Disk Evolution Study Through Imaging of Nearby Young Stars (DESTINYS): Late infall causing disk misalignment and dynamic structures in SU Aur
C. Ginski, S. Facchini, J. Huang, M. Benisty, D. Vaendel, L. Stapper, C. Dominik, J. Bae, F. Menard, G. Muro-Arena, M. Hogerheijde, M. McClure, R. G. van Holstein, T. Birnstiel, Y. Boehler, A. Bohn, M. Flock, E. E. Mamajek, C. F. Manara, P. Pinilla, C. Pinte, A. Ribas
DDraft version February 18, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Disk Evolution Study Through Imaging of Nearby Young Stars (DESTINYS):Late infall causing disk misalignment and dynamic structures in SU Aur ∗ Christian Ginski ,
1, 2
Stefano Facchini , Jane Huang ,
4, 5
Myriam Benisty ,
6, 7
Dennis Vaendel, Lucas Stapper , Carsten Dominik , Jaehan Bae ,
8, 5
Franc¸ois M´enard , Gabriela Muro-Arena , Michiel R. Hogerheijde ,
2, 1
Melissa McClure , Rob G. van Holstein ,
9, 2
Tilman Birnstiel ,
10, 11
Yann Boehler , Alexander Bohn , Mario Flock , Eric E. Mamajek , Carlo F. Manara , Paola Pinilla , Christophe Pinte , and ´Alvaro Ribas Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098XH Amsterdam, The Netherlands Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, The Netherlands European Southern Observatory, Karl-Schwarzschild-Strasse 2, 85748 Garching bei M¨unchen, Germany Department of Astronomy, University of Michigan, 323 West Hall, 1085 S. University Avenue, Ann Arbor, MI 48109, USA NHFP Sagan Fellow Unidad Mixta Internacional Franco-Chilena de Astronom´ıa, CNRS, UMI 3386 and Departamento de Astronom´ıa, Universidad de Chile,Camino El Observatorio 1515, Las Condes, Santiago, Chile Univ. Grenoble Alpes, CNRS, IPAG, 38000 Grenoble, France Department of Terrestrial Magnetism, Carnegie Institution for Science, 5241 Broad Branch Road NW, Washington, DC 20015, USA European Southern Observatory (ESO), Alonso de C´ordova 3107, Vitacura, Casilla 19001, Santiago de Chile, Chile University Observatory, Faculty of Physics, Ludwig-Maximilians-Universit¨at M¨unchen, Scheinerstr. 1, D-81679 Munich, Germany Exzellenzcluster ORIGINS, Boltzmannstr. 2, D-85748 Garching, Germany Rice University, Department of Physics and Astronomy, Main Street, 77005 Houston, USA Max Planck Institute for Astronomy, K¨onigstuhl 17, 69117 Heidelberg, Germany Jet Propulsion Laboratory, California Institute of Technology, M/S 321-100, 4800 Oak Grove Drive, Pasadena, CA 91109, USA School of Physics and Astronomy, Monash University, Clayton, Vic 3800, Australia (Received September 29, 2020; Revised January 8, 2021; Accepted January 22, 2021)
Submitted to ApJLABSTRACTGas-rich circumstellar disks are the cradles of planet formation. As such, their evolution will stronglyinfluence the resulting planet population. In the ESO DESTINYS large program, we study these diskswithin the first 10 Myr of their development with near-infrared scattered light imaging.Here we present VLT/SPHERE polarimetric observations of the nearby class II system SU Aur inwhich we resolve the disk down to scales of ∼ Corresponding author: Christian [email protected] ∗ Based on observations performed with VLT/SPHERE under pro-gram ID 1104.C-0415(E) a r X i v : . [ a s t r o - ph . E P ] F e b Ginski et al.
Keywords:
Exoplanet formation(492) — Circumstellar disks(235) — Direct imaging(387) — Polarime-try(1278) INTRODUCTIONCircumstellar disks are the birth places of planetarysystems. Thus their physical properties strongly in-fluence the outcome of the planet formation processes.In turn, massive planets dramatically impact the diskstructure. Recent scattered light observations have re-vealed a number of disks with warps or misalignmentsof inner and outer disk regions (e.g., Marino et al. 2015;Benisty et al. 2017). The interaction with either planetsor stellar companions is frequently invoked to explainthese observations (e.g., Batygin 2012; Facchini et al.2018). Recently Bi et al. (2020) and Kraus et al. (2020)showed multiple misaligned rings supporting this sce-nario in the GW Ori triple system. However, many ofthe systems with inferred misalignments are around sin-gle stars (e.g., Pinilla et al. 2018; Muro-Arena et al.2020). Similarly, spin-orbit misalignment of transitingplanets possibly inherited from the gas-rich disk phase,is common in single stellar systems (e.g., Triaud et al.2010). An alternative scenario is that disk misalign-ments are natural consequence of angular momentumtransfer due to late infall of material on the disk (Thieset al. 2011; Dullemond et al. 2019). Observations prob-ing both large and small spatial scales have the potentialto test this possibility.SU Aur is a nearby (158.4 ± log ( L/L (cid:12) ) = 0 . +0 . − . M (cid:12) and an age range of 4-5.5 Myr.SU Aur is surrounded by extended circumstellar struc-ture first resolved in near infrared scattered light (Jef-fers et al. 2014). The signal extends up to 500 au andshows a strong asymmetry along the East-West direc-tion. Subsequent scattered light observations detecteda faint dust tail extending from the main disk towardthe West (de Leon et al. 2015). A strong azimuthalbrightness asymmetry is attributed to the dust scatter-ing phase function and to a higher surface density on thenorthern side of the disk. ALMA observations show aKeplerian disk and a gas tail that extends out to severalhundreds of au to the West (Akiyama et al. 2019), thatcould either be caused by a disruption of the disk, bya perturber, or trace cloud material accreting onto the disk.In this letter, we present new observations of SU Aurobtained as part of the DESTINYS program (Disk Evo-lution Study Through Imaging of Nearby Young StarsGinski et al. 2020) that aims to study the circumstel-lar environment of nearby T Tauri stars, complementedwith VLT/NACO, HST/STIS and ALMA archival data. OBSERVATIONSWe obtained new observations of SU Aur withVLT/SPHERE (Beuzit et al. 2019), and use archivaldata taken with VLT/NACO (program ID: 088.C-0924,PI: S. Jeffers) and ALMA (program ID: 2013.1.00426.S,PI: Y. Boehler).2.1.
SPHERE observations
SU Aur was observed on 14th of December 2019 withSPHERE/IRDIS in dual-beam polarimetric imagingmode (DPI, de Boer et al. 2020; van Holstein et al. 2020)in the broad band H filter with pupil tracking setting.The central star was placed behind an apodized Lyotcoronagraph with an inner working angle of 92.5 mas.The individual frame exposure time was set to 32 s anda total of 104 frames where taken separated in 26 polari-metric cycles of the half wave plate. The total integra-tion time was 55.5 min. Observations conditions wereexcellent with an average Seeing of 0.8 (cid:48)(cid:48) and an atmo-sphere coherence time of 6.7 ms.The data were reduced using the public IRDAP pipeline(IRDIS Data reduction for Accurate Polarimetry, vanHolstein et al. 2020). The images were astrometricallycalibrated using the pixel scale and true north offsetgiven in Maire et al. (2016).Since the data were taken in pupil tracking mode wewere able to perform angular differential imaging (ADI,Marois et al. 2006) reduction in addition to the polari-metric reduction, resulting in a total intensity image anda polarized intensity image. We show the final result ofboth post-processing approaches in figure 1. Note thatinstead of polarized intensity we show the radial Stokesparameter Q φ as is now standard in most studies. Wefollow the definition in de Boer et al. (2020): Q φ = − Q cos (2 φ ) − U sin (2 φ ) (1) IWA defined as the separation at which the throughput reaches50%. hadows, spirals and dust tails in SU Aur
NACO observations
SU Aur was observed with VLT/NACO in polarimet-ric imaging mode on 2nd of November 2011 in the Ksfilter. Observing conditions were fair with an averageSeeing of 0.8 (cid:48)(cid:48) and a coherence time of 3 ms. As NACOdoes not offer a coronagraph in polarimetric mode, shortindividual frame exposure times of 0.35 s were used. Atotal of 8160 frames were taken separated in 24 polari-metric cycles. This amounts to a total integration timeof 47.6 min. The data was taken in dithering mode inorder to allow for sky background subtraction.The data reduction was performed in principle analo-gous to the SPHERE data, however without the benefitof a detailed instrument model to determine instrumentpolarization. The instrumental polarization was thusestimated from the data, by placing a small apertureat the central star location and with an aperture ra-dius smaller than one resolution element, i.e. where wewould expect the polarimetric signal to be unresolvedand thus on average small. Other data reduction stepswere performed as described in Ginski et al. (2016) foran analogous data set of HD 97048. The resulting Q φ image is shown in figure 1.2.3. Archival ALMA observations
We retrieved CO and continuum data of SU Aurobserved as part of program 2013.1.00426.S (PI: Y.Boehler) from the ALMA archive. The 880 µ m contin-uum and CO J = 3 − J = 3 − CO J = 2 −
1, and CO J = 2 − CO J = 2 − CO J = 2 − CO line are better thantheir Band 6 counterparts.The SU Aur data downloaded from the archive wereprocessed with the ALMA pipeline in CASA v. 4.5.3.Subsequent self-calibration and imaging took place inCASA v. 5.4.0. Channels with line emission wereflagged and the SPWs were spectrally averaged to formcontinuum datasets. The fixvis and fixplanets taskswere using to align the continuum peak positions of theexecution blocks within each band and to assign a com-mon phase center, respectively. One round of phase-selfcalibration was applied to the continuum data for theseparate bands. The self-calibration solutions were thenapplied to the full-resolution data. The uvcontsub taskwas used to subtract the continuum from the line spec-tral windows in the uv plane.A Briggs robust value of 0.5 was used with the ClarkCLEAN algorithm to produce the final 880 µ m contin-uum image, which has a synthesized beam of 0 . (cid:48)(cid:48) × . (cid:48)(cid:48) (26.4 ◦ ) and an rms of 0.1 mJy beam − . A Briggsrobust value of 1.0 was used with the multiscale CLEANalgorithm to produce the CO J = 3 − impbcor .The resulting syntheized beam is 0 . (cid:48)(cid:48) × . (cid:48)(cid:48) (26.4 ◦ )and the rms is 10 mJy beam − in channels 0.25 km s − wide. A Gaussian uv taper and a Briggs robust valueof 2.0 were applied to the weaker CO J = 2 − . (cid:48)(cid:48) × . (cid:48)(cid:48) (15.0 ◦ ) and the rms is 10 mJy beam − in channels 0.25 km s − wide. MORPHOLOGY IN SCATTERED LIGHTAs evidenced from figure 1, SU Aur shows a complexcircumstellar environment, with a disk, a dark lane, andlarge scale features. These features are discussed in de-tails in the following.3.1.
The circumstellar disk
The most striking morphological feature is a dark laneat a position angle of ∼ ◦ , as indicated in the upper-center panel of figure 2. This dark lane is also detectedin Subaru/HiCIAO polarimetric observations (de Leonet al. 2015) and in the NACO Ks-band (figure 1, bot-tom panel), and could trace shadowing by a misalignedinner disk. While the extent of the disk is not directlyevident from our SPHERE observations, the ALMA ob-servations indicate the presence of a circumstellar disk inthe continuum, and in the velocity channels of the CO J = 3 − ∼ (cid:48)(cid:48) (111 au). The inner Ginski et al.
SPHERE Q φ ∆ D e c [ a r c s e c ] SPHERE I ∆ RA [ arcsec ] NACO Q φ Figure 1.
SPHERE/IRDIS and NACO observations of the SU Aur system. We show SPHERE H-band Q φ polarized signalon the top, SPHERE H-band total intensity in the center panel and NACO K S -band Q φ polarized signal on the bottom. TheSPHERE coronagraph is marked with a hashed circle. disk (within 1 au) was resolved with near infrared inter-ferometry by Labdon et al. (2019). With image recon-struction they found it to be inclined by 52.8 ◦ ± ◦ with aposition angle of 140.1 ◦ ± ◦ , and the near side towardsthe West . We show the reconstructed image from theinterferometric data in figure 2, upper-left panel. Theinner disk position angle is close to the position angleof the dark lane seen in the outer disk. We comparedthe position angle of the dark lane between the archivalNACO data taken in 2011, the literature Subaru data, We note that we use the convention that at a position angle of0 ◦ the near side of a hypothetical disk is located to the West.In this convention the position angle reported by Labdon et al.(2019) is 320.1 ◦ . taken in 2014, and the new SPHERE data taken in 2019.Within this ∼ change in the orientation. A change in orientationwould be expected if the inner disk misalignment wascaused by mutual interactions with a short period bi-nary companion, a scenario found to be unlikely.We find that the disk is significantly brighter in theNorth-East than in the South-West. This was alreadyreported by de Leon et al. (2015), who interpreted thisas an azimuthally asymmetric dust distribution. How-ever this brightness asymmetry would be a natural con-sequence of the scattering phase function if we see an The calibration accuracy of SPHERE is on the order of 0.1 ◦ ,while for NACO it is on the order of 0.2 ◦ . hadows, spirals and dust tails in SU Aur ◦ ± ◦ for the outerdisk, which fits well with this interpretation. Given thisposition angle and assuming that the North-East sideis the near side of the outer disk, this implies a strongmisalignment between the inner and outer disk. Usingthe values provided by Labdon et al. (2019) for the innerdisk and our measurement for the outer disk we find arelative inclination of ∼ ◦ . This is consistent with thepresence of a narrow shadow lane.In addition to the brightness asymmetry, we see con-siderable sub-structure in the disk. In particular sev-eral spiral features are present. In figure 2, upper-right panel, we show the high-pass filtered version of theSPHERE Q φ image with the spiral features highlighted.We can clearly identify six features in the South-Westand four features in the North-East, with possible other(not highlighted) features in the East. The spirals inthe South-West have generally larger pitch angles (25 ◦ to 49 ◦ ) than the spirals in the North-East (7 ◦ to 22 ◦ ).This could be explained by projection effects on an in-clined scattering surface if the North-East side is indeedthe front side of the disk (Dong et al. 2016).3.2. The extended structure de Leon et al. (2015) reported the detection of a dusttail with Subaru/HiCIAO in scattered polarized light ex-tending from the disk around SU Aur roughly 2.5 (cid:48)(cid:48) to theWest. We find a similar tail structure in both SPHEREand NACO data. The lower signal-to-noise NACO datashow a single tail, while the SPHERE data show thatthis structure extends much further out than seen witheither HiCIAO or NACO and consists of several tailswith different curvatures and orientations. In figure 2the most prominent structures are annotated. There areat least 4 distinct tails that extend towards the West la-beled 1 to 4. The brightest tail in the SPHERE Q φ image is the southern-most of these structures, i.e., tail1. This tail is the structure seen in the NACO data andin the Subaru data. In the SPHERE image it becomesclear that this tail connects with the northern part ofthe Keplerian disk. Not only can we trace the tail struc-ture until it merges with the disk, but we can also seethe projection of the shadow lane from the outer diskon the dust tail. (see annotation in figure 2). The angleof the shadow changes as would be expected if the dusttail is approaching the disk from above the disk-plane,i.e., from in between the observer and the disk.We see several fainter tails located north of tail 1,marked with numbers 2-4. Some of them are more vis-ible in the total intensity ADI image shown in figure 1, middle panel, indicating that they likely have a low de-gree of polarization. We also see a structure extendingto the north at a significantly different angle than tail1-4, and labeled northern tail in figure 2. The northerntail appears to vanish just before it reaches the disk (seeannotation in figure and the n2), indicating that it iseither below tail 1 or behind the disk. In order to betterunderstand the geometry of the system we computed thedegree of linear polarization of the extended structures.We utilized an iterative reference differential imagingapproach (Vaendel et al. in prep.), complemented withangular differential imaging (Stapper et al., in prep.),briefly described in appendix A. The result is shown infigure 3. Both tail 1 and the northern tail stick out witha higher degree of linear polarization compared to thesurrounding structures. Assuming a standard bell curveto map the degree of linear polarization to the scatter-ing angles (e.g., Stolker et al. 2016a), both tail 1 andthe northern tail should be at intermediate scatteringangles, with an ambiguity between forward and back-scattering. However, tail 1 is significantly brighter in theSPHERE Q φ image than the northern tail (factor 1.5 to4 depending on the point of measurement). This is alsoevidenced by the fact that tail 1 is detected by SPHERE,NACO and Subaru, whereas the northern tail is only vis-ible in the highest signal-to-noise SPHERE data. Giventhat tail 1 can be smoothly traced until it connects withthe northern part of the disk (i.e., the near side), it isclear that the light from tail 1 is scattered with anglessmaller than 90 ◦ . Since the northern tail shows a simi-lar degree of polarization, but overall smaller signal, weconclude that light is scattered with angles larger than90 ◦ . This means the northern tail should be located be-hind the disk along the line of sight.In addition to the distinct tail-like structures, we see amore complex signal to the East of the disk. Followed bya zone where we see a distinct lack of signal toward theSouth-East (see annotation in figure 2). If the signal tothe East and South-East is located above the Kepleriandisk (i.e., closer to the observer), then the region with-out signal might be a natural continuation of the shadowlane visible on the disk. In particular the signal to theEast shows a very similar degree of linear polarizationto tail 1, indicating similar scattering angles. ALMA OBSERVATIONSAkiyama et al. (2019) presented ALMA Band 6 ob-servations of SU Aur, showing the dust continuum emis-sion of the Keplerian disk and revealing an extended tailstructure to the West in the gas. The dust continuumemission shows a marginally resolved disk without par-ticular features. In figure 10 we show an overlay of the
Ginski et al. ∆ RA [mas] ∆ D e c [ m a s ] -0.8 -0.6 -0.4 -0.2 0.2 0.4 0.6 0.8-0.8-0.6-0.4-0.20.00.20.40.60.8 ∆ D e c [ a r c s e c ] SPHERE Q φ -0.8 -0.6 -0.4 -0.2 0.2 0.4 0.6 0.8 ∆ RA [arcsec]∆ RA [arcsec] Labdon et al. 2019K-bandInterferometry d i s k m a j o r a x i s SPHERE, high-pass filtered s h a d o w l a n e ∆ D e c [ a r c s e c ] ∆ RA [arcsec] tail 1tail 2tail 3tail 4eastern signal northern taillack of signal shadow projected on tail 1northern tail signal vanishes Q Φ x r Figure 2.
Upper-Left:
Near-infrared interferometric image reconstructed by Labdon et al. (2019) (reproduction of theirfigure 3). We overlay the major axis of the inner disk that they recover.
Upper-Middle:
SPHERE/IRDIS Q φ image of theinnermost part of SU Aur. We overlay the major axis of the interferometric inner disk and the direction of the shadow laneseen in scattered light on the outer disk. Upper-Right:
Same as the middle panel, but after application of a high pass filter.We mark the visible spiral structures.
Bottom:
SPHERE Q φ image of SU Aur. Several large scale structures are annotated.The polarized flux was scaled by the inclination corrected square of the separation from the primary star to compensate for theillumination drop-off. mm continuum emission and the SPHERE data. Largedust particles are concentrated at the location of the diskalso seen with SPHERE but are not detected in the tailstructures to the West. The mm-emission appears lessextended than the scattered light, possibly indicating ef-ficient radial drift of the large dust particles.From a fit of a simple symmetric model to the contin-uum emission, we find a position angle of 122.9 ◦ ± ◦ and an inclination of 53.0 ◦ ± ◦ (see appendix B fordetails). Given that the gas and scattered light show ahighly asymmetric structure it is possible that this fit isaffected by systematic uncertainties.In figure 4 we show an overlay of two velocity channels ofthe CO 3-2 line emission with the SPHERE data. Thefirst channel corresponds to a velocity of ∼ − and is blue shifted relative to the intrinsic system veloc- hadows, spirals and dust tails in SU Aur ∆ RA [ arcsec ] ∆ D e c [ a r c s e c ] Figure 3.
SPHERE degree of linear polarization. The totalintensity image used is the iterative RDI reduction shown inappendix A. Regions for which the total intensity values werenot well recovered were set to 0. The inner disk is maskedsince it is dominated by artefacts in total intensity. ity ( ∼ − ; Akiyama et al. 2019), while the secondchannel corresponds to a velocity of ∼ − and isred-shifted. We show all channel maps in figure 11. Theblue-shifted frequency channels clearly trace the north-ern tail detected in the SPHERE data, while the redshifted channels trace the tails to the West, in partic-ular tail 1. Since we inferred from the scattered lightdata that tail 1 is above, and the northern tail below,the disk, we can thus conclude that both of these tailstrace material falling onto the disk from the surround-ing cloud. In order to check if the measured velocitiesare physical, we computed the free fall velocity aroundSU Aur and find values of 4-1.8 km s − for separation be-tween 200 au and 1200 au. This is compatible with theprojected velocities measured in the dust tails.Additionally, we see strong red-shifted signal to the Eastand South-East of the Keplerian disk. If the detectedscattered light signal is above the disk, then this mayindicate that we see additional in-fall of material fromthese directions. It may also be that we simply tracethe material falling in from the West as it is caught inKeplerian motion and spirals onto the disk.4.1. Possible foreground contamination
In the red-shifted velocity channels there is in generalemission with a velocity gradient from West to East,which may indicate that some of the signal is comingfrom the embedding cloud and not the infalling mate-rial. However, the blue-shifted channels are highly con-centrated on the circumstellar disk and the northern tail,thus this is less of a concern in this case. We investigatedif possible foreground contamination might be responsi-ble for some of the signatures seen in scattered light.For the shadow lane feature to be produced by optically thick foreground material, we would require a thin fila-ment aligned such that is crosses our line of sight towardsthe stellar position. To create a narrow feature like theshadow lane this filament would likely have to be closeto the star. In this case we would expect to pick up anillumination signature of such a hypothetical structurein scattered light, in particular since we would see it un-der small scattering angles and thus close to the peakof the total intensity scattering phase function. Such asignature is not visible in the SPHERE or NACO data(neither in polarized nor total intensity). Additionally ifsuch a structure was present, we would expected SU Aurto show significant extinction. The literature value forthe the extinction of SU Aur is A V ∼ Relative line fluxes in disk and tails
To test if the angular momentum transported by thedust tails is in principle sufficient to cause the misalign-ment of inner and outer disk that we discuss in sec-tion 3.1, we used the available CO and CO line datato qualitatively assess the mass ratio between the Keple-rian disk and the dust tails. Of the two line observationsone expects CO to be optically thinner than CO and
Ginski et al. thus to trace more closely the density of the gas.To estimate the mass ratio we integrated over the de-tected flux density in the moment 0 map of both datasets shown in figure 5. For the Keplerian disk area weused a circular aperture with an outer radius of 0.7 (cid:48)(cid:48) .For the integrated flux in the tails we used two ellipticalapertures in the CO data for the areas that coincidewith the northern and the western tails in the SPHEREscattered light data. In the CO data we used only oneelliptical aperture centered on the western region sincethe northern tail is not well detected in the CO emis-sion. Using this procedure we find a flux density ratioof 0.6 for the CO data and 2.9 for the CO data.Given that the Keplerian disk has a higher temperaturethan the tails, which are farther away from the star,a smaller column density of material is needed in thedisk to produce the same amount of flux compared tothe tails. For the optically thinner CO data we findroughly three times higher integrated flux in the tailsthan in the disk, which may indicate that there is sig-nificantly more mass in the tails than in the disk. Wepoint out that this assumes a constant gas-to-dust ra-tio in both areas. Also even CO may still be opticallythick in the disk (and possibly the tails) and thus we cannot directly translate the flux density ratio to a massratio. However, it is encouraging that the flux densityratio between tail and disk increases between the opti-cally thicker tracer CO and the thinner tracer CO.Given that the CO emission should be more densitydominated than the CO emission this indicates thatthere is indeed a substantial amount of material in thetails compared to the disk.For a proper measurement of the gas masses (and the an-gular momentum) in the different structures surround-ing SU Aur deep observations of optically thin tracersare needed. However the measurements extracted fromthe existing data are well compatible with a scenario inwhich the infalling material misaligns the outer disk ofSU Aur. HST/STIS OBSERVATIONSThe large scale dust tail extending from SU Aurwas previously captured in optical scattered light byGrady et al. (2001), using HST/STIS. While in the STIScoronagraphic images the disk region is masked, STISpresents a significantly larger field of view. We showthe HST/STIS data, after masking of the coronagraphicbars and combining of two telescope roll angles in fig-ure 6. We overlaid the contours of the SPHERE datafor comparison, showing the complementarity of bothinstruments.The dust tails seen in SPHERE seamlessly connect with ∆ D e c [ a r c s e c ] ∆ RA [ arcsec ] ∆ D e c [ a r c s e c ] v = 7.51 km/s Figure 4.
SPHERE Q φ image with the ALMA CO 3-2channel maps as blue and red contour overlay. Contour levelscorrespond to the 2 σ rms levels starting at 5 σ . The ALMAbeam size and orientation is indicated by the white ellipsein the lower right corner. The upper panel corresponds tosignal blue-shifted (-1.49kms − ) relative to the systemic ve-locity and the bottom panel corresponds to red-shifted signal(+1.51km s − ). ∆ D e c [ a r c s e c ] CO (3 − m J y / b e a m k m s − ∆ RA [ arcsec ] ∆ D e c [ a r c s e c ] CO (2 − m J y / b e a m k m s − Figure 5.
Moment 0 map of the CO(3-2) and CO(2-1)data of the SU Aur system. The beam size is indicated bythe white ellipse in the lower right corner. the structures visible in the STIS image. In the STISdata it becomes apparent that the tail structure is turn-ing towards the South. This is the direction in whichAB Aur is located, which shows spectacular extendedspiral features in scattered light (Boccaletti et al. 2020).The direction towards AB Aur may support the scenariodiscussed in Akiyama et al. (2019), where they speculate hadows, spirals and dust tails in SU Aur ∆ RA [ arcsec ] ∆ D e c [ a r c s e c ] Figure 6.
HST/STIS data first presented by Grady et al.(2001). We masked the coronagraph and telescope spiderfeatures and combined two roll angles. SPHERE Q φ con-tours are overlayed in blue. that cloud material initially falls towards the center ofgravity between SU Aur and AB Aur before it accretesonto either systems. DISCUSSION AND CONCLUSIONSThe observational data of SU Aur paint an intricatepicture of the formation of the system, its connection tothe surrounding molecular cloud and its evolution.6.1.
The origin of the dust tails
SPHERE and ALMA data in concert show that ma-terial is falling onto the disk surrounding SU Aur. Wecan thus likely rule out a recent close encounter orthe ejection of a dust clump for their origin (Vorobyovet al. 2020). Additionally, we checked the Gaia DR2catalog and find no obvious candidates for a close stel-lar encounter with SU Aur within 100 (cid:48)(cid:48) (i.e. sourceswith similar distance or proper motion) , neither dowe detect a point source within the 6 (cid:48)(cid:48) field of view ofSPHERE (we are generally sensitive to all stellar andbrown dwarf sources outside of ∼ (cid:48)(cid:48) ).The asymmetry of the tail structures from the largestscales seen with HST/STIS down to the smallest scalesmay suggest an interaction between SU Aur and thenearby young star AB Aur (projected separation of ∼ There are several known members of the Taurus X subgroup ofwhich SU Aur is a member (Luhman et al. 2009), however theclosest member is JH 433, which is moving tangentially to SU Aurand thus is not a candidate for a close encounter. See appendix Dfor a brief overview. that both of these systems are located in or near thesame filament structure within the L 1517 dark cloud.This indicates that we may see the late formation ofa very wide binary (or higher order) system formedby turbulent fragmentation (Padoan 1995). The arc-like structure seen around AB Aur (Grady et al. 1999)and the large-scale dust and gas tails around SU Aurmight be part of a connecting structure between the twosystems as predicted by magneto-hydrodynamic simula-tions (e.g., Kuffmeier et al. 2019). As already suggestedby Akiyama et al. (2019), material may then be fun-neled along these structures towards either system.The sharp tail structures in particular might be ex-pected from classical Bondi-Hoyle accretion (Bondi &Hoyle 1944; Bondi 1952). Following Dullemond et al.(2019), a cloudlet which undergoes a close encounterwill form a large scale arc-like structure and possiblyalso sharper dust tails. This depends largely on thesize of the cloudlet relative to the impact parameterof the encounter, but also the thermodynamics withinthe cloud. For a cloudlet with a radius larger than theimpact parameter, which also cools efficiently, they pro-duce scattered light images containing both large-scalearcs and smaller scale tails. Their synthetic scatteredlight images are reminiscent of the structure seen aroundSU Aur with the HST and the tails seen with SPHERE.Additionally, an encounter with a large cloudlet wouldalso explain that we see not only red-shifted emissionand scattered light in one direction but that it envelopsthe disk. Kuffmeier et al. (2020) also show that suchclose encounters with cloudlets can produce extendedarcs on scales of 10 au. Their simulations suggest thatthese resulting structures are long lived if the protostaris at rest relative to the surrounding gas and is encoun-tered by a cloudlet in relative motion. This may beplausible for SU Aur for two reasons. On the one hand,the fact that we indeed detect these dynamical signa-tures is itself an indication that they are long lived. Onthe other hand the systemic velocity of ∼ − fitswell with the radial velocity of the surrounding filamentas reported in Hacar & Tafalla (2011).6.2. Disk instability due to infalling material?
The new SPHERE observations allow us to trace thelarge scale structures in SU Aur seamlessly down toscales of less than 10 au. The two most striking featuresin the Keplerian disk are the multitude of spiral armsand the sharp shadow lane. Both of these can well beexplained by the infall of material. Spiral waves area common consequence of instability triggered by in-falling material (Moeckel & Throop 2009; Lesur et al.0
Ginski et al.
Disk misalignment by late infall?
The shadow lane in SU Aur is a feature now com-monly seen in scattered light images (e.g., Marino et al.2015; Stolker et al. 2016b; Benisty et al. 2017; Keppleret al. 2020) and typically explained by a misalignmentor warp between inner and outer disk. Indeed by com-paring our ALMA continuum fit with the interferomet-ric result from Labdon et al. (2019) we find a relativemisalignment of ∼ ◦ . While there is ample theory onhow such a misalignment is caused, there is little obser-vational evidence of the process. Brinch et al. (2016)reported on the misalignment of circumstellar disks inthe IRS 43 multiple system with respect to the surround-ing circumbinary disk, presumably caused by the chaoticinteraction of the stellar cores. An even more spectac-ular case of such a misalignment by multiple stars wasrecently shown for for the GW Ori system by Bi et al.(2020) and Kraus et al. (2020). Sakai et al. (2019) founda warped disk around the proto-star IRAS 04368+2557.They inferred from the absence of signs of a close stel-lar encounter that the warp should be caused by lateinfall of material, but did not find direct evidence. ForAB Aur, Tang et al. (2012) show a large scale warp inthe surrounding disk and multiple tentative spiral fea-tures and suggest that both are caused by late infall.Given the results by Boccaletti et al. (2020), who detectlarge scale spiral structures in scattered light, down todisk scales, this seems a likely scenario. However, wenote that Boccaletti et al. (2020) interpret the inner- most spiral structures as signs of a forming proto-planetrather than as instability caused by infalling material.In SU Aur we directly detect the infalling material andcan trace it from thousands of au down to disk scales.As Dullemond et al. (2019) argue, infalling material isbound to have a vastly different orientation of angularmomentum compared to the accreting disk. In section 4,we discussed that the mass estimates in the tail and diskstructure make such a scenario plausible. Thus the infallwe trace is likely causing a warp of the outer disk regions.This makes the AB Aur and SU Aur pair the best exam-ples of such effects caused by late infall. We note that itis in principle possible that the disk we currently see inscattered light around SU Aur is not primordial at all,but is actually formed as a result of a close encounterwith a cloudlet (see Dullemond et al. 2019). In this caseit would be natural that it is misaligned with respectto the (presumably) primordial inner disk detected byLabdon et al. (2019).The structures revealed around SU Aur by SPHERE andALMA form a coherent picture of late infall of materialthat dominates the evolution of the protoplanetary disk.This mechanism not only provides an additional massreservoir for forming planets (see the discussion in Ma-nara et al. 2018) but can also trigger planet formationby gravitational instability. As suggested by Thies et al.(2011), this scenario might be able to explain the spin-orbit misalignment found in evolved planetary systems.These new high-resolution observations enable detailedfuture simulations of such planet formation pathways.ACKNOWLEDGMENTSWe thank an anonymous referee for a thorough re-view that improved the paper. We would like to thankJonathan Williams and Antonio Garufi for fruitful dis-cussion. We also thank Eiji Akiyama for providing theirreduced ALMA data and Aaron Labdon and the A&Ajournal for authorizing the reprint of the near infraredinterferometric results.SPHERE is an instrument designed and builtby a consortium consisting of IPAG (Grenoble,France), MPIA (Heidelberg, Germany), LAM (Mar-seille, France), LESIA (Paris, France), Laboratoire La-grange (Nice, France), INAF - Osservatorio di Padova(Italy), Observatoire de Gen`eve (Switzerland), ETHZurich (Switzerland), NOVA (Netherlands), ONERA(France), and ASTRON (The Netherlands) in collabo-ration with ESO. SPHERE was funded by ESO, withadditional contributions from CNRS (France), MPIA(Germany), INAF (Italy), FINES (Switzerland), andNOVA (The Netherlands). SPHERE also received fund- hadows, spirals and dust tails in SU Aur Python programming language , especially the SciPy (Virta-nen et al. 2020),
NumPy (Oliphant 2006),
Matplotlib (Hunter 2007) and astropy (Astropy Collaboration et al.2013, 2018) packages. We thank the writers of thesesoftware packages for making their work available to theastronomical community.
Facilities:
VLT(SPHERE), VLT(NACO), ALMA,HST(STIS)APPENDIX Ginski et al. A. ITERATIVE FEEDBACK REFERENCE AND ANGULAR DIFFERENTIAL IMAGING ∆ D e c [ a r c s e c ] cADI x5 ∆ RA [ arcsec ] IF-ADI
IF-RDI
Figure 7.
Total intensity images of SU Aur after post-processing of the SPHERE H-band data.
Left:
Classical angulardifferential imaging reduction.
Center:
Iterative feedback angular differential imaging reduction of the same data set. Theresult after 100 iterations is shown.
Right:
Iterative feedback reference differential imaging reduction. The result after 14iterations is shown. All panels are shown in the same linear color scale. The cADI reduction was multiplied by a factor of 5 toimprove the contrast in this color scale. The innermost region is not well recovered in all cases an thus masked out.
The total intensity image was extracted from the data by using an iterative reference differential imaging (RDI)approach. In our RDI routine, scaled reference PSF images were created by projecting the data on a reference starlibrary, using the KLIP algorithm as introduced by Soummer et al. (2012). Furthermore, an iterative approach wasapplied to this KLIP RDI routine, with the aim to reduce overestimation of residual stellar light in the scaled referencePSF caused by disk signal in the data affecting the projection on the reference star library. This approach, namedIterative Disk Feedback (IDF), will be presented in a forthcoming publication by Vaendel et al. (in prep.), whoshow it to effectively minimize over-subtraction of stellar signal in the RDI reduction of data containing extended diskstructures. The iterative approach enables to subtract disk signal (selected from the resulting reduction of the previousiteration) from the image stack which is used for projection on the reference star library in the regular KLIP RDIroutine. This means that for each iteration, the KLIP RDI reduction is less affected by the presence of (identified)disk signal. Ideally, the iterative process continues until all disk signal is identified and hence only the actual stellarsignal in the data projected onto the reference star library for the creation of the scaled reference PSF. In our case thereduction did not significantly improve anymore after 14 iterations, where we stopped the IDF process.Iterative Angular Differential Imaging uses median subtraction ADI multiple times over on the same dataset. It takesthe final result coming out of the ADI pipeline and subtracts all positive signal of this result from the original datasetafter it has been rotated by the respective field rotation of each of the images. By setting the negative values tozero the self-subtracted regions are not fed back. In this way, less signal of the disk is present in the dataset. Whencomputing the median of the dataset, less of the disk signal is also present in the resulting PSF. Consequently, lessself- subtraction of the disk occurs when the median is subtracted. However, after computing the PSF, there is stillsome disk signal left in the dataset after one iteration. The more iterations, the less signal of the disk is left in thePSF and thus the better the final result is. For the dataset used in this work, this iterative process was done 100 timesover. With this technique some problems are occurring. After many iterations the star is reconstructed and ring-likestructures are being generated due to the many rotations happening during the iteration process. These effects mainlyoccur at the centre of the image due to most signal being present there. In our work, the main science occurs at theouter dust tails, where these effects are not dominating over the disk signal. We show the resulting total intensityimages for all techniques in figure 7. B. GAUSSIAN FIT OF THE ALMA VISIBILITIESIn order to derive inclination and position angle of the mm continuum disk observed by ALMA we perform a fitof the visibilities in the uv plane. The continuum visibilities of all spectral windows have been channel averaged to250 MHz wide channels to reduce the amount of data used in the fitting, without inducing any bandwidth smearing. Weassumed a simple radial Gaussian profile for the intensity distribution, with intensity normalization I and Gaussianwidth σ as free parameters. Additional four parameters are accounted for in the fitting, in particular inclination, hadows, spirals and dust tails in SU Aur GALARIO code(Tazzari et al. 2018), with the same sampling as in the ALMA observations. The parameter space is then exploredusing the emcee package (Foreman-Mackey et al. 2013), assuming wide uniform priors on all parameters. The posteriordistribution is sampled with 60 walkers for 4000 steps after a burn in of 1000 steps.The fit nicely converges to a solution, where the disk is clearly well resolved and well reproduced by the simpleGaussian model (see figure 8). Table B reports the best fit parameters, taken as the median value of the marginalizedposterior distributions shown in figure 9. No clear structure is seen in the residuals above the 3 σ limit. Alternativeparametrization for the intensity profile (as broken power laws or modified versions of it) did not improve the results. R e ( V ) ( J y ) DataModel uv-distance (k ) I m ( V ) ( J y ) Figure 8.
Visibilities of the ALMA continuum data, re-centered and de-projected using the best fit geometrical parametersdescribed in section B. The disk is well resolved and well described by a simple Gaussian approximation. I σ i PA ∆RA ∆Dec(log Jy/steradian) ( (cid:48)(cid:48) ) ( ◦ ) ( ◦ ) (mas) (mas)11 . +0 . − . . +0 . − . . +1 . − . . +1 . − . − . +0 . − . . +0 . − . Table 1.
Median of the marginalized posteriors of the fitted parameters for the continuum emission of SU Aur, with associatedstatistical uncertainties from the 16th and 84th percentiles of the marginalized distributions.C.
SPHERE DATA AND ALMA OVERLAYS D. CLUSTER MEMBERS NEAR SU AURSU Aur is located in a small embedded cluster associated with the L1517 cloud (Taurus X in Luhman et al. 2009).The distribution of members of this subgroup is studied as one of the ”NESTs” ( (cid:48)(cid:48) , 23 kau), AB Aur (184 (cid:48)(cid:48) , 29 kau), and XEST 26-052 (268 (cid:48)(cid:48) , 42 kAU).4
Ginski et al.
Figure 9.
Marginalized posterior distributions of MCMC fitting the continuum visibilities of the ALMA data.
The mean separation between the members of the group is ∼
25 kAU (Joncour et al. 2018). In figure 12 we show the2d distribution of group members and indicate their proper motions.REFERENCES
Akiyama, E., Vorobyov, E. I., Baobabu Liu, H., et al. 2019,AJ, 157, 165, doi: 10.3847/1538-3881/ab0ae4 Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,et al. 2013, A&A, 558, A33,doi: 10.1051/0004-6361/201322068 hadows, spirals and dust tails in SU Aur ∆ RA [ arcsec ] ∆ D e c [ a r c s e c ] Figure 10.
SPHERE Q φ image with the ALMA dust continuum emission in Band 7 overlayed in green contours. The ALMAbeam size and orientation is indicated by the white ellipse in the lower right corner. Contour levels are 4,8,12,16 and 20 σ .Astropy Collaboration, Price-Whelan, A. M., Sip˝ocz, B. M.,et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4fAvenhaus, H., Quanz, S. P., Schmid, H. M., et al. 2017, AJ,154, 33, doi: 10.3847/1538-3881/aa7560Bae, J., Hartmann, L., & Zhu, Z. 2015, ApJ, 805, 15,doi: 10.1088/0004-637X/805/1/15Batygin, K. 2012, Nature, 491, 418,doi: 10.1038/nature11560Benisty, M., Stolker, T., Pohl, A., et al. 2017, A&A, 597,A42, doi: 10.1051/0004-6361/201629798Beuzit, J. L., Vigan, A., Mouillet, D., et al. 2019, A&A,631, A155, doi: 10.1051/0004-6361/201935251Bi, J., van der Marel, N., Dong, R., et al. 2020, ApJL, 895,L18, doi: 10.3847/2041-8213/ab8eb4Boccaletti, A., Di Folco, E., Pantin, E., et al. 2020, A&A,637, L5, doi: 10.1051/0004-6361/202038008Bondi, H. 1952, MNRAS, 112, 195,doi: 10.1093/mnras/112.2.195 Bondi, H., & Hoyle, F. 1944, MNRAS, 104, 273,doi: 10.1093/mnras/104.5.273Brinch, C., Jørgensen, J. K., Hogerheijde, M. R., Nelson,R. P., & Gressel, O. 2016, ApJL, 830, L16,doi: 10.3847/2041-8205/830/1/L16Calvet, N., Muzerolle, J., Brice˜no, C., et al. 2004, AJ, 128,1294, doi: 10.1086/422733de Boer, J., Langlois, M., van Holstein, R. G., et al. 2020,A&A, 633, A63, doi: 10.1051/0004-6361/201834989de Leon, J., Takami, M., Karr, J. L., et al. 2015, ApJL,806, L10, doi: 10.1088/2041-8205/806/1/L10Dong, R., Fung, J., & Chiang, E. 2016, ApJ, 826, 75,doi: 10.3847/0004-637X/826/1/75Dullemond, C. P., K¨uffmeier, M., Goicovic, F., et al. 2019,A&A, 628, A20, doi: 10.1051/0004-6361/201832632Facchini, S., Juh´asz, A., & Lodato, G. 2018, MNRAS, 473,4459, doi: 10.1093/mnras/stx2523Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman,J. 2013, PASP, 125, 306, doi: 10.1086/670067 Ginski et al. ∆ D e c [ a r c s e c ] ∆ RA [ arcsec ] Figure 11.
SPHERE Q φ image with the ALMA CO 3-2 channel maps as contours. We show contours of channels blue-shiftedrelative to the systemic velocity of ∼ − in blue, red-shifted channels in red and the channels closest to the systemic velocityin white. The lowest contour represents 5 σ and subsequent contours are steps of 2 σ .Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al.2018, A&A, 616, A1, doi: 10.1051/0004-6361/201833051Ginski, C., Stolker, T., Pinilla, P., et al. 2016, A&A, 595,A112, doi: 10.1051/0004-6361/201629265Ginski, C., M´enard, F., Rab, C., et al. 2020, arXiv e-prints,arXiv:2007.05274. https://arxiv.org/abs/2007.05274 Grady, C., Stapelfeldt, K., Clampin, M., et al. 2001, inAmerican Astronomical Society Meeting Abstracts, Vol.199, American Astronomical Society Meeting Abstracts,60.15Grady, C. A., Woodgate, B., Bruhweiler, F. C., et al. 1999,ApJL, 523, L151, doi: 10.1086/312270Hacar, A., & Tafalla, M. 2011, A&A, 533, A34,doi: 10.1051/0004-6361/201117039 hadows, spirals and dust tails in SU Aur Figure 12.