they have also suggested that the presence of aplanet is required to fully explain the origin of the features in thedisk, particularly the misalignment between the inner and theouter disks (e.g. Nealon et al. 2020). There have been no directdetections of planet candidates to date.
HD 36112 (MWC 758) has a large cavity with a radius of 32 AU.Its broad outer disk has rings, clumps, and spiral arms (e.g. Donget al. 2018; Wagner et al. 2019).For the spiral structures in the disk of HD 36112 to be causedby a perturber, it is estimated that it must have a mass of ∼ Jup (Grady et al. 2013; Dong et al. 2015). However, upperlimits on companion fluxes obtained in the same works and byReggiani et al. (2018) rule out the presence of > Jup planetsbeyond 0.6 (cid:48)(cid:48) , or 94 AU. Reggiani et al. (2018) found an L band(3 . µ m) point source at 18 AU that they interpret as a planetwith a circumplanetary disk that is embedded in the disk. Wagneret al. (2019) did not find this object in the L (cid:48) and M (cid:48) bands, eventhough they achieved better sensitivities. Instead, they found apoint source at the outer end of one of the spiral arms that couldbe a planet with a CPD and could be responsible for driving thespirals.
3. Observations and data analysis
Observations of HD 100546 were obtained during April 2018,with the VLT Imager and Spectrometer for the mid-IR (VISIR,Lagage et al. 2004), and of all six disks during the science verifi-cation of its upgrade, with NEAR (Kasper et al. 2017) in Septem-ber and December of 2019. The benefit of NEAR is its use ofAO, which results in improved angular resolution, PSF stability,and sensitivities (a factor of ∼
4) across the N-band. An overviewof the observations used in this paper is presented in Table 2.For all targets, all observations were taken in the pupil track-ing mode, where the derotator is turned o ff to allow for fieldrotation during the observation sequence. For the NEAR obser-vations, AO was enabled and the targets themselves were usedas the reference star for wavefront sensing. The chopping andnodding sequence was enabled to subtract the sky background.In the VISIR data, the chop throw is 8 (cid:48)(cid:48) in the direction per-pendicular to the nodding direction; whereas, in the NEAR data,the chop throw is 4.5 (cid:48)(cid:48) in the parallel direction. Since the throwdetermines the useful field of view, the VISIR and NEAR datahave an e ff ective field of view of 16 (cid:48)(cid:48) x16 (cid:48)(cid:48) and 9 (cid:48)(cid:48) x9 (cid:48)(cid:48) , respec-tively. The VISIR data have a chopping frequency of 4 Hz and adetector integration time (DIT) of 0.012 s. The NEAR data havea chopping frequency of 8 Hz and a DIT of 0.006 s. Both NEARand VISIR have platescales of 0.0453 (cid:48)(cid:48) .The standard VISIR data reduction pipeline is not suited toreduce data taken in the pupil tracking mode, so special purposepython scripts were employed to reduce and analyse the data.VISIR and NEAR data are delivered in chop di ff erence imageswith integration times of 20-50 s each. Data from the di ff erentnod positions are subtracted from each other and the resultingimages are derotated. The beams from the chopping and nod-ding from all images are then median combined with 3 σ sigma . eso . org/sci/software/pipelines/visir/visir-pipe-recipes . html clipping into a single master image. Only the VISIR observa-tions of HD 100546 have a reliable reference star (HD 93813)with which to calibrate the result, leading to an observed fluxof 27 ± ro D i M o models (described in Section4) to calibrate the data. Since the model is fitted to SED datafrom a collection of previous observations of the targets takenwith other instruments, including data around 8-12 µ m, it is themost accurate way available to determine the brightness in theimages and this allowed us to calculate the flux in the specificwavelength ranges of the di ff erent filters (Dionatos et al. 2019;Woitke et al. 2019). The calibration is done by multiplying themodel fluxes with the filter and sky transmissions and averagingthe total flux over the required wavelength range. This is thenset as the total flux of the data. As there are no models availablefor HD 100453 and HD 36112, the averages of previous fluxmeasurements in similar filters had to be used (van Boekel et al.2005; Carmona et al. 2008; Verhoe ff <
10% of thetotal flux in the P ro D i M o models). Most of the central emissionat these wavelengths is from unresolved inner disks or inner rimsof outer disks. The master images of HD 100546 in the di ff erent filters areshown in Figure 1, along with the corresponding model imagesafter convolution with an appropriate PSF. For the J8.9 filter, thisis the PSF of reference star HD 93813. While there were no ap-propriate flux calibration observations for the other filters, pointsources were observed in the PAH1 and ARIII filters, which wereused as PSF references. For the PAH1 filter and the ARIII filter,we used our own observations of HD 163296 and HD 27639,respectively. As there were no reference PSFs available in ei-ther the PAH2 or PAH2_2 filters, we used scaled versions of theARIII reference instead. Since the di ff erent filters on the NEARinstrument result in similar sensitivities over time, and the ob-servations in the di ff erent filters have similar exposure times, allmaster images are expected to have similar sensitivities. The ex-ception are the observations with the J8.9 filter which were takenwith VISIR and where the increased observation time compen-sates for the lack of AO, meaning the final sensitivity of the mas-ter image is still expected to be similar to those in the other fil-ters. While the disk is resolved in all filters, the VISIR data areclearly more extended than the NEAR data. The J8.9 band con-tains both the PAH1 and ARIII bands and so the VISIR imagewould be expected to have a similar extent as the NEAR im-ages in these bands. Some of the di ff erence is because the AO onNEAR means the images are more compact, but mostly due tothe telescope operations during the VISIR observations. Duringthis night, there was a decrease in the precision of the altitudeaxis of the telescope, resulting in elongation of the image alongthe paralactic angle (de Wit 2020). As this was at an angle of40 degrees with the semi-major axis of the disk, the image issmeared along both axes and the smearing is not immediatelyobvious without a comparison. This is accounted for by using areference PSF of the standard star HD 93813. Since this data setwas taken immediately preceding the science observations in thesame filter, it has a similar smearing e ff ect. Article number, page 4 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars
PAH1 data0.5" PAH1 model0.5"ARIII data0.5" ARIII model0.5"PAH2 data0.5" PAH2 model0.5"PAH2_2 data0.5" PAH2_2 model0.5"J8.9 data0.5" J8.9 model0.5"
HD 100546
Fig. 1.
Master images (left) and model images (right) of HD100546 in various filters. North is up and east is left in all im-ages. The observations were scaled to have the same flux as themodel images. The PAH1, ARIII, PAH2, and PAH2_2 filter mas-ter images were taken with NEAR and show a resolved, inclineddisk. The J8.9 data were taken with VISIR and are more extendedcompared to the NEAR data due to image elongation from thetelescope resulting in a distorted and enlarged PSF. The modelimages provide a good match for the master images in each filter.
HD 163296
PAH1
HD 163296
NEAR
HD 169142 HD 169142TW Hydra TW HydraHD 100453 HD 100453HD 361120.5" HD 36112
Fig. 2.
Normalised master images of the disks observed withNEAR. North is up and east is left in all images and the scalebar in the bottom left indicates 0.5 (cid:48)(cid:48) . The left column shows thedisks in the PAH1 filter and the right column in the NEAR fil-ter. HD 163296 and TW Hydra are unresolved in both filters. HD36112 was not imaged in the PAH1 filter, but it is unresolved inthe NEAR filter. Compared to these images, it can be seen thatHD 169142 and HD 100453 are more extended in both filters.Article number, page 5 of 18 & A proofs: manuscript no. main
Table 2.
Overview of the observations used in this paper. HD 100546 was observed as part of di ff erent programmes than the other observations,leading to the di ff erence in filters and observation times. Target Instrument Date Filter λ ( µ m ) ∆ λ ( µ m ) Integration time (s)HD 100546 VISIR 28-04-2018 J8.9 8.70 0.74 3600NEAR 11-12-2019 PAH1 8.58 0.41 540ARIII 8.98 0.14 540PAH2 11.24 0.54 54012-12 2019 PAH2_2 11.68 0.37 540HD 163296 NEAR 14-09-2019 PAH1 8.58 0.41 60013-09-2019 NEAR 11.25 2.5 600HD 169142 NEAR 13-09-2019 PAH1 8.58 0.41 600NEAR 11.25 2.5 600TW Hya NEAR 13-12-2019 PAH1 8.58 0.41 60016-12-2019 NEAR 11.25 2.5 600HD 100453 NEAR 12-12-2019 PAH1 8.58 0.41 600NEAR 11.25 2.5 600HD 36112 / MWC 758 NEAR 18-12-2019 NEAR 11.25 2.5 600The central bright emission in each image is from the unre-solved inner disk, as the star is expected to be an order of mag-nitude fainter than the disk at mid-IR wavelengths based on themodel data. Beyond that, emission is expected to be dominatedby the inner rim of the outer disk, which is irradiated by the starand pu ff ed up as a result. The rest of the outer disk is not warmenough to be detected in the image.Using a Levenberg-Marquardt algorithm and least squaresstatistic to fit a simple two dimensional Gaussian to the surfacebrightness of the disk in each filter results in an average positionangle of 141 ± ◦ . Since we are fitting a two-dimensional func-tion to a three-dimensional disk, we are sensitive to projectione ff ects. This is especially the case because the inner wall of theouter disk is only visible on the far side of the disk and not onthe close side. This means what we are calculating is actuallythe position angle of the two-dimensional projection of the disk,which we call the projected position angle. We also applied thismethod to model images of HD 100546 at the same wavelengthsand found that the projected position angle is ∼ ◦ , comparedto the input of 140 ◦ , so we expect a di ff erence between the pro-jected position angle and the real position angle of roughly 10degrees. This would still be in agreement with previous positionangle values of 135-150 ◦ (Miley et al. 2019; Casassus & Pérez2019; Jamialahmadi et al. 2018; Mendigutía et al. 2017; Pinedaet al. 2014; Avenhaus et al. 2014; Walsh et al. 2014; Leinert et al.2004). A more precise determination of the disk orientation re-quires extensive modelling and is outside the scope of this paper.The deprojected disk has a full-width-half-maximum(FWHM) of 0.82 (cid:48)(cid:48) in the J8.9 filter and 0.35 (cid:48)(cid:48) -0.41 (cid:48)(cid:48) in the otherfilters. The larger size of the J8.9 image is due to the above-mentioned PSF smearing from uncertainty in the altitude axis ofthe telescope. The FWHM values for all the disks and filters arelisted in Table 3. From the disk FWHM and the PSF FWHM (thedi ff raction limit is 0 . (cid:48)(cid:48) − . (cid:48)(cid:48) depending on the filter), we cancalculate the true size of the emitting region, assuming that boththe data and the PSF are well described by Gaussian functions(e.g. Mariñas et al. 2011; van Boekel et al. 2004), as follows:FWHM disk = (cid:113) FWHM data2 − FWHM
PSF2 . (1)Due to the PSF smearing in the J8.9 image, we used the refer-ence PSF FWHM rather than the theoretical di ff raction limit forthis filter. Since the other data were observed with the NEAR in-strument, which thanks to its adaptive optics is expected to have a Strehl ratio of close to one (Kasper et al. 2017), the FWHM ofa point source PSF corresponds to the di ff raction limit. This canbe seen in the data of HD 163296, TW Hydra, and HD 36112,as is discussed in Sect. 3.2. The deconvolved FWHM of all re-solved sources and the corresponding 5 σ upper limits for un-resolved sources are also listed in Table 3. While spectroscopicdata show that the disk is more extended in PAH emission bands(van Boekel et al. 2004; Verhoe ff σ discrepancy between the J8.9 andthe PAH1 and ARIII deconvolved FWHM means the errorbarson the J8.9 image are probably underestimated, possibly due toa worsening of the smearing e ff ect as the night went on.Removing the PSF component along both axes also gives amore accurate inclination, since the semi minor axis of the diskis relatively more extended by the PSF than the semi-major axis.The calculated inclination is 47 ± ◦ . The projection e ff ect is notexpected to be as strong here, since even on the model data theresulting inclination was well within 1 σ of the input value. Theprojected inclination is in agreement with literature inclinationvalues of 42-50 ◦ . (Miley et al. 2019; Casassus & Pérez 2019;Jamialahmadi et al. 2018; Mendigutía et al. 2017; Pineda et al.2014; Avenhaus et al. 2014; Walsh et al. 2014). This value is thecombined inclination across all the available filters, except forJ8.9 due to the deformed PSF in this image. HD 163296 is unresolved in both filters and has FWHMs aroundthe di ff raction limit of the telescope which is 0.22 (cid:48)(cid:48) in the PAH1filter and 0.30 (cid:48)(cid:48) in the NEAR filter. This results in 5 σ upper lim-its of 7 AU and 6 AU, respectively. Previous mid-IR observationsbetween 8 µ m and 13 µ m have not resolved the disk, but set anupper limit on the FWHM of the emission region of 21 AU at11.7 µ m (Jayawardhana et al. 2001; van Boekel et al. 2005; Mar-iñas et al. 2011; Li et al. 2018). Our images of HD 163296 im-prove on the emission size upper limits by a factor of three. Article number, page 6 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars
Table 3.
FWHM of the disks in each filter is given in arcseconds. HD100546 is clearly resolved in all bands. HD 169142 and HD 100453 areresolved in both the PAH and NEAR bands, while HD 163296, TW Hy-dra, and HD 36112 are unresolved point sources. For resolved images,the FWHM after deconvolution is listed in AU. For unresolved images,the 5 σ upper limits are listed instead. Object Filter FWHM data ( (cid:48)(cid:48) ) FWHM disk (AU)HD 100546 J8.9 0.82 ± ± ± ± ± ± ± ± ± ± ± < ± < ± ± ± ± ± <
3. NEAR 0.297 ± < ± ± ± ± ± < ± ◦ , which is in agreement with previouslymeasured inclinations of 13 ± ◦ (Pérez et al. 2019; Pani´c et al.2008; Raman et al. 2006).TW Hydra is unresolved in our observations with upper lim-its of 3 AU in the PAH1 band and 49 AU in the NEAR band. Thehigh limit in the NEAR band is due to the data being taken withthe coronograph. While this allows for increased sensitivity forfinding planets, it also means that the extent has to be calculatedwith the o ff -axis chop and nod beams. Based on the PAH1 datataken the same night, the beams are expected to be smeared by ∼ µ m (Ratzka et al. 2007; Arnoldet al. 2012).HD 100453 is resolved in both bands. Similar to TW Hy-dra, the NEAR band images of HD 100453 were taken with thecoronograph, resulting in a 10% error in the extent of the emis-sion region. The di ff erence between the deconvolved PAH1 andNEAR band sizes suggests this might still be an underestimate.The disk has a calculated projected inclination of 35 ± ◦ , whichis in agreement with literature values of the inclination of 30-38 ◦ (Rosotti et al. 2020; Long et al. 2017; Benisty et al. 2017;Wagner et al. 2015).Finally, HD 36112 is unresolved, with a NEAR band upperlimit of the size of the emission region of 13 AU. This is an im-provement by almost a factor of 10 over previous observationswhich set an upper limit of 120 AU on the 11.7 µ m emission size(Mariñas et al. 2011).
4. Protoplanetary disk modelling with ProDiMo
We used the radiation thermo-chemical disk model P ro D i M o (Woitke et al. 2009; Kamp et al. 2010; Thi et al. 2011) tosimulate observations of the HD100546 system. P ro D i M o self-consistently and iteratively determines the physical and chem- . astro . rug . nl/~prodimo/ Table 4.
DIANA SED-fit parameters for the HD 100546 system used inthe P ro D i M o disk model. Parameters that were modified to improve themid-IR fit are included in parenthesis. Parameter Symbol ValueStellar Mass M ∗ M (cid:12) Stellar Luminosity L ∗ L (cid:12) E ff ective Temperature T e ff V . Fe . SiO . M d . × − M (cid:12) Inner Radius R in R out (cid:15) min µ mMaximum dust size a max µ mPAH abundance f PAH M d . × − M (cid:12) Inner Radius R in R out
600 AUTapering Radius R tap
100 AUCol. Density Power Index (cid:15) . min µ mMaximum dust size a max µ mPAH abundance f PAH ◦ Dust to Gas Ratio d / g ro D i M o performs a 2D continuum radia-tive transfer with a ray-based, long-characteristic, accelerated Λ -iteration method at every disk grid point to calculate the local ra-diation field J ν ( r , z ) (Woitke et al. 2009). The full radiative trans-fer methodology is described in Woitke et al. (2009). We adoptthe standard DIANA dust opacities as described in Woitke et al.(2016) and Min et al. (2016).The parameters for the HD 100546 disk model were derivedfrom the SED fitting work done as part of the European FP7project DIANA (Woitke et al. 2019). Parameters of the HD100546 disk and stellar model can be found in Table 4 and the2D gas density profile can be found in Fig. 3. The fitting was per-formed for a pre-Gaia distance of 103 pc (van den Ancker et al.1997). Further details regarding the disk modelling and SED fit-ting process can be found in Appendix A.As P ro D i M o finds formal solutions to the continuum ra-diative transfer during the calculation of the SED, the resultingmodelled intensity can be visualised as an image. P ro D i M o in-cludes only the e ff ect of isotropic scattering, and hence the pref- https: // dianaproject.wp.st-andrews.ac.uk / data-results-downloads / fortran-package / More information about the fitted stellar and disk parameters, the2D modelling results, and the predicted observables can be found at . st-and . ac . uk/~pw31/DIANA/DIANAstandard Article number, page 7 of 18 & A proofs: manuscript no. main
Fig. 3.
Gas density profile of the P ro D i M o HD 10056 disk model. Thedashed contour line traces the surface where the minimum optical ex-tinction A V in the combination of the vertical or radial direction is 1. Fig. 4.
Dust density profile of the P ro D i M o HD 100546 disk model.The light blue contour outlines the region where half of the total 9 µ memission originates. The dashed contour line traces the surface wherethe minimum optical extinction A V in the combination of the vertical orradial direction is 1. erential forward-scattering of light by larger dust grains is notrepresented realistically. As a result, the P ro D i M o model ap-pears brighter on the far side than on the near side and it can-not reproduce the observed asymmetry in brightness of actualdisks. While this e ff ect is cancelled out in the disk SED modeland radial intensity profile, it must be taken into considerationwhen comparing the model image to data on a per-pixel basis.The resulting P ro D i M o data cube was attenuated by multiplyingeach synthetic disk image with the VISIR and NEAR relative fil-ter transmission curves created with the VISIR imaging detectorand VISIR calibration unit, and then by the sky transmission ateach wavelength. Subsequently the data cube was flattened intoa single image for each filter. The images were then convolvedwith a reference PSF to simulate our observations. This was HD93813 for the J8.9 filter, HD 27639 for the ARIII filter, and theHD 163296 data for the PAH1 and NEAR filters. For the PAH2 . eso . org/observing/etc/bin/gen/form?INS . MODE=swspectr+INS . NAME=SKYCALC and PAH2_2 filters, reference PSFs were not available and thePSF from the ARIII filter was scaled to the new central wave-length and used instead.
5. Comparison to ProDiMo disk models
Figure 5 illustrates the resulting SED for variants of the fidu-cial P ro D i M o HD 100546 model between 7.5 and 10 µ m, alongwith the averaged flux of the J8.9 band observation. The VISIRobservations are included in black, as are the flux measured byAKARI and the spectrum from ISO (Malfait et al. 1998; Ishi-hara et al. 2010). Near 8.7 µ m, the observational data to whichthe SED was fit includes the ISO-SWS spectrum and a photo-metric data point from
AKARI with the S9W filter (Malfait et al.1998; Ishihara et al. 2010). While our data are in agreementwith previous observational data, the expected flux of the ba-sic P ro D i M o model falls outside the uncertainty interval. Weconsider both disk parameter modifications included and not in-cluded in the previously performed SED fitting process that mayimprove upon the local fit in the mid-IR without reducing thequality of the global fit.In our disk model, the continuum flux at 8.7 µ m is emit-ted largely from the surface of the inner disk between 1-4 AU,while in the outer disk the 8.7 µ m flux originates largely fromthe gap wall which is directly illuminated by the star and heatedto ∼
300 K (see Fig. 4). Modifying the location of the cavity’souter rim ( r ∈ of the disk outer zone) allowed us to reduce thetemperature of the gap wall and reduce the continuum emis-sion in the mid-IR. We find the optimal balance between movingthe gap outer wall further outwards and maintaining the qual-ity of the global fit occurs where the gap wall is moved out-wards from 19 to 22.3 AU. As demonstrated in Fig. 5 by theline r in = . µ m, ∼
60% has been explainedby the presence of amorphous olivine and crystalline forsteriteemission features with the remainder explained by PAHs (Mal-fait et al. 1998). We thus also consider further refinements to themid-IR fit by exploring the properties of the disk PAH popula-tion. These considerations can be found in Appendix B.Across the wavelength coverage of the
ISO-SWS spectrum,we reduced the sum of the squares of the ratio between the old fit F old ν and the new fit F new ν , that is Σ ( F new ν / F old ν ) , from 12.6 to 4.2.It should be noted that while dust settling allows for a variety ofaverage grain sizes across the vertical extent of the disk model,dust grains are not radially segregated by size in P ro D i M o , suchthat within our model’s disk zones, every grid column containsthe same underlying dust grain size distribution. Hence we cansolve for only one gap outer radius, rather than a radius for eachcorresponding grain size. Radial intensity profiles of all the disks in the sample in the dif-ferent filters were constructed by azimuthally averaging over thedeprojected disks for both the observations and the convolvedmodels and this is shown in Figs. 6 and 7. In all cases the ra-dial profile is dominated by the telescope PSF. The unresolvedsources show clear Airy rings in the images (see Fig. 2). TheAiry rings are less obvious in the resolved sources and the cen-tral disk of the Airy pattern is larger, but they are still visiblein the radial profiles. None of the profiles show signs of spirals,
Article number, page 8 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars F [ J y ] DIANA (fiducial) r gap,out = 22.3 au r gap,out + f PAH + m PAH
ISO-SWSTIMMI2VISIR J8.97.5 8.0 8.5 9.0 9.5 10.0 [ m] r e s i d u a l s ( r e l a t i v e ) Fig. 5.
Comparison between the fiducial P ro D i M o HD 100546 diskmodel (Woitke et al. 2019) and multi-parameter variants of the model.We include the observational VISIR data corrected for sky transmis-sion and additional observational data (Malfait et al. 1998; van Boekelet al. 2004; Ishihara et al. 2010). The grey filled area illustrates the J8.9filter response curve (arbitrary vertical scaling). Residuals between thevarious disk models and the
ISO-SWS spectrum are shown in the lowerpanel as the ratio between the model SED and the observed spectrum. rings, or other features in the extended disk. Although the mod-els do not include these previously observed features, this resultis still consistent with the models, which show that the mid-IRemission is dominated by the central regions and the outer re-gions where features have been detected at other wavelengthscontribute less than 5% of the flux at 8.7 µ m.For most models used in this comparison, the distance wasmeasured before the Gaia data release. With the release of theGaia data, it appears that these distances were o ff by around 10%in most cases (HD 100546, HD 163296, TW Hydra). For thesedisks, it was not necessary to rerun the model, as the di ff erencesbetween the old and new distances are small. Simply rescalingthe model to the new distance is su ffi cient to compare the extentof the disks. However, for HD 169142, the di ff erence betweenthe distance assumed in the model and the distance measured byGaia is more significant: The assumed distance is almost 30%too large. Because of this, the model was rerun with an adaptedluminosity for the new distance. The normalised radial flux distribution of both the real, depro-jected data in each filter and the corresponding simulated dataare shown in Fig. 6. The model and the data are in good agree-ment out to ∼
160 AU, where the noise starts to dominate the sig-nal. The peak in the noise in the data is caused by the sourcesubtraction in the chopping and nodding. The subtraction shad-ows are located at ∼
500 AU (4.5 (cid:48)(cid:48) ) in the four NEAR filters andat ∼
900 AU (8 (cid:48)(cid:48) ) in the J8.9 filter. In Fig. 7 we can compare thedi ff erent filters to each other for the observed and synthetic data.In both cases the shorter wavelength filters PAH1 and ARIII re-sult in narrower profiles with a smaller FWHM than the longerwavelength filters PAH2 and PAH2_2. Due to the smearing ofthe PSF, the J8.9 filter profile is much wider in both cases. Theresiduals from subtracting the model curves from the data are N o r m . f l u x PAH1 10 ARIII10 N o r m . f l u x PAH2 10 Separation (au)PAH2_210 Separation (au)10 N o r m . f l u x J8.9 data model
Fig. 6.
Radial profile of the HD 100546 protoplanetary disk in thePAH1, ARIII, PAH2, PAH2_2, and J8.9 filters. The profile from thedata is shown in blue with the 1 σ range in light blue. The profile fromsynthetic observations based on the P ro D i M o model is shown in orangewith the 1 σ range indicated in lighter orange. shown in Fig. 8. The errorbars in the image represent the 1 σ error. The residuals show that the synthetic data is a good repre-sentation of the real data. The residuals at larger separations are0 because the chopping and nodding process removes the back-ground emission from the data and the model does not includesky or instrument background emission. Previous observations in near-IR and sub-millimetre wave-lengths show that HD 163296 has multiple bright rings (e.g.Garufi et al. 2014; Isella et al. 2016; Isella et al. 2018). TheP ro D i M o model does not include rings, but instead assumes aflared, optically thick inner region up to 0.02 (cid:48)(cid:48) and a shadowedouter region beyond that. As a result, it predicts that 95% of theflux is contained within a radius of 0.01 (cid:48)(cid:48) in the PAH1 band andwithin 0.04 (cid:48)(cid:48) in the NEAR band. This makes the emitting regionmuch smaller than in the case of HD 100546, where there is acavity and the inner rim of the outer disk also contributes to theflux. It is also entirely consistent with an unresolved image. ALMA observations have detected three bright rings between0.2 (cid:48)(cid:48) and 0.6 (cid:48)(cid:48) (45 −
80 AU) in the disk around HD 169142 (Pérezet al. 2019). Again, the model does not include the rings, butinstead divides the disk into an inner and an outer zone with a gapat 0.1 (cid:48)(cid:48) (22 AU), which is consistent with the inner gap seen atother wavelengths. Assuming the observed disk is described by aGaussian function, the apparent size as defined by P ro D i M o (theradius containing 95% of the flux) corresponds to the 2 σ radiusof the Gaussian, which is larger than the FWHM, which onlycontains half the flux. After deconvolution, the apparent size ofHD 169142 is 24 ± ± Article number, page 9 of 18 & A proofs: manuscript no. main N o r m a li s e d f l u x HD 100546 data
HD 100546 model
PAH1ARIIIPAH2PAH2_2J8.9
Fig. 7.
Radial flux profile of the HD 100546 protoplanetary disk in thePAH1, ARIII, PAH2, PAH2_2, and J8.9 filters, with the real data pro-files on the left and the synthetic data profiles on the right. The shadedareas indicate 1 σ errors for the data and confidence intervals for themodels. For the model profiles, these intervals come from the PSF con-volution and the azimuthal averaging and deprojection. In both the dataand the model, it can be seen that the radial extent at 1 / th the max-imum flux is smaller for the shorter wavelength filters (PAH1, ARIII)than for the larger wavelength filters (PAH2, PAH2_2). This is expectedas the PSF is larger for larger wavelengths. The J8.9 data, both real andsynthetic, remain far more extended due to the smeared PSF. The HD 169142 model has an apparent size of 43 and 45 AUin the PAH1 and NEAR bands. While this is approximatelyconsistent with the observed apparent size in the NEAR band,there is a discrepancy with the smaller PAH band observation.This is consistent with observations by Okamoto et al. (2017),who find that the size of the emitting region is much smallerat 8.6 and 8.8 µ m than it is at 12.6 µ m. They conclude that atwavelengths smaller than 9 µ m, the inner disk and halo domi-nate; whereas, at wavelengths larger than 9 µ m, the inner wallof the disk dominates which results in a larger observed size.Modelling performed by Maaskant et al. (2014) suggests thatgas flowing through disk gaps can contribute significantly to theobserved ionised PAH emission. This could manifest as an in-crease in emission at ∼ µ m relative to ∼ µ m, correspondingto the angular size of a gap. If the neutral PAH emission primar-ily originates from the gap wall, we would expect a correspond-ingly smaller emitting region for the predominantly ∼ µ m PAHflux. This di ff erence is not reproduced by the model, leading to amismatch with the data in the PAH band. This can be due to thecomplete lack of gas and dust in the model gap and hence lackof associated emission.The previously derived inclination of 13 ± ◦ is consistentwith the model value of 13 ◦ . It is also consistent with previ-ous literature (Pérez et al. 2019; Pani´c et al. 2008; Raman et al.2006). Studies in near-IR and sub-millimetre have found six gaps lo-cated between 0.11 (cid:48)(cid:48) and 0.84 (cid:48)(cid:48) (6 −
44 AU) (Tsukagoshi et al.2016; Andrews et al. 2016; van Boekel et al. 2017). The modelassumes an optically thin inner region corresponding to the innergap and a dense outer region for the rest of the disk. All the emis-sion in both bands is predicted to be from this thin inner regionand the inner wall of the outer disk. The other gaps are not ex-pected to be visible as they are further out in the disk, where thereis no more emission. This means that there is an apparent size of3-4 AU in both filter bands and this is consistent with the obser-
HD 100546 residuals
PAH10.20.0 ARIII0.20.0 N o r m a li s e d f l u x r e s i d u a l s PAH20.20.0 PAH2_20 25 50 75 100 125 150 175 200Separation (AU)0.20.0 J8.9
Fig. 8.
Residuals from subtracting the radial profile of the syntheticdata from that of the observed data in each of the observed filters. Theerrorbars indicate 1 σ uncertainties. The residuals all being within 1 σ of 0 show that the model represents the data well. vations being unresolved. More recent observations also suggestthe central optically thin region may be much smaller than in themodel, which would shrink the expected apparent size (e.g. vanBoekel et al. 2017; Andrews et al. 2016). Deconvolving the data results in apparent sizes of 7 ± ± µ m in ISO data. We therefore expect the flux in the PAH1band to be dominated by the continuum emission. The emissionin both bands is well inside the radius where spiral arms havebeen found and this suggests that HD 100453 follows the othertargets in the sample in which the mid-IR emission is dominatedby the central regions. Since the outer disk starts at 17 AU, thePAH emission seems to come from inside the gap and the NEARband emission includes the inner wall of the disk which is heatedby the star, similar to what is seen in HD 169142.
HD 36112 has a large cavity, with an outer disk that has rings,clumps, and spiral arms (e.g. Dong et al. 2018; Wagner et al.2019). However, in our observations, the cavity is unresolved.Since the cavity has a radius of 0.2 (cid:48)(cid:48) and the upper limit for the95% flux radius is 0.07 (cid:48)(cid:48) , this means that most of the emissioncomes from inside the cavity and not from the inner rim of theouter disk, unlike the NEAR filter emission of the other sources.
6. Companions
The proposed companions of the disks in this sample are poten-tial hosts to circumplanetary disks, which thus far have only beententatively identified in the PDS 70 system (Keppler et al. 2019;Christiaens et al. 2019; Ha ff ert et al. 2019; Isella et al. 2019).To search for planetary companions and associated dust concen-trations in the disk, the circularised PSF subtraction describedin Petit dit de la Roche et al. (2020) was applied to the data.This method creates an individual reference PSF from the datafor every nod-subtracted image by azimuthally averaging it. The Article number, page 10 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars
Fig. 9.
Left:
Mapped 5 σ flux limits of the HD 100546 PAH1 data, where the disk image is the most elliptical. The shape of the emitting regiondoes not significantly influence the flux limits, especially beyond 1 (cid:48)(cid:48) where the data are background limited. Right:
HD 100546 PAH1 data withsources injected at di ff erent separations and position angles at 5 sigma. Most of the sources are clearly visible. Fig. 10. σ flux limits of potential companions to three targets compared to the CPD model described in Table 5 inserted in the circumstellar disk. Left: limits for the di ff erent observations of HD 100546 in PAH1 (blue), ARIII (orange), PAH2 (green), PAH2_2 (red), and J8.9 (purple) filters.The black line indicates the estimated flux as a function of radial separation for our fiducial CPD model as described in column 2 of Table 5. Onlyone line is included as the model values are similar across the di ff erent filters. The increase at 7 (cid:48)(cid:48) in the J8.9 filter and at 4 (cid:48)(cid:48) in the other filter are theresults of chopping and nodding shadows. Middle:
Limits for the observations of HD 163296 in the PAH1 (blue) and NEAR (pink) filters, alongwith the expected flux of the same CPD in the HD 163296 disk.
Right:
The same as the middle figure, but for HD 169142. resulting rotationally symmetric PSF was then subtracted fromthe original data to remove the radially dependent stellar flux.This was decided upon because there is not su ffi cient rotationin the images to do angular di ff erential imaging and most of thedata do not have reference stars available for standard PSF sub-traction. Standard PSF subtraction would also not subtract anyspatially extended disk emission. Subtracting a circularly sym-metric PSF from an elliptical disk image does leave residuals,but the bulk of the disk emission ( > (cid:48)(cid:48) , beyond which the background dominates andthe shape of the emitting region becomes irrelevant. An example of this can be seen in the left panel of Fig 9, where the limits aremapped for HD 100546 in the PAH1 filter, which has the mostelliptical image of our entire dataset. While none of the proposedcompanions are detected in any of the disks, it is possible to set5 σ upper limits on the fluxes of any possible companions, basedon the residual noise at each possible location. A limit of 5 σ was chosen, since injected 5 σ sources were clearly recoveredin the reduced data, as can be seen in the right panel of Figure9. The only exception is the source directly to the south of thestar, which, although still present, is less clear due to its proxim-ity to one of the shadows induced by the nodding. However, thea ff ected areas around these shadows are small.Fig. 10 and 11 show the resulting flux limits, with Fig. 10 in-cluding the flux of a model planet with a circumplanetary disk, Article number, page 11 of 18 & A proofs: manuscript no. main F l u x ( m J y ) TW Hya PAH1TW Hya NEARHD 100453 PAH1HD 100453 NEARHD 36112 NEAR
Fig. 11.
Observational limits on potential companions to TW Hya(grey), HD 100453 (yellow-green), and HD 36112 (turquoise) in thePAH1 (solid lines) and NEAR (dashed lines) filters. The increase at 4 (cid:48)(cid:48) is the result of shadows from the chopping and nodding in the observa-tions. which is discussed in the next section. The obtained limits are ofthe order of a few millijanskys between 1 (cid:48)(cid:48) and 3.5 (cid:48)(cid:48) separationup to a few tens of millijanskys at 0.5 (cid:48)(cid:48) . This is more sensitivethan any previous mid-IR imaging observations by a factor of10-100. Beyond 3.5 (cid:48)(cid:48) , the limits are dominated by the shadowsinduced by the chopping and nodding procedure in the observa-tions. The di ff ering sensitivities between objects with the sameintegration times are the result of di ff erent observing conditionsinfluencing the data quality of the di ff erent targets. The presence of planetary accretion and a CPD or circumplan-etary dust envelope can act to significantly increase the mid-IRluminosity of a putative companion (e.g. Zhu 2015). To deter-mine our own mid-IR observational limits for the planet can-didates with accompanying CPDs, we explored a grid of CPDmodels using P ro D i M o . Our model grid consists of a range ofpossible planet CPD masses, CPD dimensions, dust grain sizedistributions, and planet luminosities. We consider planetary masses of 1 to 10 M J , with correspond-ingly sized CPDs defined by the planet’s Hill radii. As CPDscould be tidally truncated to ∼ / Hill (Mitchell & Stewart 2011;Oberg et al. 2020), we set our CPD surface density tapering ra-dius to the point at which the surface density begins to declineexponentially at R
Hill / Hill .We considered a range of CPD masses relative to the planetmasses M CPD = − − − M p , and a range of planetary lu-minosities corresponding to various stages of accretion such that L p = − − − L (cid:12) (Mordasini et al. 2012). Marley et al. (2007)found that a 10 M J planet in a ’hot start’ evolution scenario candecline monotonically in luminosity from an initial ∼ × − L (cid:12) to ∼ × − L (cid:12) within 5 Myr. In the core accretion case, theyfound a peak luminosity during runaway accretion of > − L (cid:12) which lasts ∼ × yr, rapidly declining to ∼ × − L (cid:12) by 3 Myr. Given that the planetary luminosity is expected to peakonly briefly at or above L p ∼ − L (cid:12) , we consider the case of L p = − L (cid:12) to be the most optimistic detection scenario.Pressure bumps at gap edges are suspected to act as filters fordust grain size, preventing the accretion of grains significantlylarger than 10 µ m onto planets within the gap (Rice et al. 2006).We thus also considered CPDs where the dust grain size popula-tion is limited to maximum sizes of 100 and 10 µ m.A companion orbiting within an optically thin region of thecircumstellar disk can be exposed to significant UV radiationfrom its host star (Oberg et al. 2020). Photons of energy 6-13.6 eV are known as FUV and can e ffi ciently heat disk surfaces.The significant FUV luminosity of the host star can act to heatthe surface of the CPD and increase its IR luminosity. We param-eterised the FUV flux with the Draine field G = . × − ergcm − s − , which was integrated from 6-13.6 eV (Habing 1968).We extracted the G field intensity using P ro D i M o from the re-sults of the 2D radiative transfer within the DIANA circumstel-lar disk models and applied this as a UV background field toour own CPD models. Given that dust is the dominant source ofopacity in the UV, it should be noted that the gaps in the DIANAdisk models (see Fig.4 for the HD100546 dust structure) are freeof dust and do not contribute to the UV opacity. We extracted the planet and CPD flux from the SEDs producedby the P ro D i M o continuum radiative transfer and weighed itacross the filter response curves. This flux represents the ide-alised total flux emitted by the unresolved companion, uncon-volved with the observational PSF. We find that for high plane-tary luminosities ( > − L (cid:12) ), the mid-IR flux is dominated bythe planet itself, whereas the CPD only contributes 3 −
6% of thecombined emission largely independent of CPD properties.For our disk models, the size of the CPD as estimated byits Hill stability and the strength of the background FUV fieldboth vary in predictable ways. For a given CPD model, our pa-rameter grid exploration thus allowed us to fit for the resultingplanet and CPD flux given an arbitrary radial separation fromthe star. As the vertical dust opacity at arbitrary wavelengths wasalso calculated as part of our model radiative transfer for variouscircumstellar disks, we were able to determine the radial depen-dence of the extinction to the midplane as well. We solved forthe dust column density as a function of the viewing inclinationfor each radial position in the disks, and from this we derived theresulting 9 µ m optical depth. The black line in Fig. 10 representsthe resulting expected flux of the planet and CPD model in theJ8.9 filter for a 10 M J planet with a CPD of mass 10 − M p as de-scribed in Table 5. The line was derived from a fit performed tothe J8.9 flux of our model grid of CPDs in which the backgroundFUV radiation field, the disk size, and extinction to the midplanewere simultaneously varied as a function of radial separation, al-though the predicted flux is relatively flat for planets found out-side of the optically thick regions of the circumstellar disks. Forlow radial separations, the background FUV field heats the CPDsurface and results in increased mid-IR emission. The CPD sizegrows with increasing distance from the star as the companion’sHill sphere increases correspondingly; however, as the majorityof the CPD mid-IR emission originates from the innermost re-gions of the CPD, this contribution becomes negligible at largeseparation. The flux of our CPD models in the other filters is sim-ilar, varying for non-pathological model cases by at most ∼ Article number, page 12 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars
While previous estimates of the age of HD 100546 indicate anolder ( ∼
10 Myr) system (van den Ancker et al. 1997), Fairlambet al. (2015) derived an age of 7 . ± .
49 Myr and an accretionrate of ˙ M ≈ − M (cid:12) yr − . The mass of the HD 100546 innerdisk was fit to be 8 . × − M (cid:12) (Woitke et al. 2019), thus re-quiring continuous replenishment from the outer zone across thegap. The plausibility of an actively fed circumplanetary accre-tion disk is thus supported by the ongoing presence of radiallyevolving dust within the circumstellar disk (Marley et al. 2007;Mordasini et al. 2012).We considered companions placed in the midplane at mul-tiple radial separations from the star to study the influence ofthe background radiation field and circumstellar dust extinc-tion on the predicted flux. We considered the properties of theplanet candidate HD 100546b described by Quanz et al. (2015),which was found at a radial separation of 53 ± R = . + . − . R J with T = + − for a luminosity L = . + . − . × − L (cid:12) . As the addition of a CPD may produce an emis-sion signature diverging significantly from a single-temperatureblackbody, we loosened the constraints on the temperature andemitting area. For a 2.5 M (cid:12) star, a planet of 1, 5, or 10 M J at55 AU has a Hill radius of 2.77, 4.73, or 5.96 AU, respectively.We considered three cases in detail: a planet immediately inte-rior to the outer gap wall at 18 AU, a planet embedded within theouter gas and dust disk at 55 AU, and a wide-separation planet inthe optically thin region of the PPD at 100 AU, with correspond-ingly sized CPD outer radii, maximum dust grain sizes, FUVbackgrounds, and optical depths to the midplane (see Table 5).While the CPD size, as set by the Hill radius, only varies by afactor of 100 across the disk surface from 5-500 AU, the back-ground UV radiation field varies more dramatically by a factor > .At the radial location of the 55 AU planet candidate, we ex-tracted an FUV flux of G = . in the midplane from the re-sults of our circumstellar disk model radiative transfer. At 5 AUin the shadow of the inner disk, we find G = . and at 18 AU G = . . The maximum G within the gap is found to be3 × . The gas component of a CPD experiencing such irradi-ation acquires an optically thin heated envelope with a temper-ature of around 5000 K at z / r ∼ .
4. The ∼
70 K optically thicksurface below this envelope gives rise to significant re-radiatedemission peaking at 30-50 µ m. The short-wavelength tail of thiscomponent contributes non-negligibly to the J8.9 flux across theentire CPD surface for G > .From the HD 100546 disk model dust density distributionand dust opacities, we determined the optical depth to the mid-plane along the line-of-sight to the observer across the J8.9 bandto determine extinction at arbitrary radii. While emission arisingfrom planets inside the gap would be largely unextincted, imme-diately outside of the gap we find a maximum optical depth τ J8 . of 5.6. The disk becomes optically thin at 8.7 µ m only outside of82 AU. We find that at the large separations where our sensitivityis maximal at a >
160 AU, τ µ m is at most 0.18 and τ ∝ a − . .The model planet with a mass of 10 M Jup and a luminosity of10 − L (cid:12) would have been detected in the J8.9 data beyond thisradius and in PAH2 between 2 (cid:48)(cid:48) and 3 (cid:48)(cid:48) . Hence, our new mid-IRimaging data prove that no such massive, luminous planets existin the HD 100546 system at radii larger than 160 AU from thecentral star. A companion with a luminosity of 10 − L (cid:12) would bemarginally detectable at angular separations of 4-5 (cid:48)(cid:48) only. We used a single best-case representative planet and CPD to de-rive detection limits for the other observed systems as a functionof separation. The model CPD mid-IR flux levels are constant atlarge radii, because at large separations the UV radiation emittedby the star no longer significantly contributes to the heating andre-radiation of the CPD. The fact that the CPD is free to physi-cally increase in size as the planet’s Hill radius increases also nolonger acts to increase the flux, as for the optically thick CPDswe consider, the planet acts only to heat the innermost regionsof the CPD, from which the majority of the 9 µ m emission orig-inates.For HD 163296, we excluded a 10 M Jup , 10 − L (cid:12) companionbetween 1.5 (cid:48)(cid:48) and 3.5 (cid:48)(cid:48) , as it would have been observed in bothfilters. For HD 169142, TW Hydra, and HD 36112, we excludedit beyond 1 (cid:48)(cid:48) . HD 100453 is the only system in which it wouldremain undetected. In previous work, the planet candidate HD 100546 b at 55 AUseparation is the only companion that has had its putative CPDconstrained in mass to 1.44 M ⊕ (or 2 . × − M p for a planetmass 1.65 M Jup ) in the optically thin case, and a size of 0.44 AUin radius for the optically thick case, although this rests on as-sumptions regarding the grain size population of the CPD andthe ratio between planetary and CPD mass (Pineda et al. 2019).ALMA observations of HD 100546 at 870 µ m set a 3 σ limit of198 µ Jy for any planet candidate (Pineda et al. 2019) with whichwe can further constrain any CPD’s longwave emission.We find that for our fiducial CPD surrounding a 10 M
Jup planet of 10 − L (cid:12) , we overpredicted the upper limit set byALMA observations at 870 µ m by a factor of 13. When the fidu-cial CPD is modified with a maximum grain size of 10 µ m, thisoverprediction is reduced by a factor of ∼
2. Our planet and CPDmodels can be brought into agreement with the ALMA flux lim-its by reducing the mass of the CPD relative to the planet or byreducing the dust-to-gas ratio. We find that while the 9 µ m fluxof the CPDs is largely insensitive to their mass, the continuumflux in ALMA band 10 is primarily dependent on our CPD mass,radius, and dust-to-gas ratio owing to the emission region cor-responding to cooler dust at larger separation from the planet(Rab et al. 2019). For a fixed radius, dust-to-gas ratio, maxi-mum and minimum dust grain size, and grain size power law, the870 µ m flux is proportional to the CPD mass as F µ m ∝ M . for the range M CPD = − − − M p . We find that the maxi-mum CPD mass allowed by the constraint is 3 . × − M (cid:12) . Asmaller, optically thick CPD of a higher mass still satisfies theconstraint. We find that a modification to our fiducial CPD of amass > . × − M (cid:12) with a tapering radius of 0.2 AU and anouter radius of 0.6 AU has a 870 µ m flux of 190 µ Jy and wouldthus satisfy the constraint set with ALMA. This places no addi-tional constraints on our 9 µ Jy flux prediction, as the mid-IR fluxis instead primarily dependant on the planet’s luminosity and theCPD’s inner radius.
7. Discussion and conclusions
We analysed images of HD 100546 in five di ff erent mid-IR fil-ters and a further five young stellar objects in the PAH1 andNEAR infrared filters with the VISIR instrument and its upgradeNEAR. The resolved disks had their FWHMs and inclinations Article number, page 13 of 18 & A proofs: manuscript no. main
Table 5.
HD100546 candidate planets and CPD model parameters for our optimistic detection scenario (parameters listed above the first hori-zontal divider) for a variety of radial separations (parameters below the first horizontal divider) and associated J8.9 band predicted fluxes. Dustcomposition is identical to that listed in Table 4.
Parameter Symbol 18 AU 55 AU 100 AUPlanet Mass [ M J ] M p
10 10 10Planet Luminosity [ L ∗ ] L p − − − CPD Mass [ M p ] M CPD − − − CPD Inner Radius [AU] R CPD , in µ m ] a min (cid:15) d / g H . a p
18 55 100CPD Tapering Radius [AU] R tap , CPD R out , CPD µ m ] a max
10 3000 3000FUV background G . . µ m τ ∼ µ m flux (extincted) [mJy] F P , µ m flux (unextincted) [mJy] F P , PAHI ArIII J8.9 PAH2 PAH2_27.510.012.515.017.520.022.5 c o m p a n i o n f l u x [ m J y ] fiducial M CPD = 10 M p a max = 100 m a max = 10 m a max = 10 m + r in = 0.04 au a max = 10 m + G = 10 Fig. 12.
Model companion (planet and CPD) unextincted flux estimates.The ‘fiducial’ case is described by the planet and CPD parameters foundin column 2 of Table 5 at 55 AU for the HD 100546 system. We alsoconsider a variety of maximum dust grain sizes a max , CPD mass M CPD ,CPD inner radius r in , and background FUV radiation field strength G . determined. HD 100546 has a FWHM of 28-61 AU across fivedi ff erent filters, a projected inclination of 44 ± ◦ , and a projectedposition angle of 130 ◦ . HD 169142 has FWHMs of 29 AU and41 AU in the PAH1 and NEAR filter bands, respectively, anda projected inclination of 13 ± ◦ . HD 100453 has a FWHM of9 AU in the PAH1 band and 21 AU in the NEAR band and an in-clination of 35 ± ◦ . The observed values are consistent with theDIANA circumstellar disk models and previous observations ofthe sources. We set upper limits of 6 AU and 7 AU on the sizeof the emission region of HD 163296 in the PAH1 and NEARfilter bands, respectively, thus improving previous limits by afactor of three. We set upper limits of 3 AU and 7 AU on TWHydra in the same filters, which is consistent with previous ob-servations. Finally, we set an upper limit of 13 AU on the size of the NEAR filter emission of HD 36112, which is an improve-ment over previous values of a factor of 10. The fact that wedid not resolve these targets is also consistent with the DIANAP ro D i M o models (Woitke et al. 2019). Because of the methodby which the variety of observational data were weighted duringthe original fitting procedure performed by Woitke et al. (2019),and because of the non-complete set of disk model parametersfor which the fits were performed, localised improvements to theSED were still possible. After a minimal adjustment of the HD100546 disk model gap geometry, an examination of the disk ra-dial profile showed that our P ro D i M o model was a good matchfor the data and that it reproduces the radial profile of the diskto within 1 σ without the need to include a companion object. Inall cases, the mid-IR emission originates from the central area ofthe disk from the most highly irradiated areas: unresolved innerdisks and / or the inner rims of the outer disks.Given our new flux estimate for the HD 100546 system, wehave improved upon the global SED fit from 2-18 µ m by simulta-neously increasing the gap outer edge from 19.3 AU to 22.3 AU,increasing the abundance of PAHs in the outer disk relative tothe ISM from 2.8 × − to 3.4 × − , and replacing the repre-sentative PAH circumcoronene with coronene. The details of thePAH properties fitting can be found in Appendix B. Given thatthe spectral properties of alternative dust compositions have notbeen thoroughly explored nor the marginal improvement of thedetailed PAH fit, we tend to favour the simple modification ofonly the disk gap geometry. The χ statistic between the modelSED and the ISO-SWS spectrum for 2-18 µ m reduces from 588to 278 when the inner radius is increased to 22.3 AU. It shouldbe noted that increasing the model gap outer radius would act toincrease the tension with the location of the dust continuum gapedge observed with ALMA at 16-21 AU (Pérez et al. 2020), al-though as ALMA traces millimetre-sized grains, this may not beinconsistent. Additionally, the model gap outer radius is the oneparameter that we adjusted which was previously fit by meansof a genetic algorithm (Woitke et al. 2016; Kamp et al. 2017;Woitke et al. 2019; Dionatos et al. 2019).We produced planet and CPD flux estimates using the ther-mochemical disk modelling code P ro D i M o for the VISIR filters Article number, page 14 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars with a variety of CPD parameters, finding that in the absenceof extreme external FUV radiation fields, the maximum unex-tincted flux in the J8.9 band is expected to be ∼
15 mJy for aCPD with an inner radius of 0.04 AU and a maximum dust grainsize of 10 µ m. We find that this flux is largely dependent on theplanet properties and not on those of the circumplanetary disk.The CPD is found to contribute 3 − , at most, of the com-panion flux at 9 µ m. The CPD contribution at 9 µ m is greatestwhen the maximum grain size is reduced to 10 µ m and the CPDis irradiated by a significant FUV field of G ≥ .Such conditions are found within the gap of the HD 100546disk, where we determined that the G field strengths up to3 × , despite the presence of the inner disk. A planet and CPDwithin the gap at 18 AU, while more gravitationally truncatedthan our test cases at 55 and 100 AU, is unobscured by dust andwe expect F J8 . = . µ m emis-sion of the CPD is largely una ff ected for G ≤ , it rises pre-cipitously above this, and for a G = we find F J8 . = . (cid:48)(cid:48) and thus be unresolved in our observation,the contribution to the flux of the star and circumstellar disk (31 ± ff ective emission region isgreatly reduced (Oberg et al. 2020).For our a =
55 AU HD 100546 companion test case, wefind F J8 . = . a =
100 AU com-panion case, we find F J8 . = . (cid:12) ,above which we would have detected any companion.In the HD 100546 system, we rule out the presence of plan-etary mass companions with L > . (cid:12) for a >
180 AU.We find that the contribution of a planet and CPD would stillbe of the order of the uncertainties inherent in the model, as rela-tively minor modifications to the HD 100546 gap dimensions (anincrease of 2-3 AU in the outer radius) produce changes in ex-pected continuum flux of 7-10 Jy at 9 µ m. We place no stringentconstraints on the planetary mass, CPD radius, or CPD grain sizedistribution. In the HD 169142, TW Hydra, and HD 100453 sys-tems, we can exclude companions with L > − L (cid:12) beyond 1 (cid:48)(cid:48) .We consider whether the lack of detection of wide-separation( a >
50 AU) planetary mass companions (PMCs) of mass < M J in the five studied systems is remarkable. While the pres-ence of a dusty CPD may act to enhance the observability of acompanion, it has been found that rapid dust evolution in CPDsof isolated wide-separation PMCs could act to suppress the dust-to-gas ratio of CPDs on short timescales ( d / g ≤ − after 1Myr), rendering a continuum detection more challenging (Pinillaet al. 2013; Zhu et al. 2018; Rab et al. 2019). Sub-stellar com-panions have been detected in wide orbits around young stars(Neuhäuser et al. 2005; Ireland et al. 2011; Bryan et al. 2016;Naud et al. 2017; Bohn et al. 2019). It has been suggested thatsuch objects may form in situ by the fragmentation of massive,self-gravitating disks (Boss 1997, 2011; Vorobyov 2013) by thedirect collapse of molecular cloud material (Boss 2001), or bycore- or pebble accretion (Lambrechts & Johansen 2014) andsubsequent outwards scattering by an interaction with other mas- sive planets (Pollack et al. 1996; Carrera et al. 2019). In the lat-ter case, a detection of a wide-separation PMC may thus directlyimply the presence of additional massive planets in the inner sys-tem.Bowler (2016) suggests that around single, young (5-300 Myr) stars, 5–13 M J companions at separations of 30-300 AU occur 0 . + . − . % of the time. With VLT / NaCo, Vigan et al.(2017) found that 0.5-75 M J companions at separations of 20-300 AU are found around 0 . − .
7% of stars, and with the Gem-ini Planet Imager Exoplanet Survey, Nielsen et al. (2019) foundthat 5-13 M J companions with separations of 10-100 AU occuraround 9 + − % of stars. Our sensitivity at the limiting angular res-olution restricted our search to relatively wide separation com-panions ( a >
160 AU). Given the PMC occurrence rate of Bowler(2016), we expect an absolute upper bound of ∼ . + . − . % prob-ability of a single detection in our sample, assuming a perfectdetection e ffi ciency from 30-300 AU. In this context, it is di ffi -cult to make new conclusions regarding the prevalence of wide-separation PMCs in our observed systems given the relativelylow a priori likelihood of detection and the relatively large com-panion luminosity (10 − − − L (cid:12) ) necessitated. We were ableto set an upper limit to the occurrence rate for wide-separationPMCs with a luminosity ≥ − of ≤ .
2% at 68% confidence.Future observations with METIS (Brandl et al. 2018) on theELT are expected to achieve ten times better sensitivities thanNEAR and 40 times better sensitivities than VISIR at the samewavelengths, as well as improving the spatial resolution by afactor of 5, allowing for one to image more close in companions.MIRI on JWST is expected to achieve 250 times better sensitiv-ities than NEAR and 1000 times better sensitivities than VISIR.Both will be able detect the known companions to all six targets.
Acknowledgements.
This project was made possible through contributions fromthe Breakthrough Foundation and Breakthrough Watch program, as well asthrough contributions from the European Southern Observatory.
References
Andrews, S. M., Wilner, D. J., Zhu, Z., et al. 2016, ApJ, 820, L40Arnold, T. J., Eisner, J. A., Monnier, J. D., & Tuthill, P. 2012, ApJ, 750, 119Avenhaus, H., Quanz, S. P., Meyer, M. R., et al. 2014, ApJ, 790, 56Benisty, M., Stolker, T., Pohl, A., et al. 2017, A&A, 597, A42Biller, B. A., Males, J., Rodigas, T., et al. 2014, The Astrophysical Journal, 792,L22Bohn, A. J., Kenworthy, M. A., Ginski, C., et al. 2019, Monthly Notices of theRoyal Astronomical Society, 492, 431Boss, A. P. 1997, Science, 276, 1836Boss, A. P. 2001, The Astrophysical Journal, 551, L167Boss, A. P. 2011, ApJ, 731, 74Bouwman, J., de Koter, A., Dominik, C., & Waters, L. B. F. M. 2003, A&A, 401,577Bowler, B. P. 2016, PASP, 128, 102001Brandl, B. R., Absil, O., Agócs, T., et al. 2018, in Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series, Vol. 10702, Ground-based and Airborne Instrumentation for Astronomy VII, 107021UBrittain, S. D., Carr, J. S., Najita, J. R., Quanz, S. P., & Meyer, M. R. 2014, ApJ,791, 136Bryan, M. L., Bowler, B. P., Knutson, H. A., et al. 2016, ApJ, 827, 100Carmona, A., van den Ancker, M. E., Henning, T., et al. 2008, A&A, 477, 839Carrera, D., Raymond, S. N., & Davies, M. B. 2019, A&A, 629, L7Casassus, S. & Pérez, S. 2019, ApJ, 883, L41Christiaens, V., Cantalloube, F., Casassus, S., et al. 2019, ApJ, 877, L33Cugno, G., Quanz, S. P., Hunziker, S., et al. 2019, A&A, 622, A156Currie, T., Cloutier, R., Brittain, S., et al. 2015, ApJ, 814, L27Currie, T., Muto, T., Kudo, T., et al. 2014, ApJ, 796, L30de Wit, W.-J. 2020, personal communication . stsci . edu/jwst/science-planning/proposal-planning-toolbox/sensitivity-and-saturation-limits Article number, page 15 of 18 & A proofs: manuscript no. main
Dionatos, O., Woitke, P., Güdel, M., et al. 2019, A&A, 625, A66Dong, R., Liu, S.-y., Eisner, J., et al. 2018, ApJ, 860, 124Dong, R., Zhu, Z., Rafikov, R. R., & Stone, J. M. 2015, The Astrophysical Jour-nal, 809, L5Dorschner, J., Begemann, B., Henning, T., Jaeger, C., & Mutschke, H. 1995,A&A, 300, 503Doucet, C., Pantin, E., Lagage, P. O., & Dullemond, C. P. 2006, A&A, 460, 117Dullemond, C. P. & Monnier, J. D. 2010, ARA&A, 48, 205Espaillat, C., Muzerolle, J., Najita, J., et al. 2014, in Protostars and Planets VI,ed. H. Beuther, R. S. Klessen, C. P. Dullemond, & T. Henning, 497Fairlamb, J. R., Oudmaijer, R. D., Mendigutía, I., Ilee, J. D., & van den Ancker,M. E. 2015, MNRAS, 453, 976Fedele, D., Bruderer, S., van den Ancker, M. E., & Pascucci, I. 2015, ApJ, 800,23Fedele, D., Carney, M., Hogerheijde, M. R., et al. 2017, A&A, 600, A72Follette, K. B., Rameau, J., Dong, R., et al. 2017, AJ, 153, 264Francis, L. & van der Marel, N. 2020, ApJ, 892, 111Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2020, arXiv e-prints,arXiv:2012.01533Garufi, A., Quanz, S. P., Schmid, H. M., et al. 2014, A&A, 568, A40Gonzalez, J.-F., van der Plas, G., Pinte, C., et al. 2020, MNRAS, 499, 3837Grady, C. A., Muto, T., Hashimoto, J., et al. 2013, ApJ, 762, 48Grady, C. A., Woodgate, B., Heap, S. R., et al. 2005, ApJ, 620, 470Gratton, R., Ligi, R., Sissa, E., et al. 2019, A&A, 623, A140Guidi, G., Ruane, G., Williams, J. P., et al. 2018, Monthly Notices of the RoyalAstronomical Society, 479, 1505Habing, H. J. 1968, Bull. Astron. Inst. Netherlands, 19, 421Ha ff ert, S. Y., Bohn, A. J., de Boer, J., et al. 2019, Nature Astronomy, 3, 749Herczeg, G. J. & Hillenbrand, L. A. 2014, ApJ, 786, 97Honda, M., Maaskant, K., Okamoto, Y. K., et al. 2012, The Astrophysical Jour-nal, 752, 143Huang, J., Andrews, S. M., Cleeves, L. I., et al. 2018, The Astrophysical Journal,852, 122Ireland, M. J., Kraus, A., Martinache, F., Law, N., & Hillenbrand, L. A. 2011,ApJ, 726, 113Isella, A., Benisty, M., Teague, R., et al. 2019, ApJ, 879, L25Isella, A., Guidi, G., Testi, L., et al. 2016, Phys. Rev. Lett., 117, 251101Isella, A., Huang, J., Andrews, S. M., et al. 2018, ApJ, 869, L49Isella, A., Natta, A., Wilner, D., Carpenter, J. M., & Testi, L. 2010, ApJ, 725,1735Ishihara, D., Onaka, T., Kataza, H., et al. 2010, A&A, 514, A1Jamialahmadi, N., Ratzka, T., Pani´c, O., et al. 2018, ApJ, 865, 137Jayawardhana, R., Fisher, R. S., Telesco, C. M., et al. 2001, AJ, 122, 2047Joblin, C., Berné, O., Simon, A., & Mulas, G. 2009, in Astronomical Societyof the Pacific Conference Series, Vol. 414, Cosmic Dust - Near and Far, ed.T. Henning, E. Grün, & J. Steinacker, 383Kamp, I., Thi, W. F., Woitke, P., et al. 2017, A&A, 607, A41Kamp, I., Tilling, I., Woitke, P., Thi, W. F., & Hogerheijde, M. 2010, Astronomyand Astrophysics, 510, A18Kasper, M., Arsenault, R., Käufl, H. U., et al. 2017, The Messenger, 169, 16Keppler, M., Teague, R., Bae, J., et al. 2019, A&A, 625, A118Khalafinejad, S., Maaskant, K. M., Mariñas, N., & Tielens, A. G. G. M. 2016,A&A, 587, A62Kluska, J., Berger, J. P., Malbet, F., et al. 2020, A&A, 636, A116Lagage, P. O., Pel, J. W., Authier, M., et al. 2004, The Messenger, 117, 12Lambrechts, M. & Johansen, A. 2014, A&A, 572, A107Leinert, C., van Boekel, R., Waters, L. B. F. M., et al. 2004, A&A, 423, 537Li, A. & Draine, B. T. 2001, ApJ, 554, 778Li, D., Telesco, C. M., Zhang, H., et al. 2018, MNRAS, 473, 1427Ligi, R., Vigan, A., Gratton, R., et al. 2018, MNRAS, 473, 1774Liu, S.-F., Jin, S., Li, S., Isella, A., & Li, H. 2018, ApJ, 857, 87Liu, W. M., Hinz, P. M., Meyer, M. R., et al. 2003, ApJ, 598, L111Long, Z. C., Fernandes, R. B., Sitko, M., et al. 2017, The Astrophysical Journal,838, 62Maaskant, K. M., Honda, M., Waters, L. B. F. M., et al. 2013, A&A, 555, A64Maaskant, K. M., Min, M., Waters, L. B. F. M., & Tielens, A. G. G. M. 2014,A&A, 563, A78Macías, E., Anglada, G., Osorio, M., et al. 2017, The Astrophysical Journal, 838,97Malfait, K., Waelkens, C., Waters, L. B. F. M., et al. 1998, A&A, 332, L25Mariñas, N., Telesco, C. M., Fisher, R. S., & Packham, C. 2011, The Astrophys-ical Journal, 737, 57Marley, M. S., Fortney, J. J., Hubickyj, O., Bodenheimer, P., & Lissauer, J. J.2007, The Astrophysical Journal, 655, 541Martin, R. G. & Lubow, S. H. 2011, Monthly Notices of the Royal AstronomicalSociety, 413, 1447Meeus, G., Montesinos, B., Mendigutía, I., et al. 2012, A&A, 544, A78Meeus, G., Waters, L. B. F. M., Bouwman, J., et al. 2001, A&A, 365, 476Mendigutía, I., Oudmaijer, R. D., Garufi, A., et al. 2017, A&A, 608, A104Menu, J., van Boekel, R., Henning, T., et al. 2015, A&A, 581, A107 Mesa, D., Langlois, M., Garufi, A., et al. 2019, Monthly Notices of the RoyalAstronomical Society, 488, 37Miley, J. M., Pani´c, O., Haworth, T. J., et al. 2019, MNRAS, 485, 739Min, M., Rab, C., Woitke, P., Dominik, C., & Ménard, F. 2016, A&A, 585, A13Mitchell, T. R. & Stewart, G. R. 2011, AJ, 142, 168Momose, M., Morita, A., Fukagawa, M., et al. 2015, PASJ, 67, 83Mordasini, C., Alibert, Y., Klahr, H., & Henning, T. 2012, A&A, 547, A111Naud, M.-E., Artigau, É., Doyon, R., et al. 2017, AJ, 154, 129Nayakshin, S., Tsukagoshi, T., Hall, C., et al. 2020, arXiv e-prints,arXiv:2004.10094Nealon, R., Cuello, N., Gonzalez, J.-F., et al. 2020, MNRAS, 499, 3857Neuhäuser, R., Guenther, E. W., Wuchterl, G., et al. 2005, A&A, 435, L13Nielsen, E. L., De Rosa, R. J., Macintosh, B., et al. 2019, AJ, 158, 13Nomura, H., Tsukagoshi, T., Kawabe, R., et al. 2016, The Astrophysical Journal,819, L7Oberg, N., Kamp, I., Cazaux, S., & Rab, C. 2020, A&A, 638, A135Okamoto, Y. K., Kataza, H., Honda, M., et al. 2017, AJ, 154, 16Osorio, M., Anglada, G., Carrasco-González, C., et al. 2014, ApJ, 791, L36Pani´c, O., Hogerheijde, M. R., Wilner, D., & Qi, C. 2008, A&A, 491, 219Pani´c, O., Ratzka, T., Mulders, G. D., et al. 2014, A&A, 562, A101Pérez, S., Casassus, S., Baruteau, C., et al. 2019, The Astronomical Journal, 158,15Pérez, S., Casassus, S., Hales, A., et al. 2020, ApJ, 889, L24Petit dit de la Roche, D. J. M., van den Ancker, M. E., Kissler-Patig, M., Ivanov,V. D., & Fedele, D. 2020, MNRAS, 491, 1795Pineda, J. E., Quanz, S. P., Meru, F., et al. 2014, ApJ, 788, L34Pineda, J. E., Szulágyi, J., Quanz, S. P., et al. 2019, ApJ, 871, 48Pinilla, P., Birnstiel, T., Benisty, M., et al. 2013, A&A, 554, A95Pinte, C., Price, D. J., Ménard, F., et al. 2018, ApJ, 860, L13Pohl, A., Benisty, M., Pinilla, P., et al. 2017, ApJ, 850, 52Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62Quanz, S. P., Amara, A., Meyer, M. R., et al. 2015, The Astrophysical Journal,807, 64Quanz, S. P., Amara, A., Meyer, M. R., et al. 2013, ApJ, 766, L1Quillen, A. C. 2006, ApJ, 640, 1078Quillen, A. C. & Trilling, D. E. 1998, ApJ, 508, 707Rab, C., Kamp, I., Ginski, C., et al. 2019, A&A, 624, A16Raman, A., Lisanti, M., Wilner, D. J., Qi, C., & Hogerheijde, M. 2006, AJ, 131,2290Rameau, J., Follette, K. B., Pueyo, L., et al. 2017, The Astronomical Journal,153, 244Ratzka, T., Leinert, C., Henning, T., et al. 2007, A&A, 471, 173Reggiani, M., Christiaens, V., Absil, O., et al. 2018, A&A, 611, A74Reggiani, M., Quanz, S. P., Meyer, M. R., et al. 2014, ApJ, 792, L23Rice, W. K. M., Armitage, P. J., Wood, K., & Lodato, G. 2006, MNRAS, 373,1619Rosotti, G. P., Benisty, M., Juhász, A., et al. 2020, MNRAS, 491, 1335Schneider, P. C., Dougados, C., Whelan, E. T., et al. 2020, arXiv e-prints,arXiv:2006.00573Siebenmorgen, R. & Heymann, F. 2012, A&A, 539, A20Sissa, E., Gratton, R., Garufi, A., et al. 2018, A&A, 619, A160Sokal, K. R., Deen, C. P., Mace, G. N., et al. 2018, ApJ, 853, 120Teague, R., Bae, J., & Bergin, E. A. 2019, Nature, 574, 378Teague, R., Bae, J., Bergin, E. A., Birnstiel, T., & Foreman-Mackey, D. 2018,ApJ, 860, L12Thi, W. F., Woitke, P., & Kamp, I. 2011, Monthly Notices of the Royal Astro-nomical Society, 412, 711Tsukagoshi, T., Muto, T., Nomura, H., et al. 2019, ApJ, 878, L8Tsukagoshi, T., Nomura, H., Muto, T., et al. 2016, ApJ, 829, L35Vacca, W. D. & Sandell, G. 2011, ApJ, 732, 8van Boekel, R., Henning, T., Menu, J., et al. 2017, ApJ, 837, 132van Boekel, R., Min, M., Waters, L. B. F. M., et al. 2005, A&A, 437, 189van Boekel, R., Waters, L. B. F. M., Dominik, C., et al. 2004, A&A, 418, 177van den Ancker, M. E., The, P. S., Tjin A Djie, H. R. E., et al. 1997, A&A, 324,L33van der Marel, N., Williams, J. P., & Bruderer, S. 2018, The Astrophysical Jour-nal, 867, L14van der Plas, G., Ménard, F., Gonzalez, J.-F., et al. 2019, A&A, 624, A33Verhoe ff , A. 2009, PhD thesis, Sterrenkundig Instituuut “Anton Pannekoek”,University of AmsterdamVigan, A., Bonavita, M., Biller, B., et al. 2017, A&A, 603, A3Vorobyov, E. I. 2013, A&A, 552, A129Wagner, K., Apai, D., Kasper, M., & Robberto, M. 2015, The AstrophysicalJournal, 813, L2Wagner, K., Dong, R., Sheehan, P., et al. 2018, The Astrophysical Journal, 854,130Wagner, K., Stone, J. M., Spalding, E., et al. 2019, ApJ, 882, 20Walsh, C., Daley, C., Facchini, S., & Juhász, A. 2017, A&A, 607, A114Walsh, C., Juhász, A., Pinilla, P., et al. 2014, ApJ, 791, L6Weinberger, A. J., Anglada-Escudé, G., & Boss, A. P. 2013, ApJ, 762, 118Wichittanakom, C., Oudmaijer, R. D., Fairlamb, J. R., et al. 2020, MNRAS, 493,234Woitke, P., Kamp, I., Antonellini, S., et al. 2019, PASP, 131, 064301Woitke, P., Kamp, I., & Thi, W.-F. 2009, A&A, 501, 383Woitke, P., Min, M., Pinte, C., et al. 2016, A&A, 586, A103Yu, S.-Y., Ho, L. C., & Zhu, Z. 2019, ApJ, 877, 100Zhu, Z. 2015, The Astrophysical Journal, 799, 16Zhu, Z., Andrews, S. M., & Isella, A. 2018, MNRAS, 479, 1850Zubko, V. G., Mennella, V., Colangeli, L., & Bussoletti, E. 1996, MNRAS, 282,1321 Article number, page 16 of 18. J. M. Petit dit de la Roche et al.: New Mid-IR Constraints on PPDs and Companions of 6 Young Stars
Appendix A: Standard disk models and SED fittingmethodology
To perform the SED fits, a comprehensive set of publicly avail-able observational data, consisting of photometric fluxes, inter-ferometric data, low and high resolution spectra, emission linefluxes, line velocity profiles, and maps were used from which thephysical and chemical parameters of the disk could be derived(references for which can be found in Dionatos et al. (2019)).The fits were performed by iteration of parameter sampling inMCFOST radiative transfer models by means of a genetic al-gorithm. HD 100546 was fit with 120 data points, two diskzones, PAHs, and 16 free parameters total after 632 generationsand 7584 models. Further details of the standard disk models,SED fitting procedures, and the limitations of SED fitting can befound in Woitke et al. (2016), Kamp et al. (2017), Woitke et al.(2019), and Dionatos et al. (2019).
Appendix A.1: Limitations
The DIANA SED fitting procedure was performed with dustopacities corresponding to a mixture of amorphous pyroxene sil-icates and amorphous carbon (see Table 4; Dorschner et al. 1995;Zubko et al. 1996). Due to the use of standard dust opacities anda fixed PAH morphology, only the power-law of the dust sizedistribution and volume fraction of amorphous carbon was var-ied for the fit, so detailed matching of the spectral features isnot expected. The 8.6 µ m PAH complex feature, associated within-plane C-H bending modes, is not fit in detail relative to the ISO-SWS spectrum. The presence of an unidentified broad fea-ture at 7.9-8 µ m is not explained by the model, but it has beensuggested by Joblin et al. (2009) to originate from a PAH pop-ulation known as PAH x consisting of compact but large ionisedPAHs with ∼
100 or more carbon atoms not included in our ra-diative transfer modelling.We opted not to explore the parameter space of possibledust compositions to perform a detailed opacity fitting acrossthe mid-IR given that properties such as the amorphous car-bon volume fraction can have a large impact on the SED at allwavelengths, such as by changing the millimetre and centime-tre slopes (Woitke et al. 2016). While the mid-IR traces the disksurface, any features may not be indicative of the disk globaldust properties and could represent surface e ff ects, for example,PAHs confined to the surface which are generated locally. In thiscase, altering global dust properties may not be the correct ap-proach.We did not re-perform the global SED fitting procedure toaccount for the increased GAIA EDR3 distance for HD 100546,but we did consider the implications of an increased stellar lu-minosity to match the observed luminosity and new distance. Totest the sensitivity of the SED to this adjustment, we considered amodest increase in our stellar e ff ective luminosity to 34.74 L (cid:12) . Ifwe were then to scale the physical dimensions of the disk and itsgap accordingly, the resulting SED would exhibit a net decreasein mid-IR emission; across the J8.9 band, we find a deficit inemission over the fiducial model of 2 . Appendix B: HD 100546 disk model PAH propertiesexploration
Several PAH features contribute to the disk opacity near 9 µ m.The broadband filter used in these observations covers an area F [ J y ] fiducial r gap,out + f PAH + m PAH
ISO-SWSUVSPIRETD1GENEVASTROMGRENJOHNSONTYCHO2USNOB1HIPPARCOSDENIS2MASSWISEAKARIIRASgenericPACSVISIR J8.9 [ m] m o d e l / d a t a Fig. A.1.
Global SED of the HD100546 disk models and comparisonto the observational data folded into the fit. The fiducial model SED isthe orange curve and our adjusted disk gap geometry model is the bluecurve. The relative residual as defined by dividing the model by the datais shown at the bottom. around 8.6 µ m where PAH C-H in-plane bending modes can con-tribute to the continuum emission. P ro D i M o uses synthetic PAHopacities for neutral and charged PAHs as calculated accordingto Li & Draine (2001). Exploring the properties of PAHs in themodel o ff ers the possibility of modifying the disk flux across theJ8.9 filter without globally modifying the disk dust propertiesand breaking the quality of the global SED fit.The contribution of PAHs was estimated by van Boekel et al.(2004) to be around 22% of the total flux near 9 µ m. They foundthe PAH emission to be more extended than the continuum alongthe spatial dimension of their longslit spectra, with a FWHMof ≈
150 AU. Using the low resolution spectroscopic mode ofVISIR, Verhoe ff (2009) found a statistically significant increasein the spatial extent of the disk emission at 8.6 µ m over the re-solved continuum emission at a 27 σ level. While they foundthe ratio between the continuum subtracted peak flux at the8.6 µ m PAH feature over the peak flux was only 2.4%, the de-convolved FWHM size of the continuum subtracted feature was1.64 + . − . (cid:48)(cid:48) . At a distance of 108 pc, this corresponds to a diskradius of 178 + − AU. Furthermore, the variability of the 8.6 µ mfeatures between ISO and TIMMI2 spectra and their respectiveslit sizes implies that the PAH emitting region is at least 100 AUin size (Verhoe ff µ m PAH emission to be emitted primarily from angularscales corresponding to ∼
100 AU from the star.While the HD 100546 disk model PAH abundance andcharge fraction was fit for, these parameters were not varied be-tween the inner and outer disk zones. We thus considered mod-ifications to the PAH population in the outer disk, outside of r =
22 AU, specifically. The DIANA models use a single rep-resentative PAH, circumcoronene (C H ), and a constant mix-ture of charged and neutral opacities throughout the disk (Woitkeet al. 2016). For HD 100546, the abundance of PAH relative tothe ISM f PAH (defined such that in the ISM f PAH =
1) is 0.0028.The mean PAH charged fraction is 0.9. We considered both dif-fering PAH types and abundances in the inner and outer diskzones to refine our fit.We have explored a grid of a PAH abundance and morpholo-gies in an attempt to minimise the residuals with our mid-IR ob-
Article number, page 17 of 18 & A proofs: manuscript no. main servational data. Simultaneously allowing for the outer wall ofthe gap, the abundance of the PAHs, and the type of the PAHs tovary has allowed us to improve upon the standard SED fit with-out reducing the quality of the fit globally (see Fig.A.1). The re-sult of this multi-parameter exploration can be seen in the greenline in Fig. 5. We find that a smaller PAH, coronene (C H ),and a 22% increase in f PAH outside of the gap wall produce thebest agreement with an observation across the J8.9 filter.outside of the gap wall produce thebest agreement with an observation across the J8.9 filter.