Mon.Not.R.Astron.Soc.000,1–14(2015) Printed16January2017 (MNLATEXstylefilev2.2) 2MASS 0213+3648 C: A wide T3 benchmark companion to an an active, old M dwarf binary N.R. Deacon⋆1,2, E.A. Magnier3, Michael C. Liu† 3, Joshua E. Schlieder 4,2, 7 1 Kimberly M. Aller3, William M.J. Best3, Brendan P. Bowler5‡, W.S. Burgett6. 0 2 K.C. Chambers3, P.W. Draper7, H. Flewelling 3, K.W. Hodapp8, N. Kaiser 3, n N. Metcalfe7,W.E. Sweeney 3, R.J. Wainscoat 3, C. Waters3 a 1Centre for Astrophysics Research, Universityof Hertfordshire, College Lane, Hatfield, AL10 9AB, UK J 2Max Planck Institute for Astronomy, Konigstuhl 17, Heidelberg, 69117, Germany 3 3Institute for Astronomy, University of Hawaii at Manoa, 2680 Woodlawn Drive, Honolulu, HI, 96822, USA 1 4NASA Exoplanet Science Institute, Caltech, Pasadena, California, USA 5Department of Astronomy, The Universityof Texasat Austin, 2515 Speedway, Austin, TX 78712, USA R] 6Giant Magellan Telescope Observatory, USA 7Department of Physics, University of Durham, South Road, Durham DH13LE, UK S 8Institute for Astronomy, University of Hawai’i, 640 North Aohoku Place, Hilo, HI96720, USA . h p - o r t ABSTRACT s a We present the discovery of a 360AU separation T3 companion to the tight (3.1AU) [ M4.5+M6.5 binary 2MASSJ02132062+3648506. This companion was identified us- ing Pan-STARRS1 data and, despite its relative proximity to the Sun (22.2+6.4pc; 2 −4.0 Pan-STARRS1 parallax) and brightness (J=15.3), appears to have been missed by v 4 previous studies due to its position near a diffraction spike in 2MASS. The close 0 M dwarf binary has active X-ray and Hα emission and shows evidence for UV flares. 1 The binary’s weak GALEX UV emission and strong NaI 8200˚ANa absorption leads 3 us to an age range of ∼1-10Gyr. Applying this age range to evolutionary models 0 implies the wide companion has a mass of 0.063±0.009M⊙. 2MASSJ02132+3648 C 1. provides a relatively old benchmark close to the L/T transition and acts as a key, 0 older comparisonto the muchyounger early-Tcompanions HN Peg B and GU Psc b. 7 Key words: 1 : v i X r 1 INTRODUCTION last theory predicts that wide binaries originate in tertiary a systemsandthatdynamicalinteractionscandrivethesesys- Wide multiple systems are common in thesolar neighbour- temsintobeingaclosebinarypairwithawidercompanion. hood, with >25% of solar-type stars having a companion Thus wide binary components should have a large, higher widerthat100AU(Raghavan et al.2010).Howeverthefor- order multiplicity fraction (as seen in the literature review mation mechanism of these systems, especially those wider of substellar objects by Burgasser et al. 2005 and observed than∼5,000AU,presentsachallengeformodelsofstarand for M dwarfs by Law et al. 2010) and that wide compan- brown dwarf formation. Current suggestions are that such ions to close binaries should be more common than wide companions formed in situ, were captured in their birth companions to single stars (seen by Allen et al. 2012). cluster(Kouwenhovenet al.2010),orresultfromtheevolu- tionofhigher-ordermultiplesystems(Delgado-Donate et al. Wide substellar companions in particular are also use- 2004; Umbreit et al. 2005; Reipurth & Mikkola 2012). This ful as test beds for evolutionary and atmospheric models. Such objects fall in to two categories; “mass benchmarks” (Liu et al. 2008) where two of the system components are ⋆ E-mail:[email protected] close enough to have an orbit that can be measured and † VisitingAstronomerattheInfraredTelescopeFacility,whichis thereby yield a dynamical mass, and “age benchmarks”, operated by the University of Hawaii under Cooperative Agree- where the age of the primary is determined (typically from ment no. NNX-08AE38A with the National Aeronautics and activityorrotation)andthenappliedtothesecondary.Rare SpaceAdministration,ScienceMissionDirectorate,PlanetaryAs- tronomyProgram. systemsthatarebothageandmassbenchmarkshavebotha ‡ McDonaldPrizeFellow highermassprimaryforagedeterminationandlow-massob- (cid:13)c 2015RAS 2 N.R. Deacon et al. jectwithdynamicalmassmeasurements(Dupuyet al.2009; various proper motion catalogues, but unlike this previ- Crepp et al. 2012; Dupuyet al. 2014). ous work we did not exclude nearby objects with low Thepast decadehasseen anexplosion inthediscovery proper motion (<0.1′′/yr). We followed the same selec- of wide ultracool1 companions to stars and brown dwarfs. tion process as Deacon et al. (2014) selecting only ob- Reviews of this population are presented in Faherty et al. jects with proper motions measured over a >400day base- (2010) and Deacon et al. (2014). Currently there are 22 line, requiring a 5σ overall significance and two individ- known T dwarf companions with separations greater than ual detections in yP1. This process included checking the 100AU,the21listedintheliteraturereviewofDeacon et al. 2MASS Reject Table as well as the main 2MASS database (2014) plus the recently discovered companion to the for near-infrared data. One candidate we identified was exoplanet host star HIP 70849 (Lodieu et al. 2014). Of an apparent companion to a mid-M dwarf at ∼15pc these, four (Gl 570 D, Burgasser et al. 2000; Wolf 1130 B, 2MASSJ02132062+3648506( aka PSOJ33.3327+36.8105; Mace et al. 2013; Ross 458 C, Goldman et al. 2010, and ξ hereafter 2MASSJ02132+3648). This object came into our UMaE,Wright et al.2013)lieinsystemswithhigher-order input primary list from the nearby M dwarf catalogue of stellarmultiplicity.RelativelyfewknownTdwarfscompan- L´epine& Gaidos (2011). 2MASSJ02132+3648 was discov- ionslieinyoungsystems.HNPegBisaT2.5(Luhman et al. ered as an M4.5 by Riaz et al. (2006) and identified as a 2007) companion to a ∼300Myr G0 dwarf. It is well fit- close (0.217”) binary by Janson et al. (2012), we refer to ted by the spectral standards with only a possibly weaker theunresolvedMdwarfbinaryas2MASSJ02132+3648AB. 1.25µm potassium doublet hinting at reduced gravity. By Fromthecontrastratio(∆i=2.16±0.15mag.)Janson et al. contrasttheyounger(∼100Myr),lowermassT3.5GUPscb (2012) estimated a spectral type of M6.5 for the secondary (Naud et al. 2014) diverges significantly from the spectral component with the primary remaining an M4.5. The pri- standardsintheY,J andK bands.Naud et al.(2014)sug- mary+secondary pair is saturated and unresolved in Pan- gest that this may be due to a differing amount of colli- STARRS1 imaging. Our candidate common proper mo- sion induced absorption from molecular hydrogen or possi- tion companion showed contaminated 2MASS photometry bly due to an unresolved, cooler companion. These are the (flagged as c for confusion) from a nearby diffraction spike only T2.5–3.5 companions in the literature. and did not appear in survey data taken by the Widefield Here we present a wide separation T3 companion to a Infrared Survey Explorer (WISE; Wright et al. 2010), nei- known,activeMdwarfbinarysysteminthesolarneighbour- ther in the main All-WISE table nor the All-WISE Reject hooddiscoveredduringoursearchforultracool companions Table(Cutri et al.2013).Thismaybeduetotheproximity using Pan-STARRS1. of two bright stars (including the primary) within half an arcminute. The flagged 2MASS detection and WISE non- detection are the likely reasons for this otherwise bright, nearby T dwarf remaining undiscovered until now. Fig- 2 OBSERVATIONS ure 1 shows the companion in a number of different Pan- STARRS1, 2MASS and WISE filters. 2.1 Identification in Pan-STARRS1 data ThePan-STARRS1telescope(Kaiser et al.2002)isawide- 2.1.1 Probability of chance alignment field 1.8m telescope on Haleakala on Maui in the Hawaiian Islandsandhasrecentlycompleteditsthreeandahalfyear Inorder toestimate if our proposed companion and its pri- science mission. Pan-STARRS1 data are reduced and cali- mary is due to a chance alignment between unrelated ob- bratedphotometricallyandastrometrically byaprocessde- jects, we used a modified version of the test presented in scribed by Magnier (2006, 2007); Magnier et al. (2013) and L´epine& Bongiorno (2007). In the original test the posi- Schlafly et al. (2012). The telescope undertook a series sky tions in the input primary catalogue are offset by several surveysincludingthe3π Survey,coveringthe30,000sq.deg. degrees before the pairing between the primaries and with northofδ =−30◦ withtwopairsofobservationsperyearin potentialsecondariesiscarriedout.Thismeanstheresulting eachoffivefilters.Thesefilters(gP1,rP1,iP1,zP1 andyP1; datashouldonlycontaincoincidentalpairings.Thepairings Tonry et al. 2012) extend further into the red optical than arethenplottedonapropermotiondifferencevs.separation SDSS, making Pan-STARRS1 appealing for studying late- plot with a maximumseparation of10 arcminutes. Herewe typeobjectsinthesolarneighbourhood(Deacon et al.2011; takeaslightly differentapproach as wealso havea trigono- Liu et al.2011,2013;Best et al.2013,2015).Thiscombina- metric parallax for the companion (Section 2.5). Hence we tion of properties have already resulted in the identifica- assign aquantitywhich wecall”astrometric offset”. Thisis tion of thelargest numberof ultracool companions to date, a quadrature sum of the proper motion differences in each with a near-doubling of the late-M and L dwarf wide sep- axisandtheparallaxdifferenceeachdividedbytheerroron aration (>100AU) companion populations (Deacon et al. that quantity.Hence 2014) as well as the discovery of two T dwarf companions (Deacon et al. 2012a,b) . n2 = (µα1−µα2)2 + (µδ1−µδ2)2 + (π1−π2)2 (1) σ (σ2 +σ2 ) (σ2 +σ2 ) (σ2 +σ2 ) We searched the 3π dataset processed by the first ver- µα1 µα2 µδ1 µδ2 π1 π2 sion of the Pan-STARRS1 pipeline to be run on the full We drew primaries with measured photometric dis- survey area (aka. PV1) for wide companions to nearby tances within 20pc from the list of 8889 (7464 within the stars. As in Deacon et al. (2014), we included objects from Pan-STARRS1 survey area) bright M dwarfs presented in L´epine& Gaidos(2011).Thesephotometricparallaxeswere used in the calculation of the astrometric offset. No lower 1 ObjectswithspectraltypesM7orlater proper motion limit was set on top of the 40mas/yr lower (cid:13)c 2015RAS,MNRAS000,1–14 A wide T dwarf companion to an active M dwarf binary 3 g r i z y P1 P1 P1 P1 P1 J H K W1 W2 s Figure 1.DiscoveryimagesofthewideTcompanion(centre) anditsbrightprimary(totheNorthandEastofthecompanion). Each imageis60′′ acrosswithNorthupandEastleft.ThesmallcutoutsinthePS1imagesare10′′ acrosscentredonthecompanion.Notethe objectisnotdetectedinW1andW2possiblyduetocontaminationfromnearbybrightstars.Thesolidcirclemarksthe2010.0position andthedashedcirclethe2MASSposition.ThePS1imageisastackofimagestakenoverthedurationofthesurvey(2010–2013) sothe objectwillnotfallexactlyatthe2010.0positionandmayappear smearedduetopropermotion. limitinourinputprimarylist.Wethenoffsetthepositionsof using the WFCAM reduction pipeline (Irwin et al. 2004; these object by two degrees in Right Ascension and carried Hodgkin et al. 2009) by the Cambridge Astronomical Sur- out a pairing process with late-type (zP1−yP1 >0.8 ) ob- veyUnit. The resulting photometry is in Table 1. jects in the Pan-STARRS1 database. Wemade a conserva- tivecutontheχ2 (theχ2 perdegreeoffreedom)statisticof ν 2.3 NASA Infrared Telescope Facility theastrometric points around thePan-STARRS1 database spectroscopy parallax fit limiting it the be <10 and limited ourselves to objects detected in eight or more Pan-STARRS1 images OurcandidatecompanionwasobservedonUTOctoberthe (this latter cut is identical to one used in Deacon et al. 5th 2015 with the recently upgraded SpeX spectrograph 2014). The astrometry used for our pair were the proper (Rayneret al.2003)ontheNASAInfraredTelescopeFacil- motion and photometric parallax for the primary from ity. Conditions were non-photometric with periods of high L´epine & Gaidos (2011) and our own proper motion and humidityandseeingof0.6–1.0′′.Theobservation wasmade parallax for the secondary (calculated in section 2.5). Our at an airmass of 1.06 with three pairs of 50s observations results are shown in Figure 2 with plots showing nσ both noddedinanABBApattern.Thelow-resolutionprismmode with and with our the parallax term. Clearly thepairing of was used along with the 0.8′′ slit giving a spectral resolu- 2MASSJ02132+3648 AB with our proposed companion is tion of R=75, with the slit aligned to the parallactic angle distinct from the coincident pairing locus. As there are no to minimise slit losses due to atmospheric dispersion. An starsinourcoincidentpairingtestwithsimilardistancesand A0V standard star (HD 22859) was observed contempora- astrometric offsetsto2MASSJ02132+3648we areunableto neouslyatasimilarairmasstothetarget.Thespectrumwas calculate a formal chance alignment probability. The inclu- reduced using version 4.0 of theSpeXtool software package sion of parallax data also significantly reduces the number (Cushing et al.2004;Vacca et al.2003).Thefinalspectrum of chance alignment pairings. for 2MASSJ02132+3648 C is shown in Figure 9 and the analysis of this spectrum is discussed in Section 3.2. Theunresolvedprimary2MASSJ02132+3648A/Bwas 2.2 UK Infrared Telescope photometry observedonUTSeptember25th2015withSpeX.Conditions As the photometry from 2MASS was strongly affected were mostly clear with seeing of 0.5-0.8′′. The 0.3′′slit was by a diffraction spike, we obtained UKIRT/WFCAM used along with the cross-dispersed SXD mode yielding a (Casali et al. 2007) observations of 2MASSJ02132+3648 C spectral resolution of R=2000. The observation was made onUT2015 Julythe31st.Theseobservationswerereduced at an airmass of 1.05 and consisted of four pairs of 14.8s (cid:13)c 2015RAS,MNRAS000,1–14 4 N.R. Deacon et al. standards were observed under slightly better conditions with single exposures of 400s and 800s, respectively. We also obtained a set of calibration observations prior to 2MASSJ02132+3648 AB and the standards that included bias frames, continuum lamp flats, and ThAr lamp frames for wavelength calibration. The data were reduced using theobservatorypipelinedescribedinAceitunoet al.(2013), whichbiascorrects,flatfieldsandwavelengthcalibratesthe data and applies a rough flux calibration. Thereducedspectrumof2MASSJ02132+3648 ABhas a signal-to-noise-ratio (SNR) of ∼3 per pixel, the RV stan- dardshaveSNRsof∼25.Tomeasureourtarget’sradialve- locity, we selected a region of thespectrum spanning6600– 6640˚A that was relatively free of telluric lines, had some strong stellar absorption lines, and was in the part of the spectrum with the highest SNR. This 40˚A region was flat- tened by dividing by the continuum polynomial fit and re- ducedfromaresolutionof∼65000to∼48000byconvolving with a Gaussian kernel. This last step was taken to match theresolution of the late-typeRV templates used for cross- correlation (CC) observed usingFEROSon theESO/MPG 2.2m telescope at La Silla. A barycentric velocity correc- tion was also applied to the CAFE spectra. Our initial ra- dial velocities for our RV standards were discrepant from the values in Nideveret al. (2002) by −58km/s. We were unable to find the source of this shift so can only correct for it post-hoc. This results in values of −71.8±1.0km/s compared to the literature value of −71.3 ±0.1km/s for GJ 908 and 57.0±1.0km/s compared to 56.6±0.1km/s for HD 26794. This process and correction yielded a ra- dial velocity of 1.5±1.4km/s for 2MASSJ02132+3648 AB although the low signal to noise meant that the cross cor- Figure 2. Comparing our companion with a sample coin- relation power has a poor value of 20%. The error in the cident pairings generated by searching for late-type (zP1 − radial velocity is the quadraturesum of the measured error yP1 > 0.8 companions around all M dwarfs in the catalogue of on a Gaussian fit to the cross correlation peak and a sys- L´epine&Gaidos(2011)withphotometricdistanceswithin20pc. tematic error of 1km/s introduced by the use of observed ThepositionsoftheseMdwarfswereoffsetbytwodegreessoonly coincident pairings (blue dots) should be included. Top: The y- RV templates. Thereis also theadditional factor of theRV axisshowsthequadraturesumofthedifferenceinpropermotion amplitude induced by the secondary, this is approximately betweeneachpairintermsofthenumberofstandarddeviations 2km/s. Bottom:They-axisshowsthequadraturesumofthedifferences inpropermotionandparallaxbetweeneachpairintermsofthe number of standard deviations. The inclusion of parallax data 2.5 Pan-STARRS1 astrometry clearlyreduces thecontamination frombackgroundstars. Pan-STARRS1 has observed each field in the 3π survey repeatedly over several years, allowing us to measure the parallax and proper motions of 2MASSJ02132+3648 C. observations nodded in an ABBA pattern. The reduction We were unable to calculate a parallax or proper motion techniquedescribed above for thecompanion was also used for 2MASSJ02132+3648 AB due to saturation. The Pan- but with the comparison star HD 19600. The spectrum is STARRS1 astrometric analysis of 2MASSJ02132+3648 C showninFigure3.Wediscusstheuseofthe8200˚ASodium consistsofaninitialseriesoftransformationstoremovedis- doublet in this spectrum in section 3.1.1 tortions caused by the Pan-STARRS1 optical system and camera. An iterative astrometric correction process is then used to determine the conversion from chip to celestial co- 2.4 Calar Alto 2.2m spectroscopy ordinates, minimising thescatter between objects positions 2MASSJ02132+3648 AB was observed under poor condi- oneachexposure.Thisprocessresultsinanastrometricsys- tionsonthe11thofNovember2014UTusingtheCalarAlto tematic floor of ≈ 10 - 20 milliarcseconds. Fiber-Fed Echelle spectrograph (CAFE, Aceituno et al. The in-database measurement of parallax 2013) on the 2.2m telescope at Calar Alto Observatory in (0.080′′±0.010) and proper motion does not yet include a Almeria, Spain. A single 750s exposure was obtained un- robust outlier rejection scheme. All 11 PS1 measurements der clear skies with high, variable humidity and poor see- spanning 3 years, were included in this analysis. We have ing (>2′′). Along with the science target, two radial veloc- found that a careful assessment of possible outliers, and ity standards from Nideveret al. (2002) were observed for the impact of the outlier rejection, is necessary to have internal calibration checks, GJ 908 and HD 26794. Both confidence in the parallax and proper motion fits. While (cid:13)c 2015RAS,MNRAS000,1–14 A wide T dwarf companion to an active M dwarf binary 5 Figure 3. The SpeX/IRTF spectrum of 2MASSJ02132+3648 AB. Two comparison spectra of M4 and M5 standards from Kirkpatricketal.(2010)areshowninthemainplot.Thespectraarenormalisedat0.9µm.Theinsetshowsthe8200˚Afeaturecompared tothe20Myrto∼125MyroldM4.5TYC5241-986-1CDeaconetal.(2013). we are working to incorporate these checks in the auto- ategaussianswithfreeaxialtiltswereabetterfittoourpri- matic analysis, we have used an external calculation for mary.ThePSFwasfittedtothecentral10pixelsofthepri- 2MASSJ02132+3648 C for the purposeof thisarticle. mary toexcludeanyfluxfrom thesecondary.Wethensub- Figures 4 and 5 show the astrometry of 2MASS tractedtheapproximatedPSFfromthefullimagerevealing J0213+3648 C. We have used bootstrap resampling to de- thetertiaryinbettercontrast.SeeFigure6fortheresultsof termine the errors on the fitted parameters, and to assess thissubtraction.Finallyweundertookaperturephotometry whether the inclusion of any specific measurement in the onthetertiarycomponentusingtwoapertureseithersideof calculation of the fit drives the parameters to significantly theobjectbutatsimilardistancesfromtheprimarytomea- different values. For 1000 bootstrap resample tests, we also surethebackgroundflux.Wethenusedthesametechnique measured the distance of each point from the fitted path, onotherstarsinthefieldandcomparedthemeasuredfluxes scaled by theformal error on thepoint. Themedian of this to the WISE catalogue magnitudes in order to determine distributionisameasureofhowdeviantthepointisfromthe theimagezeropoints.Wefoundthat2MASSJ02132+3648C pathgiventhesetofmeasurements.Wefindthatonepoint, had W1=14.204±0.123mag. and W2=13.430±0.128mag. markedwithan‘X’onFigure4,issignificantlydeviantfrom thecollectionoffittedpaths.Themeanparallaxchangesby a substantial amount, though formally the two values are 3 RESULTS within error. We remove that single discrepant point from thedatasetandusetheresultingparallaxandpropermotion The derived properties of the components of the valuesfor2MASSJ02132+3648C.OurfinalPan-STARRS1 2MASSJ02132+3648 system are shown in Table 1. Below parallax measurement for 2MASSJ02132+3648 C is we discuss how these parameters were calculated. 0.045±0.010′′corresponding to a distance of 22.2+6.4pc. −4.0 3.1 Properties of 2MASSJ02132+3648 AB 2MASSJ02132+3648 was identified as an M dwarf in the 2.6 WISE photometry solar neighbourhood with significant X-ray emission by 2MASSJ02132+3648CdoesnotappearintheAllWISEcat- Riaz et al. (2006). Using low-resolution spectroscopy they alogue or reject table, likely due to it’s bright primary. To classified it as an M4.5 dwarf (a spectral type confirmed removetheprimaryandhenceestimatethebrightnessofthe by Alonso-Floriano et al. 2015) with a spectrophotometric CcomponentwemodelledthePSFoftheprimaryondown- distance of 11pc. Riaz et al. (2006) also reported that the loadedWISEimages.Lang et al.(2014)modelledtheWISE object showed Hα emission of 8.1˚A. Janson et al. (2012) PSFasthreeisotropic gaussians. Wefoundthattwobivari- usedLuckyImagingtoidentify2MASSJ02132+3648 Aasa (cid:13)c 2015RAS,MNRAS000,1–14 6 N.R. Deacon et al. Table1.Thepropertiesofthecomponentsofthe2MASSJ02132062+3648506 system.NotethatwedonotquotePan-STARRS1 magnitudes for2MASSJ02132+3648 ABasitisheavilysaturated. 2MASSJ02132+3648AB 2MASSJ02132+3648A 2MASSJ02132+3648B 2MASSJ02132+3648C Position 021320.63+364850.7a 021319.82+364837.5a µα (′′/yr) 0.024±0.008e 0.024±0.010 µδ (′′/yr) 0.047±0.008e 0.065±0.011 π (′′) 0.068±0.020e∗ 0.045±0.010f 0.063+0.014f∗ −0.010 Separation 0.217′′b 16.4′′a 4.8AUf 360AUf Spectral Type M4.5c T3 M4.5d M6.5d V (mag) 13.86e ... ∆i(mag) ... 2.16±0.15g ∆z (mag) ... 2.42±0.18g zP1 (ABmag) ... 19.243±0.016f yP1 (ABmag) ... 17.567±0.010f J2MASS 9.367±0.022a 15.297±0.53a† H2MASS 8.825±0.021a 14.765±0.62a‡ Ks,2MASS 8.518±0.018a 14.770±0.115a‡ YMKO ... 16.279±0.024f JMKO ... 15.158±0.013f HMKO ... 14.887±0.021f KMKO ... 14.930±0.020f W1 8.333±0.023j 14.204±0.123f,j W2 8.127±0.020j 13.430±0.128f,j W3 7.982±0.021j ... W4 8.004±0.229j ... FNUV/FJ 7.1×10−5a,f,g ... FFUV/FJ <2.2×10−5f,g,# ... FX/FJ 5.7×10−3a,f,h ... mass(M⊙) 0.26±0.06i 0.09±0.03i 0.068±0.007f age 1–1.0Gyrf a 2MASSposition,epoch1998.811Skrutskieetal.(2006) b Epoch2012.90Jansonetal.(2014) c spectrumofcombinedobjectRiazetal.(2006) d basedonfluxratioJansonetal.(2012) e L´epine&Gaidos(2011) f thiswork g Jansonetal.(2012) h Vogesetal.(2000) i Jansonetal.(2014) j Wrightetal.(2010) † 2MASSconfusionflagset ‡ markedasdiffractionspikein2MASS ∗ photometricparallax # 3σ upperlimit. close binary system with a separation of 0.181±0.002” and 3.1.1 The age of 2MASSJ02132+3648 AB an estimated orbital period of 8 years. We note here that this orbital period is too long for either component of the TheprimarypairhaveanX-raycounterpartintheROSAT binarytohaveitsactivitysignificantlyaffectedbytidalspin- Bright Source Catalogue (Voges et al. 1999) 13′′ away, up. Subsequent orbital motion combined with the compo- slightlybeyondthe1σ positionalerrorof11′′.Weestimated nent’sfluxratioledJanson et al.(2014)toestimatespectral theX-rayfluxfor2MASSJ02132+3648 ABbyapplyingthe types of M4.5 and M6.5 and masses of 0.26±0.06M⊙ and relations of Schmitt et al. (1995) to theROSAT data yield- 0.09±0.03M⊙ for theA and B components respectively. ingafluxof5.2±1.4×10−13ergs/cm2.Wethenused2MASS J-bandmagnitude of 2MASSJ02132+3648 ABto calculate an FX/FJ flux ratio of 5.7×10−3. Comparing this value to Figure3ofShkolnik et al.(2009),wefirstnotethatourtar- get is bluer in I −J than one would expect for an M4.5 (usingan I magnitude from SuperCOSMOS;Hambly et al. (cid:13)c 2015RAS,MNRAS000,1–14 A wide T dwarf companion to an active M dwarf binary 7 time (years since 2010/01/01) ∆RA (arcsec) 0.0 1.0 2.0 3.0 4.0 -0.2 0.0 0.2 ∆DEC (arcsec)-000...202 2MASS J0213+3648C -000..02.2DEC (arcsec)∆ mples 200.0 µπcµl δαip p===e d+++ f642it674 +++///--- 111100 4.0 sa 100.0 N 0.2 tim e 3.0 (ye 1500..00 ∆DEC (arcsec) 0.0 2.0 ars since 2010/0 amples 100.0 µπnµ oδα c===li p +p++i662n518g +++///--- 111282 1.0 1/01 N s 50.0 -0.2 ) 0.0 0.0 0.0 50.0 100.0 150.0 -0.2 0.0 0.2 -0.2 0.0 0.2 ∆RA (arcsec) ∆RA (arcsec) parallax (milliarcsec) Figure 4. Lower left: The relative positions of the Pan- Figure 5. Bootstrap resampling tests of the astrometric errors. STARRS1 zP1 (blue) and yP1 (red) measurements for Top: Histogram of the fitted parallax values for 3000 bootstrap 2MASSJ02132+3648C.Thebestfittedpathontheskyisdrawn resampling tests, with all points included. The blue dashed line inblack.ThesinglepointwithablackXissuspectandexcluded marksthe50%pointinthecumulativedistributionfunction.The fromthefinalfit.Upper left:ResidualDeclinationpositionsas reddashedlinesrepresentthe15.9%and84.1%points,equivalent a function of time (MJD) after the proper-motion fit has been to ±1σ. The fitted parallax and proper motion values are listed subtracted; symbolsasabove. Lower right:SameasUpper-left with the errors derived from this analysis. Bottom: Histogram forRightAscension.Upperright:ResidualDeclinationvsRight ofthefitted parallaxvalues for3000bootstrap resamplingtests, Ascension after the proper-motionfit has been subtracted; sym- withthesingleoutlierpointexcluded. Labelsasabove. bolsasabove. 2001). However this does not affect the conclusion that the object lies on the active M dwarf sequence. This sequence is defined as having similar or greater X-ray emission than the Pleiades (125Myr; Staufferet al. 1998) and the β Pic- toris young moving group (21Myr; Binks & Jeffries 2013). We note that 2MASSJ02132+3648 AB also emits a higher X-ray flux than most Hyades (625Myr; Perryman et al. 1998) members of similar spectral type. Riaz et al. (2006) find log10LX/Lbol=3.16, placing it in their saturated X-ray emission locus. Comparing this value with Fig- ure 5 of Preibisch & Feigelson (2005) we find that 2MASSJ02132+3648 AB fractional X-ray flux is similar if not higher than Hyades M dwarf members and much higher than the vast majority of field M dwarfs. Us- ing the method of Malo et al. (2014) we calculated that 2MASSJ02132+3648 A/B has logLx = 28.5 ±0.6ergs/s. Figure 6. The results of our WISE PSF subtraction with Thisis2σ lessactivethantheβ Pictoris movinggroup and 2MASSJ02132+3648 C circled. These images are 55 arcseconds 1σ less active than theABDorassociation. Wedohowever across. findthatit is1.2σ more activethanthetypicalfieldstar of similar type. As our object is a binary it is possible that the X-ray butnotinthefar-UVband.Wenotethatitdoesappearas fluxiscomingentirelyfromtheM6.5secondarycomponent. a far-UV emitter in the EUVE catalogue (Christian et al. This would not alter our conclusion that the object is X- 1999) due to a stellar flare. The GALEX near-UV flux of ray active as a recalculation of FX/FJ would result in the 20.1±2.4µJyresultsinanFNUV/FJ ratioof7.1×10−5oran objectlyingevenfurtherabovetheinactiveMdwarfregime FNUV/FKs ratioof7.7×10−5.Wenotethatthelattervalue onFigure3ofShkolnik et al.(2009).Ifwewererelyingsolely is within 0.2dex of the young (20–125Myr) M4.5 binary on X-ray emission for an age diagnostic, we would estimate TYC5241-986-1 BC (Deacon et al. 2013).Weset an upper thatthisobjectisyoungerthantheHyades(i.e.<625Myr). limitontheFUVfluxof2MASSJ02132+3648ABbyfinding 2MASSJ02132+3648 AB are also detected by the thefluxerroron theforced far-UV photometryat theposi- GALEX satellite (Martin et al. 2005) in the near-UV band tion of the GALEX NUV detection. From this we calculate (cid:13)c 2015RAS,MNRAS000,1–14 8 N.R. Deacon et al. a 3σ upper limit of FFUV/FJ < 2.2×10−5. To further ex- amine the near-UV properties of 2MASSJ02132+3648 AB, 11..44 2MJ0213+36AB we must compare with Figure 3 of Shkolnik et al. (2011). Field Here we findthat our object appears to lie between theac- M Dwarfs tive and inactive loci and that while it does not have a far- 11..33 UV, detection it could have emission above Shkolnik et al. (2009)’s FNUV/FJ > 10−5 and still remain undetected by exex GALEX. However the near-UV emission is of little use as dd nn 11..22 Ansdell et al. (2014) noted that the fully convective transi- I I β Pic I I (~20 Myr) tion in mid-M dwarfs makes it impossible to define a near- aa NN UVinactivelocusofMdwarfsbeyondM3.Hencewedonot 11..11 consider2MASSJ02132+3648AB’snear-UVemissiontobe ε Cha a reliable age diagnostic. (~5 Myr) Riaz et al. (2006) found that 2MASSJ02132+3648 AB has Hα in emission. West et al. (2008) provide a list of ac- 11..00 M Giants tivitylifetimesasafunctionofspectraltype.Astheactivity 00..55 11..00 11..55 22..00 22..55 lifetime changes with spectral type we need to know which ((RR−−II)) [[mmaagg]] componentthisemissioncomesfrom.Whilewewerenotable toresolvetheHαemission intotwoseparatecomponentsin Figure 7. The NaI index from Lyoetal. (2004) along with the our CAFE spectrum, we were able to measure its equiva- lentwidthas−6.6˚A, broadlyinagreementwithRiaz et al. measured sequences for different populations fromLawsonetal. (2006)’s value of −8.1˚A and the value of −6.2+0.4˚A from (2009). 2MASSJ02132+3648 lies just above the old field se- −0.2 quence. Alonso-Floriano et al.(2015).Thisemissioncouldintheory come from either component of the inner binary. Using the mean χ correction factors as a function of spectral type of (2012) found no disk excesses in the M dwarf population Walkowicz et al. (2004) we estimated the logLHα/Lbol to of Tucana Horologium (45±4Myr, Bell et al. 2015) and be-3.4 (from our EW of -6.6˚A) if theemission was coming AB Dor (149+51 Bell et al. 2015). Hence the lack of disk −19 fromtheAcomponent.Thisisinthemostactivequartilefor excessindicatesthat2MASSJ02132+3648 ABisolderthan starsofasimilar spectral typein Figure6ofSchmidt et al. around 20Myr. (2015) but is not outside the bounds of the activity range The Na 8200˚A doublet provides an effective grav- intheirsample.Forthesecondarycomponentweusedthei ity diagnostic for mid-late M dwarfs (Lyoet al. 2004; band fluxratio of Janson et al. (2012) and thetypicalr−i Schliederet al. 2012). We measured the equivalent width colourdifferencebetweenanM4.5andM6.5intheSEDtem- of this feature in our IRTF spectrum and found it to platesofKraus & Hillenbrand(2007)toestimateanr-band be 5.8±1.8˚A, on the upper end of the field dwarf se- contrast ratio of 2.59. We then used this value to adjust quence in Schliederet al. (2012)’s Figure 3. This points the EW of the Hα feature to take into account the lower towards an older lower age boundary of ∼1Gyr. A continuum in the r-band for an M6.5. This resulted in an visual comparison with the young (20–125Myr) M4.5 approximateequivalentwidthof72˚A iftheemissioniscom- TYC 5241-986-1 B/C (Deacon et al. 2013) also indicates ing entirely from the secondary. Again using the method that 2MASSJ02132+3648 AB is not a young object (see of Walkowicz et al. (2004) we calculated logLHα/Lbol to Figure3inset).WealsousedtheNaIspectralindexderived be -3.0 if all the Hα emission was coming from the sec- byLyo et al.(2004)andmeasuredavalueof1.25±0.15.Fig- ondarycomponent. Thisis moreactivethan anyM6or M7 ure7showsthatourobjecthasafeaturestrength(andthus in Schmidt et al. (2015)’s Figure 6 (with logLHα/Lbol de- agravity)consistentwithfielddwarfs.WhiletheM6.5com- clining with increasing spectral type). Hence we conclude ponent will have a deeper 8200˚A doublet owing to its late that it is unlikely that all the Hα emission is from the sec- spectraltype,thecontrastratioof∆i=2.16±0.15magindi- ondary component and that the M4.5 A component must catesthatitwillcontributelessthan10%ofthefluxinthis beactive. However Morgan et al. (2012) show that binaries region andthuswill not significantly changethelinedepth. in the 1–100AU separation range have significantly higher This NaI spectral index value has a relatively high error activity fractions. Hence we cannot use this activity to set dueto low S/N in thesurrounding continuum region. Both an upperage bound. the visual comparison of the 8200˚A feature and the lack of Young M dwarf stars in star forming regions typi- other youth indicators suggest that this is not a young ob- cally have infrared excesses caused by circumstellar disks. ject. Additionally wefound that thestrength of TiO (TiO5 We used the WISE photometry (Wright et al. 2010) for 0.436, M3.7; TiO6 0.66, M4.0), CaH (CaH3 0.61, M5.6), 2MASSJ02132+3648 AB to examine if it has mid-infrared VO (VO1 0.90, M5.4; VO2 0.72, M4.9) and H2O (H2OD excess indicative of a disk. Comparing with the W3 −W4 1.14) molecular indices in the optical (L´epineet al. 2013) vs. W1−W2 plot shown in Figure 1 of Simon et al. (2012) and near-infrared (McLean et al. 2003; Allers & Liu 2013) we find that our target falls close to the stellar locus sug- absorptionfeaturesof2MASSJ02132+3648A/Bareconsis- gesting thatourprimary doesnot host adisk.Asdiscussed tentwith aspectraltypeofM5.0±0.5. Thisisin agreement in Deacon et al. (2013), disks around M dwarfs are uncom- with theliterature spectral typeof M4.5 (Riazet al. 2006). monfor agesbeyond20Myr.Whilestudiesofyounger(10– One final youth indicator available to us is kinematics. 15Myr) clusters show significant amounts of disk emission We ran 2MASSJ02132+3648 AB through the BANYAN II aroundmanyMdwarfs(e.g.Currie et al.2008),Simon et al. online tool (Malo et al. 2013; Gagn´e et al. 2014) and found (cid:13)c 2015RAS,MNRAS000,1–14 A wide T dwarf companion to an active M dwarf binary 9 that its kinematics do not match any of the known young moving groups. We also plotted our object on a series of diagnostic plots comparing its Galactic XYZ positions and UVW space velocities (Figure 8). Again we find that 2MASSJ02132+3648 AB does not match any known young moving groups and lies outside the loosely defined young (.125Myr) Local Association (Zuckerman & Song 2004). We also found that 2MASSJ02132+3648 AB did not match the kinematics of other groups not shown in our plot; Octans (Murphy& Lawson 2014), Hercules-Lyra (Eisenbeiss et al. 2013), 32 Ori (Mamajek 2006), Carina Near(Zuckermanet al.2006)andUrsaMajoris(King et al. 2003). In summary we assign an upper age bound of 10Gyr for 2MASSJ02132+3648 AB based on disc kinematics and a lower age boundary of 1Gyrbased on the strength of the Na 8200˚A feature. We note that 2MASSJ02132+3648 AB has X-ray emission indicative of it being younger than the Hyades.Howeverwecautionthattherelativelylatespectral type of this object and its multiplicity make it difficult to Figure 9. Our spectrum of 2MASSJ02132062+3648506 C, we classify this is object as a T3±0.5. Shown for com- apply some age diagnostics such as UV emission with suffi- parison are the spectra for the young (∼100Myr) T3.5 cient certainty. GU Psc b (Naudetal. 2014), the young (∼300Myr) T2.5 HN Peg B (Luhmanetal. 2007), the alternative T3 dwarf 3.2 Properties of 2MASSJ02132+3648 C standard SDSSJ120602.51+281328.7 (Chiuetal. 2006) (as suggested by Liuetal. 2010), the previous T3 standard The observed near-infrared spectrum of 2MASS J120956.13−100400.8 (Burgasseretal. 2004, 2006; now 2MASSJ02132+3648 C is shown in Figure 9. We spec- knowntobeabinary)andtheblueT dwarf(Burninghametal. trally classified 2MASSJ02132+3648 C using the flux 2010; Maroccoetal. 2015). All spectra are smoothed to similar resolutionsandnormalisedtotheJ-bandpeak. indices of Burgasser et al. (2006) and the polynomial relations of Burgasser (2007). The individual indices and the derived spectral types are shown in Table 2. We also To compare with our trigonometric parallax measure- compared visually with the standards of Burgasser et al. ment from Pan-STARRS1, we calculated a photometric (2006), finding a best fit of T3 (note the T3 standard of distance for 2MASSJ02132+3648 C using only our WF- Burgasser et al.2006wasfoundtobeabinarybyLiu et al. CAM photometry. This was done in a similar manner to 2010, see later discussion on the comparison with the Deacon et al. (2014) using the relations of Dupuy& Liu newer standard). As this compares well with the spectral (2012), by calculating distances in each band along with types derived from the indices we adopt this as our final the error on these distances caused by photometric mea- spectral type. Figure 9 also shows a comparison with the surementerrorandtheerrorinspectraltype.Wethenused young (∼ 100Myr) T3 GU Psc b (Naud et al. 2014), the the quadrature sum of these error terms to estimate a dis- young (∼300Myr) T2.5 HN Peg B (Luhmanet al. 2007), tance based on the J, H & K bands. Our final distance is the original T3 standard (2MASS J120956.13−100400.8, 15.08+3.0pc which includes both the calculated error due Burgasser et al. 2006) and the alternative T3 standard −2.5 to photometric and spectroscopic measurement uncertainty (SDSS J120602.51+281328.7) suggested by Liu et al. and the scatter around the absolute magnitude relations. (2010) after 2MASS J120956.13−100400.8 was found It compares well with the photometric distance of the AB to be a binary. We note that 2MASSJ02132+3648 C component14.7+6.2pc(L´epine& Gaidos2011)andthepar- has a narrower J band peak and deeper J-band water −3.3 allactic distance of 22.2+6.3pc. absorption than SDSS J120602.51+281328.7 but matches −4.0 both HN Peg B and 2MASS J120956.13−100400.8 well in this regime, the H and K-bands the spectrum are a 3.2.1 Comparison with evolutionary models better fit to SDSS J120602.51+281328.7. GU Psc b is unique amongst our comparison objects in that it shows We calculated the bolometric correction for a distinct slope in the Y and J-bands compared to the 2MASSJ02132+3648 C using the polynomial relations other objects. We do not see sufficient evidence to suggest of Liu et al. (2010). This resulted in an MKO J-band that 2MASSJ02132+3648 C is spectrally peculiar and correction of BCJ=2.24±0.14. We combined this with our note that it does not have the enhanced K-band emission WFCAM observations to yield an apparent bolometric that Naud et al. (2014) attributed to reduced collision magnitude of 17.40±0.50mag (including the error in our induced absorption resulting from the lower surface gravity parallax measurements). To determine the physical prop- (i.e. younger age) of GU Psc b. 2MASSJ02132+3648 C erties of 2MASSJ02132+3648 C, we ran a Monte Carlo is the only known T3 benchmark companion known comparisontotheBaraffe et al.(2003)evolutionarymodels. (see Figure 10). Our measured WISE photometry yields An age for the system was drawn from a flat distribution W1 − W2=0.77, in line with the expect colour for an with between our upper and lower age boundaries (10Gyr early-mid T dwarf (Kirkpatrick et al. 2011). and 1Gyr) and the apparent bolometric magnitude and (cid:13)c 2015RAS,MNRAS000,1–14 10 N.R. Deacon et al. 0 5 5 −5 0 −10 0 −5 m/s) −15 m/s) −5 m/s) −10 V (k −20 W (k −10 W (k −15 −25 −15 −20 −30 −20 −25 −35 −30−25−20−15−10 −5 0 5 −30 −25 −20 −15 −10 −5 0 −40 −30 −20 −10 0 U (km/s) U (km/s) V (km/s) BetaPic TucHor Columba Carina TWHya EtaCha Oct Arg ABDor 50 50 0 0 0 c) −50 c) c) p p p Y ( Z ( −50 Z ( −50 −100 −100 −150 −100 −100 −50 0 50 100 150 −100 −50 0 50 100 150 −150 −100 −50 0 50 X (pc) X (pc) Y (pc) Figure 8. A plot showing the positions of known young moving groups and our target system 2MASSJ02132+3648 (red circle). The dottedboxshowsthebroadlyyoungLocalAssociation(Zuckerman&Song2004). Table 2.Spectralclassificationof2MASSJ02132+3648 C.Shownarethespectralindices(Burgasseretal.2006)andtheimplied spectralclassifications frompolynomialrelations(Burgasser2007)foreachindex. H2O-J(SpT) CH4-J(SpT) H2O-H(SpT) CH4-H(SpT) CH4-K(SpT) Visual Final Type Type 0.395±0.013(T3.7) 0.450±0.019(T4.3) 0.435±0.026(T3.7) 0.709±0.027(T2.5) 0.432±0.034(T3.1) T3 T3 parallax were both offset by a random gaussian offset with parallax for the unresolved inner pair. These are for the a standard deviation equal to the measurement errors. trigonometric parallax MJ = 13.424 ± 0.483mag, MH = This resulted in an age and absolute bolometric magnitude 13.153±0.483mag & MK =13.196±0.483mag and MJ = for each realisation. The mass, effective temperature and 14.321 ±0.639mag, MH = 14.050 ±0.639mag & MK = gravity corresponding to that absolute bolometric magni- 14.093±0.639mag with the error estimates dominated by tude and age were then determined from the Baraffe et al. the uncertainties on the distance measurements. Compar- (2003) model grid. The results are shown in Figure 11 ing with the absolute magnitude to spectral type plots of with values of Teff = 1641 ± 167K, m = 68 ± 7MJ Figure 27 of Dupuy& Liu (2012) we find that theabsolute and logg = 5.45±0.08dex (cgs). Note this relatively high magnitudes calculated with thephotometric parallax lie on temperature(foraT3)islikelyduetoourdistancefromthe themain locus of early T dwarfs. The absolute magnitudes trigonometric parallax being larger than our photometric based on the trigonometric parallax are approximately one distance. magnitude over-luminous or approximately 2σ. This sug- gests that either the trigonometric parallax is a chance un- derestimateor perhapsthatthesecondary isover-luminous 3.2.2 The absolute magnitude of 2MASSJ02132+3648 C duetobinarity.Bothoftheseexplanationswouldalsofitour anomalously hoteffectivetemperatureestimatefrom evolu- We calculated absolute magnitudes in the MKO system tionary models. based on both our trigonometric parallax for the tertiary component and the L´epine & Gaidos (2011) photometric (cid:13)c 2015RAS,MNRAS000,1–14