Circumbinary, not transitional: On the spiral arms, cavity, shadows, fast radial flows, streamers and horseshoe in the HD142527 disc
Daniel J. Price, Nicolas Cuello, Christophe Pinte, Daniel Mentiplay, Simon Casassus, Valentin Christiaens, Grant M. Kennedy, Jorge Cuadra, Sebastian Perez M., Sebastian Marino, Philip J. Armitage, Alice Zurlo, Attila Juhasz, Enrico Ragusa, Guillaume Laibe, Giuseppe Lodato
MMNRAS , 1–16 (2017) Preprint 8 March 2018 Compiled using MNRAS L A TEX style file v3.0
Circumbinary, not transitional: On the spiral arms, cavity, shadows,fast radial flows, streamers and horseshoe in the HD142527 disc
Daniel J. Price (cid:63) , Nicol´as Cuello , , , Christophe Pinte , , Daniel Mentiplay , SimonCasassus , , Valentin Christiaens , , Grant M. Kennedy , Jorge Cuadra , , , SebastianPerez M. , , Sebastian Marino , Philip J. Armitage , , Alice Zurlo , , Attila Juhasz ,Enrico Ragusa , Guillaume Laibe and Giuseppe Lodato Monash Centre for Astrophysics (MoCA) and School of Physics and Astronomy, Monash University, Clayton Vic 3800, Australia Instituto de Astrof´ısica, Pontificia Universidad Cat´olica de Chile, Santiago, Chile Millennium Nucleus ‘Protoplanetary discs’, Chile N´ucleo Milenio de Formaci´on Planetaria (NPF), Chile Univ. Grenoble Alpes, CNRS, IPAG / UMR 5274, F-38000 Grenoble Departamento de Astronom´ıa, Universidad de Chile, Casilla 36-D, Santiago, Chile Institute of Astronomy, University of Cambridge, Madingley Rd, Cambridge, CB3 0HA, UK JILA, University of Colorado & NIST, UCB 440, Boulder, CO 80309-0440, USA Department of Astrophysical and Planetary Sciences, University of Colorado, 391 UCB, Boulder, CO 80309-0391, USA Dipartimento di Fisica, Universit`a Degli Studi di Milano, Via Celoria, 16, Milano, I-20133, Italy Univ Lyon, Univ Lyon1, Ens de Lyon, CNRS, Centre de Recherche Astrophysique de Lyon UMR5574, F-69230, Saint-Genis-Laval, France
ABSTRACT
We present 3D hydrodynamical models of the HD142527 protoplanetary disc, a bright andwell studied disc that shows spirals and shadows in scattered light around a 100 au gas cavity, alarge horseshoe dust structure in mm continuum emission, together with mysterious fast radialflows and streamers seen in gas kinematics. By considering several possible orbits consistentwith the observed arc, we show that all of the main observational features can be explained byone mechanism — the interaction between the disc and the observed binary companion. Wefind that the spirals, shadows and horseshoe are only produced in the correct position anglesby a companion on an inclined and eccentric orbit approaching periastron — the ‘red’ familyfrom Lacour et al. (2016). Dust-gas simulations show radial and azimuthal concentration ofdust around the cavity, consistent with the observed horseshoe. The success of this model inthe HD142527 disc suggests other mm-bright transition discs showing cavities, spirals anddust asymmetries may also be explained by the interaction with central companions.
Key words: protoplanetary discs — planet-disc interactions — binaries: general — submil-limetre: planetary systems — accretion, accretion discs
Around the young star HD142527 lies an enigmatic and spectacularprotoplanetary disc. Cycle 0 observations with the Atacama LargeMillimetre/submillimetre Array (ALMA) by Casassus et al. (2013)revealed a ‘horseshoe’ of dust continuum emission (first detectedby Ohashi 2008) surrounding a ∼ (cid:63) [email protected] The star itself is a Herbig Fe star of spectral type F6 IIIe(M ≈ (cid:12) ) at a distance of +7 − pc (Gaia Collaboration et al.2016) in the Sco-Cen association (Biller et al. 2012; Mendigut´ıaet al. 2014). Modelling of the Spectral Energy Distribution (SED)suggested a disc gap between 30 and 130 au (Verhoeff et al. 2011),confirmed by the initial ALMA observations (Casassus et al. 2013).Earlier mid-infrared observations (van Boekel et al. 2004; Fujiwaraet al. 2006) and more recent scattered light images (Avenhaus et al.2017) found a small inner disc inside the cavity of ∼
10 au in radius.The high accretion rate onto the central star ( ≈ × − M (cid:12) /yr;Garcia Lopez et al. 2006; Mendigut´ıa et al. 2014) implies that thesmall inner disc must be refilled from the outer disc, most likely inan episodic manner (Casassus et al. 2013).Even before ALMA, the disc around HD142527 had provedspectacular, with a spiral arm detected at R (cid:38) au in scattered © 2017 The Authors a r X i v : . [ a s t r o - ph . S R ] M a r Price et al. light observations in the near IR by Fukagawa et al. (2006). Fur-ther observations revealed a wealth of spiral structure, includingsmall near IR spirals at the edge of the cavity (Casassus et al. 2012;Canovas et al. 2013; Avenhaus et al. 2014, 2017), a counterpart tothe Fukagawa et al. (2006) spiral seen in the CO emission in furtherALMA observations, as well as two further large scale ( ∼
500 au)spiral arms by Christiaens et al. (2014).Fukagawa et al. (2006) first suggested that the spirals mightbe caused by an inner companion, with the presence of a 0.1–0.4M (cid:12) companion with a projected separation of ∼ au confirmedby Biller et al. (2012) using Sparse Aperture Masking (SAM) withthe NACO instrument on the Very Large Telescope (VLT). Morerecent observations have refined both the orbit and the companionmass (Close et al. 2014; Lacour et al. 2016) (hereafter L16).Scattered light images also revealed shadows cast across theouter disc (Fukagawa et al. 2006; Avenhaus et al. 2014). Marinoet al. (2015a) satisfyingly reproduced the observed illuminationpattern by assuming a compact ∼ au inclined and thin innerdisc casting shadows on the outer disc, with a relative inclinationof ∼ ◦ . This is consistent with the estimated inclination of theinner binary with respect to the disc (L16).Perhaps the greatest surprise were the complexities found indetailed kinematic studies using CO(6–5) molecular line data byCasassus et al. (2015a), with evidence for a warped disc and nearfree-fall motions within the central cavity, suggested to be relatedto theoretical models of ‘disc tearing’ in warped accretion discs(Nixon et al. 2013; Nealon et al. 2015).Understanding the origin of features in the HD142527 discis important because several of these appear common to a broadclass of discs. In particular, dust horseshoes or rings surround-ing the central cavities in mm-bright transition discs appear to bewidespread (van der Marel et al. 2016; Canovas et al. 2016; see re-view by Casassus 2016), with horseshoes normally interpreted as‘dust traps’ caused by a vortex at the edge of a planet-induced gap(e.g. Pinilla et al. 2012; van der Marel et al. 2013; Baruteau & Zhu2016; Marino et al. 2015b). Spirals are also observed in an increas-ing number of discs (Garufi et al. 2013; Benisty et al. 2015; P´erezet al. 2016; Benisty et al. 2017). Thus, unlocking the mystery ofHD142527 may help to shed light on the origin of these features inthis wider class of discs.Our approach is to perform 3D hydrodynamical and dust-gassimulations of the disc-binary interaction, to try to explain the ob-served features in the HD142527 disc. We describe the key obser-vations in Section 2, methods and initial conditions in Section 3.Results are in Section 4. We discuss in Section 5 and summarise inSection 6. The sheer volume of data collected on HD142527 provides tightconstraints on theoretical models, and it seems optimistic to assumethat a single model could explain multiple observed features simul-taneously. From a dynamical perspective the main puzzles are:(i) Shadows. The close agreement found by Marino et al.(2015a) between their radiative transfer model and the scatteredlight images means these shadows are almost certainly cast by aninner circumprimary disc inclined to the outer disc plane by ∼ ◦ .The size of such a disc would be consistent with the infrared obser-vations (Verhoeff et al. 2011). The constraints from ALMA COobservations of the inner disc suggests that it is small and with un- usual (non-Keplerian) kinematics between the inner and outer disc(Perez et al. 2015; Casassus et al. 2015a).(ii) Fast radial flows. Casassus et al. (2015a) suggested an expla-nation for fast radial flows in terms of disc tearing by an inclinedinner binary (e.g. Nixon et al. 2013). However, subsequent sim-ulations by Dunhill and collaborators (reported in Cuadra 2016)found that such a binary tended to simply break the disc into twodistinct sections, as found by Facchini et al. (2013) and consistentwith expectations of warp dynamics in thick ( H/R ∼ . ) proto-planetary discs. Rosenfeld et al. (2014) also found that the fast ra-dial flows were better explained by free-fall radial velocities ratherthan a warp, suggesting material is somehow able to shed angularmomentum and plunge into the central regions. Those authors pro-posed gravitational torques from giant planets or brown dwarfs asa possible solution.(iii) Spiral arms. Spirals are seen immediately outside the cavityin both scattered light (Fukagawa et al. 2006; Rodigas et al. 2014;Avenhaus et al. 2014, 2017) and in CO emission (Christiaens et al.2014). These are usually attributed to the presence of either com-panions orbiting interior or exterior to the arms (Dong et al. 2015),to a gravitationally unstable disc (Dipierro et al. 2015a), or to somecombination of both (Pohl et al. 2015). Quillen (2006) offered anexplanation of similar spirals seen in the disc around HD100546in terms of a precessing, warped disc driven by misaligned embed-ded protoplanets. Montesinos et al. (2016) also showed that spiralstructure could be induced by temperature fluctuations caused byshadows.(iv) Cavity. The origin of central cavities in transitional discsis not yet certain (Williams & Cieza 2011). Traditionally centralholes were thought to arise from either photoevaporation of gas bythe central star or depletion of gas and/or dust due to formationof planetary mass companions (Williams & Cieza 2011; Andrewset al. 2011; Espaillat et al. 2014; Owen 2016). Zhu et al. (2012)found that dust filtering by giant planets (e.g. Rice et al. 2006)could partially explain the mm-dust holes seen in many transitiondiscs, but that the depletion in small particles was insufficient to ex-plain the near-IR deficit in the Spectral Energy Distribution (SED).An alternative possibility is the tidal truncation of the disc froma central binary (Artymowicz & Lubow 1994). While a low masscompanion has been found in HD142527 (see below), previous au-thors (e.g. L16) have assumed that the projected orbital separationof ∼ au is too small to tidally truncate the disc out to 100 au —the size of the observed CO cavity (Perez et al. 2015; Muto et al.2015; Boehler et al. 2017).(v) Dust horseshoe. Currently the main accepted model for pro-ducing dust horseshoes in mm-bright transition discs involves dusttrapping by a gap-edge vortex (Pinilla et al. 2012; Lyra & Lin2013). Ataiee et al. (2013) considered an alternative model wheredust horseshoes could be produced by eccentric circumbinary cavi-ties, but dismissed this model based on their simulations. However,dust evolution was added only in post-processing which neglectsthe role of tidal torques on the dust and backreaction, leading topotentially misleading conclusions. Recently, Ragusa et al. (2017)showed that central binaries can produce both rings and horse-shoes in dust emission. Indeed, eccentricities around the cavityedge are a common feature in hydrodynamical simulations of cir-cumbinary discs (Kley & Dirksen 2006; Farris et al. 2014; Ragusaet al. 2016). Ragusa et al. (2017) showed that more massive com-panions produce progressively more asymmetric structures, withsimulated ALMA observations closely matching observed discs.(vi) Gap-crossing filaments. Casassus et al. (2013) attributedthese to flows of gas through a planetary gap, as observed com- MNRAS000
500 au)spiral arms by Christiaens et al. (2014).Fukagawa et al. (2006) first suggested that the spirals mightbe caused by an inner companion, with the presence of a 0.1–0.4M (cid:12) companion with a projected separation of ∼ au confirmedby Biller et al. (2012) using Sparse Aperture Masking (SAM) withthe NACO instrument on the Very Large Telescope (VLT). Morerecent observations have refined both the orbit and the companionmass (Close et al. 2014; Lacour et al. 2016) (hereafter L16).Scattered light images also revealed shadows cast across theouter disc (Fukagawa et al. 2006; Avenhaus et al. 2014). Marinoet al. (2015a) satisfyingly reproduced the observed illuminationpattern by assuming a compact ∼ au inclined and thin innerdisc casting shadows on the outer disc, with a relative inclinationof ∼ ◦ . This is consistent with the estimated inclination of theinner binary with respect to the disc (L16).Perhaps the greatest surprise were the complexities found indetailed kinematic studies using CO(6–5) molecular line data byCasassus et al. (2015a), with evidence for a warped disc and nearfree-fall motions within the central cavity, suggested to be relatedto theoretical models of ‘disc tearing’ in warped accretion discs(Nixon et al. 2013; Nealon et al. 2015).Understanding the origin of features in the HD142527 discis important because several of these appear common to a broadclass of discs. In particular, dust horseshoes or rings surround-ing the central cavities in mm-bright transition discs appear to bewidespread (van der Marel et al. 2016; Canovas et al. 2016; see re-view by Casassus 2016), with horseshoes normally interpreted as‘dust traps’ caused by a vortex at the edge of a planet-induced gap(e.g. Pinilla et al. 2012; van der Marel et al. 2013; Baruteau & Zhu2016; Marino et al. 2015b). Spirals are also observed in an increas-ing number of discs (Garufi et al. 2013; Benisty et al. 2015; P´erezet al. 2016; Benisty et al. 2017). Thus, unlocking the mystery ofHD142527 may help to shed light on the origin of these features inthis wider class of discs.Our approach is to perform 3D hydrodynamical and dust-gassimulations of the disc-binary interaction, to try to explain the ob-served features in the HD142527 disc. We describe the key obser-vations in Section 2, methods and initial conditions in Section 3.Results are in Section 4. We discuss in Section 5 and summarise inSection 6. The sheer volume of data collected on HD142527 provides tightconstraints on theoretical models, and it seems optimistic to assumethat a single model could explain multiple observed features simul-taneously. From a dynamical perspective the main puzzles are:(i) Shadows. The close agreement found by Marino et al.(2015a) between their radiative transfer model and the scatteredlight images means these shadows are almost certainly cast by aninner circumprimary disc inclined to the outer disc plane by ∼ ◦ .The size of such a disc would be consistent with the infrared obser-vations (Verhoeff et al. 2011). The constraints from ALMA COobservations of the inner disc suggests that it is small and with un- usual (non-Keplerian) kinematics between the inner and outer disc(Perez et al. 2015; Casassus et al. 2015a).(ii) Fast radial flows. Casassus et al. (2015a) suggested an expla-nation for fast radial flows in terms of disc tearing by an inclinedinner binary (e.g. Nixon et al. 2013). However, subsequent sim-ulations by Dunhill and collaborators (reported in Cuadra 2016)found that such a binary tended to simply break the disc into twodistinct sections, as found by Facchini et al. (2013) and consistentwith expectations of warp dynamics in thick ( H/R ∼ . ) proto-planetary discs. Rosenfeld et al. (2014) also found that the fast ra-dial flows were better explained by free-fall radial velocities ratherthan a warp, suggesting material is somehow able to shed angularmomentum and plunge into the central regions. Those authors pro-posed gravitational torques from giant planets or brown dwarfs asa possible solution.(iii) Spiral arms. Spirals are seen immediately outside the cavityin both scattered light (Fukagawa et al. 2006; Rodigas et al. 2014;Avenhaus et al. 2014, 2017) and in CO emission (Christiaens et al.2014). These are usually attributed to the presence of either com-panions orbiting interior or exterior to the arms (Dong et al. 2015),to a gravitationally unstable disc (Dipierro et al. 2015a), or to somecombination of both (Pohl et al. 2015). Quillen (2006) offered anexplanation of similar spirals seen in the disc around HD100546in terms of a precessing, warped disc driven by misaligned embed-ded protoplanets. Montesinos et al. (2016) also showed that spiralstructure could be induced by temperature fluctuations caused byshadows.(iv) Cavity. The origin of central cavities in transitional discsis not yet certain (Williams & Cieza 2011). Traditionally centralholes were thought to arise from either photoevaporation of gas bythe central star or depletion of gas and/or dust due to formationof planetary mass companions (Williams & Cieza 2011; Andrewset al. 2011; Espaillat et al. 2014; Owen 2016). Zhu et al. (2012)found that dust filtering by giant planets (e.g. Rice et al. 2006)could partially explain the mm-dust holes seen in many transitiondiscs, but that the depletion in small particles was insufficient to ex-plain the near-IR deficit in the Spectral Energy Distribution (SED).An alternative possibility is the tidal truncation of the disc froma central binary (Artymowicz & Lubow 1994). While a low masscompanion has been found in HD142527 (see below), previous au-thors (e.g. L16) have assumed that the projected orbital separationof ∼ au is too small to tidally truncate the disc out to 100 au —the size of the observed CO cavity (Perez et al. 2015; Muto et al.2015; Boehler et al. 2017).(v) Dust horseshoe. Currently the main accepted model for pro-ducing dust horseshoes in mm-bright transition discs involves dusttrapping by a gap-edge vortex (Pinilla et al. 2012; Lyra & Lin2013). Ataiee et al. (2013) considered an alternative model wheredust horseshoes could be produced by eccentric circumbinary cavi-ties, but dismissed this model based on their simulations. However,dust evolution was added only in post-processing which neglectsthe role of tidal torques on the dust and backreaction, leading topotentially misleading conclusions. Recently, Ragusa et al. (2017)showed that central binaries can produce both rings and horse-shoes in dust emission. Indeed, eccentricities around the cavityedge are a common feature in hydrodynamical simulations of cir-cumbinary discs (Kley & Dirksen 2006; Farris et al. 2014; Ragusaet al. 2016). Ragusa et al. (2017) showed that more massive com-panions produce progressively more asymmetric structures, withsimulated ALMA observations closely matching observed discs.(vi) Gap-crossing filaments. Casassus et al. (2013) attributedthese to flows of gas through a planetary gap, as observed com- MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc D e c ( m a s ) May 2014March 2012 x [au] -200-1000100200 y [ m a s ] y [ a u ] Figure 1.
Binary orbit. Left: Orbital fits for HD142527B (Credit: Lacour et al. 2016, reproduced with permission ©ESO). Right: Selected trial orbits forHD142527B used in this paper, corresponding to the orbital elements listed in Table 1. We assume the GAIA distance of 156pc. The star gives the location ofthe primary. Motion is clockwise. monly in simulations of gap-opening by embedded planets in discs(e.g. Bryden et al. 1999; Lubow & D’Angelo 2006). However, it isnot clear how the ‘filaments’ might be obviously related to eitherthe companion or any other putative planets, nor how they relate tothe fast radial flows, or whether they may be illumination effectsdue to shadowing from the tilted inner disc.A clue to solving the puzzle is that all of the above featuresmay in principle be caused by the interaction with an (inclined)central binary (Fukagawa et al. 2006). The main issue is how abinary with a ∼ au projected separation could carve a ∼ – au cavity. This led L16 to conclude that the cavity could notbe created by the binary. However, the orbit is poorly constrained,with best fitting orbits from L16 suggesting significant eccentricity. We perform 3D hydrodynamics simulations of the disc-binary in-teraction using the P
HANTOM smoothed particle hydrodynamics(SPH) code (Price et al. 2018; for reviews of SPH see Monaghan2005; Price 2012). We also perform several models with dust andgas to see if we can simultaneously produce the dust horseshoe.Our approach to dust-gas modelling is similar to that used in ourprevious papers (Dipierro et al. 2015b, 2016), where we model thedust using a separate set of particles coupled to the gas via a dragterm (Laibe & Price 2012a,b; Price et al. 2018).
We model the binary using sink particles that interact with the gasdisc only via gravity and by accreting gas (Bate, Bonnell & Price1995). The sink particles are free to migrate and also change massand orbital parameters due to interaction with the gas disc. We setthe primary mass to 1.8 M (cid:12) (from the spectral type), and con-strained by the total mass estimate of ∼ . ± . M (cid:12) from theKeplerian motion of the outer disc (Casassus et al. 2015a) we setthe companion mass to 0.4 M (cid:12) (the latest observational estimatesbased on spectral fitting are in the range 0.2–0.4 M (cid:12) ; Christiaenset al. 2018). Since our chosen mass of 0.4 M (cid:12) is at the higher end of Orbit a e i Ω ω f P θ B1 26.5 0.24 119.9 349.7 218.0 25.93 91.8 40.4B2 28.8 0.40 120.4 340.3 201.5 33.78 104 39.6B3 34.3 0.50 119.3 159.2 19.98 35.04 135 80.7R1 31.4 0.74 131.3 44.95 27.88 249.3 118 43.1R2 38.9 0.61 120.3 19.25 354.0 268.3 164 45.3R3 51.3 0.70 119.3 201.4 173.3 270.4 247 76.3
Table 1.
Orbital elements for HD142527B for 6 trial orbits, drawn fromfits to the observed arc with
IMORBEL . From left to right: semi-major axis( a , au), eccentricity ( e ), inclination ( i , deg), position angle of ascendingnode ( Ω , deg; East of North), argument of pericentre ( ω , deg), true anomaly( f , deg), orbital period ( P , yrs) and relative angle between disc and binary( θ , deg). this range, we also performed an additional set of calculations em-ploying lower mass companions. We found that our results are notstrongly sensitive to the secondary mass, with companion massesas low as 0.1M (cid:12) producing the same cavity but with lower ampli-tude perturbations around the cavity edge. The change in mass dueto accretion is negligible in our simulations.We fix the accretion radii for both sinks to au in order toresolve the circumprimary disc, if present. After a few trial simulations with binaries of various semi-majoraxes and eccentricities, and from previous attempts at modellingHD142527 (see Cuadra 2016), we realised the importance of theknown observational constraints on the orbit. To this end we em-ployed
IMORBEL to fit the orbit, written by Tim Pearce and GrantKennedy (Pearce et al. 2015). The same code was also used in L16.Rather than taking the orbital fits directly from L16, we pro-duced a revised set of orbital fits using the GAIA distance measure-ment and with our assumed primary and secondary masses. IMOR - BEL produces a plot similar to the one shown in figure 5 of L16, http://github.com/drgmk/imorbel MNRAS , 1–16 (2017)
Price et al.
B1 B2 B3R1 R2 100 au 0.51 c o l u m n d e n s it y [ g / c m ] R3 Figure 2.
Gas surface densities after 20 orbital periods of the binary for the calculations with R in = 50 au, showing the initial dynamical carving of thecavity for binary orbits shown in Figure 1 and listed in Table 1. Top row shows the ‘blue’ orbits (binary just past periastron) while bottom row corresponds tothe ‘red’ orbits (binary approaching periastron) from L16. Cavity size scales with apastron separation, with more eccentric binaries (R2, R3) producing cavitysizes consistent with those observed in HD142527. Transient circumprimary discs are visible in all calculations except B1. from which one may interactively select orbits with given parame-ters that match the observational constraints. Our guiding wisdomin selecting trial orbits was i) to examine orbits similar to the ‘blue’and ‘red’ orbit families found in the Monte Carlo fitting shown inFigure 6 of L16; and ii) to ensure an apoastron separation for thebinary large enough to plausibly explain the cavity size.Table 1 lists the orbital elements used for the six representativeorbits shown in this paper, with the resultant orbits shown in Fig-ure 1. For clarity we show three representative “blue” orbits (B1,B2, B3) and three “red” orbits (R1, R2, R3) listed in order of in-creasing semi-major axis. The most tightly constrained parameteris the inclination, which is i = 120 ◦ for all but one of the orbits.We choose eccentricities ranging from 0.24 to 0.7. The two classesof orbits are ‘families’ only in a projected sense, as the values of Ω and ω (and hence the orientation perpendicular to the sky plane)change dramatically. The main similarity between each family issimply the location of the periastron in position angle on the skyand thus whether the binary is approaching (red) or receding (blue)from periastron. The last column in Table 1 gives the angle betweenthe disc and binary angular momentum vectors.The Monte Carlo fitting performed by L16 found semi-majoraxes < a < au, eccentricities e = 0 . ± . and inclinations ± degrees to within 1 σ probability. Thus all our chosenorbits except R3 fall are probable to within 1 σ of these constraints.Finally, we adopted slightly different conventions for the or-bital elements to those used in Pearce et al. (2015) and L16. In particular, we removed the assumption by Pearce et al. (2015) thatthe angle between the binary and the sky is restricted to be lessthan ◦ (thus flipping the observer from + z to − z dependingon the orbit). Since the absolute orientation of the disc and binaryare irrelevant in SPH, for convenience we defined the observer tobe viewing the disc down the z − axis (i.e. from z = ∞ ). That is,we tilted both the disc and binary in our initial setup. This simpli-fied our analysis since replicating the observers view of HD142527simply meant making an x - y plot in our computational coordinates.Finally, we adapted the setup routine in P HANTOM to use orbital el-ements from
IMORBEL in the form given in Table 1, such that thepublic codes are mutually compatible.
We performed two sets of calculations, both with a 0.01 M (cid:12) cir-cumbinary gas disc but with inner radius set initially to either 50 or90 au. The R in = 50 au calculations ensure that clearing of the in-ner regions is entirely due to tidal effects, while starting with R in =90 au avoids transient formation of circumprimary and circumsec-ondary discs. We set the outer radius to R out = 350 au (purelyfor computational efficiency; the real disc extends to ∼ au).We assume an initially power law surface density profile Σ ∝ R − and model the disc with 10 SPH particles assuming a total gasmass of 0.01 M (cid:12) — see Price et al. (2018) for details of the discsetup used in P
HANTOM . We adopt a mean Shakura-Sunyaev disc
MNRAS000
MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc l og Σ [ g / c m ] -3 -2 -1 l og Σ [ g / c m ] R [au]10
50 200 50010 -3 -2 -1
50 200 500R2 R [au]10
50 200 500R3
Figure 3.
Cavity. Surface density as a function of cylindrical radius (in the sky plane) from the R in = 90 au calculations after 50 binary orbits. Red dottedline shows the model adopted by Perez et al. (2015) from radiative transfer fitting to CO gas. Simulation R2 shows the closest match to the observed cavitysize, with R cav within a few percent of the gas cavity size inferred from the CO data. viscosity α SS ≈ . by setting a fixed artificial viscosity param-eter α AV = 0 . in the code and using the ‘disc viscosity’ flag(see Lodato & Price 2010). Since the artificial viscosity is explic-itly specified in SPH, the value of α SS is directly related to α AV according to α SS = 110 α AV (cid:104) h (cid:105) H , (1)where the ratio (cid:104) h (cid:105) is the mean resolution length at a given radiusand H is the disc scale height. There are no additional sources ofnumerical diffusion in SPH (see Lodato & Price 2010 for detailedcalibration and tests). For our chosen resolution the mean h/H inthe outer disc is 0.2 (i.e., 5 resolution lengths per scale height),hence the values of α SS and α AV given above.We prescribe temperature as a function of (spherical) radiusaccording to T = 28K( r/r in ) − . , consistent with the disc modelfit by Casassus et al. (2015a) to the observational temperature pro-file (we use spherical rather than cylindrical radius to avoid confu-sion when the disc is warped or inclined; see Lodato & Price 2010).This corresponds to H/R = 0 . at R = R in and H/R = 0 . at R = R out . More recent observations (Muto et al. 2015; Boehleret al. 2017) found higher temperatures which may suggest a thickerdisc. Further modelling of the temperature changes in the disc dueto the companion is beyond the scope of this paper but may be im-portant (Verhoeff et al. 2011). The disc was oriented by i = 160 ◦ with respect to the z = 0 plane, rotated about a Position Angle of − ◦ . That is, our discis inclined by ◦ to the line of sight but with the disc rotatingclockwise on the sky, consistent with (e.g.) Casassus et al. (2015a). To perform a direct comparison with observations of HD142527 wepost-processed a subset of our simulations using the Monte Carloradiative transfer code
MCFOST (Pinte et al. 2006, 2009).
MCFOST is particularly suited to post-processing data from SPH codes be-cause it uses a Voronoi tesselation where each cell corresponds tothe position of an SPH particle. Properties such as density, temper-ature and velocity may then be mapped directly from the particlesto the radiative transfer grid without interpolation.To irradiate the disc we used the two sink particles as isotropicsources with stellar spectral models appropriate to their mass.Specifically, we adopt 3Myr isochrones from Siess et al. (2000),with T eff = 4800 K and a luminosity of 2.7 L (cid:12) for the primaryand T eff = 3700 K and L = 0.22 L (cid:12) for the secondary. We as-sume astronomical silicates (Draine 2003), with a grain size distri-bution ranging from 0.03 µ m to 1 mm, and a slope of − . . Opticalproperties are calculated using Mie theory. We then used MCFOST to predict line emission from our gas-only simulations for the iso-topologues of carbon monoxide including CO, CO and C O MNRAS , 1–16 (2017)
Price et al.
Figure 4.
Gas inside the cavity. Predicted CO J=2–1 from simulation R2 (right), compared to ALMA observations (left; credit: Figure 1 of Perez et al.(2015) ©AAS, reproduced with permission). Sufficient gas remains inside the cavity to produce optically thick CO emission, as observed.M ( < au) ( × − M (cid:12) ) Pitch ang.(PA ◦ ) Pitch ang.(1st half) Pitch ang.(2nd half) SWFork?Obs . ± . ◦ -3.6 ◦ ◦ YesB1 2.0 1.5 ◦ ◦ ◦ NoB2 1.9 1.3 ◦ ◦ ◦ NoB3 1.6 5.1 ◦ ◦ ◦ NoR1 2.1 -1.8 ◦ -4.0 ◦ ◦ YesR2 1.5 6.0 ◦ ◦ ◦ NoR3 1.4 3.6 ◦ ◦ ◦ No Table 2.
Gas mass interior to 90 au (first column); pitch angles along theouter spiral arm at position angle of ◦ (second column) and withinthe first and second half of the spiral in position angle (third and fourthcolumns); comparing our simulations to the observations (top row) after 50orbits of the binary. The final column indicates whether or not the spiralarms show a bifurcation or ‘fork’ towards the south-west. All simulationsobtain residual mass interior to the cavity consistent with the observations.Simulations B3, R2 and R3 show pitch angles most consistent with the data. as well as HCO + . We assume a uniform abundance of × − , × − , × − and − for CO, CO, C O, and HCO + ,respectively. We assume the gas is in local thermodynamic equilib-rium with T gas = T dust . We predicted only the line emission, notthe continuum, from our gas-only simulations under the assump-tion that the dust follows the gas. This is a good assumption forgrain sizes (cid:46) µ m but for larger grains this assumption no longerholds (see Section 4.5). Figure 2 shows the column density view of the six calculations cor-responding to the orbits listed in Table 2, shown after 20 binaryorbital periods with initial R in = 50 au. Transient circumprimarydiscs are present in all calculations except B1. For both ‘blue’ and ‘red’ sets of orbits the cavity size scales with semi-major axis (leftto right; Artymowicz & Lubow 1994). Cavity size also increaseswith eccentricity due to the increased apocentre separation (but de-creases with inclination; Miranda & Lai 2015). Comparison withthe cavity size seen in scattered light, we can exclude binary or-bits with a (cid:38) au. That with plausible orbits for the observedbinary we can already produce a cavity size too big solves one ofthe major mysteries of HD142527 — how a binary with a projectedseparation of 13 au can produce a 100 au cavity.Figure 3 shows the surface density binned as a function ofcylindrical radius in the sky plane ( z = 0 ). The red dotted lineshows the parameterised model adopted by Perez et al. (2015) fromfits to the CO emission (specifically, we plot equation 6 from theirpaper, with parameters taken from their best fit model corrected forthe GAIA distance, giving R cav = 100 au). Simulation R2 pro-duces the closest match to the data, with R cav within a few percentof the observational fit. The remarkable agreement demonstratesthat orbits consistent with those inferred by L16 indeed producecavities of the correct size. The drop in surface density seen in Figure 3 indicates that the cav-ity interior is not completely devoid of gas. To quantify this, Table 2lists the mass interior to 90 au in each of our simulations, comparedto the observational measurement from Perez et al. (2015). Regard-less of our choice of binary orbit the residual mass inside the cav-ity agrees with the observational estimate to within the error bars.For example, we measure a residual gas mass of . × − M (cid:12) within 90 au in simulation R2 after 50 orbits, within 11% of the (1 . ± . × − M (cid:12) measured by Perez et al. (2015).Figure 4 shows the predicted CO J=2–1 emission from sim-ulation R2 (right panel), convolved to a beam size consistent withthe Perez et al. (2015) Cycle 0 ALMA data (shown in left panel).We find remarkable agreement with the observed CO emissionfrom inside the cavity. In particular there is sufficient gas inside the
MNRAS000
MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc B1 B2 B3R1 R2 100 au 0.51 c o l u m n d e n s it y [ g / c m ] R3 Figure 5.
Spirals. Gas surface densities after 50 binary orbital periods in calculations with initial R in = 90 au, showing the spiral arms. As in Figure 2, top rowshows the ‘blue’ orbits (binary just past periastron) while bottom row corresponds to the ‘red’ orbits (binary approaching periastron) from L16. Comparisonwith the observed spiral structure (left panel of Figure 6) favours the latter. Comparison with Fig. 2 shows the cavity size is independent of the initial R in . Redlines show polynomial fits to ‘inner’ and ‘outer’ spiral arms tracing a position angle range similar to the scattered light spirals (Fig. 6). cavity such that the CO is optically thick, as observed. The bi-nary scenario thus naturally explains the residual gas found withinthe cavity around HD142527.
Figure 5 shows the column density from the R in = 90 au calcu-lations after 50 binary orbital periods. In every case three or moreprominent spiral arms are seen around the cavity. The companionthus already explains why spiral arms are present around the cavitywithout needing to invoke gravitational instability or other physics.Comparing the R in = 50 au calculations shown in Figure 2 showsthat both calculations produce the same cavity sizes, with the cavitysize approximately constant after (cid:38) orbits of the central binary.Comparing our results with scattered light (Avenhaus et al.2014, 2017), 2 micron (Casassus et al. 2012, 2013) images ofHD142527 shows that the blue orbits (top row) tend to producespirals inconsistent with the observations. For a more quantitativecomparison, we fit the spiral structure with a polynomial of theform r ( θ ) = (cid:80) i =0 a i θ i . Aside from the formula used to fit thespirals, the fitting procedure is otherwise identical to that describedin Christiaens et al. (2014). That is, we trace pixels along the max-ima in column density and fit a polynomial to the resulting pointsusing a least squares fit. Using this procedure we fit ‘inner’ and‘outer’ spiral arms spanning the position angle ranges [260 ◦ , ◦ ] and [180 ◦ , ◦ ] , respectively, corresponding to the two main spi-rals seen to the south west of the cavity in the scattered light image. Red lines in Figure 5 show the corresponding fits. For orbits B1 andB2 no inner spiral could be fitted, so we only show the outer arm.Table 2 lists the resulting pitch angles for the outer spiral mea-sured at a position angle (PA) of ◦ , and with the first and secondhalf of the spiral (in position angle), corresponding to PA ranges [180 ◦ , ◦ ] (third column) and [225 ◦ , ◦ ] (fourth column). Or-bits B1 and B2 in particular show pitch angles too small and little orno asymmetry in the gas distribution around a cavity which is closeto circular. Orbit B3 shows promising spirals to the north-east ofthe cavity, but the spiral arms to the south-west show a series ofalmost-circular tightly wrapped arms not seen in the observations.The red orbits, by contrast, produce spiral structure and asym-metry in both the cavity and the gas distribution more similar towhat is observed (bottom row). In particular, orbit R1 is the onlysimulation to show a bifurcation in the spiral arms to the south-westof the cavity, as seen in the scattered light image (see Table 2). Or-bit R2 is closest to the observed cavity size (Fig. 3) and developsan eccentric cavity similar to those found in previous circumbinarydisc simulations (e.g. Farris et al. 2014; D’Orazio et al. 2016; Ra-gusa et al. 2016) and used by Ragusa et al. (2017) to explain dusthorseshoes. The spiral arms appear fixed in relation to the binaryorbit, but superimposed on this is a precessing overdensity whichprecesses on a timescale of 2–3 orbital periods, as found by Dun-hill, Cuadra & Dougados (2015). Orbit R3 shows the most openspiral arms (Table 2) but the overdensity around the cavity is lessprominent, making it less promising for the observed dust struc-tures. The cavity size is also too big (Fig. 3). MNRAS , 1–16 (2017)
Price et al. t=2358 yrs100 au
Figure 6.
Shadows. Column density in the R1 orbit simulated with initial R in = 50 au after 20 orbits at the observed orbital phase (right), showing theorientation of the (transient) circumprimary disc, compared to the scattered light (600-900nm) ZIMPOL polarisation image (left; taken from Figure 1 ofAvenhaus et al. (2017) ©AAS, reproduced with permission). Dotted line indicates the expected shadow from our simulated inner disc (right), which lies closeto the orientation of the observed shadows (left). ]R2 + large dust -3 ]R2 + medium dust Figure 7.
Dust. Dust column density in dust-gas simulations using orbit R2 with mm grains (left panel) and 100 µ m-sized grains (right panel). Our ‘large’(mm) grains are close to Stokes number of unity, and hence quickly migrate to form a thin ring at the cavity edge. Decreasing the grain size by a factor of ten(right panel) produces a more radially extended dust structure. Figure 2 shows the formation of transient circumprimary discs inthe R in = 50 au calculations during the first 20 orbits, caused bythe clearing of the inner disc material. The orientation of these innerdiscs are highly sensitive to the orbit of the companion. For exam-ple, the circumprimary disc in calculation B3 is formed with majoraxis aligned east-west in our images (i.e. horizontal in Figure 2) ,while orbits R1 and R2 produce a disc aligned north-south (i.e. ver- tical in Figure 2) — a second piece of evidence favouring the redorbit family. Caution is required, however, since the inner disc pre-cesses with time, though on a timescale longer than our simulations( ∼ . Myr; e.g. Owen & Lai 2017). Furthermore, in our R in = 90 au calculations we find rotationally supported circumprimary discsonly with the R3 orbit at this resolution (Figure 5; this mainly indi-cates that the disc mass in other calculations is too low or that thenumerical viscosity drains the disc too fast at this resolution; notthat circumprimary discs do not exist). MNRAS000
Dust. Dust column density in dust-gas simulations using orbit R2 with mm grains (left panel) and 100 µ m-sized grains (right panel). Our ‘large’(mm) grains are close to Stokes number of unity, and hence quickly migrate to form a thin ring at the cavity edge. Decreasing the grain size by a factor of ten(right panel) produces a more radially extended dust structure. Figure 2 shows the formation of transient circumprimary discs inthe R in = 50 au calculations during the first 20 orbits, caused bythe clearing of the inner disc material. The orientation of these innerdiscs are highly sensitive to the orbit of the companion. For exam-ple, the circumprimary disc in calculation B3 is formed with majoraxis aligned east-west in our images (i.e. horizontal in Figure 2) ,while orbits R1 and R2 produce a disc aligned north-south (i.e. ver- tical in Figure 2) — a second piece of evidence favouring the redorbit family. Caution is required, however, since the inner disc pre-cesses with time, though on a timescale longer than our simulations( ∼ . Myr; e.g. Owen & Lai 2017). Furthermore, in our R in = 90 au calculations we find rotationally supported circumprimary discsonly with the R3 orbit at this resolution (Figure 5; this mainly indi-cates that the disc mass in other calculations is too low or that thenumerical viscosity drains the disc too fast at this resolution; notthat circumprimary discs do not exist). MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc Dust trapping in HD 142527 9
Figure 6.
Statistics of the multi-frequency morphological di↵erences inferred from Monte-Carlo (MC) simulations. All imageshave been restored from uvmem models. a) : restored ATCA image at 34 GHz. b) : average of ATCA observations with 200realizations of noise on the band 7 data. c) : average of the MC simulations for band 9 d)-g) : di↵erent realizations of noise. h) : rms scatter of the band 9 simulations. i) : rms scatter of the band 7 simulations. the following temperature profile: T = 70 K , if r <
30 AU , or (C1) T = 70 K(r / (30 AU) . if r
30 AU . However, setting the gas temperature to a representativedust species in the outer disk (we chose the average popu-lation of amorphous carbon grains) attenuates the depthof the decrements - they are restricted to the outline of the underlying optical depth field. The decrements arereproduced by the above prescription for the gas temper-ature because the gas temperatures in the crescent, at r = 140 au, are ⇠
30 K, and cooler than the continuumtemperatures in the model, which reach 50 to 60 K. Thusin this case the gas acts like a cold foreground on a hot-ter continuum background, which after continuum sub-traction appears like foreground absorption. In fact, COat higher altitudes could be hotter than the background ]R2 + large dust -3 ]R2 + medium dust Figure 8.
Horseshoe. Dust column density comparing observations (top) to our simulations (bottom). We show orbit R2 with grain sizes of 1 mm (‘large’)and 100 µ m (‘medium’) (bottom panels, left and right respectively) smeared to a beam size of 20 au (white circle). The larger grains (bottom left) show dustcolumn densities consistent with the ATCA observations at cm wavelengths while the smaller grains (bottom right) produce an overdense ‘horseshoe’ in thedust surface density at the correct position angle to explain the ALMA observations (top right). Top row credit: Figure 6 of Casassus et al. (2015b) ©AAS,reproduced with permission. Figure 6 shows the resultant inner disc structure after 20 orbits(left), shown alongside the scattered light image taken from Aven-haus et al. (2017) (left). Despite the remaining uncertainty in theorbital dynamics the position angle of the expected shadow is con-sistent with the northern shadow seen in the scattered light image,and within 10 ◦ of the southern shadow. A difference of 10 ◦ is notsignificant — shadows do not exactly fall in the projected planeof the inner disc due to the vertical extension of the discs (Min et al. 2017). We do not imply that this is the only possible orbitalconfiguration which can explain the shadows — nor even the mostprobable — merely that it is possible to produce a satisfactory ori-entation of the inner disc to produce the correct shadow from ourcalculations. We also demonstrate that the orbital dynamics of thecompanion naturally produces a circumprimary disc with an ori-entation and size consistent with the structure invoked by Marinoet al. (2015a) to explain the observed shadowing. MNRAS , 1–16 (2017) Price et al.
Figure 9.
Gas. Predicted CO J=3–2 and C O J=3–2 emission (moment 0) from simulation R2 after 50 orbits (bottom two panels), compared to Boehleret al. (2017) ALMA data (top two panels; credit: Figure 1 of Boehler et al. (2017) ©AAS, reproduced with permission). The ring-like feature seen in bothspectral lines is reproduced in our simulations. The ‘broken ring’ effect seen in the observations is not reproduced in our models because we do not accountfor the temperature dip caused by the shadows from the circumprimary disc.
Ragusa et al. (2017) found that dust horseshoes similar to those ob-served by ALMA could be naturally produced by eccentric cavitiesin gas and dust around binary stars. However, that paper assumedtightly coupled dust grains such that the dust structures merely re-flected those in the gas. Whether or not this is the case for the mmcontinuum emission in HD142527 (see discussion in Casassus et al.2015b) depends on the Stokes number — the ratio of the dust stop-ping time to the disc orbital period.Assuming subsonic Epstein drag the Stokes number dependsonly on the gas surface density, according to (Dipierro et al. 2015b) S t = 1 (cid:18) Σ0 . g cm − (cid:19) − (cid:18) ρ grain g cm − (cid:19)(cid:18) s grain mm (cid:19) , (2)where ρ grain is the intrinsic grain density and s grain is the grainsize. Modelling dust-gas dynamics in discs is usually uncertain be-cause the gas surface density is poorly constrained, being measured only from multiplying the dust continuum emission by a factor(typically 100). For HD142527 we are already confident in our as-sumed surface density profile because of our match to the measuredgas mass inside the cavity and to the surface density profile (Fig. 3).However, there remains uncertainty in the assumed CO-to-H con-version.Our assumed gas disc mass of 0.01 M (cid:12) corresponds to Σ =0.6 g cm − at R = R in , giving a Stokes number for mm grainsof ≈ ≈ at 300 au (see Fig. 3). Thus we expectdecoupling of grains in the outer disc, since S t = 1 corresponds tomaximal radial drift (Weidenschilling 1977).Since orbit R2 produces a cavity close to the observed sizewith a prominent asymmetry seen in the gas in the position anglesimilar to the observed mm horseshoe, we computed two additionalsimulations using grain sizes of mm and 100 µ m, respectively.We set up dust disc initially between R in = 120 and R out = 250 au using . × dust particles, with an (arbitrary) initial dust-to-gas ratio of 0.01 and properties otherwise reflecting the gas disc(composed of the usual SPH particles). The smaller dust disc is
MNRAS000
MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc merely to avoid numerical problems during the initial disc responseto the binary. This also ensures that any dust migration to the cavityedge occurs naturally rather than as a result of the initial conditions.Figure 7 shows the dust column density in these two simula-tions (left and right, respectively), shown after 62 and 63 binaryorbits, respectively, such that the orbital phase of both the binaryand the dust structures are consistent with the observations. Wefind that the dust in both simulations drifts radially, concentrating atthe cavity edge within a few tens of orbits. The larger grains, with S t ∼ , form a thin ring around the cavity edge (left panel). We alsoobserve dust grains collecting into azimuthally distinct structures,trapped by the pressure bumps at the locations where spiral armsin the gas cross the dust ring. Although these grains are nominally‘mm-sized’ in our simulations, the resulting dust structures appearmore similar to what is seen at cm wavelengths. Figure 8 showsa direct comparison with the ATCA 34 GHz image from Casassuset al. (2015b) (top left shows the ATCA image; bottom left showsour simulation). To make this comparison we simply convolved ourdust column density image to a beam size of 20 au (0.12”). In hind-sight this is not surprising, since the peak emission in wavelengthis roughtly π times the grain size.Using grains ten times smaller produces slower migration tothe cavity edge and, as a result, a more radially extended dust dis-tribution (right panel of Figure 7). The asymmetry driven by thebinary in the gas produces a horseshoe remarkably similar to theobserved mm horseshoe. Figure 8 makes the direct comparison.The predicted continuum emission is shown in red in Figure 11.The main disagreement between the simulations and observa-tions concerns the prominent dip seen in the mm-horseshoe at a po-sition angle of ∼ ◦ (top right). Caution is required in making thiscomparison since our visualisation shown in Figure 8 assumes op-tically thin dust emission where spectral index variations show thatthe continuum emission is optically thick (Casassus et al. 2015b).Our models also do not account for any azimuthal variation in tem-perature. This is interesting, because a temperature decrement isindeed observed at this position angle caused by the shadow fromthe circumprimary disc. Such a temperature decrement will affectthe dust emission and would need to be accounted for in makingaccurate radiative transfer predictions from our simulations. More-over, it suggests that thermodynamic effects from the shadow maybe important (see Montesinos et al. 2016).Two conclusions stand out: i) decoupled dust dynamics arounda binary-carved cavity can naturally explain the observed asymme-tries in HD142527 without recourse to vortices or additional com-panions; and ii) the distinct ‘blobs’ seen in the radio emission maybe real and not just artefacts of noisy observations. Casassus et al. (2015b) note that some of the asymmetry seen inthe dust emission is also seen in the gas. To quantify this, Figure 9compares the predicted CO J=3–2 and C O J=3–2 (moment 0)emission maps from simulation R2 (bottom row) to the ALMA ob-servations recently published by Boehler et al. (2017) (top row; seealso Muto et al. 2015). For both spectral lines we find a bright,asymmetric ring of emission surrounding the cavity at a radius be-tween 0.5 and 1” from the central source, as seen in the observa-tions.The main source of disagreement is that we do not reproducethe two dips in emission at the position angles coincident with thescattered light shadows. This again suggests that the shadowing ofthe outer disc by the circumprimary disc needs to be accounted for r [au] v r [ k m / s ]
10 10
20 500510
Figure 10.
Fast radial flows. Maximum inflow velocity on the SPH particlesbinned as a function of radius, in our model R2 (red), compared to the radialvelocity model used to fit the kinematic data (from Casassus et al. 2015b;solid blue line). Fast radial flows of order 10 km/s occur naturally in themodels caused by the streamers which penetrate the cavity. in the CO emission. Our models suggest that the underlying gasdensity structure is more axisymmetric. If one neglects the dipscaused by the shadows, then the contrast in emission around thecavity is similar between our simulations and the observations (i.e.roughly a factor of two).Again, the close match with the observed line emission doesnot suggest alternative hypotheses other than the binary are needed.
Can the streamers seen in Figure 2 explain the fast radial flows?Figure 10 shows the radial velocity of the SPH particles, binned asa function of radial distance from the centre of mass (red line), com-pared to the model used to fit the kinematic data by Casassus et al.(2015b) (blue line). Inflow speeds can be seen to reach 10 km/s at adistance of 20–30 au, consistent with the ‘fast radial flows’ neededto fit the kinematic data. This suggests that these flows indeed orig-inate in the streams of material that feed the inner disc.
One of the great mysteries in HD142527 concerns the origin ofthe ‘filaments’ of gas seen across the cavity in HCO+ emission byCasassus et al. (2013). These were seen only in HCO+ emission ina particular range of velocity channels. Figure 11 compares the pre-dicted HCO+ J=4–3 emission from our simulations (right panel) tothe corresponding figure from Casassus et al. (2015a) (left panel).As in the observational figure, we show our predicted continuummap for the mm grains (extrapolated from our ”medium grains”dust simulation) in red, with the predicted HCO+ emission in se-lected velocity channels in green and the predicted CO emissionin blue.Our predicted HCO+ emission shows a thin, faint filamentcrossing the cavity (right panel), remarkably similar to what is ob-served (left panel). This suggests that this feature is indeed of phys-ical origin, originating in the streams of material that feed the cir-cumprimary disc across the cavity. This serves as further confir- MNRAS , 1–16 (2017) Price et al. −
2 0 2 − Δ RA ["] Δ D e c [ " ] Intra-cavity kinematics 9
Filaments?
Figure 8.
Summary of Cycle 0 band 7 observations, from MEMmaps, with continuum in red, HCO + (4-3) in green, and CO(3-2)in blue. Velocities have been restricted to highlight the fainterstructures seen in HCO + , which are otherwise dwarfed by the fastHCO + central emission. ing the UV radiation required to produce HCO + in thecavity, resulting in HCO + filaments with wide openingangles, as opposed to the thin protoplanetary stream-ers predicted in hydrodynamic simulations. A problemwith this interpretation is that the outer ring is fairlyround in HCO + , there are no obvious counterparts inHCO + of the scattered-light shadows. Perhaps this re-flects a di↵erent production chain for HCO + in the outerdisk, where it could be driven by charge-exchange withcosmic-ray induced H +3 . CONCLUSION
The new CO(6-5) data, along with the orientation ofthe disk inferred from the scattered light shadows, have allowed us to understand the intra-cavity kinematics inHD 142527. Stellocentric accretion, starting from theouter disk and reaching close to free-fall velocities, witha steady-state mass flow fixed at the observed stellaraccretion, is consistent with the bulk properties of theavailable CO isotopologue data. The observed HCO + flows are also consistent with this stellocentric accretion,but with an emissivity that is somehow modulated in az-imuth. Fine structure in CO(6-5) also suggests non-axialsymmetry inside the cavity.While the data are consistent with a continous, yetabrupt and fast warp linking the two non-coplanar disks,further observations are required to understand the de-tailed structure of the intra-cavity kinematics and theinner warp. For instance, the large relative inclinationobserved between the inner and outer disks suggests thatthe fast accretion could be due to disk tearing. If thelow-mass companion is contained in the inner disk andis breaking the disks, then its orbit would also be highlyinclined with respect to the outer massive disk. The com-panion may then undergo Kozai oscillations, with high-eccentricity periods that may perhaps explain the largecavity.Based on observations acquired at the ALMA Observa-tory, through program ALMA . ALMA isa partnership of ESO, NSF, NINS, NRC, NSC, ASIAA.The Joint ALMA Observatory is operated by ESO,AUI/NRAO and NAOJ. Financial support was providedby Millennium Nucleus RC130007 (Chilean Ministryof Economy), and additionally by FONDECYT grants1130949, 1141175, 3140601, 3140634, 3140393. S.M.acknowledges CONICYT-PCHA / Magister Nacional /2014-22140628. PR and VM acknowledge support fromCONICYT-ALMA grant alma-conicyt 31120006. AD ac-knowledges CONICYT-ALMA grant 31120007. PJA ac-knowledges support from NSF award AST 1313021. MMacknowledges CONICYT-Gemini grant 32130007. Thiswork was partially supported by the Chilean supercom-puting infrastructure of the NLHPC (ECM-02).
APPENDIX
OBSERVATIONS
ALMA CO(6-5) data
Details on the instrumental setup are described in a companion article on the continuum emission (Casassus et al.2015). We used self-calibration to improve the dynamic range of the continuum images. Applying the same gaincorrections to the line data also resulted in improved dynamic range. Before self calibration, the peak signal in thesystemic velocity channel reached 0.42 Jy beam , in natural weights (beam of 0 . ⇥ . ), with a noise level of0.08 Jy beam clearly dominated by systematics rather than thermal noise. After self calibration, the peak signalincreased to 0.67 Jy beam , while the rms noise level decreased to 0.04 Jy beam .Continuum subtraction under CO(6-5) was performed with a first-order fit to the continuum in the visibility domain.The subset of channels neighboring the line was then split o↵ into another datafile, and subsequently re-sampled infrequency into the LSRK frame.The CO(6-5) dataset is presented in channel maps in Figs. 9 and 10, with three-channel bins. We chose to presentboth the restored image and the underlying MEM model. Observations Simulation Filaments
Figure 11.
Streamers. Predicted HCO+ emission from simulation R2 (right; HCO+ in green), compared to the corresponding Cycle 0 image (left; credit:Figure 8 of Casassus et al. (2015a) ©AAS, reproduced with permission). Both images show the mm dust horseshoe in red with HCO+ emission shown ingreen, with CO emission in blue. A filament can be seen crossing the cavity in our simulations, similar to what is observed (albeit with a different positionangle, indicating our orbit is not the correct one). We thus identify this feature with the streams of material feeding the primary across the circumbinary cavity. mation that the dynamical interaction with the binary companionis the source of almost all of the mysterious features present inHD142527. It also suggests that HCO+ should be more widely em-ployed to detect intra-cavity flows in circumbinary discs. The maindiscrepancy our comparison with observations in Figure 11 is thatthe position angle of the stream differs from our simulation. Thismainly indicates that orbit R2 is not the true orbit.
Finally, Figure 12 compares the predicted moment 1 maps fromsimulation R2 in CO and C O emission (bottom row) with theequivalent maps taken from Boehler et al. (2017) using ALMA data(top; see also Muto et al. 2015). Although the broad pattern is sim-ilar, there is a small discrepancy between the inner disc kinematicsin the observations (top) compared to our simulations (bottom). Weattribute this to the secular change in the disc orientation caused bythe torque from the binary. This occurs because orbit R2 is not astable binary-disc configuration. In particular, the torque from thebinary on the outer disc vanishes only when the binary is eitherplanar or perpendicular to the disc. It may therefore be fruitful infuture investigations to search for orbits which match the observeddata but are consistent with one of these arrangements (see discus-sion below).
If the observed companion can explain the otherwise unexplainedfeatures observed in HD142527 disc, this has potential implicationsfor our understanding of mm-bright transition discs in general (c.f. Casassus 2016; Owen 2016). These represent the fraction of tran-sition discs with large, mm-bright cavities and high mass accretionrates. Observations of transition discs with mass flow onto the cen-tral star continuing unabated despite the presence of a large cavityalready led numerous authors to conclude that the transition discpopulation is not a homogeneous class (Najita et al. 2007; Alexan-der & Armitage 2007; Owen & Clarke 2012) and that the likelyexplanation for the mm-bright subclass is the presence of ‘objectsmassive enough to alter the accretion flow’ (Najita et al. 2015).Such an explanation was offered as far back as Skrutskie et al.(1990).Cavities in transition discs are thought to be explained by ei-ther photoevaporation or companions — the latter either by dynam-ical clearing, similar to gap opening, or by having simply usedup the dust to form planets. We have shown that the cavity inHD142527 can be satisfyingly explained by dynamical interac-tion with the observed binary companion, hereby reclassifying itas a circumbinary disc. The streams of gas across the cavity pro-vide a natural explanation for the high accretion rate. Indeed, theaverage accretion rate in circumbinary discs with
H/R ∼ . is expected to be similar (within a factor of a few) to that in adisc around a single object (e.g. Farris et al. 2014; Ragusa et al.2016). In our simulations we measure a mass accretion rate of ≈ − M (cid:12) / yr onto the primary, consistent with observational es-timates of ± × − M (cid:12) / yr (Mendigut´ıa et al. 2014).Interestingly, HD142527 is not the first transition disc to havebeen reclassified as circumbinary. Similar reclassifications weremade for CoKu Tau/4 (Ireland & Kraus 2008) (inspiring the firstpart of our title) and CS Cha (Espaillat et al. 2007; Guenther et al.2007). This is largely due to the difficulty in detecting close-in com-panions. HD142527 demonstrates that it is easy to hide even rela- MNRAS000
H/R ∼ . is expected to be similar (within a factor of a few) to that in adisc around a single object (e.g. Farris et al. 2014; Ragusa et al.2016). In our simulations we measure a mass accretion rate of ≈ − M (cid:12) / yr onto the primary, consistent with observational es-timates of ± × − M (cid:12) / yr (Mendigut´ıa et al. 2014).Interestingly, HD142527 is not the first transition disc to havebeen reclassified as circumbinary. Similar reclassifications weremade for CoKu Tau/4 (Ireland & Kraus 2008) (inspiring the firstpart of our title) and CS Cha (Espaillat et al. 2007; Guenther et al.2007). This is largely due to the difficulty in detecting close-in com-panions. HD142527 demonstrates that it is easy to hide even rela- MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc V e l o c i t y V e l o c i t y D i s p e r s i on Figure 5 . Top: Maps of the velocity centroid for the CO and C O J=3-2 lines plotted between 0.8 to 6.5 kms . Contoursstart from 1.75 km s and are spaced by 0.5 kms . The systemic velocity is 3.75 km s . Bottom: map of the velocitydispersion plotted between 0 to 0.65 km s . As a reference, contours have been drawn at 0.2, 0.4 and 0.6 km s . This region coincides with the faint compact continuumsource discussed above (source c). It is also in this regionthat Casassus et al. (2013) claimed to have observed gasstreamers in HCO+.A CO J=3-2 line spectrum extracted from the regionof large velocity dispersion is shown on the top panel ofFigure 7. The spectrum presents a large shoulder inthe velocity range -1.150 km s to 0.3 km s . Theintegrated intensity map of the line emission in this ve-locity range is displayed on the bottom panel of Figure 7. Due to its association with the local dust continuumemission c , we argue that this emission might arise froma planetary disk or streamers accreted through a giantplanet. ANALYSIS4.1.
On the anti-correlation between dust andmolecular line emission
A decrease in the molecular line intensity from the diskregion corresponding to the dust continuum crescent has −2 0 2−2 0 2 −2 0 2 CO ∆ RA ["] ∆ D e c [ " ] C O ∆ RA ["] v [ k m . s − ]
1 2 3 4 5 6
Figure 12.
Kinematics. Predicted moment 1 emission maps in CO 3–2 and C O from our simulation R2 (bottom left and right, respectively) compared tothe observations with ALMA (top; credit: Boehler et al. (2017) ©AAS, reproduced with permission). tively massive companions inside discs, given that the companionin this case is massive (stellar!) but was only discovered in 2012due to its close proximity to the central star (Biller et al. 2012).In a recent study, Ragusa et al. (2017) found that circumbi-nary discs could naturally explain the cavity-edge rings, asymme-tries and horseshoes seen in ALMA images of mm-bright transitiondiscs including HD135344B, HD142527, SR21, DoAr 44 and LkH α MNRAS , 1–16 (2017) Price et al. ets, however, could perform a similar role in carving central holesin discs, as shown by Ragusa et al. (2017).No modelling is perfect, and despite some success, there area number of remaining caveats to our modelling of HD142527.The main one from our perspective is the secular change in boththe binary and outer disc orientation on long timescales ( (cid:38)
100 or-bits). That is, the binary and the disc in our simulations are notin a steady configuration. Recent work by Aly et al. (2015), Mar-tin & Lubow (2017) and Zanazzi & Lai (2018) suggests that theequilibrium configuration for eccentric binaries inclined by morethan ∼ ◦ is for the disc to align perpendicular to the binary (i.e.in a polar alignment). An interesting follow up would be to try toreproduce the disc features with a binary in such an equilibriumconfiguration, to see whether it can be ruled in or out (a binary or-bit with HD142527B at 90 ◦ to the outer disc seems possible; seeL16). An equilibrium configuration is likely given the typical align-ment timescale of order one thousand orbital periods (Martin &Lubow 2017) — years in HD142527 — is much shorter thanthe disc lifetime ( ∼ Myrs). Using better orbital constraints to de-termine whether the binary is in or out of equilibrium with the disccould therefore place strong constraints on formation models.A second caveat is the mismatch in the orientation of thestreams seen in HCO+ emission. Better orbital constraints shouldalso help to solve this problem, since there are currently a widerange of possible orbits. The observed orientation of the streamersitself presents a powerful constraint on the orbit.If large cavities are produced by eccentric massive compan-ions in the disc plane they should be observed, on average, at largeprojected separations, possibly even embedded in the disc (as ap-pears to be the case in HD100546). Polar orbits could help tosolve this coincidence problem since for a face on disc (such asHD142527) the range in projected separation on the sky is muchsmaller. For the orbits we chose for HD142527, the companionspends 48%, 43% and 31% of its period at projected separationsless than 20 au (0.13”) for orbits B1, B2 and B3, respectively. Inorbits R1, R2 and R3 this fraction is 19%, 22% and 15%, respec-tively. So the coincidence problem is not too severe with our chosenorbits.The most valuable observational follow up would thus be tobetter constrain the binary orbit and the companion mass, since thisdirectly influences the modelling. Better observations of the innerdisc similar to recent observations by Avenhaus et al. (2017) wouldbe very valuable in helping to constrain its geometry, mass andorientation. Finally, we found that kinematic data provides a richsource of information on the cavity dynamic. In particular, HCO+is a powerful probe of the streams of gas crossing the cavity. De-tection of similar streamers in other mm-bright transitional discswould be a powerful way to infer the presence of hidden compan-ions.That such eccentric and inclined binaries appear to exist in na-ture opens many fruitful avenues for theoretical investigation intowarped and inclined discs around binaries. For example, Owen &Lai (2017) proposed that large misalignments of inner and outerdiscs as in HD142527 may occur through a resonance between theprecession period of the inner disc and the precession of the sec-ondary. However, they predict that the orbital plane of the binaryshould be aligned with the outer disc, which does not appear to betrue in HD142527 based on L16 and our models.Another interesting direction (in our view) would be to try toinfer the binary orbit from the spiral pattern induced around thecavity. This should be possible given sufficient observational con-straints on the temperature structure of the disc. Current analytic prescriptions for spiral arms from companions work only in thelinear regime (i.e. for low mass companions; see Ogilvie & Lubow2002; Rafikov 2002), leading to potentially misleading conclusionsregarding the number and mass of the required companions (e.g.Stolker et al. 2016).The dynamical and thermodynamical influence of the circum-primary disc shadow on the outer disc in HD142527 also presentsan interesting avenue for further investigation. For example, Mon-tesinos et al. (2016) showed that shadows can trigger additionalspiral arms in the outer disc. (i) The cavity, spiral arms, shadows, dust horseshoe, gap-crossing filaments and fast radial flows seen in HD142527 can allbe explained, in part or in full, by the interaction with the observedcentral binary companion. HD142527 should therefore be firmlyreclassified as a circumbinary rather than transitional disc.(ii) Orbits drawn from the best fitting orbits considered by L16readily produce a cavity of the required size in HD142527, imply-ing that the observed binary is likely the origin of the large ≈
90 audust cavity in this disc. Constraints from the cavity size imply abinary semi-major axis a (cid:46) au.(iii) Comparison of the spiral structure and shadows with theobservations favours the ‘red’ family of orbits considered by L16with the binary approaching periastron, with large eccentricity e =0 . – . , and almost polar inclination with respect to the outer disc.(iv) Binary orbits from L16 with the companion approachingperiastron naturally produce an inclined circumprimary disc withmajor axis oriented north-south, fed by streams from the outer disc.This orientation of the inner disc is consistent with the radiativetransfer model used by Marino et al. (2015a) to fit the shadows.(v) We find radial velocities across the cavity of order 10 km/s,consistent with the observed ‘fast radial flows’. We thus offer anexplanation for fast radial flows in terms of the streams of materialconnecting the inner and outer discs.(vi) We find gas and dust are decoupled in HD142527. Dust mi-gration to the cavity edge produces features consistent with the ob-served dust emission, including the prominent mm-horseshoe.Given that all of the features present in HD142527 are presentin some or all mm-bright transition discs, explaining them may helpto explain this class of discs in general. For example, spirals andshadows seen in scattered light around the ∼ au dust cavity inthe disc of HD135344B (Garufi et al. 2013) are best explained byan ‘inner dust ring’ inclined by 22 ◦ and an ‘accretion funnel flow’onto the star (Stolker et al. 2016). In the context of the model wehave presented in this paper, these phrases sound eerily familiar. ACKNOWLEDGEMENTS
This project was initiated during the workshop on ‘Planet forma-tion in the era of ALMA and extreme AO’ in Santiago, Chile. Sec-tion 2 documents a lengthy discussion between many of us whilesharing a sushi lunch. DJP thanks S. Casassus, J. Cuadra, Univer-sidad de Chile, Pontificia Universidad Cat´olica de Chile and theMillenium Nucleus for their hospitality and financial support dur-ing two visits to Santiago. We thank Yann Boehler, Rebecca Mar-tin and Rebecca Nealon for useful discussions. DJP is funded byan Australian Research Council Future Fellowship FT13010034
MNRAS000
MNRAS000 , 1–16 (2017) ircumbinary dynamics in the HD142527 disc and Discovery Projects DP130102078 and DP180104235. We ac-knowledge CPU time on Gstar/SwinStar at Swinburne University,funded by the Australian Government, and on the MonARCH clus-ter at Monash. NC acknowledges financial support from FONDE-CYT grant 3170680. JC and NC acknowledge Millenium Nucleusgrant RC130007 (Chilean Ministry of Economy). GMK is sup-ported by the Royal Society as a University Research Fellow. JC ac-knowledges support from CONICYT-Chile through FONDECYT(1141175) and Basal (PFB0609) grants. We thank the anonymousreferee for comments which have improved the manuscript. REFERENCES
Alexander R. D., Armitage P. J., 2007, MNRAS, 375, 500Aly H., Dehnen W., Nixon C., King A., 2015, MNRAS, 449, 65Andrews S. M., et al., 2011, ApJ, 732, 42Artymowicz P., Lubow S. H., 1994, ApJ, 421, 651Ataiee S., Pinilla P., Zsom A., Dullemond C. P., Dominik C., Ghanbari J.,2013, A&A, 553, L3Avenhaus H., Quanz S. P., Schmid H. M., Meyer M. R., Garufi A., Wolf S.,Dominik C., 2014, ApJ, 781, 87Avenhaus H., et al., 2017, AJ, 154, 33Baruteau C., Zhu Z., 2016, MNRAS, 458, 3927Bate M. R., Bonnell I. A., Price N. M., 1995, MNRAS, 277, 362Benisty M., et al., 2015, A&A, 578, L6Benisty M., et al., 2017, A&A, 597, A42Biller B., et al., 2012, ApJ, 753, L38Boehler Y., Weaver E., Isella A., Ricci L., Grady C., Carpenter J., Perez L.,2017, ApJ, 840, 60Brittain S. D., Carr J. S., Najita J. R., Quanz S. P., Meyer M. R., 2014, ApJ,791, 136Bryden G., Chen X., Lin D. N. C., Nelson R. P., Papaloizou J. C. B., 1999,ApJ, 514, 344Canovas H., M´enard F., Hales A., Jord´an A., Schreiber M. R., Casassus S.,Gledhill T. M., Pinte C., 2013, A&A, 556, A123Canovas H., Caceres C., Schreiber M. R., Hardy A., Cieza L., M´enard F.,Hales A., 2016, MNRAS, 458, L29Casassus S., 2016, PASA, 33, e013Casassus S., Perez M. S., Jord´an A., M´enard F., Cuadra J., Schreiber M. R.,Hales A. S., Ercolano B., 2012, ApJ, 754, L31Casassus S., et al., 2013, Nature, 493, 191Casassus S., et al., 2015a, ApJ, 811, 92Casassus S., et al., 2015b, ApJ, 812, 126Christiaens V., Casassus S., Perez S., van der Plas G., M´enard F., 2014, ApJ,785, L12Christiaens V., et al., 2018, A&A, submittedClose L. M., et al., 2014, ApJ, 781, L30Cuadra J., 2016, in Resolving planet formation in the era of ALMA andextreme AO. Proceedings of the conference held 16-20 May, 2016 inSantiago, Chile, id.30. p. 30, doi:10.5281/zenodo.58297Currie T., et al., 2014, ApJ, 796, L30Currie T., Cloutier R., Brittain S., Grady C., Burrows A., Muto T., KenyonS. J., Kuchner M. J., 2015, ApJ, 814, L27D’Orazio D. J., Haiman Z., Duffell P., MacFadyen A., Farris B., 2016, MN-RAS, 459, 2379Dipierro G., Pinilla P., Lodato G., Testi L., 2015a, MNRAS, 451, 5493Dipierro G., Price D., Laibe G., Hirsh K., Cerioli A., Lodato G., 2015b,MNRAS, 453, L73Dipierro G., Laibe G., Price D. J., Lodato G., 2016, MNRAS, 459, L1Dong R., Zhu Z., Rafikov R. R., Stone J. M., 2015, ApJ, 809, L5Draine B. T., 2003, ApJ, 598, 1017Dunhill A. C., Cuadra J., Dougados C., 2015, MNRAS, 448, 3545Espaillat C., Calvet N., D’Alessio P., Hern´andez J., Qi C., Hartmann L.,Furlan E., Watson D. M., 2007, ApJ, 670, L135Espaillat C., et al., 2014, Protostars and Planets VI, pp 497–520Facchini S., Lodato G., Price D. J., 2013, MNRAS, 433, 2142 Farris B. D., Duffell P., MacFadyen A. I., Haiman Z., 2014, ApJ, 783, 134Fujiwara H., et al., 2006, ApJ, 644, L133Fukagawa M., Tamura M., Itoh Y., Kudo T., Imaeda Y., Oasa Y., HayashiS. S., Hayashi M., 2006, ApJ, 636, L153Gaia Collaboration et al., 2016, A&A, 595, A2Garcia Lopez R., Natta A., Testi L., Habart E., 2006, A&A, 459, 837Garufi A., et al., 2013, A&A, 560, A105Guenther E. W., Esposito M., Mundt R., Covino E., Alcal´a J. M., CusanoF., Stecklum B., 2007, A&A, 467, 1147Ireland M. J., Kraus A. L., 2008, ApJ, 678, L59Kley W., Dirksen G., 2006, A&A, 447, 369Kohn S. A., Shkolnik E. L., Weinberger A. J., Carlberg J. K., Llama J.,2016, ApJ, 820, 2Lacour S., et al., 2016, A&A, 590, A90Laibe G., Price D. J., 2012a, MNRAS, 420, 2345Laibe G., Price D. J., 2012b, MNRAS, 420, 2365Lodato G., Price D. J., 2010, MNRAS, 405, 1212Lubow S. H., D’Angelo G., 2006, ApJ, 641, 526Lyra W., Lin M.-K., 2013, ApJ, 775, 17Marino S., Perez S., Casassus S., 2015a, ApJ, 798, L44Marino S., Casassus S., Perez S., Lyra W., Roman P. E., Avenhaus H.,Wright C. M., Maddison S. T., 2015b, ApJ, 813, 76Martin R. G., Lubow S. H., 2017, ApJ, 835, L28Mendigut´ıa I., Fairlamb J., Montesinos B., Oudmaijer R. D., Najita J. R.,Brittain S. D., van den Ancker M. E., 2014, ApJ, 790, 21Min M., Stolker T., Dominik C., Benisty M., 2017, A&A, 604, L10Miranda R., Lai D., 2015, MNRAS, 452, 2396Monaghan J. J., 2005, Reports on Progress in Physics, 68, 1703Montesinos M., Perez S., Casassus S., Marino S., Cuadra J., Christiaens V.,2016, ApJ, 823, L8Muto T., et al., 2015, PASJ, 67, 122Najita J. R., Strom S. E., Muzerolle J., 2007, MNRAS, 378, 369Najita J. R., Andrews S. M., Muzerolle J., 2015, MNRAS, 450, 3559Nealon R., Price D. J., Nixon C. J., 2015, MNRAS, 448, 1526Nixon C., King A., Price D., 2013, MNRAS, 434, 1946Ogilvie G. I., Lubow S. H., 2002, MNRAS, 330, 950Ohashi N., 2008, Ap&SS, 313, 101Owen J. E., 2016, PASA, 33, e005Owen J. E., Clarke C. J., 2012, MNRAS, 426, L96Owen J. E., Lai D., 2017, MNRAS, 469, 2834Pearce T. D., Wyatt M. C., Kennedy G. M., 2015, MNRAS, 448, 3679Perez S., et al., 2015, ApJ, 798, 85P´erez L. M., et al., 2016, Science, 353, 1519Pinilla P., et al., 2012, A&A, 538, A114Pinte C., M´enard F., Duchˆene G., Bastien P., 2006, A&A, 459, 797Pinte C., Harries T. J., Min M., Watson A. M., Dullemond C. P., Woitke P.,M´enard F., Dur´an-Rojas M. C., 2009, A&A, 498, 967Pohl A., Pinilla P., Benisty M., Ataiee S., Juh´asz A., Dullemond C. P., VanBoekel R., Henning T., 2015, MNRAS, 453, 1768Pott J.-U., Perrin M. D., Furlan E., Ghez A. M., Herbst T. M., Metchev S.,2010, ApJ, 710, 265Price D. J., 2012, J. Comp. Phys., 231, 759Price D. J., et al., 2018, preprint, ( arXiv:1702.03930 )Quanz S. P., Amara A., Meyer M. R., Kenworthy M. A., Kasper M., GirardJ. H., 2013, ApJ, 766, L1Quanz S. P., Amara A., Meyer M. R., Girard J. H., Kenworthy M. A., KasperM., 2015, ApJ, 807, 64Quillen A. C., 2006, ApJ, 640, 1078Rafikov R. R., 2002, ApJ, 572, 566Raghavan D., et al., 2010, ApJS, 190, 1Ragusa E., Lodato G., Price D. J., 2016, MNRAS, 460, 1243Ragusa E., Dipierro G., Lodato G., Laibe G., Price D. J., 2017, MNRAS,464, 1449Reggiani M., et al., 2017, preprint, ( arXiv:1710.11393 )Rice W. K. M., Armitage P. J., Wood K., Lodato G., 2006, MNRAS, 373,1619Rodigas T. J., Follette K. B., Weinberger A., Close L., Hines D. C., 2014,ApJ, 791, L37MNRAS , 1–16 (2017) Price et al.
Rosenfeld K. A., Chiang E., Andrews S. M., 2014, ApJ, 782, 62Ru´ız-Rodr´ıguez D., Ireland M., Cieza L., Kraus A., 2016, MNRAS, 463,3829Sallum S., et al., 2015, Nature, 527, 342Siess L., Dufour E., Forestini M., 2000, A&A, 358, 593Skrutskie M. F., Dutkevitch D., Strom S. E., Edwards S., Strom K. M.,Shure M. A., 1990, AJ, 99, 1187Stolker T., et al., 2016, A&A, 595, A113Strom K. M., Strom S. E., Edwards S., Cabrit S., Skrutskie M. F., 1989, AJ,97, 1451Verhoeff A. P., et al., 2011, A&A, 528, A91Wagner K., Apai D., Kasper M., Robberto M., 2015, ApJ, 813, L2Weidenschilling S. J., 1977, MNRAS, 180, 57Williams J. P., Cieza L. A., 2011, ARA&A, 49, 67Zanazzi J. J., Lai D., 2018, MNRAS, 473, 603Zhu Z., Nelson R. P., Dong R., Espaillat C., Hartmann L., 2012, ApJ, 755,6van Boekel R., et al., 2004, Nature, 432, 479van der Marel N., et al., 2013, Science, 340, 1199van der Marel N., van Dishoeck E. F., Bruderer S., Andrews S. M., Pontop-pidan K. M., Herczeg G. J., van Kempen T., Miotello A., 2016, A&A,585, A58 MNRAS000