Search for substellar-mass companions and asymmetries in their parent discs
M. Willson, S. Kraus, J. Kluska, J. D. Monnier, M. Ireland, A. Aarnio, M. L. Sitko, N. Calvet, C. Espaillat, D. J. Wilner
AAstronomy & Astrophysics manuscript no. aa c (cid:13)
ESO 2018November 5, 2018
Sparse aperture masking interferometry survey of transitionaldiscs (cid:63)
Search for substellar-mass companions and asymmetries in their parent discs
M. Willson, S. Kraus, J. Kluska, J. D. Monnier, M. Ireland, A. Aarnio, M. L. Sitko, , , N. Calvet, C. Espaillat, D. J. Wilner University of Exeter, Astrophysics Group, School of Physics, Stocker Road, Exeter EX4 4QL, UK Department of Astronomy, University of Michigan, 311 West Hall, 1085 South University Ave, Ann Arbor, MI 48109, USA Research School of Astronomy & Astrophysics, Mount Stromlo Observatory Cotter Road, Weston Creek, ACT 2611, Australia Department of Physics, University of Cincinnati, Cincinnati, OH 45221, USA Space Science Institute, 475 Walnut St., Suite 205, Boulder, CO 80301, USA Visiting Astronomer, NASA Infrared Telescope Facility Department of Astronomy, Boston University, 725 Commonwealth Avenue, Boston, MA 02215, USA Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, MS-78, Cambridge, MA 02138, USAReceived ; accepted
ABSTRACT
Context.
Transitional discs are a class of circumstellar discs around young stars with extensive clearing of dusty material within theirinner regions on 10s of au scales. One of the primary candidates for this kind of clearing is the formation of planet(s) within the discthat then accrete or clear their immediate area as they migrate through the disc.
Aims.
The goal of this survey was to search for asymmetries in the brightness distribution around a selection of transitional disctargets. We then aimed to determine whether these asymmetries trace dynamically-induced structures in the disc or the gap-openingplanets themselves.
Methods.
Our sample included eight transitional discs. Using the Keck / NIRC2 instrument we utilised the Sparse Aperture Masking(SAM) interferometry technique to search for asymmetries indicative of ongoing planet formation. We searched for close-in compan-ions using both model fitting and interferometric image reconstruction techniques. Using simulated data, we derived diagnostics thathelped us to distinguish between point sources and extended asymmetric disc emission. In addition, we investigated the degeneracybetween the contrast and separation that appear for marginally resolved companions.
Results.
We found FP Tau to contain a previously unseen disc wall, and DM Tau, LkH α M c ˙ M c and found values in the range of 10 − − − M J yr − . Conclusions.
We found significant asymmetries in four targets. Of these, three were consistent with companions. We resolved apreviously unseen gap in the disc of FP Tau extending inwards from approximately 10 au.
Key words.
Techniques: interferometric – Planetary systems – Planets and satellites: formation – Planets and satellites: detection –Protoplanetary disks – Stars: pre-main sequence
1. Introduction
It is thought that planet formation is a direct result of aggre-gation and growth of dust particles within the protoplanetarydiscs that form around accreting protostars during the star for-mation process (Pollack et al. 1996). Within the spectral energydistributions (SEDs) of some more evolved discs there are dra-matic drops in the near-infrared (NIR) to mid-infrared (MIR)flux from the disc compared to a classical T Tauri-type disc.This drop in flux is typically interpreted as being caused by theclearing of dust grains through mechanisms such as grain growth(Brauer et al. 2008; Birnstiel et al. 2011), photo-evaporation ofdust grains by the stellar radiation field (Armitage 2011), or in-teractions between a forming giant planet and the disc. These (cid:63)
Based on observations made with the Keck observatory (NASAprogram IDs N104N2 and N121N2). objects are classified as transitional discs (in case of discs withan inner dust-cleared cavity) or pre-transitional discs (in case ofa gapped disc structure) and thought to be the sites of ongoingplanet formation (Espaillat et al. 2014).During the earliest stages of their lives, planetary cores arehighly challenging to detect as they are deeply embedded withinthe dusty material of their parent discs. However, once they havegained su ffi cient mass to clear a gap (at the transitional or pre-transitional disc stage) they become accessible to high resolu-tion imaging observations. This phase likely coincides with thehydrodynamic collapse and oligarchic growth of proto-Jupitersand proto-brown dwarfs and is likely associated with the forma-tion of an extended, hot circumplanetary disc that feeds mate-rial onto the accreting core (Pollack et al. 1996; Ayli ff e & Bate2012). Once protoplanetary cores have cleared most of their im-mediate disc environment, they can continue to accrete signif- Article number, page 1 of 14 a r X i v : . [ a s t r o - ph . E P ] A ug & A proofs: manuscript no. aa icant amounts of mass from material flowing through the gap(10 − M (cid:12) yr − , Najita et al. 2007; Varnière et al. 2006). There-fore, it is expected that protoplanets would appear as strong NIRsources within cleared gap regions.Spatially resolving such systems proves a challenge as theclose angular separation between the protoplanets and theirparent stars and that the parent star is likely to be substan-tially brighter than even a rapidly accreting protoplanet. Spa-tially resolved evidence for protoplanetary companions couldonly be obtained for a small sample of objects so far: Corona-graphic imaging has revealed ring-like and spiral-like structureson scales of 50-200 au (e.g. Subaru / SEEDS survey; Hashimotoet al. 2011; Grady et al. 2013) and sparse aperture mask-ing interferometry (SAM) has resulted in the detection ofsmall-scale asymmetries in the brightness distribution aroundyoung stars that have been interpreted as low-mass companions(T Cha: Huélamo et al. 2011; LkCa 15: Kraus & Ireland 2012;HD 142527: Biller et al. 2012) or as disc emission of a heatedwall in a centro-symmetric disc seen under intermediate incli-nation (FL Cha: Cieza et al. 2013; T Cha: Olofsson et al. 2013).Only in the case of LkCa 15 has this continuum detection beenconfirmed as an accreting companion, with subsequent observa-tions using a combination of SAM and H α spectral di ff erentialimaging performed by Sallum et al. (2015) to demonstrate forthe first time unambiguous evidence for the accreting nature ofthe companion.Besides the emission associated with the protoplanets them-selves and their associated circumplanetary discs, asymmetriescan also be caused by dynamically-induced disc features suchas spiral arms, disc warps (Alencar et al. 2010; Muzerolle et al.2009), or disc physics-related processes such as gravitational in-stabilities, and density waves (Bouvier et al. 2007). Addition-ally a highly inclined disc can induce strong asymmetries in anaxial-symmetric disc owing to forward-scattered light from theilluminated inner rim of disc (Olofsson et al. 2013; Cheethamet al. 2015). Most of the evidence for these processes comes fromphotometric or spectroscopic monitoring investigations. For in-stance, it was found that the variability shows an anti-correlatedbehaviour at NIR and MIR wavelengths. In order to explain boththe timescale and spectral behaviour of the variability, Espaillatet al. (2011) proposed shadowing e ff ects from co-rotating discwarps at the inner dust rim triggered by orbiting planets. Suchwarps are also predicted by hydrodynamic simulations of discswith embedded planets (e.g. Fouchet et al. 2010) and would re-sult in a highly asymmetric brightness distribution. The warpemission will be extended and more complex in geometry than acompanion point source.Here we report on a survey of eight transitional disc targetsobserved using the Keck-II / NIRC2 instrument over the course ofthree years of observations. We discuss the observations in detailwithin Chapter 2; in Chapters 3 and 4 the methodology of fittingfor a companion in the closure phase data and the procedure forproducing reconstructed images is explained; we show a seriesof simulations in Chapter 5 representing di ff erent scenarios anduse these to develop criteria for classifying non-detections, de-tections and potential disc features; in Chapter 6 we present theresults and classifications for each of the data sets and finally inChapter 7 we outline our conclusions.
2. Observations
Our high-angular resolution observations were conducted usingthe NIRC2 instrument at the 10m Keck-II telescope located onthe summit of Mauna Kea on Hawaii. We employed the sparse aperture masking technique, which allows us to remove atmo-spheric phase noise through the use of the closure phase tech-nique. We employed the nine hole mask on NIRC2, which o ff ersa good compromise between sensitivity and uv-coverage. Thechosen wavebands were H, K’ and L’ band filters as we expectan accreting companion to emit strongly in these bands. A list ofour target stars can be found in Table 1.Our data set was obtained during five nights between Jan-uary 2012 and June 2014 (Table 2). We observed most targets ina single epoch with the K’ filter (2.124 ± µ m) to search fordirect emission from any close protoplanetary candidates. Weobserved FP Tau and LkH α
330 again during the same epochbut in additional wavebands, L’ (3.776 ± µ m) and H (CH4s;1.633 ± µ m) respectively. We obtain an additional observa-tion in the same epoch in the case of TW Hya.The NIRC2 data were reduced using the pipeline describedpreviously in Ireland & Kraus (2008) and Kraus et al. (2013),providing calibrated closure phases. In order to record the instru-ment transfer function, we bracketed the science star observa-tions with observations of two unresolved calibrators. We aimedto alternate between two (ideally three) di ff erent calibrator stars,which allows us to still calibrate our data even if a calibrator isfound to be a previously unknown binary. A calibrator with spa-tially resolved structure will induce erroneous phase signals inour data, masking any companion signal or inducing a false sig-nal. We test for multiplicity in our calibrators by calibrating themagainst each other. In the cases where we have three calibratorswe can identify which calibrators are binaries by fitting a bi-nary model. In this way we find a 5 σ binary signal in HD 95105during the observations of TW Hya on 2012-01-10. We addition-ally observed this calibrator on in the same epoch on 2012-01-08and observed a binary signal from the same region of the sky, al-though substantially weaker during the first night at 3.5 σ . Theposition angle for both nights was found to be 219 ± ◦ while theseparations were found to be 113 ± ± ± ± ff ectedby degenerating factors such as short coherence times and vibra-tion in the optical system, we plotted histograms of the raw clo-sure phase measurements of the individual interferograms. Wefit gaussians to the closure phase distribution and derive the vari-ance ω . For the calibrator stars, ω provides a measure for theresidual phase noise as these point sources should not exhibit anintrinsic non-zero phase signal. We list the measured variancesin Table 3. The high variability seen in the L-band observationsis related to the misalignment of the IR dichroic which has sincebeen corrected.Our observations represent "snapshots" of the targets withoften little field rotation. This means that we are likely to su ff erfrom hole aliasing problems creating strong artefacts. For thisreason we only consider the most significant fit in the follow sec-tions and do not discuss apparent additional asymmetric signalswithout complementary observations.
3. Model fitting
We fit a star + companion model (Kraus & Ireland 2012) to themeasured closure phases. The free parameters are separation ( ρ ),positional angle (PA) and contrast ( f ). The results from this fit-ting are listed in Table 4. Article number, page 2 of 14illson et al.: Sparse aperture masking interferometry survey of transitional discs
Table 1.
Target list
Name Association Distance SpecType A v T e f f M ∗ L ∗ R ∗ References[pc] [K] [M (cid:12) ] [L (cid:12) ] [R (cid:12) ]DM Tau Taurus 140 M1 0.0 3705 0.47 0.37 1.3 6,8FP Tau Taurus 140 M5 0.3 3125 0.22 0.62 2.5 6,8LkH α
330 Perseus 250 G3 1.8 5830 1.25 2.78 1.5 2,5RXJ1615.3-3255 Oph 185 K4 1.00 4590 1.28 1.01 1.59 11,13RXJ1842.9-3532 CrA 136 K2 1.1 4900 1.33 1.29 1.5 14,15TW Hya TWHya 56 M0 1.0 3850 0.57 0.64 1.7 3,12V2062 Oph Oph 125 K3 2.3 4730 1.4 1.3 1.7 1V2246 Oph Oph 121.9 K0 6.2 5016 2.2 20.5 — 10,1,4,7
Notes.
Column 1: Target; Column 2: Association; Column 3: Distance; Column 4: Spectral Type; Column 5: Visual Extinction; Column 6:E ff ective Temperature; Column 7: Stellar Mass; Column 8: Stellar Luminosity; Column 9: Stellar Radius; Column 10: Literature References: (1)Bouvier & Appenzeller (1992), (2) Brown et al. (2009), (3) Calvet et al. (2004), (4) Chen et al. (1995), (5) Fernandez et al. (1995), (6) Furlan et al.(2006), (7) Jensen et al. (2009), (8) Kenyon et al. (1998), (9) Kraus et al. (2013), (10) Loinard et al. (2008), (11) Merín et al. (2010), (12) Reipurthet al. (1996), (13) Andrews et al. (2011), (14) Silverstone et al. (2006), (15) White et al. (2007) Table 2.
Log for our Keck / NIRC2 observations
Target Filter Date N visits
On Sky Rotation Calibrator[dd / mm / yy] [ ◦ ]DM Tau K’ 08 / /
12 3 1 HD 285938FP Tau K’ 20 / /
13 3 24 HD 283420, HD 283477L’ 20 / /
13 2 63 HD 283420, HD 283477LkH α
330 CH4s 16 / /
13 3 32 HD 22781, HD 281309K’ 08 / /
12 3 18 HD 22781, HD 2813092M034 + + / /
14 4 15 HD 146569, HD 146593,HD 148806RXJ1842.9-3532 K’ 09 / /
14 5 13 HD 171450, HD 171574,HD 176999TW Hya K’ 08 / /
12 3 8 HD 94542, HD 95105K’ 10 / /
12 3 11 HD 94542, HD 95105,HD 97507V2062 Oph K’ 09 / /
14 5 16 HD 147681, HD 148212,HD 148562V2246 Oph K’ 09 / /
14 1 2 HD 147742, HD 148352 N u m b e r o f C a li b r a t o r s F r a c t i o n o f C a li b r a t o r s Fig. 1. Left:
Distribution of significances for binary model fits to ourcalibrators.
Right:
Cumulative distribution of the fitted poisson distri-bution. We expect no significant asymmetric signal within these stars souse them to form a crude test statistic for our threshold of 4 σ . Fittinga poisson distribution we find a confidence level of 88% for a 4 σ andmore than 95% for 4.5 σ . We calibrate our detection threshold by investigating the bestfit significance distribution in our sample of 24 calibrator stardata sets. We fit binary models to the calibrated closure phasesof the calibrator stars, using the other calibrators observed inthe same area of the sky and close in time to build our sam- ple. This leads to a correlation between the values obtained ad-versely a ff ecting the shape of the distribution and the accuracyof our detection thresholds. Only one data set showed an asym-metric signal with a significance above 4.0 σ , and none above4.5 σ , setting a simplistic confidence level of greater than 95%for a significance of 4.0 σ and 99% for significances above 4.5 σ .Fitting a poisson distribution to the data set, we find more so-phisticated confidence levels of 88.3% for 4.0 σ and 95.4% for4.5 σ in agreement with our cruder approximation. Values ≥ . σ represent confidence levels of greater than 98%. This method isinaccurate as the low number of targets within each bin and thecross calibration of pairs or triples of calibrators results in valuesfor the significance which are only semi-independent. We see nostrong correlations between targets of similar R magnitudes orwithin the same night. This provides an intuitive interpretationfor the meaning of the significance values we calculate (See Fig-ure 1) and quantifies our ability to di ff erentiate from false posi-tives.Additionally we consider the sample of 54 M-dwarfs ob-served with SAM as part of an investigation into M dwarf multi-plicity by Gaidos et al. (2016) using Keck-II / NIRC2. They foundapproximately 25 % of science targets displayed asymmetric sig-nals within their closure phases in the 4-5 σ range. Furthermore, Article number, page 3 of 14 & A proofs: manuscript no. aa
Table 3.
Phase noise variance ω for uncalibrated closures phases. Science Target Filter Date Calibrator ω [dd / mm / yy] [ ◦ ]DM Tau K’ 08 / /
12 HD285938 4.51FP Tau K’ 20 / /
13 HD283420 5.85K’ 20 / /
13 HD283477 5.86L’ 20 / /
13 HD283420 11.20L’ 20 / /
13 HD283477 12.35LkH α
330 H 16 / /
13 2M0340 + / /
13 2M0400 + / /
13 HD22781 4.18H 16 / /
13 HD23849 6.47K’ 08 / /
12 HD22781 4.0K’ 08 / /
12 HD281309 4.7RXJ1615.3-3255 K’ 09 / /
14 HD146369 10.92K’ 09 / /
14 HD146593 10.78K’ 09 / /
14 HD148806 9.01RXJ1842.9-3532 K’ 09 / /
14 HD171450 10.8K’ 09 / /
14 HD171574 11.2K’ 09 / /
14 HD176999 12.3TW Hya K’ 08 / /
12 HD94542 5.8K’ 08 / /
12 HD95105 5.4K’ 10 / /
12 HD94542 9.3K’ 10 / /
12 HD95105 8.6K’ 10 / /
12 HD97507 9.3V2062 Oph K’ 09 / /
14 HD147681 11.80K’ 09 / /
14 HD148212 11.38K’ 09 / /
14 HD148562 11.17V2246 Oph K’ 09 / /
14 HD147742 18.4K’ 09 / /
14 HD148352 15.45 % were found to possess signals between 5-6 σ . This placeslower confidence levels on our 4 σ threshold but this sample oftargets is likely to be strongly a ff ected by systematics caused bythe observation strategy of a single visit and the use of the LaserGuide Star. This will result in little on-sky rotation and inferiorcalibration so is not as applicable to our data set, except in a caseswhere we too have little on-sky rotation (i.e. DM Tau). Measuredon-sky rotations are displayed in Table 2.The fitting procedure outlined above is suitable if the bright-ness distribution resembles a binary, but it may be inadequate formore complex distributions such as triple systems, or complexdisc features. To explore this we create maps of the significances( σ ) produced in a binary fit. We use the following equation toproduce complex visibilities for theoretical binary models fromwhich we construct model closure phases to fit to our measuredclosure phases V ( u , v ) = + f exp(2 π i ( u α + v β ))1 + f , (1)Here, f denotes the flux ratio of the model companion and theparent star, u and v are the Fourier plane coordinates and α and β are the angular coordinates of the companion within the model.We then construct a grid of positions in RA and DEC with aresolution of 1 mas, covering an area of 400 ×
400 mas with theparent star located in the centre of the field. At each positionwe fit for the best contrast and convert the calculated χ s intoa significance to form a map which enables us to make qualita-tive judgements about whether the detection resembles likely acompanion or a more complex brightness distribution. The sig- nificance is estimated using: σ = (cid:113) χ null − χ , (2)where the χ null is calculated using Eq. 1 taking an unresolved sin-gle point source. We enforce within our fits positive flux and fluxratios less than 1.0, physically representing that the companioncannot be brighter than the parent star.This modelling approach allows us to search for pointsource-like asymmetries consistent with a gap-clearing compan-ion. However we are unable to distinguish companions fromother potential sources of asymmetry which could mimic a pointsource in our data sets such as disc over-densities, accretionstreams and other complex structure. For this reason we onlyconsider significant detections to be companion candidates inneed of further observation rather than confirmed companions.To establish their nature as protoplanets or substellar compan-ions, evidence for orbital motion and ongoing accretion is re-quired. Detections of companions with separations ρ (cid:46) λ/ D are prob-lematic to fit because of a degeneracy that appears between theseparation and contrast. In this separation regime, the phasesdo not sample the full sinusoidal modulation that is requiredto constrain the companion contrast and separation separately.This makes our fits highly sensitive to the signal-to-noise ratio(SNR) of the closure phases, resulting in a range of separationsand contrasts that can reproduce the measured non-zero closure Article number, page 4 of 14illson et al.: Sparse aperture masking interferometry survey of transitional discs P h a s e [ ◦ ] ρ = 26 mas, ∆ M K = 6 . magρ = 39 mas, ∆ M K = 6 . magρ = 52 mas, ∆ M K = 7 . mag /f )0102030405060708090 S e p a r a t i o n ( m a s ) Fig. 2.
Degeneracy plots of the detection in DM Tau.
Left:
Phases forthree companions at three separations and their contrasts according toEq. 5.
Right:
Fit of the degeneracy profile using the shortest projectedbaseline length. The fit using the longer baseline well describes the pro-file at larger separations but poorly describes the shorter separationsas the contrast ratio asymptotically goes to 1.0 as the separation ap-proaches λ/ D . The shortest baseline poorly follows the structure atlarger separations but does follow closely the profile at closer separa-tions as a result of its ability to probe the more SNR sensitive regionclose to the λ/ D resolution limit. phases equally well (see Figure 2). We therefore find that a sim-ilarly good fit can be obtained for di ff erent separation / contrastpairings. This is most clearly seen within the significance mapsthemselves, producing lobe-like structures in the region between λ/ D and λ/ D .To explore this degeneracy and allow one to translate fromone separation / contrast pair to another we take two approaches.The first approach is to plot the degeneracy directly. We plot agrid of contrasts against separations along the non-degeneratebest-fit position angle and construct a significance map in thesame manner as outlined above (see Figure 2). In our secondapproach, we aim to derive an analytic expression for the sepa-ration / contrast degeneracy. For this purpose, we start from Eq. 1and retrieve the phase component, φ ,tan φ = f sin( − π b ρ )1 + f cos( − π b ρ ) , (3)where ρ is the scalar companion separation for our best fit posi-tion and b is the projected length of the baseline along the vectorseparation. Rearranging we find: f = sin φ sin( − φ − π b ρ ) , (4)For small values of φ (i.e. values of φ < π/
4) this second equa-tion can be further simplified using the small angle approxima-tion: f ≈ − φφ + π b ρ , (5)To most accurately trace the profile of the degeneracy, wewould need to use every u projected baseline and weight accord-ing to their associated uncertainties. However, using simply theshortest projected baseline was found to be e ff ective for trac-ing the degeneracy to smaller separations. Within our degener-acy plots, the physical degenerate region is shown by the blackcontour defining the ∆ σ = . χ . In these cases we set the sep-aration to the resolution limit of our observations; λ/ B , where λ is the wavelength of the observations and B is the longest base-line from our mask. However the separations are not well con-strained and solutions at larger separations and higher contrastswould result in good fits of similar significances. Our analyticalsolution enables us to calculate the contrast at a di ff erent separa-tion from our fit here.
4. Image Reconstruction
In order to retrieve the brightness distribution of the observed ob-jects in a model-independent way, we use image reconstructiontechniques developed for infrared long-baseline interferometryon our measured calibrated closure phases and visibilities.The image estimation from the discrete points in the Fourierplane (the aperture masking measurements) can be considered asan inverse problem. Given that there are more pixels than mea-surements, the problem is ill-posed and solving it requires one toadopt a Bayesian approach. This amounts to minimising a globalcost function ( F ) defined as: F = F data + µ F rgl , (6)where F data is the likelihood term (here the χ ), F rgl is the reg-ularisation term and µ the regularisation weight (see Thiébaut2008; Renard et al. 2011, for more background information).The likelihood term ensures that the image is reproducing thedata whereas the regularisation term helps to fill the gaps in theFourier space by interpolating it in a specific way defined bythe user. This term helps also to converge to the most likely a-posteriori estimate of the image.To perform our image reconstructions we have chosen theMiRA algorithm (Thiébaut 2008). This algorithm is minimisingthe cost function ( F ) with a downhill gradient method. In ourobjects the central star is spatially unresolved. In order to imageits environment we have therefore modelled it as a point sourceand reconstruct an image of the environment only, using the ap-proach outlined in Kluska et al. (2014).The images are defined to have 128 ×
128 pixels each. Forthe pixel size, we chose 5, 7 and 11 mas for H , K and L -bandsrespectively. We have chosen to use the quadratic smoothnessregularisation (Renard et al. 2011). We employed the L-curvemethod (see Renard et al. 2011; Kluska et al. 2014, for moredetails) to determine the weight of the regularisation for all datasets and then used the average weight of all the L-curves whichis µ = .To define the fraction of the stellar flux in the parametricmodel, we made a grid of reconstructions with di ff erent flux ra-tios for the star. Because we are minimising the global cost func-tion F we should have chosen the images having the minimum F value. Because of the regularisation e ff ects, these images stillhave flux at the star position which is not physical. Therefore wedecided to keep the flux ratio for which the image has the smallerlikelihood term ( F data ). These images do not di ff er significantlyfrom the images with smaller F except in correcting this e ff ect. Article number, page 5 of 14 & A proofs: manuscript no. aa
Table 4.
Binary fit results
Target Filter Date Contrast Significance Comment[dd / mm / yy] [mag]DM Tau K’ 08 / /
12 6.8 ± / /
13 4.4 ± / /
13 4.1 ± α
330 CH4s 16 / /
13 5.7 ± / /
12 5.6 ± / /
14 4.9 ± / /
14 5.5 ± / /
12 5.6 ± / /
12 3.7 ± / /
14 5.00 ± / /
14 4.3 ± Notes. () Criterion for detection or non-detection is based on a σ of greater than 4.0 representing a confidence level greater than 88%. However thereare e ff ects which can mimic a detection of between 4-5 σ so for these cases which look individually at each target and try attempt to rule them outthrough inspection of their reconstructed images, significance maps, uv-coverage and visibilities. Detections close to 6.0 σ ( >
5. Simulations - reference models
Within many of our reconstructed images and significance mapswe see patterns or structures which are not consistent with sim-ple point source companions. To aid our understanding of thesestructures we simulated a range of possible scenarios. We simu-late companions with di ff erent separations, position angles andcontrasts in order to understand potential e ff ects that might becaused by the imperfect uv-coverage and to investigate how thestructure of the significance maps changes within the fully re-solved and partially resolved regimes described in Section 3.2.While we expect these scenarios to cover most structures likelyto be seen, this is an incomplete set and other scenarios may oc-cur.For our simulations we model data sets that correspond to theK-band and the NIRC2 9-hole mask. We add phase noise witha variance of ω = ◦ , which resembles good conditions in ourobservations. In data sets where the companion or disc wall was positioned atseparations at or below λ/ D we see that the images and signif-icance maps become dominated by the gaussian noise placed inthe models (see Figure 3). In all the cases shown, the artificialcompanion has a contrast of f = .
1. We find a much reducedsignificance compared to a similar companion at larger separa-tions. We therefore consider any companion with a separationbelow λ/ D to be unresolved. In cases where our uv-coverageis sparser caused by the flagging of one or multiple holes duringthe data reduction process, we often see this noise as periodicsignals in the background distribution. The strength of these pe-riodic signals is dependent on the precise uv-coverage and thelevel of noise in the closure phases.Within the reconstructed images, we find that data sets withan unresolved companion will simply be dominated by randomlydistributed noise peaks (i.e. TW Hya, K’-band). We also en-countered cases, where the image reconstruction algorithm at-tributed the flux elements of the companion to the central star(e.g. FP Tau, L’-band). Both are shown in Figure 11. In thesecases we are limited to placing lower limits on the possible con-trast for a companion around these targets at separations within
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) /f )0102030405060708090 S e p a r a t i o n ( m a s ) Fig. 3. Left:
Significance map.
Right:
Degeneracy plot. Simulated dataof a a companion located at a separation of 10 mas, a position angle of90 ◦ , and contributed 10% of the total flux (white triangle). The whitecircle shows our best fit position. Within the background it is possibleto see noise artefacts caused by holes within the uv-coverage. Theseholes create periodic signals within the background and may take ongeometric patterns.
200 mas. This limit is set at the 99% confidence level which isdetermined by the individual noise properties of the data.
To study a marginally resolved regime, we simulated data witha companion at a separation of 30 mas. We observe the "stronglobe" structure characteristic of this regime (Figure 4). In thecase of a low-contrast companion ( f = . To simulate a fully-resolved companion we computed modelswith a companion located at a separation of 60 mas, just beyond λ/ D . At these separations we can see that the degenerate regionhas largely disappeared allowing the position of the companionto be well constrained (see Figure 5). Article number, page 6 of 14illson et al.: Sparse aperture masking interferometry survey of transitional discs
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) /f )0102030405060708090 S e p a r a t i o n ( m a s ) Fig. 4. Left:
Significance map.
Right:
Degeneracy plot. Simulated dataof a companion located at a separation of 30 mas, a position angle of90 ◦ , and contributes 10% of the total flux (white triangle). The whitecircle shows our best fit position. We see the distinctive lobing of apartially resolved companion. In this case with excellent SNR we areable to accurately identify the location of the companion but in practicethis is not always the case.
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) /f )0102030405060708090 S e p a r a t i o n ( m a s ) Fig. 5. Left:
Significance map.
Right:
Degeneracy plot. Simulated datafor a companion located at a separation of 60 mas, a position angle of90 ◦ , and contributed 10% of the total flux (white triangle). The whitecircle shows our best fit position. Here the companion is fully resolvedand the position and contrast are well constrained. Asymmetries in the brightness distribution can also be causedby disc-related structures, producing closure phase signals thatmight be di ffi cult to discern from those produced by close-incompanions (Cheetham et al. 2015). To investigate this scenario,we produced synthetic images that are intended to mimic therim of a disc seen under intermediate inclination (60 ◦ from theface-on orientation) with a radius of 30 mas, which correspondsto ∼ λ/ D . The image shown was produced by simulating askewed ring with a gaussian profile and a width of 15 mas, askewness of 0.8, and whose major axis is oriented along positionangle 0 ◦ . The flux of the disc represents 1% of the total flux inthe frame.Within all the significance maps from this scenario we ob-serve that the significance contours take on double-lobed struc-tures (Figure 6). This is in agreement with previous work per-formed by Cheetham et al. (2015), who showed that an innerwall of a optically thick disc will appear as two point-sourcelike structures co-locational with the illuminated rim of the disc,bisected by the center of the disc wall. We find that these alsoappear within our significance maps and reconstructed images.Extending the semi-major axis such that the ring appears out-side of the degenerate region we begin to resolve the shape of thedisc wall. This structure tends to be comparable in strength tothe artefacts however and is unseen unless the flux contributionof the disc is not enhanced. This is the result of the flux becom-
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) −200−100 0 100 200−200−100 0 100 200 Da (mas) Dd ( m a s ) Fig. 6. Left:
Significance map for a simulation with a partially resolveddisc wall. The resulting significance maps show two strong detectionslocated at the disc wall. At increasing separations and resolution the twopoint sources begin to merge.
Right:
Input intensity distribution. Greenstar indicates the position of the parent star. ing more spread out within the frame, inducing smaller phasesignals.To make a comparison to a more physical model we createa disc model using the radiative transfer code, TORUS (Harries2014). Here we can include e ff ects such as forward scatteringfrom the near edge of the disc. We scale this model to have semi-major axes of 30, 45, 60, 90, 120, and 180 mas. The results areshown in Figure 7. The forward scattered component, while con-taining more flux, is closer to the central star than the thermalcomponent so only appears at larger separations. It also appearsas a single lobe as a resut of flux being most concentrated at thecentre of the arc whereas in the thermal case, the flux is moreevenly spread across the disc wall. At larger separations this sin-gle lobe becomes more resolved, similar to the thermal emissionseen in the bottom left frame of Figure 7 and similarly di ffi cultto distinguish from artefacts. To investigate scenarios in which an asymmetry is caused by anover density within the disc we simulate a disc feature with acontrast of 5:1 to the rest of the disc. We take a similar approachto Section 5.4 but skew the ring in such a way to resemble pos-sible asymmetries such as those found in simulations by de Val-Borro et al. (2007). These are extreme cases as it is di ffi cult tophysically create such a strong contrast particularly in contin-uum emission (Juhász et al. 2015).In Figure 8 we show two cases representing partially re-solved and fully resolved cases. Both strongly resemble thestructures seen in our companion detection simulations in Sec-tions 5.2 and 5.3. This should be kept in mind when consid-ering our companion detections without complementary multi-wavelength observations.
6. Results
We identify potential candidates through a combination of set-ting a threshold on the significance of the binary fit and inspec-tion of reconstructed images and significance maps. In Table 5we list data sets in which we find significant closure phase asym-metries excluding false positives and each case is discussed in-dividually in detail below.To calculate the semi-major axis for our candidates we as-sume circular orbits coplanar with the outer disc. Where disc in-clination / position angle information is unavailable, we assume aface-on disc. To estimate the companions’ absolute magnitudes, Article number, page 7 of 14 & A proofs: manuscript no. aa −50 0 50−50 0 50
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Fig. 7. Top Left:
Base model for physical disc simulations. We scalethis model for each semi-major axis case. Thermal emission from thefar side of the disc is seen in the right side of the frame and forwardscattered light in the top left.
Top Right:
30 mas case. Here we see thedouble lobe structure caused by the far side of the inner disc wall.
Bot-tom Left:
60 mas case. The arc of the far side of the disc wall is clearlyseen but is comparable in strength to the artefacts within the frame andso would be di ffi cult to identify in practice. Bottom Right:
90 mas case.The forward scattered component is now the dominant feature withinthe frame. It forms a single lobe owing to the greater concentration offlux in the centre of the arc than in the thermal emission from the oppo-site side of the disc wall. we used the reddening law outlined in Cardelli et al. (1989).From the dereddened absolute magnitudes, we then estimate val-ues of M c ˙ M c using the accreting protoplanetary disc models de-scribed by Zhu (2015). We match our dereddened absolute mag-nitudes to the table within Zhu (2015), assuming an inner cir-cumplanetary disc radius of 2R J . This is a highly unknown quan-tity with a significant e ff ect on the resultant values of M c ˙ M c . Wearbitrarily chose our inner disc radius to be the same value asthat assumed by Sallum et al. (2015) for the purposes of com-parison. When matching the absolute magntiudes are di ffi cult tomatch we linearly interpolate. We are unable to directly estimatethe mass of a potential companion as these objects are thoughtto likely possess extended, accreting circumplanetary discs thatdominate the infrared excess emission; this prevents us from sep-arating the mass M c and accretion rate ˙ M C .In addition, we place lower limits on the contrast at the 99%confidence level. For the data sets that were recorded under detri-mental conditions, our sensitivity is reduced from typical K-bandcontrasts of ∆ m λ = ∆ m λ > + dust masses and the disc po-sition angles are measured East-of-North along the major axis.Besides the significance maps derived from closure phase fit-ting, we also show the visibility amplitudes derived from our ob-servations.
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) −200−100 0 100 200−200−100 0 100 200 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) −200−100 0 100 200−200−100 0 100 200 ∆α (mas) ∆ δ ( m a s ) Fig. 8. Left:
Significance map for a simulation with a partially resolvedand fully resolved disc asymmetry. The resulting significance mapsshows structure similar to a companion detection.
Right:
Input inten-sity distribution. Green star indicates the position of the parent star.
The structure of the inner 10s of au around DM Tau is complexand di ffi cult to constrain with SED-based models alone. Study-ing the Spitzer
IRS spectrum, Calvet et al. (2005) modelled theSED of DM Tau inferring the presence of a 3 au inner cavity inthe disc. In contrast, Andrews et al. (2011) used SMA data tospatially resolve an inner disc cavity with a radius of 19 ± µ m observations. Neither model however explains simul-taneously the IR and sub-mm spectra suggesting the inner discis potentially populated by a species of small dust grains (Cal-vet et al. 2005). Andrews et al. (2011) additionally estimated thetotal disc mass to be 0.04 M (cid:12) and measured the inclination andposition angles of the disc to be 35 ◦ and 155 ◦ respectively.The result of our simple binary model indicates a com-panion at 43 mas ( ≈ M K = ± σ (93% confi-dence level) (see Figure 9). This places the companion candidatewithin the disc cavity resolved by Andrews et al. (2011) and out-wards of the ring of small dust grains suggested by Calvet et al.(2005). We find a value of M c ˙ M c = − M J yr − . The sourceof the asymmetric signal is located within the partially resolvedregion nevertheless the SNR in the closure phases is su ffi cient toconstrain the separation through our binary fitting. The net re-sult is an inflation in the uncertainties within the separation andcontrast (see Table 5).We set limits on the contrast of a companion within 200 masat ∆ m K > ∆ m K > ∼ ◦ ) and low strength of the detection. Thesmall on-sky rotation in particular makes this case vulnerable tosystematics which may mimic a detection. Multiple visits to thetarget however should aid in reducing such e ff ects but the confi-dence levels established by comparison to the sample collected Article number, page 8 of 14illson et al.: Sparse aperture masking interferometry survey of transitional discs
Table 5.
Companion candidates
Identifier ρ PA Contrast Sig Semi-Major Axis R In M K M c ˙ M ca [mas] [ ◦ ] [mag] [ σ ] [AU] [AU] [mag] [10 − M J yr − ]DM Tau 43 ± ± ± ± ± / (b) ± α
330 132 ± ± ± ± ± ± ± ± ± ± ± ± Notes.
Columns are organised by Identifier, angular separation, position angle, significance, orbital separation based on previous observationsof the disc inclination and position angle (error when distance error neglected), inner radius of optically thick disc, absolute magnitude of thecompanion, stellar accretion rate and companion mass. Dereddening was performed as described in Cardelli et al. (1989) ( a ) Values derived fromZhu (2015) assuming circumplanetary disc radii extending from 2 R J outwards. ( b ) First value derived from fitting IR spectra, second from fittingto sub-mm data
DMTau − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) /f )0102030405060708090 S e p a r a t i o n ( m a s ) Sq u a r e d V i s i b ili t y LkHa330 − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) /f )020406080100120140160180 S e p a r a t i o n ( m a s ) Sq u a r e d V i s i b ili t y TWHya − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) /f )020406080100120140160180 S e p a r a t i o n ( m a s ) Sq u a r e d V i s i b ili t y Fig. 9.
Potential Candidate Detections:
Left:
Reconstructed Image.
Middle Left:
Computed significance map.
Middle Right:
Degeneracy plot
Right: V plots. First row:
DM Tau, K-band,
Second row: , LkH α Third row: , TW Hya, K’-band
FPTau − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y Fig. 10.
Potential disc feature detections
Left:
Reconstructed Image
Right:
Computed significance map
Right: V plot. FP Tau, K-band.Article number, page 9 of 14 & A proofs: manuscript no. aa
FPTau − Lband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y LkHa330 − Hband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y TWHya − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y V2246Oph − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y Fig. 11.
Data sets where we see no significant emission
Left:
Reconstructed Images
Middle:
Computed significance maps
Right: V plot. Firstrow:
FP Tau, L-band,
Second row:
LkH α Third row: , TW Hya, K’-band,
Fourth row: , V2246 Oph, K’-band. In these cases we setlimits on the contrast of a potential candidate or disc feature. by Gaidos et al. (2016) are likely to be more applicable to thiscase ( ∼ Furlan et al. (2005) classified FP Tau as a Class II object basedon
Spitzer mid-infrared spectra and inferred the presence of anextended gap within the disc from the lack of near-infrared ex-cess flux. This was further supported by later analysis by Currie& Sicilia-Aguilar (2011) but neither characterised the spatial ex-tent of the gap. They did however measure the disc mass to be2.5 × − M (cid:12) .We see clear disc structures within FP Tau, with the K-banddata set displaying both "dual lobing" in the significance map and dual point source-like emission in the image (Figure 10),which we identified in Section 5.4 as likely indicators for disc-related asymmetries. From this we conclude the inner edge of theouter disc of FP Tau to be likely moderately inclined and locatedbetween 26-52 mas. The SNR of the closure phase is insu ffi cientto find a solution for the separation and we are forced to fix theseparation to λ /
2D within our fit. This corresponds to ∼
26 masand leads to an inner edge located at 10.0 ± ∆ m K > σ significance threshold, which is consis-tent with the reconstructed images where we see little o ff -centre Article number, page 10 of 14illson et al.: Sparse aperture masking interferometry survey of transitional discs
RXJ1615.3−3255 − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y RXJ1842.9−3532 − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y V2062Oph − Kband −200−100 0 100 200−200−100 0 100 200 00.10.20.30.40.50.60.70.80.91 ∆α (mas) ∆ δ ( m a s )
200 100 0 -100 -200 ∆ RA (mas)-200-1000100200 ∆ D e c ( m a s ) Sq u a r e d V i s i b ili t y Fig. 12.
Data sets where we detect significant asymmetric signal but rule it to be a false positive through similarity in structure to other data setstaken during the same night.
Left:
Reconstructed Images
Middle:
Computed significance maps.
Right: V plots. First row:
RXJ1615.3-3255,K-band,
Second row:
RXJ1842.9-3532, K-band,
Third row: , V2062 Oph, K’-band. We rule out these asymmetries because of the similarity inthe structure of their artefacts. All three targets were observed on the same night and the same structure is also observed in some of the calibrators.The position angle of the structures di ff er by the same angle as the on sky rotation of the 9-hole mask (See Sections 6.5 and 6.7). The cause of thise ff ect is unknown. flux (see Figure 11). Between 50-200 mas we set an lower limiton the contrast of a companion as ∆ m L > . ± ◦ and25 ± ◦ respectively. We find a semi-major axis of 60 mas in thecase of the ring model and a FWHM of 80 mas in the gaussianprofile case. We find the K-band visibilities to be consistent withan unresolved object within the measurement uncertainties, in-dicating a more compact emitting region. α LkH α
330 has been extensively studied in unresolved spec-troscopy and through interferometry in the millimetre. Brownet al. (2007) inferred a disc gap between 0.7-50 au through SEDmodelling. This was in agreement with later modelling of theSED by Andrews et al. (2011). They resolved the gap cavityin sub-mm SMA observations, inferring in the process that theinfrared emission had its origin within the cleared region gap.They attribute this infrared emission to an additional popula-tion of small dust grains located in the gap. The disc was found to have an inclination of 35 ◦ and to be oriented along positionangle 80 ◦ . They estimated the disc mass to be 0.025 M (cid:12) . Isellaet al. (2013) carried on further study of the outer disc through theSMA data. They identified a "lopsided" ring in the 1.3 mm ther-mal dust emission at a radius of 100 au. Through hydrodynamicsimulations they find this asymmetric ring to be consistent withperturbations in the surface density of the disc caused by an un-seen companion. They set limits on the mass and orbital radiusof this companion to > J and <
70 au respectively.Our observations of LkH α
330 were performed in K- and H-band at two epochs separated by 678 days. We see no significantsignal within the H-band data set but do see a strong asymmetricsignal within the K-band closure phases indicative of a compan-ion detection.The contrast of the best-fit companion candidate was foundto be ∆ m K = . ± . σ = . M c ˙ M c to be 10 − M J yr − .Amongst our companion candidate detections, the K-bandobservations of LkH α
330 display the most pronounced visibil-ity drop ( ∼ . − . Article number, page 11 of 14 & A proofs: manuscript no. aa by holes within our uv-coverage. With the existing data set, wecannot rule out that the aforementioned asymmetric signal maybe associated with these artefacts. Additionally our calculatedupper limits on the contrast of a companion were found to be ∆ m K < . J yr − values we would expect contrasts of 6.0-6.2 mag in H-band, well below the 99% confidence level prevent-ing us from ruling out the K-band detections using the H-bandobservations. We place upper limits on a companion contrast at ∆ m H < . Previous, resolved observations of RXJ1615.3-3255 are limited.Makarov (2007) linked the object kinematically to the Lupus as-sociation at a distance of approximately 185 pc. Henize (1976)and Krautter et al. (1997) classified RXJ1615.3-3255 as a weak-line T Tauri star, whereas Merín et al. (2010) classified it as apotential transitional disc based on
Spitzer spectra.Andrews et al. (2011) resolved the disc at 880 µ m withSMA observations and found that the emission from the disc ishighly extended suggesting a large disc extending out to 115 au,and they measure a particularly low-density cavity extending to30 au. The low density of the cavity forced them to remove alldust from their models from within 0.5 au of the star. The lowfar-infrared flux of the source was interpreted by them to be asa result of the e ff ects of dust settling in the outer regions of thedisc. This leads to a high estimate for the disc mass of 0.13M (cid:12) ,that is ∼
12% of the stellar mass. They estimated the disc incli-nation to be 4 ◦ with position angle 143 ◦ .We observed RXJ1615.3-3233 at a single epoch in theK-band and detected a significant asymmetry in the closurephases. However, inspecting the significance maps we seestrong similarity between RXJ1615.3-3233, RXJ1842.9-3532and V2062 Oph. All three targets were observed on the samenight (09 / / ff er froman systematic e ff ect that results in close to identical structure.The rotation of the structure is equal to the on-sky rotation of themask. We are not able to identify the precise cause of this sys-tematic e ff ect, but note that the night su ff ered from poor atmo-spheric conditions and variable wind speeds, which might haveinduced vibrations and degraded the AO performance (thesepoor conditions also reflect in a high variance in the individ-ual uncalibrated closure phase; see Table 3). The visibilities arealso strongly a ff ected by this systematic, showing similar strongdrops and structure.We set a lower limit for the contrast of a potential binaryto ∆ m K > ∆ m K > ff ect adversely the accuracy of the limits weset in these cases. Hughes et al. (2010) used a combination of resolved SMA obser-vations and SED modelling to infer the presence of an opticallythin region inwards from 5 au with a narrow ring of opticallythick material at ∼ . − . (cid:12) and measure the inclinationto be 54 ◦ with a position angle of 32 ◦ .We detect no significant asymmetric signal in the closurephases but see the same systematic structure in the significancemaps as in RXJ1615.3-3255 and V2062 Oph (see Section 6.4).We set lower limits on the contrast of a companion at ∆ m K > Estimates by Calvet et al. (2002) found that the optically-thickdisc of TW Hya extends from 4 to 140 au, with a mass of 0.06 M (cid:12) for a 10 Myr old disc. They additionally found that the inner re-gion of the disc is not fully cleared. A population of 1 µ m dustgrains is required within the optically thin inner 4 au to prop-erly fit the SED in agreement with observed continued accretiononto TW Hya. This interpretation is supported by recent ALMAobservations by Andrews et al. (2016) which probed, through870 µ m emission, the distribution of millimeter-sized grains tospatial scales on the order of an au. They observed ring struc-tures suggestive of ongoing planet formation, in particular anunresolved inner disc within 0.5 au and a bright ring at 2.4 auseparated by a dark annulus centred at 1 au.Radial velocity studies of this object performed by Setiawanet al. (2008) provided evidence for the presence of a 9.8 ± . M J planet on an orbit with a semi-major axis of 0.041 ± ∼ (cid:48)(cid:48) . (cid:48)(cid:48) .
2, correspond-ing to distances of 2.75 to 11 au.Using VLT / NACO, Vicente et al. (2011) searched for a po-tential companion in 1.75 µ m and 2.12 µ m. They employed theLOCI PSF removal algorithm and detected no companion moremassive than 0.11 M (cid:12) outward of 5.5 au (0 (cid:48)(cid:48) .
1) or brown dwarfcompanion outward of 7 au (0 (cid:48)(cid:48) .
13) or planetary mass outwardof 13 au (0 (cid:48)(cid:48) .
24) at a contrast of 2 mag. Outward of 87 au theyachieve their maximum contrast sensitivity of 8 mag allowingthem to rule out companions above 7 M J . Evans et al. (2012)observed TW Hya with Keck-II / CONICA in L-band in March,2009 and observed no significant asymmetric signal within200 mas and set lower limits on the contrast of a companion.We observed TW Hya twice in the K-band on non-consecutive nights in the same epoch. In the data from the firstnight, we see a significant asymmetric signal corresponding toa contrast of ∆ m K = ± ∼ M c ˙ M c = − M J yr − . We set limits on the contrast of a companion at ∆ m K > ff ected. Between 80-160 maswe set contrast limits of ∆ m K > Article number, page 12 of 14illson et al.: Sparse aperture masking interferometry survey of transitional discs
Comparing our result from the first night to the limits in L-band set by Evans et al. (2012), we use the circumplanetary discmodels in Zhu (2015) to estimate the L-band absolute magnitudean accreting companion of this absolute magnitude would dis-play. We find the expected contrast to be ∆ m L ≈ ∆ m L > M onto TW Hya isknown to be highly variable with values fluctuating at least byan order of magnitude (Alencar & Batalha 2002). This variationoccurs on a time scale of years and we would expect the accre-tion rate onto a companion to be related to the amount of materialflowing through the disc so we cannot rule out this companioncandidate based on previous SAM observations.Another possible cause is that the origin of this asymmetryis not protoplanetary in nature but instead from a another po-tential source of asymmetry such as an accretion stream or discasymmetry. ALMA observations carried out by Andrews et al.(2016) at ∼ ∼ ∼ Espaillat et al. (2010) modelled the
Spitzer
SED and found adisc cavity extending to 36 au containing some optically thindust consistent with other resolved observations of V2062 Oph.Andrews et al. (2011) finds a cavity in the disc extending out to30 au. They additionally constrained the inclination and positionangle to 35 ◦ and 80 ◦ respectively and they estimate the disc massto be 0.007 M (cid:12) .We observed V2062 Oph in the K-band within a singleepoch. Observing conditions were not ideal, but we detect a sig-nificant asymmetry signal in the closure phases. As mentionedpreviously in Section 6.4, the produced structures are also re-produced within the RXJ1615.3-3255 and RXJ1842.9-3532 datasets leading to the conclusion that these are false positives alongwith V2062 Oph.We set lower limits on the contrast of a companion at ∆ m K > > Mid-infrared 9-18 µ m Gemini observations by Jensen et al.(2009) resolved V2246 Oph at subarcsecond resolution andfound very little mid-infrared excess within 100 au. Beyond thisregion they observed strongly extended and asymmetric emis-sion out to 100s of au. The asymmetric emission forms a halfring structure to the north west, at an angular separation of 1 (cid:48)(cid:48) .
1. Vicente et al. (2011) observed V2246 Oph as part of theirVLT / NACO high resolution observations. They reached sensi-tivities of 15 M J and 6 M J past separations of 3 au and 192 aurespectively. They found no evidence for a companion withinthese limits.We observed V2246 Oph in the K-band in a single epoch.Poor observing conditions severely limited the sensitivity of ourobservations. We place limits on the contrast of a potential com-panion at ∆ m K > >
7. Conclusions
In this paper we presented results of five nights of Keck sparseaperture masking observations on eight targets in K-band, onein L-band (FP Tau), one in H-band (LkH α ff ect that a ff ected one of our observ-ing nights. The remaining detected asymmetries indicate eitherthe presence of complex disc structures and / or the presence ofcompanions. We conducted detailed simulations in order to un-derstand the signature that these di ff erent scenarios produce inour phase measurements and investigated the degeneracies thatoccur between the derived separation and contrast parameters inthe case of marginally resolved companions.Using both modelling and image reconstruction methods, weinvestigated the likely origin of the asymmetries for each targetstar. We estimate confidence levels for our companion detectionsthrough fitting companion models to a sample of 24 calibratorsstars known to be point source-like. We use the resultant ditri-bution to form our confidence levels. We report companion de-tections at a confidence level of >
99% ( > . σ ) in LkH α >
95% ( > . σ ). For the detections, wederive M c ˙ M c values of 10 − M J yr − (LkH α − M J yr − (DM Tau) and, 10 − M J yr − (TW Hya). Additionally we inferthrough comparison to limits previously set on the contrast ofa companion in L-band that the origin of the asymmetry signalwithin the TW Hya data set would require an increase in the ac-cretion rate of an order of magnitude within a few years for it tobe consistent with an accreting protoplanet, assuming accuratescaling from L- to K-band. Observations by Alencar & Batalha(2002) indicate TW Hya to be a highly variable disc with val-ues of ˙ M varying by an order of magnitude over time scales of ayear, adding support to this scenario.In LkH α
330 and DM Tau the gap properties have been char-acterised by earlier observations and we find the companion can-didates to be located within the disc gaps, suggesting that theyare orbiting within the cleared regions of the disc. In the caseof TW Hya we find the companion candidate to be located onthe outer edge of the bright annulus located at 2.4 au in recent350GHz ALMA observations by Andrews et al. (2016). Fur-thermore we find that separation of our companion candidate tolie within the shallow gap at 6 au observed by Tsukagoshi et al.(2016) in 138 and 230GHz ALMA observations.We interpret the asymmetries in FP Tau be associated withdisc emission, most likely a disc wall between 20-40 mas, similarto the asymmetries seen in T Cha (Cheetham et al. 2015) andFL Cha (Cieza et al. 2013). This is supported through strongdrops in the visibilties in both the K- and L-band observationsof this target. Fitting geometric disc models to the data sets we
Article number, page 13 of 14 & A proofs: manuscript no. aa find find visibilities consistent with a compact emitting region inK-Band and an extended component in L-band with a positionangle of 350 ± ◦ and an inclination of 25 ± ◦ . Finally, for theremaining data sets we detect no significant asymmetries and setlower limits on the contrast of potential companions.With the detection of significant asymmetries in four outof eight target stars, our detection frequency is relatively high(50%). This is higher than the detection rate that was found insurveys of other object classes (14%: Kraus et al. 2008; 20%:Kraus et al. 2011) conducted with Keck / NIRC2 SAM interfer-ometry with a same observational setup and a similar data anal-ysis scheme. This demonstrates that transitional discs indeedtrace a particularly interesting phase in disc evolution and high-lights the need for further studies on these object classes with theunique observational window that SAM provides, both with thecurrent-generation telescopes and the upcoming generation ofExtremely Large Telescopes. Besides further continuum imag-ing, it is promising to image these objects in accretion tracingspectral lines such as H α , in order to confirm that these objectsare sites of continued accretion and to ultimately establish theirclassification as protoplanets. Acknowledgements.
We acknowledge support from a STFC Rutherford Fellow-ship and Grant (ST / J004030 /
1, ST / K003445 / References
Alencar, S. H. P. & Batalha, C. 2002, ApJ, 571, 378Alencar, S. H. P., Teixeira, P. S., Guimarães, M. M., et al. 2010, A&A, 519, A88Andrews, S. M., Wilner, D. J., Espaillat, C., et al. 2011, ApJ, 732, 42Andrews, S. M., Wilner, D. J., Zhu, Z., et al. 2016, ArXiv e-printsArmitage, P. J. 2011, ARA&A, 49, 195Ayli ff e, B. A. & Bate, M. R. 2012, MNRAS, 427, 2597Biller, B., Lacour, S., Juhász, A., et al. 2012, ApJ, 753, L38Birnstiel, T., Ormel, C. W., & Dullemond, C. P. 2011, A&A, 525, A11Bouvier, J., Alencar, S. H. P., Boutelier, T., et al. 2007, A&A, 463, 1017Bouvier, J. & Appenzeller, I. 1992, A&AS, 92, 481Brandeker, A., Jayawardhana, R., & Najita, J. 2003, AJ, 126, 2009Brauer, F., Dullemond, C. P., & Henning, T. 2008, A&A, 480, 859Brown, J. M., Blake, G. A., Dullemond, C. P., et al. 2007, ApJ, 664, L107Brown, J. M., Blake, G. A., Qi, C., et al. 2009, ApJ, 704, 496Calvet, N., D’Alessio, P., Hartmann, L., et al. 2002, ApJ, 568, 1008Calvet, N., D’Alessio, P., Watson, D. M., et al. 2005, ApJ, 630, L185Calvet, N., Muzerolle, J., Briceño, C., et al. 2004, AJ, 128, 1294Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245Cheetham, A., Huélamo, N., Lacour, S., de Gregorio-Monsalvo, I., & Tuthill, P.2015, MNRAS, 450, L1Chen, H., Myers, P. C., Ladd, E. F., & Wood, D. O. S. 1995, ApJ, 445, 377Cieza, L. A., Lacour, S., Schreiber, M. R., et al. 2013, ApJ, 762, L12Currie, T. & Sicilia-Aguilar, A. 2011, ApJ, 732, 24de Val-Borro, M., Artymowicz, P., D’Angelo, G., & Peplinski, A. 2007, A&A,471, 1043Espaillat, C., D’Alessio, P., Hernández, J., et al. 2010, ApJ, 717, 441Espaillat, C., Furlan, E., D’Alessio, P., et al. 2011, ApJ, 728, 49Espaillat, C., Muzerolle, J., Najita, J., et al. 2014, Protostars and Planets VI, 497Evans, T. M., Ireland, M. J., Kraus, A. L., et al. 2012, ApJ, 744, 120Fernandez, M., Ortiz, E., Eiroa, C., & Miranda, L. F. 1995, A&AS, 114, 439Fouchet, L., Gonzalez, J.-F., & Maddison, S. T. 2010, A&A, 518, A16Furlan, E., Calvet, N., D’Alessio, P., et al. 2005, ApJ, 628, L65Furlan, E., Hartmann, L., Calvet, N., et al. 2006, ApJS, 165, 568Gaidos, E., Mann, A. W., Kraus, A. L., & Ireland, M. 2016, MNRAS, 457, 2877Grady, C. A., Muto, T., Hashimoto, J., et al. 2013, ApJ, 762, 48 Harries, T. 2014, TORUS: Radiation transport and hydrodynamics code, Astro-physics Source Code LibraryHashimoto, J., Tamura, M., Muto, T., et al. 2011, ApJ, 729, L17Henize, K. G. 1976, ApJS, 30, 491Huélamo, N., Figueira, P., Bonfils, X., et al. 2008, A&A, 489, L9Huélamo, N., Lacour, S., Tuthill, P., et al. 2011, A&A, 528, L7Hughes, A. M., Andrews, S. M., Wilner, D. J., et al. 2010, AJ, 140, 887Ireland, M. J. & Kraus, A. L. 2008, ApJ, 678, L59Isella, A., Pérez, L. M., Carpenter, J. M., et al. 2013, ApJ, 775, 30Jensen, E. L. N., Cohen, D. H., & Gagné, M. 2009, ApJ, 703, 252Juhász, A., Benisty, M., Pohl, A., et al. 2015, MNRAS, 451, 1147Kenyon, S. J., Brown, D. I., Tout, C. A., & Berlind, P. 1998, AJ, 115, 2491Kluska, J., Malbet, F., Berger, J.-P., et al. 2014, A&A, 564, A80Kraus, A. L. & Ireland, M. J. 2012, ApJ, 745, 5Kraus, A. L., Ireland, M. J., Martinache, F., & Hillenbrand, L. A. 2011, ApJ, 731,8Kraus, A. L., Ireland, M. J., Martinache, F., & Lloyd, J. P. 2008, ApJ, 679, 762Kraus, S., Ireland, M. J., Sitko, M. L., et al. 2013, ApJ, 768, 80Krautter, J., Wichmann, R., Schmitt, J. H. M. M., et al. 1997, A&AS, 123, 329Loinard, L., Torres, R. M., Mioduszewski, A. J., & Rodríguez, L. F. 2008, ApJ,675, L29Makarov, V. V. 2007, ApJ, 658, 480Merín, B., Brown, J. M., Oliveira, I., et al. 2010, ApJ, 718, 1200Muzerolle, J., Flaherty, K., Balog, Z., et al. 2009, ApJ, 704, L15Najita, J. R., Strom, S. E., & Muzerolle, J. 2007, MNRAS, 378, 369Olofsson, J., Benisty, M., Le Bouquin, J.-B., et al. 2013, A&A, 552, A4Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62Reipurth, B., Pedrosa, A., & Lago, M. T. V. T. 1996, A&AS, 120, 229Renard, S., Thiébaut, E., & Malbet, F. 2011, A&A, 533, A64Sallum, S., Follette, K. B., Eisner, J. A., et al. 2015, Nature, 527, 342Setiawan, J., Henning, T., Launhardt, R., et al. 2008, Nature, 451, 38Silverstone, M. D., Meyer, M. R., Mamajek, E. E., et al. 2006, ApJ, 639, 1138Thiébaut, E. 2008, in Society of Photo-Optical Instrumentation Engineers (SPIE)Conference Series, Vol. 7013, Society of Photo-Optical Instrumentation En-gineers (SPIE) Conference Series, 1Tsukagoshi, T., Nomura, H., Muto, T., et al. 2016, ArXiv e-printsVarnière, P., Blackman, E. G., Frank, A., & Quillen, A. C. 2006, ApJ, 640, 1110Vicente, S., Merín, B., Hartung, M., et al. 2011, A&A, 533, A135White, R. J., Gabor, J. M., & Hillenbrand, L. A. 2007, AJ, 133, 2524Zhu, Z. 2015, ApJ, 799, 16e, B. A. & Bate, M. R. 2012, MNRAS, 427, 2597Biller, B., Lacour, S., Juhász, A., et al. 2012, ApJ, 753, L38Birnstiel, T., Ormel, C. W., & Dullemond, C. P. 2011, A&A, 525, A11Bouvier, J., Alencar, S. H. P., Boutelier, T., et al. 2007, A&A, 463, 1017Bouvier, J. & Appenzeller, I. 1992, A&AS, 92, 481Brandeker, A., Jayawardhana, R., & Najita, J. 2003, AJ, 126, 2009Brauer, F., Dullemond, C. P., & Henning, T. 2008, A&A, 480, 859Brown, J. M., Blake, G. A., Dullemond, C. P., et al. 2007, ApJ, 664, L107Brown, J. M., Blake, G. A., Qi, C., et al. 2009, ApJ, 704, 496Calvet, N., D’Alessio, P., Hartmann, L., et al. 2002, ApJ, 568, 1008Calvet, N., D’Alessio, P., Watson, D. M., et al. 2005, ApJ, 630, L185Calvet, N., Muzerolle, J., Briceño, C., et al. 2004, AJ, 128, 1294Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245Cheetham, A., Huélamo, N., Lacour, S., de Gregorio-Monsalvo, I., & Tuthill, P.2015, MNRAS, 450, L1Chen, H., Myers, P. C., Ladd, E. F., & Wood, D. O. S. 1995, ApJ, 445, 377Cieza, L. A., Lacour, S., Schreiber, M. R., et al. 2013, ApJ, 762, L12Currie, T. & Sicilia-Aguilar, A. 2011, ApJ, 732, 24de Val-Borro, M., Artymowicz, P., D’Angelo, G., & Peplinski, A. 2007, A&A,471, 1043Espaillat, C., D’Alessio, P., Hernández, J., et al. 2010, ApJ, 717, 441Espaillat, C., Furlan, E., D’Alessio, P., et al. 2011, ApJ, 728, 49Espaillat, C., Muzerolle, J., Najita, J., et al. 2014, Protostars and Planets VI, 497Evans, T. M., Ireland, M. J., Kraus, A. L., et al. 2012, ApJ, 744, 120Fernandez, M., Ortiz, E., Eiroa, C., & Miranda, L. F. 1995, A&AS, 114, 439Fouchet, L., Gonzalez, J.-F., & Maddison, S. T. 2010, A&A, 518, A16Furlan, E., Calvet, N., D’Alessio, P., et al. 2005, ApJ, 628, L65Furlan, E., Hartmann, L., Calvet, N., et al. 2006, ApJS, 165, 568Gaidos, E., Mann, A. W., Kraus, A. L., & Ireland, M. 2016, MNRAS, 457, 2877Grady, C. A., Muto, T., Hashimoto, J., et al. 2013, ApJ, 762, 48 Harries, T. 2014, TORUS: Radiation transport and hydrodynamics code, Astro-physics Source Code LibraryHashimoto, J., Tamura, M., Muto, T., et al. 2011, ApJ, 729, L17Henize, K. G. 1976, ApJS, 30, 491Huélamo, N., Figueira, P., Bonfils, X., et al. 2008, A&A, 489, L9Huélamo, N., Lacour, S., Tuthill, P., et al. 2011, A&A, 528, L7Hughes, A. M., Andrews, S. M., Wilner, D. J., et al. 2010, AJ, 140, 887Ireland, M. J. & Kraus, A. L. 2008, ApJ, 678, L59Isella, A., Pérez, L. M., Carpenter, J. M., et al. 2013, ApJ, 775, 30Jensen, E. L. N., Cohen, D. H., & Gagné, M. 2009, ApJ, 703, 252Juhász, A., Benisty, M., Pohl, A., et al. 2015, MNRAS, 451, 1147Kenyon, S. J., Brown, D. I., Tout, C. A., & Berlind, P. 1998, AJ, 115, 2491Kluska, J., Malbet, F., Berger, J.-P., et al. 2014, A&A, 564, A80Kraus, A. L. & Ireland, M. J. 2012, ApJ, 745, 5Kraus, A. L., Ireland, M. J., Martinache, F., & Hillenbrand, L. A. 2011, ApJ, 731,8Kraus, A. L., Ireland, M. J., Martinache, F., & Lloyd, J. P. 2008, ApJ, 679, 762Kraus, S., Ireland, M. J., Sitko, M. L., et al. 2013, ApJ, 768, 80Krautter, J., Wichmann, R., Schmitt, J. H. M. M., et al. 1997, A&AS, 123, 329Loinard, L., Torres, R. M., Mioduszewski, A. J., & Rodríguez, L. F. 2008, ApJ,675, L29Makarov, V. V. 2007, ApJ, 658, 480Merín, B., Brown, J. M., Oliveira, I., et al. 2010, ApJ, 718, 1200Muzerolle, J., Flaherty, K., Balog, Z., et al. 2009, ApJ, 704, L15Najita, J. R., Strom, S. E., & Muzerolle, J. 2007, MNRAS, 378, 369Olofsson, J., Benisty, M., Le Bouquin, J.-B., et al. 2013, A&A, 552, A4Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62Reipurth, B., Pedrosa, A., & Lago, M. T. V. T. 1996, A&AS, 120, 229Renard, S., Thiébaut, E., & Malbet, F. 2011, A&A, 533, A64Sallum, S., Follette, K. B., Eisner, J. A., et al. 2015, Nature, 527, 342Setiawan, J., Henning, T., Launhardt, R., et al. 2008, Nature, 451, 38Silverstone, M. D., Meyer, M. R., Mamajek, E. E., et al. 2006, ApJ, 639, 1138Thiébaut, E. 2008, in Society of Photo-Optical Instrumentation Engineers (SPIE)Conference Series, Vol. 7013, Society of Photo-Optical Instrumentation En-gineers (SPIE) Conference Series, 1Tsukagoshi, T., Nomura, H., Muto, T., et al. 2016, ArXiv e-printsVarnière, P., Blackman, E. G., Frank, A., & Quillen, A. C. 2006, ApJ, 640, 1110Vicente, S., Merín, B., Hartung, M., et al. 2011, A&A, 533, A135White, R. J., Gabor, J. M., & Hillenbrand, L. A. 2007, AJ, 133, 2524Zhu, Z. 2015, ApJ, 799, 16