Ringed Substructure and a Gap at 1 AU in the Nearest Protoplanetary Disk
Sean M. Andrews, David J. Wilner, Zhaohuan Zhu, Tilman Birnstiel, John M. Carpenter, Laura M. Perez, Xue-Ning Bai, Karin I. Oberg, A. Meredith Hughes, Andrea Isella, Luca Ricci
DDraft version April 1, 2016
Preprint typeset using L A TEX style AASTeX6 v. 1.0
RINGED SUBSTRUCTURE AND A GAP AT 1 AU IN THE NEAREST PROTOPLANETARY DISK
Sean M. Andrews , David J. Wilner , Zhaohuan Zhu , Tilman Birnstiel , John M. Carpenter ,Laura M. P´erez , Xue-Ning Bai , Karin I. ¨Oberg , A. Meredith Hughes , Andrea Isella , & Luca Ricci Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA; [email protected] Department of Astrophysical Sciences, 4 Ivy Lane, Peyton Hall, Princeton University, Princeton, NJ 08544, USA Max-Planck-Institut f¨ur Astronomie, K¨onigstuhl 17, 69117 Heidelberg, Germany Joint ALMA Observatory (JAO), Alonso de Cordova 3107 Vitacura -Santiago de Chile Max-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 69, 53121 Bonn, Germany Department of Astronomy, Wesleyan University, Van Vleck Observatory, 96 Foss Hill Drive, Middletown, CT 06457, USA Department of Physics and Astronomy, Rice University, 6100 Main Street, Houston, TX, 77005, USA
ABSTRACTWe present long-baseline Atacama Large Millimeter/submillimeter Array (ALMA) observations ofthe 870 µ m continuum emission from the nearest gas-rich protoplanetary disk, around TW Hya, thattrace millimeter-sized particles down to spatial scales as small as 1 AU (20 mas). These data reveala series of concentric ring-shaped substructures in the form of bright zones and narrow dark annuli(1–6 AU) with modest contrasts (5–30%). We associate these features with concentrations of solidsthat have had their inward radial drift slowed or stopped, presumably at local gas pressure maxima.No significant non-axisymmetric structures are detected. Some of the observed features occur neartemperatures that may be associated with the condensation fronts of major volatile species, but therelatively small brightness contrasts may also be a consequence of magnetized disk evolution (theso-called zonal flows). Other features, particularly a narrow dark annulus located only 1 AU from thestar, could indicate interactions between the disk and young planets. These data signal that orderedsubstructures on ∼ AU scales can be common, fundamental factors in disk evolution, and that highresolution microwave imaging can help characterize them during the epoch of planet formation.
Keywords: protoplanetary disks — planet-disk interactions — stars: individual (TW Hydrae) INTRODUCTIONThe disks around young stars are the formation sitesof planetary systems. However, the smooth, monotonicradial distributions of gas temperatures and densities as-sumed in most theoretical models of planetary formationcreate a fundamental dilemma. The millimeter-sized par-ticles needed to assemble larger planetesimals (Ormel &Klahr 2010; Lambrechts & Johansen 2012) experienceaerodynamic drag with the gas that results in their rapidmigration toward the host star (Weidenschilling 1977;Takeuchi & Lin 2002). Yet, this predicted depletion ofsolids is not commensurate with observations that rou-tinely detect microwave continuum emission from suchparticles extending over a large range of disk radii, outto tens or hundreds of AU (see reviews by Williams &Cieza 2011 or Andrews 2015). The hypothesized solu-tion to this conflict invokes substructure in the form oflocal gas pressure maxima, which slow or stop the migra-tion of these particles and concentrate the solid densitiesto levels that might trigger efficient planetesimal growth(e.g., Whipple 1972; Pinilla et al. 2012). The young solar analog TW Hya is an especially in-teresting target to characterize disk substructures, forthree primary reasons. First, it is the closest gas-richdisk to Earth (54 ± (cid:38) cm) particles (Wilner et al.2005; Menu et al. 2014) and a sharp radial decrease in thesolids-to-gas mass ratio (Andrews et al. 2012; Birnstiel& Andrews 2014) that announce substantial growth andinward migration. The predicted depletion of microwaveemission due to radial drift (Takeuchi & Lin 2005) shouldbe especially prominent at the relatively advanced age ofTW Hya ( ∼
10 Myr; e.g., Weinberger et al. 2013); butsince that is not what is observed, the case for relaxingthe assumption of a globally negative pressure gradientis bolstered. And third, there is already tantalizing ev-idence for substructure in this disk, including a centraldepletion (Calvet et al. 2002; Hughes et al. 2007) and ten-tative signatures of “gaps” or “breaks” in the infrared a r X i v : . [ a s t r o - ph . E P ] M a r scattered light emission (Akiyama et al. 2015; Rapsonet al. 2015; Debes et al. 2013, 2016).In this Letter, we present and analyze observationsthat shed new light on the substructure in the TW Hyadisk. We have used the long baselines of ALMA to mea-sure the 870 µ m continuum emission from this disk atan unprecedented spatial resolution of ∼ OBSERVATIONS AND DATA CALIBRATIONTW Hya was observed by ALMA on 2015 Novem-ber 23, November 30, and December 1. The array in-cluded 36, 31, and 34 antennas, respectively, configuredto span baseline lengths from 20 m to 14 km. The corre-lator processed four spectral windows centered at 344.5,345.8, 355.1, and 357.1 GHz with bandwidths of 1875,469, 1875, and 1875 MHz, respectively. The observa-tions cycled between the target and J1103-3251 with a1 minute cadence. Additional visits to J1107-3043 weremade every 15 minutes. J1037-2934, J1058+0133, andJ1107-4449 were briefly observed as calibrators. The pre-cipitable water vapor (PWV) levels were ∼ ∼ ∆ α [arcseconds]1.00.50.00.51.0 ∆ δ [ a r c s e c o n d s ]
10 AU 0.00.51.01.52.0 s u r f a c e b r i g h t n e ss [ m J y / b e a m ] Figure 1 . A synthesized image of the 870 µ m continuum emissionfrom the TW Hya disk with a 30 mas FWHM (1.6 AU) circularbeam. The RMS noise level is ∼ µ Jy beam − . The inset showsa 0.2 (cid:48)(cid:48) -wide (10.8 AU) zoom using an image with finer resolution(24 ×
18 mas, or 1 . × . of 0.5–1 mm. The combined on-target integration timewas 95 minutes. The basic calibration was as describedabove. As a check, we compared the amplitudes fromeach individual dataset on overlapping spatial frequen-cies and found exceptional consistency.The calibrated visibilities from each observation wereshifted to account for the proper motion of the targetand then combined after excising channels with potentialemission from spectral lines. Some modest improvementswere made with a round of phase-only self-calibration.Continuum images at a mean frequency of 345.9 GHz(867 µ m) were generated by Fourier inverting the visi-bilities, deconvolving with a multi-scale, multi-frequencysynthesis version of the CLEAN algorithm, and then restor-ing with a synthesized beam. All calibration and imagingwas performed with the
CASA package (v4.5.0).After some experimentation, we settled on an analysisof two images made from the same composite dataset.The first used a Briggs weighting (with a robust param-eter of 0) to provide a 24 ×
18 mas synthesized beam (atP.A.=78 ◦ ). While this provides enhanced resolution, itcomes at the cost of a dirty beam with ∼
20% sidelobes(due to the sparse coverage at long baselines) that de-grades the image quality. A second image was made witha robust parameter of 0.5 and an elliptical taper to createa circular 30 mas beam with negligible sidelobes. Bothimages are consistent (within the resolution differences)and have RMS noise levels around 35 µ Jy beam − . RESULTSFigure 1 shows a high resolution map of the 870 µ mcontinuum emission from the TW Hya disk, revealing a a z i m u t h [ d e g r ee s ] b r i g h t n e ss t e m p e r a t u r e [ K ] s u r f a c e b r i g h t n e ss [ m J y / b e a m ] Figure 2 . ( top ) The high resolution (24 ×
18 mas beam) synthe-sized image described in Sect. 2, deprojected into a map in polar co-ordinates to more easily view the disk substructure. ( bottom ) Theazimuthally-averaged radial surface brightness profile. For refer-ence, the dashed red curve shows the midplane temperature profilederived from a representative model disk. The gray curve in thebottom left reflects the profile of the synthesized beam. series of concentric bright and dark rings out to a radialdistance of 60 AU from the host star with a nearly pole-on viewing geometry. To aid in the visualization of thissubstructure, Figure 2 shows the image transformed intopolar coordinates and azimuthally averaged into a radialsurface brightness profile.The inner disk includes an unresolved ( < . ± .
04 mJy source coincident with the stellarposition and a bright ring that peaks at 2.4 AU; betweenthem is a dark annulus centered at 1 AU. The brightring and dark annulus are unresolved ( < . (cid:48)(cid:48) Debeset al. (2013) reported an additional annular deficit in thescattered light around 80 AU, but there is too little mi-crowave continuum emission that far from the star toidentify any related features in the ALMA data.Based on the data in Figure 2, comparisons of the ra-dial profile along small ranges of polar angles in the diskplane demonstrate that there are no statistically signifi-cant azimuthal variations in the observed emission. Thetypical deviations correspond roughly to the RMS noiselevel; the median fractional deviation is only 7%. Wefitted ellipses to the prominent bright rings at 2.4 and40 AU (the latter between two dark annuli) and the darkannulus at 22 AU to estimate the projected viewing ge-ometry of the disk. Each feature was consistent with thestandard geometry inferred from molecular line observa-tions (e.g., Qi et al. 2004; Andrews et al. 2012): the jointconstraints suggest an inclination of 7 ± ◦ and a majoraxis position angle of 155 ± ◦ .On broad angular scales, the overall continuum bright-ness distribution is roughly consistent with the brokenpower-law model of Hogerheijde et al. (2016): the sur-face brightness falls off like r − . inside 50 AU and thendrops precipitously, like r − , at larger radii. A refineddescription would characterize the emission inside 50 AUusing two different segments, where the linear slope in-side 20 AU is about 6 × steeper than from 20–50 AU. Thefact that this slope change occurs near the 22 AU darkannulus is certainly (at least partially) associated withthe feature noticed by Zhang et al. (2016) and Nomuraet al. (2016): after all, the dark annulus itself would notbe detectable at resolutions coarser than ∼ . (cid:48)(cid:48) RADMC-3D (Dullemond 2012). Wefollowed the same basic assumptions as Andrews et al.(2012), with a few distinctions: (1) we employed a brokenpower-law dust surface density prescription (Hogerheijdeet al. 2016); (2) in the very inner disk, we included a beltof dust that has a Gaussian density profile with a widthof 0.4 AU (to insure it would be unresolved); (3) to mimicthe observed dark annulus, the outer disk is truncated at2.4 AU (rather than the previously assumed 4 AU). Withonly minor tweaks to the original parameters, this modelreproduces well the broadband spectral energy distribu-tion. It also accounts for the ∼ ∼
200 K) dust adjacent to the host star.However, it still has difficulty reproducing the observedbrightness profile in detail. The midplane temperatureprofile is shown in the bottom panel of Figure 2. While Nomura et al. (2016) identify a similar feature, but the extremespatial filtering applied in their imaging introduces severe artifactsthat masquerade as “gaps” and muddle the interpretation. we aim to refine this model in future work, for now itserves as a crude reference for the disk temperatures thatwill aid in our discussion of the potential mechanisms re-sponsible for the observed substructure. DISCUSSIONCurrently only one other disk, around the muchyounger source HL Tau, is known to exhibit ringed sub-structure like we find here (ALMA Partnership 2015).But the apparent similarities are superficial. The darkannuli in the TW Hya disk are considerably less deepand narrower: if TW Hya were located at the same dis-tance as HL Tau (140 pc), the same data would at bestonly be able to (tentatively) identify the 22 AU feature.Unlike in the HL Tau disk, the bright zones we observehave only marginal optical depths (given our represen-tative temperature model and others like it). The slopetransition noted at ∼
20 AU seems to mark the change toan optically thick inner disk (aside from the 1 AU darkannulus). In any case, it is interesting that this type ofordered substructure is observed in these two very dif-ferent disks, which bracket a factor of ∼
10 in both theirages and microwave luminosities (the TW Hya disk beingolder and intrinsically less luminous).While the new data presented here corroborate theemerging concept that well-ordered, azimuthally sym-metric substructures are prevalent and important forcesin disk evolution, they also extend the interesting diver-sity we have seen in terms of the amplitudes and physicalscales on which that substructure is manifested. Vari-ous mechanisms to produce ringed substructure in diskshave been proposed. They can be broadly categorizedinto magnetic, chemical, or dynamical origins: here weconsider each in the context of the TW Hya disk.Magnetized disks may exhibit radial pressure varia-tions known as zonal flows (Johansen et al. 2009), nat-ural outcomes of the turbulence driven by the magneto-rotational instability (Balbus & Hawley 1991). Numer-ical simulations in a local box show that zonal flowpressure maxima are separated by a few scale heightsand can have amplitudes as high as ∼ ∼
20 AU are consistent withthis scenario. While global simulations with realisticdisk physics and magnetic fields remain computationallychallenging, low-amplitude surface density variations arepresent in the recent global simulations assuming idealMHD by Suzuki & Inutsuka (2014). The current obser-vational constraints on turbulent linewidths in the TWHya disk (Hughes et al. 2011) are sensitive to much largerradii; it remains unclear if there is sufficient MHD tur- bulence in the 20–50 AU range to produce zonal flowsconsistent with the ALMA image. However, Bai (2015)found that even a thin turbulent layer at the disk surface(owing to far-ultraviolet ionization; e.g., Perez-Becker &Chiang 2011) appears sufficient to drive such flows downto the midplane, even if it is largely laminar. Ring andgap features have also been found in simulations thatconsider the dynamical effects of radial variations in thedisk resistivity (e.g., Flock et al. 2015; Lyra et al. 2015). Overall, low-amplitude radial surface density variationsmay be a generic consequence of magnetized disk evolu-tion. On the other hand, it is less clear whether magneticmechanisms alone could account for the ring at 2.4 AUand the dark annulus at 1 AU, where the more complexdisk microphysics (thermodynamics, ionization, and non-ideal MHD effects) is not very well understood.The chemical explanation is elegant and should be uni-versal. As migrating solids approach the condensationfronts of major volatile species, they shed their corre-sponding ices back into the gas phase. This and anysubsequent re-condensation can modify the solid opaci-ties, and thereby the associated continuum emission (e.g.,Cuzzi & Zahnle 2004). This sublimation may also makethe particles brittle enough that collisions become de-structive, leaving small fragments that are better cou-pled to the gas, and therefore migrate much more slowly.The net result is a sequence of traffic jams that locally en-hance the solid densities and would appear as bright con-tinuum rings at a set of specific temperatures (Okuzumiet al. 2015). This scenario is different than the magneticand dynamical mechanisms, in that it is not precipitatedby substructure in the gas disk and therefore does not“trap” particles. Nevertheless, it is perhaps appealing inthe outer disk of TW Hya, where the carbon monoxide(CO) and molecular nitrogen (N ) condensation frontsat temperatures of ∼
20 and 17 K lie outside the 22 and37 AU dark annuli, respectively, in our reference model.The former also overlaps with the inner edge of the N H + ring found by Qi et al. (2013) and associated with the COsnowline. However, it is worth a reminder that tempera-ture models are subject to uncertainties associated withdetailed assumptions about the disk structure and grainproperties: small changes to the optical depth profile cansubstantially shift the radii that correspond to the rele-vant volatile condensation temperatures.It may be tempting to consider the H O snowline asa potential cause for the substructure observed at ∼ Although we note that other non-ideal MHD effects were notincluded in these studies, and that this scenario may not be ableto account for multiple ring-like features. ∼
150 K, at radii < O snowline itself would result in thesubstantial density depletion of a wide range of particlesizes ( µ m–cm) that would be required to explain boththe pronounced dip in the mid-infrared spectrum (Cal-vet et al. 2002) and the 1 AU dark annulus in the ALMAimage. Moreover, we would expect a much higher thanobserved incidence of TW Hya-like infrared spectra (i.e.,classical “transition” disks) if this were the case, sinceessentially all disks around young solar analogs shouldhave H O snowlines at similar locations.The dynamical alternative instead postulates that thedark annuli are true gaps that have had their densitiesdepleted by interactions with (as yet unseen) planetary-mass companions (Lin & Papaloizou 1986; Kley & Nel-son 2012). Given the scale heights implied by our repre-sentative model temperatures, even relatively low-mass( ∼ several M ⊕ ) planets can open narrow gaps (Zhu et al.2013; Duffell & MacFadyen 2013; Fung et al. 2014) andtrap solid particles (Paardekooper & Mellema 2006; Zhuet al. 2014; Picogna & Kley 2015) to produce featuressimilar to the substructure that we observe here. Evenwith such low masses, it seems difficult to explain boththe features at 37 and 43 with two planets since theywould tend to open a common gap without the brightring observed to bisect them. In principle, a singlelow-mass planet at ∼
40 AU could potentially open a‘double-gap’ feature if the disk is particularly inviscid(e.g., Goodman & Rafikov 2001; Dong et al. 2011; Duf-fell & MacFadyen 2012). Or, interactions with a planetat ∼
37 AU might also perturb the distribution of solidsout of the disk plane beyond it; the shadowing inducedby such surface variations can generate temperature vari-ations that might mimic the 43 AU feature (e.g., Jang-Condell 2009). The 1 AU dark annulus is an especiallycompelling case study. If a more comprehensive modelingof the ALMA data confirms that the observed low bright-ness contrast indeed corresponds to a density depletionfactor like that expected from a young super-Earth, itcould serve as a touchstone for modeling the formationof the large population of such planets identified with the
Kepler mission (e.g., Howard 2013).Regardless of which of these mechanisms are at work,the new ALMA data we have presented suggest thatsymmetric, well-ordered, substructure is prevalent in thedisks around young stars, down to very small physicalscales (comparable to or smaller than the local scaleheight) and with a range of amplitudes. Such featuresare the observational hallmarks of the long-speculatedsolution to the fundamental problem of the fast migra- tion of disk solids that has subverted the standard theoryof the planet formation process for decades.We thank Ian Czekala for advice on the figures,the NRAO/NAASC staff for their assistance withthe data calibration, and Ilse Cleeves and RuobingDong for helpful conversations. This research greatlybenefitted from the
Astropy (Astropy Collaborationet al. 2013) and
Matplotlib (Hunter 2007) softwarepackages. This paper makes use of the followingALMA data: ADS/JAO.ALMA