Accepted for publication in the Astrophysical Journal on December 28, 2015 PreprinttypesetusingLATEXstyleemulateapjv.04/17/13 TWO SMALL TEMPERATE PLANETS TRANSITING NEARBY M DWARFS IN K2 CAMPAIGNS 0 AND 1*,†,‡ Joshua E. Schlieder1,2, Ian J. M. Crossfield3,17, Erik A. Petigura4,18, Andrew W. Howard5, Kimberly M. Aller5,2, Evan Sinukoff5, Howard T. Isaacson6, Benjamin J. Fulton5, David R. Ciardi7, Mickae¨l Bonnefoy8, Carl Ziegler9, Timothy D. Morton10, Se´bastien Le´pine11, Christian Obermeier12, Michael C. Liu5, Vanessa P. Bailey13, Christoph Baranec14, Charles A. Beichman7, Denis Defre`re15, Thomas Henning12, Philip Hinz15, Nicholas Law9, Reed Riddle4, Andrew Skemer15,16,18 Accepted for publication in the Astrophysical Journal on December 28, 2015 ABSTRACT 6 TheprimeKepler missionrevealedthatsmallplanets(<4R )arecommon, especiallyaroundlow-massMdwarfs. ⊕ 1 K2, the re-purposed Kepler mission, continues this exploration of small planets around small stars. Here we combine 0 K2 photometry with spectroscopy, adaptive optics imaging, and archival survey images to analyze two small planets 2 orbiting the nearby, field age, M dwarfs K2-26 (EPIC 202083828) and K2-9. K2-26 is an M1.0±0.5 dwarf at 93±7 n pc from K2 Campaign 0. We validate its 14.5665 d period planet and estimate a radius of 2.67+0.46 R . K2-9 is an −0.42 ⊕ a M2.5±0.5 dwarf at 110±12 pc from K2 Campaign 1. K2-9b was first identified by Montet et al. (2015); here we J present spectra and adaptive optics imaging of the host star and independently validate and characterize the planet. 2 Our analyses indicate K2-9b is a 2.25+0.53 R planet with a 18.4498 d period. K2-26b exhibits a transit duration 1 that is too long to be consistent with−a0c.9i6rcul⊕ar orbit given the measured stellar radius. Thus, the long transits are likely due to the photoeccentric effect and our transit fits hint at an eccentric orbit. Both planets receive low incident ] flux from their host stars and have estimated equilibrium temperatures <500 K. K2-9b may receive approximately P Earth-like insolation. However, its host star exhibits strong GALEX UV emission which could affect any atmosphere E it harbors. K2-26b and K2-9b are representatives of a poorly studied class of small planets with cool temperatures . h that have radii intermediate to Earth and Neptune. Future study of these systems can provide key insight into trends p in bulk composition and atmospheric properties at the transition from silicate dominated to volatile rich bodies. - Subject headings: eclipses - stars: individual (K2-26, K2-9) — techniques: photometric — techniques: o spectroscopic r t s a [ *Based in part on data obtained at the LBT. The LBT is 1 an international collaboration amonginstitutions in the United v States, Italy and Germany. LBT Corporation partners are: 6 The University of Arizona on behalf of the Arizona university system; Istituto Nazionale di Astrofisica, Italy; LBT Beteili- 0 gungsgesellschaft, Germany, representing the Max-Planck Soci- 7 ety,theAstrophysicalInstitutePotsdam,andHeidelbergUniver- 2 sity;TheOhioStateUniversity,andTheResearchCorporation, 0 on behalf of The University of Notre Dame, University of Min- . nesotaandUniversityofVirginia. 1 †Some of the data presented herein were obtained at the W. 0 M. Keck Observatory, which is operated as a scientific partner- 6 shipamongtheCaliforniaInstituteofTechnology,theUniversity 1 of California, and the National Aeronautics and Space Admin- istration. The Observatory was made possible by the generous : v financialsupportoftheW.M.KeckFoundation. i ‡Based on observations collected at the European Organiza- X tionforAstronomicalResearchintheSouthernHemisphere,La SillaObservatory,ChileduringprogramID194.C-0443. r 1NASAPostdoctoralProgramFellow,NASAAmesResearch a Center, Space Science and Astrobiology Division, MS 245-6, MoffettField,CA94035,USA;[email protected] 2VisitingAstronomer,NASAInfraredTelescopeFacility 3Lunar&PlanetaryLaboratory,UniversityofArizona,1629 tonNJ,08544,USA E.UniversityBlvd.,Tucson,AZ,USA 11Department of Physics & Astronomy, Georgia State Uni- 4California Institute of Technology, Pasadena, CA 91125, versity,Atlanta,GA,USA USA 12Max-Planck-Institutfu¨rAstronomie,K¨onigstuhl17,69117, 5Institute for Astronomy, University of Hawai‘i, 2680 Wood- Heidelberg,Germany lawnDrive,Honolulu,HI,USA 13Kavli Institute for Particle Astrophysics and Cosmology, 6Astronomy Department, University of California, Berkeley, StanfordUniversity,Stanford,CA94305,USA CA,USA 14Institute for Astronomy, University of Hawai‘i at M¯anoa, 7NASA Exoplanet Science Institute, California Institute of Hilo,HI96720-2700,USA Technology,770S.WilsonAve.,Pasadena,CA91125,USA 15Steward Observatory, Department of Astronomy, Univer- 8Universit´e Grenoble Alpes, IPAG, 38000, Grenoble, 38000, sityofArizona,933N.CherryAve,Tucson,AZ85721,USA Grenoble;CNRS,IPAG,38000Grenoble,France 16Department of Astronomy and Astrophysics, University of 9UniversityofNorthCarolinaatChapelHill,ChapelHill,NC California,SantaCruz,CA95064,USA 27599,USA 17SaganFellow 10Department of Astrophysics, Princeton University, Prince- 18HubbleFellow 2 J. E. Schlieder et al. 1. INTRODUCTION ground based spectroscopy and imaging. We detail our analyses of these observed data and the results in §3. §4 Planets are commonplace in the Galaxy. In the provides a discussion of the properties of these planets last 20 years, knowledge of planet demographics, ar- in the context of known demographics and §5 provides chitectures, and frequencies has expanded beyond the concluding remarks. eight primary bodies in our solar system to thousands of planets orbiting thousands of stars. A workhorse 2. OBSERVATIONSANDDATAREDUCTION of this exoplanet revolution is the Kepler space tele- 2.1. K2 Target Selection, Photometry, and Transit scope. Transit data collected during the prime mis- Search sion of Kepler revealed that small planets: Earth ana- logues, super-Earths, and sub-Neptunes (R < 4 R ), We identified the high proper motion stars PM p ⊕ are abundant around Sun-like stars (Petigura et al. I06168+2435 and PM I11450+0000 (LP 613-39, NLTT 2013a). Statistical studies focusing on the few M 28423) as candidate M dwarf targets for our K2 Cam- dwarfs (T (cid:46)4000 K, M (cid:46)0.6 M ) that Kepler ob- paign0(C0: GO0120-PIL´epine)andCampaign1(C1: eff ∗ (cid:12) served (3900 stars) revealed that small planets exist GO1036 - PI Crossfield) proposals, respectively. The around nearly all M dwarfs (Dressing & Charbonneau stars were also proposed as targets in C0 and C1 by 2013, 2015). several other groups (C0: GO0111 - PI Sanchis Ojeda, ThesmallradiiandmassesofMdwarfs,combinedwith GO0119 - PI Montet; C1: GO1052 - PI Robertson, their sheer numbers (∼70% of all stars, Bochanski et al. GO1053 - PI Montet, GO1059 - PI Stello, GO1062 - PI 2010),providethebestopportunitiestodetectandchar- Anglada-Escude). Weselectedthesetargetsascandidate acterize small planets in the Solar neighborhood. Be- nearbyMdwarfsfromtheSUPERBLINKpropermotion cause of the large numbers of M dwarfs and the high survey (L´epine & Shara 2005; L´epine & Gaidos 2011) frequency of planets around them, the closest Earth-size followingthephotometricandpropermotioncriteriade- planetsinthehabitablezonealmostcertainlyorbitthese scribed in Crossfield et al. (2015). A coordinate cross- low-mass stars. Planets of a given radius transiting M match of PM I06168+2435 and PM I11450+0000 with dwarfs exhibit deeper transit signatures and planets of a the K2 Ecliptic Plane Input Catalog (EPIC) returned givenmassproducelargerstellarreflexmotions(Howard matches with the sources EPIC 202083828 and EPIC etal.2012). Additionally,theatmospheresofsmallplan- 201465501, respectively. EPIC 201465501 was given the etsorbitingMdwarfsaremoreamenabletotransmission K2 identifier K2-9 by NExScI1 after validation of its spectroscopy studies (e.g. Kreidberg et al. 2014) due to planetinMontetetal.(2015). EPIC202083828wasdes- the favorable star-to-planet radius ratio. However, since ignated K2-26 after the validation of its planet in § 3.4 Keplerobservedrelativelyfewofthesestars,thenumber of this work. K2-26 was observed by K2 in long-cadence ofsmallplanetsdetectedandconfirmedintransitaround mode during C0 from 2014 March 08 to May 27 and Mdwarfsremainssmall. Subsequently,theirdemograph- K2-9 was observed using the same mode during C1 from ics,formationscenarios,andtheevolutionoftheirorbits 2014 May 30 to August 21. We provide basic identifying remain poorly constrained. information and available photometry for these stars in We are pursuing a program to identify and character- Table 1. ize additional small planets transiting M dwarfs using The degraded pointing precision of K2 due to the loss data from K2, the 2 reaction wheel, ecliptic plane sur- of 2 reaction wheels leads to telescope drift in the form vey of NASA’s re-purposed Kepler spacecraft (Howell ofarollaroundthetelescopeboresight. Thisdriftiscor- et al. 2014). The K2 M Dwarf Program (K2-MDP) is rectedusingthrusterfireswhenthespacecraftreachesa a comprehensive approach to select M dwarf targets in predetermined roll limit; approximately every 6 h. The each K2 field, generate calibrated light curves and iden- periodicdriftandcorrectionofastarover∼1pixelleads tify candidate transiting planets, and obtain follow-up tosystematicbrightnessvariationsof∼0.5%. Thesevari- observations to validate and characterize the planetary ations are roll angle dependent and must be corrected in systems. The first discoveries from the K2-MDP are the light curve extraction process. Our approach to cor- K2-3 and K2-21, M0 dwarfs within 100 pc each hosting recting these effects and extracting calibrated photome- multiple transiting super-Earths (Crossfield et al. 2015; try from the raw K2 pixel data is identical to that de- Petiguraetal.2015). K2-3bcd,K2-21bc,andotherearly scribedinCrossfieldetal.(2015). Ingeneral,weperform K2 discoveries (Vanderburg et al. 2015a; Montet et al. a frame by frame median flux subtraction and compute 2015)haveprovidedplanetsthatoccupypoorlyexplored the the raw photometry by summing the single frame regions of the planetary mass-radius-temperature dia- flux within a circular aperture centered on the target. gram (R <4R ,T <600K), some ideal early targets We then compute the principal components of the row p ⊕ eq for spectroscopic follow-up with the James Webb Space and column centroids and fit a Gaussian process (GP) Telescope (JWST, Batalha et al. 2015; Beichman et al. toremovethesystematicvariations. Inpractice, theGP 2014), and some truly novel systems (i.e. WASP-47bcd, fitting is iterative and updates are made to the GP pa- Beckeretal.2015);allwellbeforethelaunchoftheTran- rameters to minimize the rms fit residuals. The flux ex- siting Exoplanet Survey Satellite (TESS, Ricker et al. tractionandGPfittingproceduresarerepeatedforaper- 2014). tures of varying size until an aperture size is found that Herewepresentthediscoveryandvalidationofasmall, minimizesthermsresidualsinthecalibratedlightcurve. cool planet orbiting the nearby M dwarf K2-26 and an The extraction apertures for K2-26 and K2-9 were soft- independent validation and detailed characterization of edged, circular apertures having radii of 2 and 3 pixels, the known planet transiting the M dwarf K2-9. In §2 we describe the observations of these systems using K2 and 1 http://exoplanetarchive.ipac.caltech.edu/docs/ K2Numbers.html Two Small Temperate Planets from K2 3 TABLE 1 Summary of Stellar Properties Parameter K2-26 K2-9 Reference α(hh:mm:ss) 06:16:49.579 11:45:03.472 1 δ (dd:mm:ss) +24:35:47.08 +00:00:19.08 1 µα (masyr−1) −27.8±4.1 −171.6±3.8 2 µ (masyr−1) −117.9±4.1 32.1±3.8 2 δ RV(kms−1) 95.34±0.15 -31.02±0.15 1 d (pc) 93±7 110±12 1 phot Kep(mag) 14.00 14.96 1 B(mag) 15.97±0.13 ... 3 V(mag) 14.53±0.03 15.63a 3 BPOSSI (mag) 16.16 16.55 2 RPOSSI (mag) 13.14 14.41 2 g(cid:48) (mag) 15.296±0.023 16.652±0.117 3 r(cid:48) (mag) 13.927±0.080 15.218±0.018 3 i(cid:48) (mag) 13.421±0.493 14.147±0.095 3 J (mag) 11.350±0.024 12.451±0.024 4 H (mag) 10.762±0.022 11.710±0.022 4 Ks (mag) 10.530±0.018 11.495±0.023 4 W1(mag) 10.422±0.023 11.348±0.022 5 W2(mag) 10.349±0.021 11.214±0.021 5 W3(mag) 10.409±0.086 11.354±0.193 5 SpectralType M1.0V±0.5 M2.5V±0.5 1 T (K) 3785±185 3390±150 1 eff [Fe/H](dex) −0.13±0.15 −0.25±0.20 1 Radius(R(cid:12)) 0.52±0.08 0.31±0.11 1 Mass(M(cid:12)) 0.56±0.10 0.30±0.14 1 Luminosity(L(cid:12)) 0.049±0.023 0.012±0.010 1 Density(gcm−3) 3.92±1.43 9.88±4.25 1 Age(Gyr) (cid:38)1 (cid:38)1 1 Note. — 1 - This Work; 2 - Roeser et al. (2010, PPMXL); 3 - Henden et al. (2012, APASS DR9); 4 - Cutri et al. (2003, 2MASS); 5 - Cutri&etal.(2013,ALLWISE);a -EstimatedusingphotometricrelationsinL´epine&Shara(2005) Fig. 1.— Top: Calibrated K2 photometry for K2-26 (EPIC 202083828). Vertical ticks indicate the locations of the transits. Bottom: Phase-foldedphotometryandbest-fitlightcurve. The∼4.7htransitdurationislikelytheresultofaneccentricorbit. respectively. After correcting for spacecraft roll, both residual variability is also removed. The calibrated K2 stars exhibit smooth, low-amplitude, slowly modulating, light curves for the stars are shown in top panels of Fig- photometric variations on the order of 1%. These fea- ures 1 and 2 and are available upon request. tures could be related to intrinsic stellar variability or We searched the calibrated and detrended light curves unaccounted for spacecraft systematics; possibly small of K2-26 and K2-9 using the TERRA algorithm; an au- focus changes due to thermal expansion and contraction tomated, grid-based, transit search pipeline (Petigura & over the course of a K2 observing campaign. If the ob- Marcy2012;Petiguraetal.2013b). OurTERRAsearchof served residual variability is at least in part intrinsic to theK2-26photometryidentifiedacandidateplanetwith the stars, i.e. star spots, it is very low-level and indica- a period of P ≈ 14.567 d and SNR ≈ 34. A candidate tive of slow rotation rates in both cases (∼weeks). Prior wasalsodetectedtransitingK2-9withP ≈18.450dand to searching for transit events in the light curves, this SNR ≈ 24. Each of these transit signals was fit with a 4 J. E. Schlieder et al. 1.003 x 1.002 u Fl 1.001 d 1.000 ze0.999 ali0.998 m0.997 or0.996 N 0.995 1980 1990 2000 2010 2020 2030 2040 2050 BJD_TBD - 2454833 x 1.003 u 1.002 b Fl 1.001 d 1.000 e z0.999 ali0.998 m0.997 r0.996 o0.995 N 4 3 2 1 0 1 2 3 4 Hours From Mid-Transit Fig. 2.—Top: CalibratedK2photometryforK2-9. Verticalticksindicatethelocationsofthetransits. Bottom: Phase-foldedphotometry andbest-fitlightcurve. Mandel & Agol (2002) transit model which we show in along the slit following an AB pattern for sky subtrac- the bottom panels of Figures 1 and 2. We then masked tion. TheK2-26observingsequenceconsistedof8×75s out the in-transit observations of each planet candidate exposures for a total integration time of 600 s. K2-9 was and searched for additional transit signals with TERRA. observed for 24×120 s for a total time of 2880 s. We This subsequent search of each light curve returned no also observed an A0 standard and flat and arc lamp ex- further candidates above our SNR threshold of 12. We posuresimmediatelyaftereachstarfortelluriccorrection note that the maximum likelihood periods for the can- and wavelength calibration. didates transiting K2-26 and K2-9 are close to integer The data were reduced using the SpeXTool package multiples of the observing cadence. We consider this a (Vacca et al. 2003; Cushing et al. 2004). SpeXTool per- prioriunlikely,andhypothesizethatresidualsystematics forms flat fielding, bad pixel removal, wavelength cali- exist in the K2 photometry after our photometric pro- bration, sky subtraction, spectral extraction and com- cessing. TheK2 C0-C3candidatecatalogofVanderburg bination, telluric correction, flux calibration, and order et al. (2015b) includes K2-26 and K2-9 with periods of merging. Thefinalcalibratedspectrahadsignal-to-noise 14.5670 d and 18.4487 d (no uncertainties), respectively. ratios (SNR) of ∼80 per resolution element in the H- TheVanderburgetal.(2015b)photometrywasextracted (∼1.6 µm) and K-bands (∼2.2 µm). The spectral qual- usinganindependentanalysisandtheyfindperiodscon- itydecreasesrapidlytowardbluerwavelengthswithSNR sistent with ours at the ∼0.3σ and ∼0.8σ levels. Montet ∼60 in the J-band (∼1.25 µm) and ∼10 at 0.75 µm. et al. (2015) also report an independent period for K2-9 The JHK-band spectra are compared to late-type stan- of 18.44883±0.00137 d, consistent with our estimate at dards from the IRTF Spectral Library2 (Cushing et al. the ∼1σ level. Thus, we conclude that any systematic 2005; Rayner et al. 2009) in Figures 3 and 4. K2-26 is errors that favor periods that are near-integer multiples a best visual match to the M1 standard across the near- of the K2 long cadence are second order and have mini- IR bands. K2-9 is later-type and matches well with the malimpactonourreportedparameters. Tovalidateand M2/M3 standards. characterize these candidate planets, we obtain and an- alyze spectroscopic and imaging data, perform detailed 2.2.2. NTT/EFOSC2 checksoftheK2pixellevelphotometryandlightcurves, and estimate false positive probabilities. These observa- On UT 2015 January 11, we observed K2-26 using tionsandanalysesaredescribedinthefollowingsections. the ESO Faint Object Spectrograph and Camera (v.2) (EFOSC2,Buzzonietal.1984)mountedtotheNasmyth B focus of the 3.6m ESO New Technology Telescope 2.2. Follow-up Spectroscopy (NTT). These observations were made as part of our 70 2.2.1. IRTF/SpeX nightK2follow-upprogram(PID194.C-0443,PI:I.J.M. We observed K2-26 and K2-9 using the near-infrared Crossfield). Thestarwasobservedundergoodconditions cross-dispersedspectrograph(SpeX, Rayneretal.2003) withaverageseeing∼1(cid:48).(cid:48)0withatotalintegrationtimeof on the 3.0m NASA Infrared Telescope facility on 2015 270 s. We used EFOSC2 in spectroscopic mode with the May 02 UT and 2015 April 16 UT, respectively. K2-26 0(cid:48).(cid:48)3 slit and grism 16 to provide a resolution R ∼ 1600 was observed under clear skies with an average seeing from 0.6-1.0 µm. We also obtained standard bias, flat, of ∼0(cid:48).(cid:48)6. K2-9 was observed under poorer conditions andHeArlampcalibrationframesimmediatelyafterob- with thin, variable cirrus, high humidity, and seeing be- serving K2-26 along with observations of spectrophoto- tween 1(cid:48).(cid:48)0 - 1(cid:48).(cid:48)2. We used the instrument in short cross metricstandardsforfluxcalibration(Bohlinetal.2001). dispersed mode using the 0.3×15(cid:48)(cid:48) slit which provides wavelength coverage from 0.68 to 2.5 µm at a resolution 2 http://irtfweb.ifa.hawaii.edu/~spex/IRTF_Spectral_ of R ≈ 2000. The stars were dithered to two positions Library/ Two Small Temperate Planets from K2 5 4 EPIC202083828 K I Si I Standards K I Na I Mg I K4V K4V K4V Na I K5V K5V K5V 3 K7V K7V K7V . t s n M0V M0V M0V o c M1V M1V M1V + (cid:104) f M2V M2V M2V d 2 e z i M3V M3V M3V l a m M4V M4V M4V r o N M5V M5V M5V 1 M6V M6V M6V 1.20 1.25 1.30 1.48 1.59 1.69 1.80 2.1 2.2 2.3 2.4 (cid:104) [µm] Fig. 3.—JHK-bandIRTF/SpeXspectraofK2-26(EPIC202083828)comparedtolate-typestandardsfromtheIRTFspectrallibrary. Allspectraarenormalizedtothecontinuumineachoftheplottedregions. ThestarisabestvisualmatchtospectraltypeM1acrossthe threenear-IRbands. Thisisconsistentwiththeresultsfromouranalysesusingspectroscopicindices. TheEFOSC2datawasreducedusingstandardIRAF3 the red side. The target was observed at an airmass routines that included bias subtraction, flat fielding, of 1.0 and conditions were generally favorable with see- wavelength calibration, and spectral extraction. The ing of about 1.5-1.8(cid:48)(cid:48). Standard IRAF functions (apall, spectrum was then flux calibrated using a standard ob- standard, sensfunc)4 were used to calibrate the data served close in time. The final calibrated spectrum had including: bias frame subtraction and flat-fielding using a SNR ∼50 per resolution element. dome flats, wavelength calibration using Fe-Ar arcs in the blue and He-Ne-Ar arcs in the red, and initial flux 2.2.3. Palomar Hale 5.0m/Double Spectrograph calibrationwithrespecttostandardHiltner600(Hamuy We observed K2-26 using the Double Spectrograph etal.1994). AseparateIDLroutinewasusedtostitchto- (DBSP, Oke & Gunn 1982) at the Palomar observatory gethertheredandbluespectra. A4000-10000˚Aportion Hale5.0mtelescopeon2015February12UT.Ontheblue is shown in Figure 5. For comparison, we also show the side of the spectrogaph, the 600 l/mm grating blazed spectrum of the M1 standard star GJ 229 (Kirkpatrick at 3780 ˚A was used at a setting of 29.5◦. On the red et al. 1991; Maldonado et al. 2015) observed using the side, the 600 l/mm grating blazed at 9500 ˚A was used same DBSP settings. at an angle of 32.5◦. The star was observed with a 1(cid:48)(cid:48) 2.2.4. Keck/HIRES slit which provided a spectral resolution of R∼2400 and wavelength coverage from ∼4000 - 7000 ˚A on the blue We observed both stars using the High Resolution sideandR∼3000andcoveragefrom∼7000-10000˚Aon Echelle Spectrometer (HIRES, Vogt et al. 1994) on the 10.0m Keck I telescope. We observed the stars fol- lowing standard California Planet Search (CPS, Marcy 3 IRAF is distributed by the National Optical Astronomy Ob- servatories, which are operated by the Association of Universities forResearchinAstronomy,Inc.,undercooperativeagreementwith 4 http://www.twilightlandscapes.com/IRAFtutorial/ theNationalScienceFoundation. IRAFintro_06.html 6 J. E. Schlieder et al. 4 K2−9 K I Si I Standards K I Na I Mg I K4V K4V K4V Na I K5V K5V K5V 3 K7V K7V K7V . t s n M0V M0V M0V o c M1V M1V M1V + (cid:104) f M2V M2V M2V d 2 e z i M3V M3V M3V l a m M4V M4V M4V r o N M5V M5V M5V 1 M6V M6V M6V 1.20 1.25 1.30 1.48 1.59 1.69 1.80 2.1 2.2 2.3 2.4 (cid:104) [µm] Fig. 4.— JHK-band IRTF/SpeX spectra of K2-9 compared to late-type standards from the IRTF spectral library. All spectra are normalized to the continuum in each of the plotted regions. The star is a best visual match to spectral type M2/M3 across the three near-IRbands. Thisisconsistentwiththespectraltypederivedfromspectroscopicindexbasedmethods. et al. 2008) procedures using the C2 decker and the ular Telescope (LBT) and works in conjunction with the 0(cid:48).(cid:48)87 × 14(cid:48).(cid:48)0 slit. The 0(cid:48).(cid:48)87 slit provides wavelength cov- deformablesecondaryLBTAdaptiveOpticssystem(LB- erage from ∼3600 - 8000 ˚A at a resolution R ≈ 60000. TIAO, Esposito et al. 2010, 2011; Riccardi et al. 2010; NoIodinecellwasusedfortheseobservations, thewave- Bailey et al. 2014) to deliver high-resolution near-IR length scale was calibrated using the standard HIRES imaging. For our observations, we only used the right reference. K2-26 was observed on UT 2015 February 5 side of the LBT. K2-26 was observed using the Ks-band and UT 2015 November 15 under good conditions with filter (λc = 2.16 µm, ∆λ = 0.32 µm) following a two ∼1(cid:48).(cid:48)0 seeing for a total of 565 s on each night. K2-9 was point dither pattern for sky subtraction. We obtained observed on UT 2015 July 12 under clear skies with 1(cid:48).(cid:48)4 40×0.15s exposures using the target as a natural AO seeing for a total of 1200 s. These data were reduced us- guidestarforatotalintegrationtimeof6s. Ourdatare- ingthestandardpipelineoftheCPS(Marcyetal.2008). duction included corrections for detector bias, sky back- TheresultingspectraofK2-26andK2-9hadSNR’s∼30 ground, and bad pixels followed by frame re-centering and 25 per pixel at 5500 ˚A, respectively. Examples of and averaging. The reduced image has a field-of-view the HIRES spectra for both stars are shown in Figure 8. (FOV) 10(cid:48).(cid:48)9 and a plate scale of 10.707 ± 0.012 mas pixel−1 (Maire et al. 2015). To ensure optimal back- ground subtraction and contrast, the final image of K2- 2.3. Adaptive Optics and Archival Imaging 26istrimmedtoa4(cid:48).(cid:48)0regionoffullditheroverlap. This 2.3.1. LBT - LBTI/LMIRcam is shown in the inset of the left panel of Figure 6. K2-26 was observed on 2015 January 07 UT using the 2.3.2. Palomar 60 Inch/Robo-AO L/M-band Infrared Camera (LMIRcam, Skrutskie et al. 2010; Leisenring et al. 2012) of the LBT Interferometer Weacquiredvisible-lightadaptiveopticsimagesofK2- (LBTI,Hinzetal.2008). LBTI/LMIRcamismountedat 26usingtheRobo-AOsystem(Baranecetal.2013,2014) the bent Gregorian focus of the dual 8.4m Large Binoc- on the 60-inch Telescope at Palomar Observatory. On Two Small Temperate Planets from K2 7 Fig. 5.—Left: SpectraofK2-26(EPIC202083828)takenwiththeblueandredsidesoftheDoubleSpectrograph(DBSP)atthePalomar Hale5.0m. Thefluxunitsarearbitrary. Right: AcomparisonspectrumoftheM1standardstarGJ229. Theoverallcontinuumshapeand strength of the deep, broad molecular features (TiO, CaH, VO) of K2-26 are an excellent match to the M1 standard across the observed wavelengthrange. 33 EPIC 202083828 − LBT/LMIRcam EPIC 202083828 − Robo−AO K2−9 − Keck/NIRC2 33 K 11 LP600 K 44 s! ! p! 44 g)g) 22 (mag) (mag)KK(cid:54)(cid:54)ss 6565 (ma (maLP600LP600 33 N (mag) (mag)KK(cid:54)(cid:54)pp 6565 (cid:54)(cid:54) 1”! 44 4”! E 2”! 77 77 55 88 00..55 11..00 11..55 22..00 22 44 66 00..55 11..00 11..55 SSeeppaarraattiioonn ((aarrccsseecc)) SSeeppaarraattiioonn ((aarrccsseecc)) SSeeppaarraattiioonn ((aarrccsseecc)) Fig. 6.— AO images and contrast curves for K2-26 (EPIC 202083828) and K2-9 Left: K2-26 LBT/LMIRcam Ks-band image (inset) andcontrastcurve. Noadditionalstarsaredetectedwithin2(cid:48).(cid:48)0. Center: K2-26Robo-AOLP600-bandimage(inset)andcontrastcurve. An additional star is detected with ∆LP600=5 mag at 5.5(cid:48)(cid:48) separation (red circle). The small box in the upper left of the inset shows K2-26 without the hard stretch necessary to reveal the faint companion. Right: K2-9 Keck/NIRC2 Kp-band image and contrast curve. NoadditionalstarsaredetectedintheNIRC2fieldofview. 2015 March 8 UT, we observed K2-26 with a long-pass servedintheK -bandfilter(λ =2.124µm,∆λ=0.351 p c filter cutting on at 600 nm (LP600) as a sequence of µm) using the narrow camera setting with a pixel scale full-frame-transfer detector readouts from an electron- of 9.942 mas pixel−1. To avoid the noisier lower-left multiplying CCD at the maximum rate of 8.6 Hz for a quadrantoftheNIRC2array,weemployedathree-point total of 120 s of integration time. The individual images ditherpatternwith11×10sintegrationsperditheryield- arecorrectedfordetectorbiasandflat-fieldingeffectsbe- ing a total on-source integration time of 330s. Individ- fore being combined using post-facto shift-and-add pro- ual frames were flat-fielded and sky-subtracted and then cessing using K2-26 as the tip-tilt star with 100% frame shiftedandcoaddedtoproducethefinal10(cid:48).(cid:48)2image. Our selectiontosynthesizealong-exposureimage(Lawetal. NIRC2 image of K2-9 is shown in the inset of the right 2014). The resulting reduced image has a nominal FOV panel of Fig. 6. of 44(cid:48).(cid:48)0 and plate scale of 0(cid:48).(cid:48)0216 pixel−1 (Baranec et al. 2014). A 15(cid:48).(cid:48)5 portion of the Robo-AO image centered 2.3.4. DSS and SDSS Archival Imaging onK2-26isshownintheinsetofthecenterpanelofFig- K2-26 and K2-9 were both observed in two different ure6. Afaint,widelyseparatedcompanionwasdetected photometric bands (blue and red; B and R) during The in the Robo-AO image and is described in § 3.3.1. NationalGeographicSociety-PalomarObservatorySky Survey (POSS I, Minkowski & Abell 1963) and the Sec- 2.3.3. Keck/NIRC2 ond Palomar Observatory Sky Survey (POSS II, Reid We observed K2-9 using the Near Infrared Camera 2 et al. 1991) using the 1.2m Samuel Oschin Telescope. (NIRC2) and laser guide star AO (LGS AO, van Dam TheoriginalPOSSphotographicplateswerescannedand et al. 2006; Wizinowich et al. 2006) on the 10.0m Keck- digitizedbytheSpaceTelescopeScienceInstituteandare II telescope on 2015 April 07 UT. The target was ob- now available for flexible download as the Digitized Sky 8 J. E. Schlieder et al. Survey (DSS)5. The digitized POSS I and II plates have (2013b). In these works, metallicity is estimated using plate scales of 1(cid:48).(cid:48)01 pixel−1. K2-26 was observed in the spectroscopic index and equivalent width based meth- B andR-bandsduringPOSSIon1954November22UT ods (Rojas-Ayala et al. 2012; Terrien et al. 2012; Mann and in the B-band during POSS II on 1996 January 13 et al. 2013b) that were calibrated using a sample of M UT. K2-9 was observed in both POSS I bands on 1952 dwarfs having wide, co-moving FGK companions with January31UTandinthePOSSIIB-bandon1995Jan- welldetermined[Fe/H].WeuseIDLsoftwaremadepub- uary 01 UT. Both stars were also observed in five pho- licly available by A. Mann8 to calculate the metallicities tometric bands (u,g,r,i,z) during the Sloan Digital Sky of K2-26 and K2-9 in the H and K bands. We average Survey (SDSS, York et al. 2000) using the 2.5m Sloan theH andK metallicitiesandaddthemeasurementand Foundation Telescope (Gunn et al. 2006). SDSS images systematic uncertainties in quadrature to arrive at the haveaplatescaleof0(cid:48).(cid:48)396pixel−1. K2-26andK2-9were final values. We find K2-26 has [Fe/H]=−0.13±0.15 observedduringtheSDSSon2006November11UTand and K2-9 has [Fe/H]=−0.25±0.20. Thus neither star 2006 January 06 UT, respectively. The total time base- is metal-rich. linesbetweenthePOSSIandSDSSepochsforeachstar Effectivetemperature, radius, andmassarecalculated are 52 and 54 years, respectively. We obtained the pub- using temperature sensitive spectroscopic indices in the liclyavailableimagingdataintheformof1.(cid:48)0POSSIB, JHK-bands(Mannetal.2013a)and empiricalrelations POSSIIB,andSDSSDR7(Abazajianetal.2009)g im- calibrated using nearby, bright M dwarfs with interfer- ages of the stars centered on their 2015 epoch positions ometrically measured radii (Boyajian et al. 2012). We using the NASA/IPAC Infrared Science Archive (IRSA) calculateTeff intheJHK-bandsandaveragetheresults. Finder Chart web interface6. These data are presented Conservative Teff uncertainties are estimated by adding in Figure 7. in quadrature the rms scatter in the JHK-band values and the systematic error in the empirical fits for each 3. ANALYSESANDRESULTS band (Mann et al. 2013a). The stellar radii, masses, 3.1. Spectroscopic Analyses luminosities, and densities are computed using publicly available software from A. Mann9. The resulting fun- 3.1.1. Medium-Resolution Spectroscopy damental parameters are listed in Table 1. The larger Weusemolecularbandindicesintheopticalandnear- relative uncertainties in R and M for K2-9 are a result IRtoestimatespectraltypes(SpTy)forK2-26andK2-9. ∗ ∗ of the poorer empirical fits in the Mann et al. (2013a) The EFOSC2 spectrum of K2-26 provides access to TiO relations due to having relatively few calibrators at low andCaHmolecularbandsthataretemperaturesensitive temperatures. Vanderburg et al. (2015b) estimated T and calibrated to provide SpTy’s for stars ∼K7-M6. We eff andR forK2-26usingitsV −K colorandfoundresults specificallyusetheTiO5,CaH2,andCaH3indices(Reid ∗ consistent with ours. They also estimated T and R etal.1995;Gizis1997)toestimatethestar’sSpTyusing eff ∗ for K2-9, this time using its H−K color, and found val- the calibrated relations in L´epine et al. (2013). We find ues consistent with ours at ∼2σ. Additionally, Montet K2-26 has an optical SpTy of M1.0. The L´epine et al. et al. (2015) estimated the fundamental parameters of (2013) relations have an accuracy of 0.5 subtypes. We K2-9 using broadband photometry and model fits. Our also compare the EFOSC spectrum to optical M dwarf spectroscopicparametersareconsistentwiththeirsinall standard spectra from Kirkpatrick et al. (1991) and find cases within 1σ uncertainties, but our nominal values a best visual match to types M1.0/M1.5. We perform a of mass, radius, and metallicity are all systematically similarcomparisonofourDBSPspectratoMdwarfstan- larger. Photometric distances to the stars are estimated dards and find a consistent best match to the M1 stan- bycalculatingthedistancemodulifortheirspectraltypes dard GJ 229 (Fig. 5). In the near-IR K-band, the H 0- 2 fromthecolor-temperatureconversiontableofPecaut& K2 index measures temperature sensitive water opacity Mamajek (2013). We estimate K2-26 lies at 93±7 pc and is calibrated for SpTy’s M0-M9 (Rojas-Ayala et al. and K2-9 at 110±12 pc, just at the outer boundary of 2012). We use our SpeX spectra to measure this index the extended solar neighborhood. in K2-26 and K2-9 and find SpTy’s of M1.0 and M2.5, respectively. The H 0-K2/SpTy relation has a system- 2 3.1.2. High-Resolution Spectroscopy atic scatter of 0.6 subtypes. Following these results, we adopt a SpTy of M1.0 ± 0.5 for K2-26 and M2.5 ± 0.5 We searched for tight, spectroscopic binary compan- for K2-9. Our index based measurements are consistent ions or background stars at very close angular separa- with the visual best matches to M dwarf standards (e.g. tions in our HIRES spectra using the methodology of Figures 3 and 4) and are also consistent with SpTy esti- Kolbl et al. (2015). The Kolbl et al. (2015) algorithm matesusingthestars’opticalandnear-IRcolors(Pecaut uses a library of more than 600 HIRES spectra of stars & Mamajek 2013)7. witharangeofT ’s, log(g)’s, andmetallcitiestomodel eff Following Crossfield et al. (2015), we use our SpeX thespectrumofthetargetstarasthesumoftwolibrary spectratoestimatethefundamentalparametersofmetal- templates and search for secondary lines. For high SNR licity ([Fe/H]), effective temperature (T ), radius (R ), targets,thismethodcandetectcompanionswithin∼0(cid:48).(cid:48)8 eff ∗ and mass (M ) for K2-26 and K2-9 using the meth- of the primary, with as little as ∼1% of primary’s flux ∗ ods presented in Mann et al. (2013a) and Mann et al. in the V-band, and ∆RV > 10 km s−1. The algorithm also measures the barycentric corrected primary RV via 5 http://stdatu.stsci.edu/cgi-bin/dss_form comparison to a standard solar spectrum. 6 http://irsa.ipac.caltech.edu/applications/finderchart/ 7 Throughout this work, we use the expanded table available on Eric Mamajek’s webpage: http://www.pas.rochester.edu/ 8 https://github.com/awmann/metal ~emamajek/EEM_dwarf_UBVIJHK_colors_Teff.txt 9 https://github.com/awmann/Teff_rad_mass_lum Two Small Temperate Planets from K2 9 EPIC 202083828! N POSS I-B: 1954-11-22! POSS II-B: 1996-01-13! E 15”! SDSS-g: 2006-11-20! K2-9! POSS I-B: 1952-01-31! POSS II-B: 1995-01-01! SDSS-g: 2006-01-06! Fig. 7.—1.(cid:48)0×1.(cid:48)0archivalsurveyimagesofK2-26(EPIC202083828)andK2-9. Top: K2-26DSSandSDSSimagesdisplaying>6(cid:48)(cid:48) of transversemotionover52years. Thefaintstartothesouth-southeastineachimageisthesameasrevealedinourRobo-AOimages. The starisnotco-movingwithK2-26. Bottom: DSSandSDSSimagesofK2-9showing>9(cid:48)(cid:48) oftransversemotionover54years. Averyfaint sourceisdetected∼3(cid:48).(cid:48)5tothenortheastofK2-9intheSDSSimageandappearsasanextensionoftheK2-9intensitydistribution. This source is not detected with confidence in the shallower POSS images. Further details are provided in section 3.3.2. The POSS I images revealnostarsatthecurrentpositionsofK2-26orK2-9(magentacircles)downtothephotometriclimitsofthatsurvey. NeitherofourK2-26spectranorthespectrumofK2-9 companions that includes (cid:38)2 M at (cid:46)0.25 AU and (cid:38)5 Jup exhibit evidence for a close in, spectroscopic compan- M at (cid:46)1.5 AU. Jup ion. Our analyses exclude tight companions as faint as 3% of the primary flux in the approximate V-band with 3.2. Activities, Kinematics, Ages, and Surface Gravities ∆RV > 10 km s−1. The lower flux limit corresponds The ∼6000 - 9000 ˚A (0.6 - 0.9 µm) region of M dwarf to companions with ∆V ≈ 3.8. Assuming circular or- spectraprovideaccesstoseveralfeaturessensitivetosur- bits and using the photometric relations in Pecaut & facegravityandmagneticactivity. Priortothetransition Mamajek (2013), our ∆V and ∆RV limits allow us to to fully convective interiors ((cid:46)M4), M dwarfs lose angu- rule out companions ∼M4.5 and earlier at (cid:46)0.7 AU sep- larmomentumtoasteadystellarwindandtheirrotation arations for K2-26 and companions ∼M5.0 and earlier rates decrease over time. This decrease in rotation rate at (cid:46)0.7 AU separations for K2-9. We measure RV = leadstoalossindynamodrivenmagneticactivityanda 95.34±0.15 km s−1 and RV = 95.33±0.15 km s−1 for subsequent loss of high energy emission over time. Here K2-26inFebruaryandNovember2015,respectively. We wefocusonemissionfromthe6563˚AHαlineasanactiv- also measure RV = −31.02±0.15 km s−1 for K2-9. Our ityindicator(Westetal.2008,2011)toplaceconstraints two RV measurements for K2-26 are separated by 281 on the ages of K2-26 and K2-9. Our HIRES spectra of days and are consistent within the 150 m s−1 measure- both K2-26 and K2-9 provide access to the Hα line at mentuncertainty. TheseRVmeasurementsarealsocon- highresolutionwhereitisseeninabsorptioninbothstars sistent with the multi-epoch measurements spanning 28 (Figure 8). We used the IDL software line eqwidth10 days in Vanderburg et al. (2015b). The consistency of toestimateequivalentwidths(EWs)of0.46±0.02˚Aand these long term RV measurements allows us to rule out RV accelerations due to companions below the sensitiv- 0.34±0.01 ˚A for K2-26 and K2-9, respectively. The Hα ity of our initial secondary line search. If we adopt our EWs suggest that both stars are relatively inactive; con- 150 m s−1 HIRES measurement uncertainty as the max- sistent with field age early M dwarfs (West et al. 2008). imum possible acceleration and assume circular orbits, Our EFOSC2 and DBSP spectra of K2-26 cover the Hα ourmulti-epochmeasurementsruleoutstellarmasscom- line where it is also seen in absorption. panions at separations that overlap the limits from our As an additional check of the stars’ activity levels, we LBT AO imaging (see §3.3.1) and a range of gas-giant 10 http://fuse.pha.jhu.edu/analysis/fuse_idl_tools.html 10 J. E. Schlieder et al. young. Wenotethatthe∼80dayK2lightcurveofK2-9 22..55 H(cid:95) does not exhibit convincing evidence for magnetic spot modulated variability or strong flares. st.st. EPIC 202083828 We investigate the kinematics of K2-26 and K2-9 by nn oo 22..00 calculating their UVW Galactic velocities following the cc methods outlined in Johnson & Soderblom (1987) up- ++ datedtoepochJ2000.0. Weadoptasolarcentriccoordi- ff(cid:104)(cid:104) 11..55 natesystemwhereU ispositivetowardtheGalacticcen- dd ee K2−9 ter, V is positive in the direction of solar motion around zz aliali 11..00 theGalaxy, andW ispositivetowardthenorthGalactic mm pole. Using the measured proper motions and RV’s and rr estimated distances, we calculate UVW = (-91.7, oo K2−26 NN 00..55 -52.0, -28.7) ± (2.1, 3.6, 2.2) km s−1 and UVW = K2−9 (-86.3, -8.1, -41.6) ± (9.6, 3.1, 2.0) km s−1. These es- 00..00 timates yield total Galactic velocities S = 109.3 K2−26 66554488 66555544 66556600 66556666 66557722 66557788 km s−1 and SK2−9 = 96.1 km s−1. Both stars have S consistentwiththestatisticallyolder, kinematicallyhot- (cid:104)(cid:104) [[AAnngg..]] ter, thick disk population following the kinematic sub- Fig. 8.— RV corrected HIRES spectra of K2-26 (EPIC divisions of Bensby & Feltzing (2010). Their kinematics 202083828)andK2-9centeredontheHαlineat6563˚A.Theline are thus consistent with other old M dwarfs. Alone, nei- is seen in absorption in both stars, indicating they are relatively therthelackofHαemissionnorthelargeGalacticveloc- inactiveandlikely(cid:38)1Gyrold. ities of K2-26 and K2-9 place strong constraints on their ages. However, when combined, these observations sug- searched for excess ultraviolet (UV) emission using data gestbothstarsare(cid:38)1Gyrold. TheobservedUVexcess from the NASA GALEX satellite (Martin et al. 2005). of K2-9 warrants further consideration and is detailed in Like Hα, UV is tracer of magnetic activity in late-type the context of its planet in § 4. starsandcanbeusedtoplacelimitsontheirages(Shkol- Spectroscopic akali lines are sensitive to electron pres- nik et al. 2011; Stelzer et al. 2013; Jones & West 2015). sure in stellar atmospheres in the sense that increases We searched the GALEX data using a 10(cid:48)(cid:48) radius cen- in pressure lead to lines with broader wings (pressure tered on our targets in the GalexView Web Tool11. No broadening), thus in low pressure (low gravity) atmo- GALEX observations are available for K2-26. K2-9 was spheres, alkali lines are comparatively weak (Schlieder observed by GALEX during the Medium Imaging Sur- et al. 2012a). To verify that K2-26 and K2-9 are dwarf vey (MIS, Martin et al. 2005) in the far- and near-UV stars with high surface gravities we investigate the grav- (FUV, NUV) bands. The star was detected at ∼4σ and ity sensitive Na I doublet near 8190 ˚A. After reducing ∼5σ intheFUVandNUV,respectively. TheMISobser- our spectra to a resolution of R∼900, we measured the vations spanned 2006 October to 2009 April. Following Na I index as defined by Lyo et al. (2004) and compared Shkolnik et al. (2011) and Schlieder et al. (2012b), we to samples with known surface gravities. For K2-26 we calculatetheratiooftheFUVandNUVfluxdensitiesto use the EFOSC2 spectrum to measure an Na I index of the2MASSJ andK -bandfluxdensitiesandcompareto s 1.064±0.023. WeprefertheEFOSC2spectrumoverthe samples of M dwarfs with known ages. The flux density SpeX spectrum for this measurement since the SNR is ratios indicate that K2-9 has FUV and NUV emission ∼6×greaterat8200˚A.ForK2-9,wehaveonlytheSpeX consistent with stars of similar SpTy having relatively spectrum with SNR∼20 at 8200 ˚A and measure an Na I large excess emission and young ages. When compared index of 1.144 ± 0.129. The Na I uncertainties are esti- to the M dwarf samples with known ages in Shkolnik & matedusingMCmethods. Whencomparedtothedwarf, Barman (2014), the fractional UV excesses suggest that young, and giant samples in Lawson et al. (2009), both K2-9 is at most as old as the Hyades (600-800 Myr, Per- K2-26 and K2-9 are consistent with the field M dwarf rymanetal.1998;Brandt&Huang2015a,b). Itispossi- sequence within uncertainties. This is reinforced by our blethatoneoftheepochsoftheGALEX measurements initial target selection using reduced proper motion dia- caught the star during a flare or other transient period grams (which removes giants) and the visual matches of of heightened activity. UV flare events were observed in theirSpeXspectratoMdwarfstandards(Figs.3and4). ∼3% of field M dwarfs in the variability survey of Welsh etal.(2007)andFranceetal.(2013)observedflareswith 3.3. Imaging Analyses timescales of 100-1000s in Hubble Space Telescope UV spectra of field age M dwarf planet hosts. These events 3.3.1. Adaptive Optics Imaging arealikelycontributortothemeasuredrangesofUVac- In our LBT/LMIRcam imaging, K2-26 was measured tivity in M dwarf samples of known age (up to 2 orders with a resolution of 0(cid:48).(cid:48)116 (FWHM) and appears single of magnitude, Shkolnik & Barman 2014). Therefore, the at that limit. No other stars were detected in the 4(cid:48).(cid:48)0 evidence for at least transient strong UV emission from region of full dither overlap. We estimate the sensitivity K2-9 is intriguing and may affect the properties of its to faint companions and background stars by injecting planet, but when considered along with the large scat- fake point sources with SNR=5 into the final image at ter of M dwarf UV excesses and the star’s lack of Hα separations N× FWHM where N is an integer. The 5σ emission, it does not indicate that the star is strikingly sensitivities as a function of separation are shown in the left panel of Fig. 6. Our LMIRcam imaging is sensitive 11 galex.stsci.edu/GalexView/ to stars with K -band contrast ∆K = 3.1 mag at 0(cid:48).(cid:48)1 s s