Exoplanets around Low-mass Stars Unveiled by K2
Teruyuki Hirano, Fei Dai, Davide Gandolfi, Akihiko Fukui, John H. Livingston, Kohei Miyakawa, Michael Endl, William D. Cochran, Francisco J. Alonso-Floriano, Masayuki Kuzuhara, David Montes, Tsuguru Ryu, Simon Albrecht, Oscar Barragan, Juan Cabrera, Szilard Csizmadia, Hans Deeg, Philipp Eigmüller, Anders Erikson, Malcolm Fridlund, Sascha Grziwa, Eike W. Guenther, Artie P. Hatzes, Judith Korth, Tomoyuki Kudo, Nobuhiko Kusakabe, Norio Narita, David Nespral, Grzegorz Nowak, Martin Pätzold, Enric Palle, Carina M. Persson, Jorge Prieto-Arranz, Heike Rauer, Ignasi Ribas, Bun'ei Sato, Alexis M. S. Smith, Motohide Tamura, Yusuke Tanaka, Vincent Van Eylen, Joshua N. Winn
aa r X i v : . [ a s t r o - ph . E P ] J a n Preprint typeset using L A TEX style AASTeX6 v. 1.0
EXOPLANETS AROUND LOW-MASS STARS UNVEILED BY K2
Teruyuki Hirano , Fei Dai , Davide Gandolfi , Akihiko Fukui , John H. Livingston , Kohei Miyakawa ,Michael Endl , William D. Cochran , Francisco J. Alonso-Floriano , Masayuki Kuzuhara , DavidMontes , Tsuguru Ryu , Simon Albrecht , Oscar Barragan , Juan Cabrera , Szilard Csizmadia , HansDeeg , Philipp Eigm¨uller , Anders Erikson , Malcolm Fridlund , Sascha Grziwa , Eike W. Guenther ,Artie P. Hatzes , Judith Korth , Tomoyuki Kudo , Nobuhiko Kusakabe , Norio Narita , DavidNespral , Grzegorz Nowak , Martin P¨atzold , Enric Palle , Carina M. Persson , JorgePrieto-Arranz , Heike Rauer , Ignasi Ribas , Bun’ei Sato , Alexis M. S. Smith , Motohide Tamura ,Yusuke Tanaka , Vincent Van Eylen , Joshua N. Winn Department of Earth and Planetary Sciences, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan Department of Physics, and Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA02139, USA Department of Astrophysical Sciences, Princeton University, 4 Ivy Lane, Princeton, NJ 08544, USA Dipartimento di Fisica, Universit´a di Torino, via P. Giuria 1, 10125 Torino, Italy Okayama Astrophysical Observatory, National Astronomical Observatory of Japan, Asakuchi, Okayama 719-0232, Japan Department of Astronomy, Graduate School of Science, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo, 113-0033, Japan Department of Astronomy and McDonald Observatory, University of Texas at Austin, 2515 Speedway, Stop C1400, Austin, TX 78712, USA Leiden Observatory, Leiden University, 2333CA Leiden, The Netherlands Departamento de Astrof´ısica y Ciencias de la Atm´osfera, Facultad de Ciencias F´ısicas, Universidad Complutense de Madrid, 28040 Madrid,Spain Astrobiology Center, NINS, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan National Astronomical Observatory of Japan, NINS, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan SOKENDAI (The Graduate University for Advanced Studies), 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan Stellar Astrophysics Centre, Department of Physics and Astronomy, Aarhus University, Ny Munkegade 120, DK-8000 Aarhus C, Denmark Institute of Planetary Research, German Aerospace Center, Rutherfordstrasse 2, 12489 Berlin, Germany Instituto de Astrof´ısica de Canarias, C/ V´ıa L´actea s/n, 38205 La Laguna, Spain Departamento de Astrof´ısica, Universidad de La Laguna, 38206 La Laguna, Spain Department of Space, Earth and Environment, Chalmers University of Technology, Onsala Space Observatory, 439 92 Onsala, Sweden Rheinisches Institut f¨ur Umweltforschung an der Universit¨at zu K¨oln, Aachener Strasse 209, 50931 K¨oln, Germany Th¨uringer Landessternwarte Tautenburg, Sternwarte 5, D-07778 Tautenberg, Germany Subaru Telescope, National Astronomical Observatory of Japan, 650 North Aohoku Place, Hilo, HI 96720, USA Center for Astronomy and Astrophysics, TU Berlin, Hardenbergstr. 36, 10623 Berlin, Germany Institut de Ci`encies de l’Espai (CSIC-IEEC), Carrer de Can Magrans, Campus UAB, 08193 Bellaterra, Spain
ABSTRACTWe present the detection and follow-up observations of planetary candidates around low-mass starsobserved by the K2 mission. Based on light-curve analysis, adaptive-optics imaging, and opticalspectroscopy at low and high resolution (including radial velocity measurements), we validate 16planets around 12 low-mass stars observed during K2 campaigns 5–10. Among the 16 planets, 12 arenewly validated, with orbital periods ranging from 0.96–33 days. For one of the planets (K2-151b)we present ground-based transit photometry, allowing us to refine the ephemerides. Combining our K2 M-dwarf planets together with the validated or confirmed planets found previously, we investigatethe dependence of planet radius R p on stellar insolation and metallicity [Fe/H]. We confirm that forperiods P . R p & R ⊕ are less common than planets with a radiusbetween 1–2 R ⊕ . We also see a hint of the “radius valley” between 1.5 and 2 R ⊕ that has beenseen for close-in planets around FGK stars. These features in the radius/period distribution couldbe attributed to photoevaporation of planetary envelopes by high-energy photons from the host star,as they have for FGK stars. For the M dwarfs, though, the features are not as well defined, and wecannot rule out other explanations such as atmospheric loss from internal planetary heat sources, ortruncation of the protoplanetary disk. There also appears to be a relation between planet size and Hirano et al. metallicity: those few planets larger than about 3 R ⊕ are found around the most metal-rich M dwarfs. Keywords: methods: observational – techniques: high angular resolution – techniques: photometric –techniques: radial velocities – techniques: spectroscopic – planets and satellites: detection INTRODUCTIONM dwarfs have some advantages over solar-type(FGK) stars in the detection and characterization oftransiting planets. Their smaller sizes lead to deepertransits for a given planet radius. In addition, theirhabitable zones occur at shorter orbital periods, facili-tating the study of terrestrial planets in the habitablezone. These advantages are now widely appreciated.Many observational and theoretical studies have focusedon M-dwarf planets, including their potential habit-ability and detectable biosignatures (e.g., Scalo et al.2007; Shields et al. 2016). However, the number of cur-rently known transiting planets around low-mass starsis much smaller than that for solar-type stars, be-cause low-mass stars are optically faint. In particular,the number of mid-to-late M dwarfs ( T eff . Kepler sample (Dressing & Charbonneau 2013,2015; Morton & Swift 2014; Mulders et al. 2015a,b;Ballard & Johnson 2016), the distribution and proper-ties of mid-to-late M-dwarf planetary systems are stillrelatively unexplored.
Kepler ’s second mission, K2 (Howell et al. 2014), hasalso contributed to the search for transiting planetsaround M dwarfs. Hundreds of stars have been identi-fied as candidate planet-hosting stars (e.g., Montet et al.2015; Vanderburg et al. 2016; Crossfield et al. 2016;Pope et al. 2016), many of which have been validated(e.g., Dressing et al. 2017b). Moreover, K2 has ob-served young stars in stellar clusters (e.g., the Hyades,Pleiades, and Beehive), including many low-mass stars.Several transiting planet candidates around these havealready been reported (Mann et al. 2016a,b, 2017b,2018; Ciardi et al. 2017). These planets are potentiallypromising targets for follow-up studies such as Dopplermass measurement and atmospheric characterization.We have been participating in K2 planet detection andcharacterization in the framework of an internationalcollaboration called KESPRINT . Making use of ourown pipeline to reduce the K2 data and look for transit [email protected] In 2016, the two independent K2 follow-up teams KEST(Kepler Exoplanet Science Team) and ESPRINT ( Equipo deSeguimiento de Planetas Rocosos Intepretando sus Transitos )merged and became the larger collaboration “KESPRINT”. signals, we have detected 30-80 planet candidates in eachof the K2 campaign fields. Through intensive follow-upobservations using various facilities all over the world,we have validated or confirmed many transiting plan-ets (e.g., Sanchis-Ojeda et al. 2015; Fridlund et al. 2017;Gandolfi et al. 2017; Guenther et al. 2017). In this pa-per, we focus on planetary systems around M dwarfsfound by the KESPRINT project.The rest of the paper is organized as follows. In Sec-tion 2, we describe the reduction of the K2 data and de-tection of the planet candidates by our pipeline. Next,we report our follow-up observations, including low-and high-resolution optical spectroscopy, high-contrastimaging, and ground-based follow-up transit observa-tions (Section 3). Section 4 presents the analysis ofthe follow-up observations, through which we validate15 planets around M dwarfs. Individual systems of spe-cial interest are described in Section 5. In Section 6we examine the properties of all the transiting planetscurrently known around M dwarfs, with a focus on theplanetary radius. Our conclusions are in Section 7. K2 PHOTOMETRY AND DETECTION OFPLANET CANDIDATES2.1.
K2 Light Curve Reduction
Due to the loss of two of its four reaction wheels, the
Kepler spacecraft can no longer maintain the pointingstability required to observe its original field of view.The
Kepler telescope was re-purposed for a new seriesof observations under the name K2 (Howell et al. 2014).By observing in the ecliptic, the torque by solar radi-ation pressure is minimized, significantly improving itspointing stability. The spacecraft must also switch toa different field of view about every three months tomaintain pointing away from the Sun. In this opera-tional mode, the photometry is strongly affected by therolling motion of the spacecraft along its boresight andthe variation of pixel sensitivity. To reduce this effect,we adopted an approach similar to that described byVanderburg & Johnson (2014).We now briefly describe our light-curve productionpipeline. We downloaded the target pixel files from theMikulski Archive for Space Telescopes. We then putdown circular apertures surrounding the brightest pixelwithin the collection of pixels recorded for each target.We fitted a 2-D Gaussian function to the intensity dis- https://archive.stsci.edu/k2. alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 X and Y positions of the Gaussian function, as a function of time,allowed us to track the rolling motion of the spacecraft.To reduce the intensity fluctuations associated with thismotion, we divided the apparent flux variation by thebest-fitting piecewise linear relationship between appar-ent flux and the coordinates X and Y . The system-atic correction was described in more detail by Dai et al.(2017). 2.2. Transit Detection
To remove any long-term systematic or instrumentalflux variations that may complicate the search of transitsignals, we fitted the K2 light curve with a cubic splinewith a timescale of 1.5 days. The observed light curvewas then divided by the spline fit. The smoothing inter-val of 1.5 days was chosen to be much longer than theexpected duration of planetary transits, which are mea-sured in hours for for short-period planets around dwarfstars. We then searched for periodic transit signals withthe Box-Least-Squares algorithm (Kov´acs et al. 2002).We employed a modification of the BLS algorithm, us-ing a more efficient nonlinear frequency grid that takesinto account the scaling of transit duration with orbitalperiod (Ofir 2014). To quantify the significance of atransit detection, we adopted the signal detection effi-ciency (SDE) (Ofir 2014) which is defined by the am-plitude of peak in the BLS spectrum normalized by thelocal standard deviation. A signal was considered sig-nificant if the SDE is greater than 6.5. To search forany additional planets in the system, we re-computedthe BLS spectrum after removing the transit signal thatwas detected in the previous iteration, until the maxi-mum SDE dropped below 6.5.2.3. Initial Vetting
After the transit signals were identified, we performeda quick initial vetting process to exclude obvious falsepositives. We sought evidence for any alternation in theeclipse depths or a significant secondary eclipse, eitherof which would reveal the system to be an eclipsing bi-nary (EB). Such effects should not be observed if thedetected signal is from a planetary transit. We fitteda Mandel & Agol (2002) model to the odd- and even-numbered transits separately. If the transit depths dif-fered by more than 3 σ , the system was flagged as a likelyfalse positive.We also searched for any evidence of a secondaryeclipse. First we fitted the observed transits with aMandel & Agol (2002) model. The fit was used as a tem-plate for the secondary eclipse. We allowed the eclipsedepth and time of opposition to float freely; all the otherrelevant parameters were held fixed based on the transitmodel. If a secondary eclipse was detected with more than 3 σ significance, we then calculated the geometricalbedo implied by the depth of secondary eclipse. If theimplied albedo was much larger than 1, we concludedthe eclipsing object is likely to be too luminous to bea planet. Typically, in each of the K2 Campaigns 5, 6,7, 8, and 10, approximately 5 −
10 M-dwarf planetarycandidates survived this initial vetting process. OBSERVATIONS AND DATA REDUCTIONSWe here report the follow-up observations for theplanet candidates around M dwarfs detected by ourpipeline. The complete list of our candidates will bepresented elsewhere (Livingston et al. and other pa-pers in preparation). We attempted follow-up obser-vations for as many M-dwarf planet hosts as possible.Our selection of targets included all planet candidatesthat had not already been validated (to our knowledge),with a preference for northern-hemisphere targets forwhich our follow-up resources are best suited. Specifi-cally, we report on the candidates around K2-117, K2-146, K2-122, K2-123, K2-147, EPIC 220187552, EPIC220194953, K2-148, K2-149, K2-150, K2-151, K2-152,K2-153, and K2-154, for which we conducted both high-resolution imaging and optical spectroscopy. This list ofM dwarfs covers about half of all candidate planet-hostsin the K2 Campaign fields 5, 8, and 10. Campaign fields6 and 7 are located in the southern hemisphere whereour telescope resources are limited. The M-dwarf sys-tems we did not follow up are generally fainter objects(
V >
15) for which follow-up observations are difficultand time-consuming.3.1.
Low Dispersion Optical Spectroscopy
We conducted low dispersion optical spectroscopywith the Calar Alto Faint Object Spectrograph(CAFOS) on the 2.2 m telescope at the Calar Alto ob-servatory. We observed planet-host candidates in K2 campaign fields 5 and 8 (K2-117, K2-146, K2-123, EPIC220187552, EPIC 220194953, K2-149, K2-150, K2-151)on UT 2016 October 28 and 29, and three stars in field10 (K2-152, K2-153, K2-154) on UT 2017 February 21 .Following Alonso-Floriano et al. (2015), we employedthe grism “G-100” setup, covering ∼ − R ∼ >
600 s), we split theexposures into several small ones so that we can min-imize the impact of cosmic rays in the data reduction.For the absolute flux calibration, we observed Feige 34 As we describe in Section 4.2.1, K2-148 (EPIC 220194974)turns out to be the planet host, although at first we misidentifiedEPIC 220194953 to be the host of transiting planets and obtainedthe optical spectrum for EPIC 220194953 with CAFOS.
Hirano et al. no r m a li z ed f l u x wavelength [angstrom]K2-117K2-146EPIC220187552EPIC220194953K2-149K2-150K2-151K2-152K2-153K2-154 Figure 1 . Wavelengh-calibrated, normalized optical spectraobserved by CAFOS. Later M dwarfs are plotted towards thebottom. as a flux standard on each observing night. We did notobserve K2-147 because this target never rises above 25 ◦ elevation at Calar Alto.We reduced the data taken by CAFOS in a stan-dard manner using IRAF packages; bias subtraction,flat-fielding, sky-subtraction, and extraction of one-dimensional (1D) spectra. Wavelength was calibratedusing the revised line list of the comparison lamp (Hg-Cd-Ar) spectrum (Alonso-Floriano et al. 2015). Finally,we corrected the instrumental response and convertedthe flux counts into the absolute fluxes using the ex-tracted 1D spectrum of Feige 34. The data for one ofthe targets, K2-123, were not useful because the signal-to-noise ratio (SNR) of the spectrum turned out to betoo low. Figure 1 plots the reduced, normalized spectraobserved by CAFOS. 3.2. High Dispersion Spectroscopy
In order to estimate stellar physical parameters andcheck binarity, we obtained high resolution optical spec-tra with various spectrographs. K2-117, K2-146, K2-123, K2-147, EPIC 220187552, EPIC 220194953, K2-148, K2-149, K2-150, K2-151, and K2-153 were observedby High Dispersion Spectrograph (HDS; Noguchi et al.2002) on the Subaru 8.2 m telescope between 2015 falland 2017 summer. For all HDS targets except K2-146,we adopted the standard “I2a” setup and Image Slicer ∼ − R ∼ . ′′ R ∼ & − ) caused bystellar companions (i.e., EB scenarios). Except K2-150,the multi-epoch spectra were taken with the iodine (I )cell; the stellar light, transmitted through the cell, is im-printed with the iodine absorption lines which are usedfor the simultaneous precise calibration of wavelength(e.g., Butler et al. 1996). By using the I cell, we canimprove the RV precision by more than tenfold, and cannot only rule out the EB scenario but also put a con-straint on planetary masses, provided that the spectraare obtained at appropriate orbital phases. The onlydrawback is that we need to take one additional I − freespectrum as a template in the RV analysis for each tar-get.Two-dimensional (2D) HDS data in echelle formatwere reduced in the standard manner, including flat-fielding, scattered-light subtraction, and extraction of1D spectra for multiple orders. Wavelength was cal-ibrated based on the Th-Ar emission lamp spectra ob-tained at the beginning and end of each observing night.Typical SNR’s of the resulting 1D spectra were ∼ − cell (K2-123,EPIC 220187552, K2-149, and K2-151), we put the re-duced 1D spectra into the RV analysis pipeline devel-oped by Sato et al. (2002) and extracted relative RVvalues with respect to the I -out template spectrum foreach target. Among the four targets, the RV fit did notconverge for EPIC 220187552, which turns out to be aspectroscopic binary (see Sections 3.3 and 4.1). The re-sults of RV measurements are summarized in Table 1.Figure 2 plots the relative RV variation as a functionof orbital phase of each planet candidate; the absenceof significant RV variations, along with the typical RV alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 Table 1 . Results of RV MeasurementsBJD
TDB
RV RV error RV Type Instrument( − .
0) (km s − ) (km s − ) K2-122 − . − . − . − . − . − . − . − . − . − . K2-123 . − . . K2-147 − . − . K2-149 . . − . K2-150 .
748 0.171 absolute HDS7921.089719 4 .
850 0.339 absolute HDS
K2-151 . − . − . K2-152 − .
153 0.133 absolute Tull7954.629452 − .
643 0.614 absolute Tull precision of 10 −
20 m s − for I − in spectra, completelyrules out the presence of stellar companions in close-inorbits.We performed the RV follow-up observations of K2-122 and K2-147 using the FIbre-fed ´Echelle Spectro-graph (FIES; Frandsen & Lindberg 1999; Telting et al.2014) mounted at the 2.56 m Nordic Optical Telescope(NOT) of Roque de los Muchachos Observatory (LaPalma, Spain). We collected 4 high-resolution spec-tra ( R ∼ , R ∼ , -8.4-8.2-8-7.8-7.6-7.4-7.2-7 -0.4 -0.2 0 0.2 0.4 ab s o l u t e R V [ k m s - ] orbital phaseK2-152b Tull-20-1001020 r e l a t i v e R V [ m s - ] K2-151b HDS4.44.54.64.74.84.955.15.25.3 ab s o l u t e R V [ k m s - ] K2-150b HDS-60-50-40-30-20-10010203040 r e l a t i v e R V [ m s - ] K2-149b HDS-24.96-24.95-24.94-24.93-24.92-24.91-24.9-24.89-24.88 K2-147b FIES-40-30-20-10010203040 r e l a t i v e R V [ m s - ] K2-123b HDS-50-40-30-20-1001020304050 r e l a t i v e R V [ m s - ] K2-122b FIESHARPS-N ab s o l u t e R V [ k m s - ] Figure 2 . RV values folded by the orbital period of eachtransiting planet. Relative RV values are plotted for K2-122, K2-123, K2-149, and K2-151, while absolute RV valuesare shown for K2-147, K2-150 and K2-152. Note that for K2-122, the systemic velocity was subtracted from each datasetto take into account the small RV offset between the FIESand HARPS-N datasets.
Hirano et al. part of the observing programs P52-201 (CAT), P52-108 (OPTICON), and P55-019. Three consecutive ex-posures of 900-1200 s were secured to remove cosmicray hits, leading to an SNR of 25-30 per pixel at5800 ˚A. We followed the observing strategy described inBuchhave et al. (2010) and Gandolfi et al. (2013), andtraced the RV intra-exposure drift of the instrument byacquiring long-exposed (T exp = 35 s) Th-Ar spectra im-mediately before and after each observation. The datareduction was performed using standard IRAF and IDLroutines, which include bias subtraction, flat fielding,order tracing and extraction, and wavelength calibra-tion. The RVs were determined by multi-order cross-correlation against a spectrum of the M2-dwarf GJ 411that was observed with the same instrumental set-upsas the two target stars, and for which we adopted anabsolute RV of − .
689 km s − .We also acquired 6 high-resolution spectra ( R ∼ , R = 60 , Kea code(Endl & Cochran 2016) to determine stellar parameters.
Kea is not well suited to derive accurate parameters forcooler stars, but the results showed that both stars arecool ( T eff ∼ High Contrast Imaging
In transit surveys, typical false positives arise frombackground or hierarchical-triple EBs. High resolu-tion imaging is especially useful to constrain back-ground EB scenarios, and thus has intensively been usedfor planet validations (e.g., Dressing et al. 2017b). Tosearch for nearby companions that could be could bethe source of the observed transit-like signal, we con-ducted high resolution imaging using the adaptive-opticssystem (AO188; Hayano et al. 2010) with the High Con-trast Instrument (HiCIAO; Suzuki et al. 2010) for K2-146 and K2-122 and the Infrared Camera and Spectro-graph (IRCS; Kobayashi et al. 2000) for the other sys-tems, both mounted on the Subaru telescope between2015 winter and 2017 summer.For the HiCIAO observation, we adopted the sameobserving scheme as described in Hirano et al. (2016b),except that we employed the angular differential imaging(ADI; Marois et al. 2006) for K2-146. With the three-point dithering and H − band filter, a total of 11 unsat-urated frames after co-addition were obtained with AOfor K2-146, resulting in the total exposure time of 1135s. For K2-122, we obtained three saturated frames (af-ter co-addition) with two-point dithering, correspondingto the total exposure time of 450 s. We also took twounsaturated frames for absolute flux calibration using aneutral-density filter.HiCIAO data were reduced with the ACORNSpipeline developed by Brandt et al. (2013) for the re-moval of biases and correlated noises, hot pixel mask-ing, flat-fielding, and distortion correction. We thenaligned and median-combined the processed frames toobtain the highest contrast image. The resulting fullwidth at half maximum (FWHM) of the combined im-ages were ∼ . ′′
07. We visually inspected the combinedimages for K2-146 and K2-122, and found two neighbor-ing faint companions to the northwest of K2-146. Thebrighter of the two is located 9 . ′′ m H = 6 . . ′′ m H = 7 . m r = 6 . K2 lightcurve, but the optical and near infrared magnitudes im-ply that these cannot produce the deep transit signaldetected for K2-146. We detected no nearby companionin the combined image of K2-122.Regarding IRCS observations, we conducted AOimaging using each target itself as the natural guide forAO with the H − band filter. Adopting the fine sam-pling mode (1 pix = 0 . ′′ alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 ∼
360 s for a m H = 10 mag star. The saturation radii were less than0 . ′′
05 for all frames. As the second sequence, we alsotook unsaturated frames with much shorter exposures,and used these frames for absolute flux calibrations.Following Hirano et al. (2016a), we reduced the rawIRCS data: subtraction of the dark current, flat fielding,and distortion correction, before aligning and median-combining the frames for each target. The combinedimages were respectively generated for saturated andunsaturated frames. We visually checked the combinedsaturated image for each target, in which the field-of-view (FoV) is ∼ ′′ × ′′ . Most importantly, we foundthat EPIC 220187552 consists of two stars of similarmagnitude separated by ∼ . ′′ ∼ ′′ away from EPIC 220187552 with ∆ m H ∼ m H ∼ . . ′′
6. We foundno bright nearby stars in the FoV for the other targets.To estimate the detection limit of faint nearby sourcesin the combined images, we drew 5 σ contrast curve foreach object. To do so, we first convolved the satu-rated images with each convolution radius being halfof FWHM. We then calculated the scatter of the fluxcounts in the narrow anulus as a function of angular sep-aration from the target’s centroid. Finally, we obtainedthe target’s absolute flux by aperture photometry usingthe unsaturated frames for each target with aperturediameter being FWHM, and normalized the flux scat-ter in the anulus by dividing by the photometric valueafter adjusting the exposure times for saturated and un-saturated combined images. Figure 3 displays the 5 σ contrast curves for all objects, along with the 4 ′′ × ′′ combined images of the targets in the insets. Note thatas we show in Section 4.2.1, EPIC 220194953 and K2-148 are imaged in the same frame, but since K2-148 islikely the host of transiting planets, we show the con-tract curve around it.3.4. Follow-up Transit Observations
OAO 188cm/MuSCAT
On 2016 September 20, we conducted a photo-metric follow-up observation of a transit of K2-151b with the Multi-color Simultaneous Camera forstudying Atmospheres of Transiting exoplanets (MuS-CAT; Narita et al. 2015) on the 1.88 m telescope atOkayama Astronomical Observatory (OAO). MuSCATis equipped with three 1k ×
1k CCDs with a pixel scale of0 . ′′
36 pixel − , enabling us to obtain three-band imagessimultaneously through the SDSS 2nd-generation g ′ , r ′ , and z s -band filters. We set the exposure times to 60, 10,and 25 s for the g ′ , r ′ , and z s bands, respectively. Weobserved the target star along with several bright com-parison stars for ∼ ∼ ∼ g ′ , r ′ , and z s bands, respectively, through clear skies.The observed images were dark-subtracted, flat-fielded, and corrected for non-linearlity of each detec-tor. Aperture photometry was performed with a cus-tomized pipeline (Fukui et al. 2011) for the target starand three similar-brightness stars for comparison, oneof which, however, was saturated on the g ′ -band imagesand omitted from the rest of the analysis for this band.The aperture radius for each band was optimized so thatthe apparent dispersion of a relative light curve (a lightcurve of the target star divided by that of the compar-ison stars) was minimized. As a result, the radii of 11,13, and 12 pixels were adopted for the g ′ , r ′ , and z s bands, respectively.3.4.2. IRSF 1.4 m/SIRIUS
On 2016 October 5 UT, we also conducted a follow-up transit observation with the Simultaneous InfraredImager for Unbiased Survey (SIRIUS; Nagayama et al.2003) on the IRSF 1.4 m telescope at South AfricanAstronomical Observatory. SIRIUS is equipped withthree 1k ×
1k HgCdTe detectors with the pixel scale of0 . ′′
45 pixel − , enabling us to take three near-infrared im-ages in J , H , and K s bands simultaneously. Setting theexposure times to 30 s with the dead time of about 8 sfor all bands, we continued the observations for ∼ J -, H -, and K s -band data, respectively. Weapplied aperture photometry for the target and two com-parison stars for all bands. However, we found that thebrighter comparison star was saturated in the H − banddata and was thus useless. With only the fainter com-parison star, we could not achieve a sufficiently highphotometric precision to extract the transit signal, andtherefore we decided to ignore the H -band data from thesubsequent analyses. We selected 9 pixels as the optimalaperture radii for both J and K s band data. DATA ANALYSES AND VALIDATION OFPLANET CANDIDATES4.1.
Estimation of Spectroscopic Parameters
Hirano et al. ∆ m H [ m ag ] angular separation [arcsec]K2-117 012345678910 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-146 2468101214 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-122012345678910 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-123 01234567891011 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-147 01234567891011 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]EPIC 220187552012345678910 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-148 012345678910 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-149 0123456789 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-150012345678910 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-151 012345678910 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-152 0123456789 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-1530123456789 0 0.5 1 1.5 2 2.5 3 3.5 ∆ m H [ m ag ] angular separation [arcsec]K2-154 Figure 3 . 5 σ contrast curves in the H band as a function of angular separation from the centroid for K2 planet-host candidates.The insets display the saturated combined images with FoV of 4 ′′ × ′′ . EPIC 220187552 is clearly a multiple-star system, andwe conclude that the candidate is a false positive. alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 Table 2 . Spectral Indices by CAFOS Spectroscopy.Star TiO 2 TiO 5 PC1 VO-7912 Color-MK2-117 0.826 0.662 1.037 0.998 0.752K2-146 0.641 0.423 1.157 1.072 1.045K2-123 1.061 0.998 0.935 0.980 0.556
EPIC 220187552
EPIC 220194953
Spectral Types
Based on the low resolution spectra obtained byCAFOS, we measured the spectral types (SpT) for thetarget stars. Following Alonso-Floriano et al. (2015),we measured a suite of (31) spectral indices for eachCAFOS spectrum. Alonso-Floriano et al. (2015) foundthat five indices (TiO 2, TiO 5, PC1, VO-7912, andColor-M) amongst all have the best correlations withSpT and thus we converted each of the measured fiveindices listed in Table 2 into SpT through the polynomi-als given by Alonso-Floriano et al. (2015), with revisedcoefficients (Alonso-Floriano 2015). We then took theweighted mean of the calculated SpT values to obtainthe final value for each target and round those meanspectral types to the nearest standard subtypes (e.g.,M0.0, M0.5, M1.0, · · · ), which are listed in Table 3. Thescatter of the calculated SpT values from the five indicesfor each object is generally less than 0.5 subtype, whichis comparable to the fiducial measurement error in SpTby the present method. The converted SpT values forK2-117 have a relatively large scatter (standard devia-tion = 0 .
523 subtype), which might be due to passageof clouds or other bad weather conditions.We also checked if the target stars are dwarf starsand not M giants, by inspecting the index “Ratio C”(Kirkpatrick et al. 1991), which is a good indicator ofsurface gravity. As described in Alonso-Floriano et al.(2015), stars with a low surface gravity should have avalue of Ratio C lower than ∼ .
07, but all the targetslisted in Table 3 show higher Ratio C values, by whichwe safely conclude that those stars observed by CAFOSare all M dwarfs.4.1.2.
Atmospheric and Physical Parameters
In order to estimate the precise atmospheric and phys-ical parameters of the target stars, we analyzed high res- olution optical spectra obtained in Section 3.2. We madeuse of
SpecMatch-Emp developed by Yee et al. (2017).
SpecMatch-Emp uses a library of optical high resolutionspectra for hundreds of well-characterized FGKM starscollected by the California Planet Search; it matchesan observed spectrum of unknown propety to librarystars, by which the best-matched spectra and their stel-lar parameters (the effective temperature T eff , stellar ra-dius R s , and metallicity [Fe/H]) are found for the inputspectrum while the RV shift and rotation plus instru-mental line-broadening are simultaneously optimized. SpecMatch-Emp is particularly useful for late-type stars,for which spectral fitting using theoretical models oftenhas large systematics due to imperfection of the molec-ular line list in the visible region.Since
SpecMatch-Emp is developed for optical spec-tra obtained by Keck/HIRES, we converted our spec-tra taken by Subaru/HDS, etc, into the same for-mat as HIRES. To check the validity of applying
SpecMatch-Emp to those spectra taken by other in-struments, for which spectral resolutions and pixel-samplings are slightly different from those of HIRES,we put several spectra collected by Subaru/HDS inthe past campaigns (e.g., Hirano et al. 2014) into
SpecMatch-Emp and compared the outputs with liter-ature values. Consequently, we found that the output T eff , R s , and [Fe/H] are all consistent with the literaturevalues within 2 σ (typically within 1 σ ), and we justifiedthe validity of applying SpecMatch-Emp to our new spec-tra.Inputting our high resolution spectra to
SpecMatch-Emp , we obtained the stellar spectro-scopic parameters. We discarded EPIC 220187552from this analysis, since EPIC 220187552 was foundto be a double (in fact triple) star revealed by the AOimaging (Section 3.3). The output parameters ( T eff , R s , and [Fe/H]) are listed in Table 3. To estimatethe other stellar parameters (i.e., stellar mass M s ,surface gravity log g , and luminosity L s ), we adoptedthe empirical formulas derived by Mann et al. (2015),who gave empirical relations of stellar mass and radiusas a function of the absolute K s − band magnitudeand [Fe/H]. Assuming that SpecMatch-Emp ’s outputparameters follow independent Gaussians with their σ being the errors returned by SpecMatch-Emp , weperformed Monte Carlo simulations and converted T eff , R s , and [Fe/H] into M s , log g , and L s through theabsolute K s − band magnitude. Those estimates arealso summarized in Table 3. In the same table, we alsolist the distance d calculated from the apparent andabsolute K s − band magnitudes.4.1.3. Cross-correlation Analysis Hirano et al.
Table 3 . Stellar Parameters by Optical Low and High Resolution Spectroscopy.EPIC ID K2 ID SpT T eff (K) [Fe/H] (dex) R s ( M ⊙ ) M s ( M ⊙ ) log g (dex) L s ( L ⊙ ) d (pc)211331236 K2-117 M1 .
0V 3676 ± − . ± .
12 0 . ± .
051 0 . ± .
056 4 . ± .
046 0 . ± .
009 100 ± .
0V 3385 ± − . ± .
12 0 . ± .
035 0 . ± .
042 4 . ± .
041 0 . ± .
003 86 ± − ±
70 0 . ± .
12 0 . ± .
061 0 . ± .
061 4 . ± .
051 0 . ± .
017 74 ± − ± − . ± .
12 0 . ± .
059 0 . ± .
060 4 . ± .
049 0 . ± .
016 156 ± − ±
70 0 . ± .
12 0 . ± .
055 0 . ± .
059 4 . ± .
048 0 . ± .
011 88 ± − M0 . − − − − − − − − M0 .
5V 3854 ± − . ± .
12 0 . ± .
058 0 . ± .
059 4 . ± .
049 0 . ± .
014 121 ± − ± − . ± .
12 0 . ± .
063 0 . ± .
061 4 . ± .
051 0 . ± .
022 121 ± .
0V 3745 ±
70 0 . ± .
12 0 . ± .
057 0 . ± .
059 4 . ± .
048 0 . ± .
011 118 ± .
5V 3499 ±
70 0 . ± .
12 0 . ± .
044 0 . ± .
051 4 . ± .
043 0 . ± .
006 110 ± .
5V 3585 ± − . ± .
12 0 . ± .
043 0 . ± .
050 4 . ± .
043 0 . ± .
006 62 . ± . .
0V 3940 ±
70 0 . ± .
12 0 . ± .
063 0 . ± .
061 4 . ± .
051 0 . ± .
019 112 ± .
0V 3720 ± − . ± .
12 0 . ± .
050 0 . ± .
055 4 . ± .
045 0 . ± .
009 126 ± .
0V 3978 ±
70 0 . ± .
12 0 . ± .
065 0 . ± .
061 4 . ± .
052 0 . ± .
021 133 ± In addition to estimating stellar parameters from thehigh resolution spectra, we also analyzed the line pro-file for each target. In the case that a transit-like signalis caused by an eclipsing spectroscopic binary of simi-lar size, we expect to see a secondary line or distortionof the profile in the spectra, depending on the orbitalphase of the binary. Using the cross-correlation tech-nique, we computed the averaged spectral line profilesso that we can check for the presence of line blending.In doing so, we cross-correlated each observed spectrum(without the I cell) with the numerical binary mask(M2 mask; see e.g., Bonfils et al. 2013) developed for theRV analysis of HARPS-like spectrographs. From eachobserved spectrum, we extracted the spectral segmentswhose wavelengths are covered by the binary mask, andcross-correlated each segment with the mask as a func-tion of Doppler shift (RV). We then took a weightedaverage of the cross-correlation profiles to get the nor-malized line profile for each object.Figure 4 displays the line profiles for the observedstars. For the targets with multi-epoch observations, weshow the cross-correlation profiles with the highest SNR.Except EPIC 220187552, all stars exhibit single-lineprofiles, though the cross-correlation continuum looksnoisier for particularly cool stars (K2-146 and K2-150),which is most likely due to the more complicated molec-ular absorption features. EPIC 220187552 clearly showsthe secondary line in the cross-correlation profile, as weexpected from Figure 3; due to the small angular separa-tion ( ∼ . ′′ K2 light curve. Therefore, we concludedthat EPIC 220187552 is a hierarchical triple system, inwhich two stars among the three are an EB. We willrevisit this system in Section 5.From the cross-correlation profile, we also measuredthe absolute RV for each target. Since Subaru/HDS(without the I cell) and McDonald 2.7m/Tull are nei-ther stabilized spectrographs nor do they obtain si-multaneous reference spectra like HARPS/HARPS-N,it is difficult to trace the small wavelength drift dur-ing a night, which prohibits accurate RV measurements.In order to correct for the wavelength drift of eachspectrum, we extracted the spectral segment includ-ing strong telluric absorption lines (primarily 6860 − . − (less than half a pixel for HDS). Regard-ing K2-150 and K2-152, we obtained multiple spectra forabsolute RV measurements, which are plotted in Figure2 as a function of the candidates’ phase; no significantRV variation is seen for both objects.4.2. Light Curve Analysis
Fitting K2 Light Curves alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 no r m a li z ed c r o ss - c o rr e l a t i on radial velocity [km s -1 ]K2-117K2-146K2-122K2-123 K2-147EPIC 220187552EPIC 220194953K2-148K2-149K2-150K2-151K2-152K2-153K2-154 Figure 4 . Averaged and normalized cross-correlations be-tween the observed spectra and M2 binary mask. Cross-correlations based on the HDS, HARPS-N, Tull spectra areshown in blue, green, and red, respectively. The Earth’s mo-tion is corrected and RV value is given with respect to thebarycenter of the solar system.
In order to estimate the most precise parameters ofeach planet candidate, we compared the light curvesfor the same objects produced by three differentpipelines: our own light curves (Section 2.1), ones byVanderburg & Johnson (2014), and ones by EVEREST(Luger et al. 2016, 2017). As a result, we found that forour sample, the EVEREST light curves generally pro-vided the best precision in terms of the scatter of thebaseline flux. We thus used EVEREST light curves toestimate the final transit parameters. For the three tar- gets in K2 field 10, since EVEREST light curves havenot been published yet, we employed the light curves byVanderburg & Johnson (2014).We reduced the light curves in the following steps.First, using the reduced light curve products, we spliteach target’s light curve into segments, each spanning6 − . a/R s ,transit impact parameter b , limb-darkening coefficients u and u for the quadratic law, and planet-to-star ra-dius ratio R p /R s . We fixed the orbital eccentricity at e = 0. In addition to these, we introduced the param-eters describing the flux baseline, for which we adopteda linear function of time, and time of the transit cen-ter T c for each transit (segment). To take into accountthe long cadence of K2 observation, we integrated thetransit model by Ohta et al. (2009) over 29 . χ statistic by Powell’s conjugate direction method (e.g.,Press et al. 1992) to obtain the best-fit values for all theparameters, and fixed the baseline parameters for eachsegment at these values. We then implemented MarkovChain Monte Carlo (MCMC) simulations to estimatethe posterior distribution of the remaining fitting pa-rameters. We imposed Gaussian priors on u + u and u − u based on the theoretical values by Claret et al.(2013); the central values for u and u were derived byinterpolation for each target using the stellar parameterslisted in Table 3, and we adopted the dispersion of Gaus-sians as 0.1. At first we assigned an uncertainty to each K2 data point equal to the observed scatter in neigh-boring flux values, which sometimes led to a very smallor large reduced χ , presumably due to non-stationarynoise. To obtain reasonable uncertainties in the fittedparameter values, we rescaled the flux uncertainties suchthat the reduced χ was equal to unity, before perform-ing the MCMC analysis. We adopted the median, and15.87 and 84.13 percentiles of the marginalized poste-rior distribution as the central value and its ± σ foreach fitting parameter.EPIC 220194953 and K2-148 are separated by ∼ . ′′ Hirano et al. y ( P i x e l s ) l o g ( C o un t s ) r e l a t i v e f l u x time from T c [day]0.99920.99940.99960.999811.00021.0004-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]0.99920.99940.99960.999811.00021.0004-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day] AB C
N E
Figure 5 . EVEREST light curves (left panels and top right panel) produced by different apertures (central panel) for EPIC220194953 and K2-148 (EPIC 220194974). The light curves are folded by the period of K2-148c (= 6 .
92 days). The rightbottom panel shows a high-resolution image with FoV of 15 ′′ × ′′ taken by Subaru/IRCS; the upper right and lower left starscorrespond to EPIC 220194953 and K2-148, respectively. EST light curves for those objects involve at least a partof both stars. In order to identify which of the two starsis the source of transit signals, we analyzed three differ-ent light curves provided by EVEREST: the EVERESTversion 2.0 light curves for K2-148 (EPIC 220194974)(A) and EPIC 220194953 (B), and EVEREST version1.0 light curve for EPIC 220194953 (C). The aperturesused to produce the three light curves are shown in thecentral panel of Figure 5. As a result of analyzing andfitting each light curve folded by the period of K2-148c,we found that light curves based on apertures A andB exhibit similar depths in the folded transits, but theone with aperture C shows a much shallower transit (al-most invisible; Figure 5). Since a significant fraction oflight from K2-148 is missing for aperture C, K2-148 islikely the host of the transiting planet candidates . Wethus performed the further analysis below based on thisassumption. Note that we found a similar trend whenthe light curve was folded by the period of K2-148b, butwith a lower SNR.To estimate the planetary parameters for K2-148bto K2-148d, we need to know the contamination (di-lution) factor from EPIC 220194953 for the photomet-ric aperture we adopt. In doing so, we estimatedthe flux ratio between EPIC 220194953 and K2-148in the Kepler ( Kp ) band by the following procedure We also analyzed our own light curves using customized aper-tures with smaller numbers of pixels, but the transit signals be-came invisible owing to the larger scatter in flux. . Adopting the PHOENIX atmosphere model (BT-SETTL; Allard et al. 2013), we first computed the ab-solute fluxes by integrating the grid PHOENIX spectrafor T eff = 3600 , , , , , , , Kp − band. We then performed a MonteCarlo simulation, in which T eff and R s were randomlyperturbed for both of EPIC 220194953 and K2-148 as-suming Gaussian distributions based on the values inTable 3, and absolute fluxes were interpolated and con-verted into the photon count ratio between the two stars.Consequently, we found the relative flux contributionfrom EPIC 220194953 is 0 . ± .
075 while that of K2-148 is 0 . ± .
075 in the Kp − band.The actual flux contribution from each star dependson which aperture we use. We used aperture A for thelight curve fitting (Figure 5). In order to estimate therelative contributions from EPIC 220194953 and K2-148for this aperture, we summed the total flux counts inthe postage stamp ( N tot ), the counts in the pixels in theupper half of the postage stamp which are “not” in theaperture ( N ), and the counts in the pixels in the lowerhalf of the postage stamp which are not in the aperture( N ). The resulting ratios N /N tot and N /N tot canapproximately be considered as the relative flux ratiosfrom EPIC 220194953 and K2-148 that are not inside the The Kp magnitudes are reported to be 12.856 and 12.975for EPIC 220194953 and K2-148, respectively. However, the K2 pixel image ( Kp − band) and our AO image by IRCS (Fig-ure 5; H − band) both imply that K2-148 is brighter than EPIC220194953, suggesting EPIC 220194953 is a later-type star theK2-148 and the reported Kp magnitudes are inaccurate. alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 . ± . . ± .
077 for EPIC 220194953 and K2-148, re-spectively. In fitting the transit light curve, we took thisdilution factor into account for K2-148.After fitting the light curve segments for each planetcandidate, we obtained the transit parameters summa-rized in Table 4. Figure 6 plots the folded K2 dataaround the transits (black points) along with the best-fitlight curve models (red solid lines) for individual planetcandidates. For K2-117, double transit events, in whichtwo planets transit the host star simultaneously, werepredicted and identified in two light curve segments,and we fitted these segments separately with only T c and baseline coefficients floating freely (Figures 7 and8). Using the optimized T c datasets, we fitted the ob-served T c ’s for each candidate with a linear ephemerisand estimated the orbital period P and transit-centerzero point T c, , which are also listed in Table 4. Wenote that in Figure 6, the data for some of the planetcandidates exhibits a larger scatter in the residuals dur-ing the transits, compared to the data outside of tran-sits. This increased scatter during transits could be as-cribed to spot-crossings for relatively active stars (e.g.,Sanchis-Ojeda & Winn 2011), but the large outliers areprobably the instrumental artifacts and were clippedin the light curve analysis. In order to check the ab-sence/presence of TTVs, we plot the observed minuscalculated ( O − C ) diagrams of T c for each candidatein Figures 9–12. Visual inspection suggests that K2-146exhibits a strong TTV while the other candidates showno clear sign of TTVs. Based on the stellar and tran-sit parameters, we also estimate the planet radius R p ,semi-major axis a , and insolation flux from the host star S , as also shown in Table 4.4.2.2. Fitting Ground-based Transits
Because the transit signals of K2-151b are difficultto detect in the ground-based light curves, not all thetransit parameters can be constrained from these lightcurves alone. We therefore fitted these light curves byfixing a/R s and b at the values determined from the K2 light curves. We also fixed the limb-darkening parame-ters at the theoretical values of ( u , u ) = (0 . , . . , . . , . . , . − . , . g ′ , r ′ , z s , J , and K s bands, respectively. Foreach transit, we fitted the multi-band data simultane-ously by allowing R p /R s for each band and a common T c to be free. In addition, we simultaneously modeledthe baseline systematics adopting a parameterization in-troduced by Fukui et al. (2016), which takes account ofthe second-order extinction effect. The applied function is m t ( t ) = M tr + k + k t t + k c m c ( t ) + Σ k i X i , (1)where m t and m c are the apparent magnitudes of thetarget star and comparison stars, respectively, M tr is atransit model in magnitude scale, t is time, X i is aux-iliary observables such as stellar displacements on thedetectors, sky backgrounds, and FWHM of the stellarPSFs, and k , k t , k c , and k i a re coefficients to be fitted.For the auxiliary observables, we included only the onesthat show apparent correlations with the light curves;the stellar displacements in X direction and sky back-grounds (in magnitude scale) were included for the J -band light curve and none was included for the otherlight curves.To obtain the best estimates and uncertainties of thefree parameters, we performed an MCMC analysis us-ing a custom code (Narita et al. 2013). We first opti-mized the free parameters using the AMOEBA algo-rithm (Press et al. 1992), and rescaled the error barof each data point so that the reduced χ becomesunity. To take into account approximate time-correlatednoises, we further inflated each error bar by a factor β ,which is the ratio of the standard deviation of a binnedresidual light curve to the one expected from the un-binned residual light curve assuming white noises alone(Pont et al. 2006; Winn et al. 2008). We then imple-mented 10 and 50 independent MCMC runs with 10 steps each for the MuSCAT and SIRIUS data, respec-tively, and calculated the median and 16 (84) percentilevalues from the merged posterior distributions of the in-dividual parameters. The resultant values are listed inTable 5 and the systematics-corrected light curves alongwith the best-fit transit models are shown in Figures 13and 14.We note that the detections of these transit signalsare marginal. The χ improvement by the best-fit tran-sit model over a null-transit one ( R p /R s are forced to bezero) for the MuSCAT data is 58.7, to which 6.4, 37.8,and 14.5 are contributed from the g ′ -, r ′ -, and z s -banddata, respectively, corresponding to the 6.5 σ significancegiven the number of additional free parameters of four.In the same way, the χ improvement for the SIRIUSdata is 24.2, to which 15.6 and 6.6 are contributed fromthe J - and K s -band data, respectively, corresponding tothe 4.2 σ significance given the number of additional freeparameters of three. Nevertheless, as discussed below,all the R p /R s values are largely consistent with eachother and all the T c values are well aligned, both sup-porting that these transit detections are positive.Based on the results of the ground-based transit ob-servations, we compare the transit depths in differentbandpasses. In Figure 15, the R p /R s value for each bandis plotted as a function of wavelength. The blue hori-4 Hirano et al. r e l a t i v e f l u x time from T c [day]K2-117b 0.9975 0.998 0.9985 0.999 0.9995 1 1.0005 1.001 1.0015-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-117c 0.995 0.996 0.997 0.998 0.999 1 1.001 1.002-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-146b 0.9993 0.9994 0.9995 0.9996 0.9997 0.9998 0.9999 1 1.0001 1.0002 1.0003-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-122b 0.9975 0.998 0.9985 0.999 0.9995 1 1.0005-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-123b 0.9988 0.999 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004 1.0006-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-147b 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004 1.0006 1.0008-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-148b 0.9986 0.9988 0.999 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004 1.0006 1.0008-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-148c 0.9988 0.999 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004 1.0006 1.0008-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-148d 0.9988 0.999 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004 1.0006 1.0008-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-149b 0.997 0.9975 0.998 0.9985 0.999 0.9995 1 1.0005 1.001-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-150b 0.9986 0.9988 0.999 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004 1.0006-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-151b 0.9975 0.998 0.9985 0.999 0.9995 1 1.0005-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-152b 0.9965 0.997 0.9975 0.998 0.9985 0.999 0.9995 1 1.0005 1.001 1.0015-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-153b 0.9984 0.9986 0.9988 0.999 0.9992 0.9994 0.9996 0.9998 1 1.0002 1.0004-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-154b 0.9985 0.999 0.9995 1 1.0005 1.001-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 r e l a t i v e f l u x time from T c [day]K2-154c Figure 6 . K2 light curves around transits for individual candidates folded by their periods. Possible TTVs are corrected andall the transits are aligned in these light curves. For K2-148, the flux contamination from EPIC 220194953 is taken into accountand the dilution factor is corrected. The best-fit transit curves are shown by the red solid lines. a l i d a t i o n o f M - d w a r f P l an e t s i n K C a m p a i g n F i e l d s Table 4 . Planetary ParametersPlanet FPP P (days) T c, (BJD − a/R s R p /R s R p ( R ⊕ ) a (AU) S ( S ⊕ )K2-117b 4 . × − . ± . . ± . . +0 . − . . +0 . − . . +0 . − . . ± . . ± . < − . ± . . ± . . +2 . − . . +0 . − . . +0 . − . . ± . . ± . < − . ± . . ± . . +0 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +1 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +6 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +1 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +3 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +3 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +6 . − . . +0 . − . . +0 . − . . ± . . ± . < − . ± . . ± . . +3 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +3 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +2 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +5 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +3 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +1 . − . . +0 . − . . +0 . − . . ± . . ± . . × − . ± . . ± . . +2 . − . . +0 . − . . +0 . − . . ± . . ± . Hirano et al. r e l a t i v e f l u x BJD - 2457150
Figure 7 . First double transit event observed for K2-117.The best-fit model is shown by the red solid line. r e l a t i v e f l u x BJD - 2457150
Figure 8 . Seond double transit event observed for K2-117.The best-fit model is shown by the red solid line.
Table 5 . Results of Follow-up Transit Observations for K2-151bandpass R p /R s T c (BJD − (MuSCAT observation) . ± . g ′ . +0 . − . r ′ . +0 . − . z s . +0 . − . (SIRIUS observation) . +0 . − . J . +0 . − . K s . +0 . − . zontal line indicates R p /R s in the Kp band, for whichthe ± σ errors are shown by the blue shaded area. Thetransit depths in the g ′ , r ′ , z s , and K s bands are con-sistent with the K2 result within 2 σ , while the J − bandresult exhibits a moderate disagreement. But as is seenin Figure 13, the J − band light curve seems to sufferfrom a systematic flux variation, which has not beencorrected by our light-curve modeling. A more sophisti-cated light-curve analysis using e.g., Gaussian processes(see e.g., Evans et al. 2015) may be able to settle this -10-8-6-4-2 0 2 4 6 8 10 2310 2320 2330 2340 2350 2360 2370 2380 O - C [ m i n ] BJD - 2454833K2-123b-20-15-10-5 0 5 10 15 20 O - C [ m i n ] K2-122b-20-10 0 10 20 30 40 50 O - C [ m i n ] K2-146b-30-20-10 0 10 20 30 40 O - C [ m i n ] K2-117c-40-30-20-10 0 10 20 30 O - C [ m i n ] K2-117b
Figure 9 . O − C diagrams for mid-transit times for K2 cam-paign field 5 planets. -60-40-20 0 20 40 60 80 2470 2480 2490 2500 2510 2520 2530 2540 2550 O - C [ m i n ] BJD - 2454833K2-147b
Figure 10 . O − C diagram for mid-transit times for K2-147b. issue.In the absence of the follow-up transit observations,we obtained the orbital period as P = 3 . ± . K2 data alone. Our ground-based tran-sit observations were conducted >
180 days after the K2 observation for campaign 8 was over, as shown inFigure 16. These follow-up observations improved theprecision in the orbital period of K2-151b by a factor of >
6. Figure 16 also implies that the mid-transit times alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 -20-15-10-5 0 5 10 15 20 2560 2570 2580 2590 2600 2610 2620 2630 2640K2-150b O - C [ m i n ] BJD - 2454833-20-15-10-5 0 5 10 15 K2-149b O - C [ m i n ] -30-20-10 0 10 20 30 K2-148d O - C [ m i n ] -40-30-20-10 0 10 20 30 40 50 60 K2-148c O - C [ m i n ] -80-60-40-20 0 20 40 60 80 K2-148b O - C [ m i n ] Figure 11 . O − C diagrams for mid-transit times for K2 campaign field 8 planets. observed by K2 are consistent with the follow-up tran-sit observations, and no clear sign of TTV is seen forK2-151b. 4.3. Validating Planets
We used the open source vespa software package(Morton 2015b) to compute the false positive proba-bilities (FPPs) of each planet candidate. Similar toprevious statistical validation frameworks (Torres et al.2011; D´ıaz et al. 2014), vespa relies upon Galaxy modelstellar population simulations to compute the likeli-hoods of both planetary and non-planetary scenar-ios given the observations. In particular, vespa usesthe
TRILEGAL
Galaxy model (Girardi et al. 2005) andconsiders false positive scenarios involving EBs, back-ground EBs (BEBs), as well as hierarchical triple sys-tems (HEBs). vespa models the physical properties ofthe host star taking into account broadband photometryand spectroscopic stellar parameters using isochrones (Morton 2015a), and compares a large number of simu- -20-15-10-5 0 5 10 15 2750 2760 2770 2780 2790 2800 2810 2820 O - C [ m i n ] BJD - 2454833K2-154c-25-20-15-10-5 0 5 10 15 20 O - C [ m i n ] K2-154b-50-40-30-20-10 0 10 20 O - C [ m i n ] K2-153b-10-5 0 5 10 O - C [ m i n ] K2-152b
Figure 12 . O − C diagrams for mid-transit times for K2 campaign field 10 planets. Figure 13 . Ground-based transit observation for K2-151 byOAO/MuSCAT (grey dots). The binned flux data for g ′ − , r ′ − , and z s − bands are shown by the blue circles, green tri-angle, and red squares, respectively. The black solid linesindicate the best-fit transit models for individual bands. Hirano et al.
Figure 14 . Ground-based transit observation for K2-151 byIRSF/SIRIUS (grey dots). The binned flux data for J − , and K s − bands are shown by the dark-red circles, brown trian-gles, respectively. The black solid lines indicate the best-fittransit models for individual bands. R p / R s wavelength [nm]g r z s,2 J K s MuSCATSIRIUS
Figure 15 . Observed R p /R s values of for K2-151b in differ-ent bandpasses. The blue horizontal line and its upper andlower shaded areas indicate R p /R s and its ± σ errors in the K p band. -30-20-10 0 10 20 30 2550 2600 2650 2700 2750 2800 2850 O - C [ m i n ] BJD - 2454833K2-151b K2MuSCATSIRIUS
Figure 16 . O − C diagram for mid-transit times for K2-151b. Ground-based transit observations are shown by thegreen square (MuSCAT) and red triangle (SIRIUS). lated scenarios to the observed phase-folded light curve.Both the size of the photometric aperture and contrastcurve constraints are accounted for in the calculations,as well as any other observed constraints such as themaximum depth of secondary eclipses allowed by thedata. Finally, vespa computes the FPP for a givenplanet candidate as the posterior probability of all non-planetary scenarios.Inputting all available information (e.g., folded K2 light curves, contrast curves from AO imaging, con-straint on the depths of secondary eclipses, and spectro-scopic parameters of the target stars) from our follow-upobservations and analyses, we ran vespa and calculatedFPP for each planet candidate. Table 4 summarizes thusderived FPP for our planet candidates; all the FPP val-ues are well below the fiducial criterion of planet valida-tion (FPP < ∼ ′′ × ′′ . Moreover, the targets were not im-aged at the exact center of the detector, and nearbystars within K2 photometric apertures may be miss-ing in our high resolution images. In order to ensurethat such missing stars are not sources of false posi-tive (i.e., BEBs), we checked the archived catalogs (e.g.,Zacharias et al. 2005; Ahn et al. 2012) to look for faintnearby sources for each target. As a consequence, wefound that K2-146, K2-147, K2-148, and K2-150 havenearby faint stars, which could be inside the K2 photo-metric apertures ( ∼ ′′ × ′′ ) . The delta magnitudesof these nearby stars are larger than ∆ m r = 5 mag, butsmaller than those corresponding to the observed transitdepths. Among the four systems, however, the nearbystars around K2-146, K2-148, and K2-150 are locatedaround the edge of the K2 photometric apertures (sep-aration larger than 10 ′′ ), and so a significant fraction oflight from those faint stars should be missing in the K2 photometry ( > . ′′ m R =6 . . ′′ m R = 6 . ∼ . Here, the faint star around K2-146 is different from thetwo faint sources that we identified in the HiCIAO image. Thefaint nearby source around K2-148 is also different from EPIC220194953. alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 K2 light curves using customized apertures for this ob-ject, which excluded the pixels around those faint stars.This analysis revealed that the transits are indeed repro-duced even after excluding these faint stars, by which weconcluded that K2-147 is the source of transits.Finally, we checked if the stellar densities estimatedvia transit fitting are consistent with the spectroscopi-cally estimated densities, in order to make sure that theplanets are indeed transiting the low-mass host stars.As a result, we found that the stellar densities from thetransit modeling all have super-solar densities, suggest-ing that the planets are transiting low-mass stars, andare in good agreement with spectroscopic values within1 σ except K2-117b, for which the two densities are com-patible within 2 σ . Based on all these facts above as wellas the vespa calculations and absence of large RV vari-ations for a fraction of systems, we conclude that thecandidates in Table 4 are all bona-fide planets . INDIVIDUAL SYSTEMS5.1.
K2-117
The planet candidate K2-117b ( P = 1 .
29 days, R p =2 . R ⊕ ) was first reported by Pope et al. (2016) and re-cently Dressing et al. (2017b) validated this candidatealong with the additional planet K2-117c of similar size( R p = 1 . R ⊕ ), orbiting the same star with P = 5 . O − C dia-gram. The two planets exhibit moderate transit depths( ∼ . K2-146
K2-146 is the coolest star in our sample, for whichwe obtain T eff = 3385 K. Pope et al. (2016) andDressing et al. (2017b) reported that K2-146 hosts amini-Neptune candidate in a 2 . − day orbit with apossible TTV. We have performed a global fit to the K2 light curve allowing every transit center to float freely,and confirmed the TTV as shown in Figure 9. As aresult of inputting the TTV-corrected transit curve to vespa , we were able to validate K2-146b as a bona-fideplanet. The strong TTV ( >
30 minutes) suggests that We note that false positives of an instrumental origin are veryunlikely, since our candidates do not include one whose period isclose to the known periods associated with instrumental artifacts(e.g., the 6-hour rolling motion). the object causing TTV is either a very massive planetor has an orbit very close to the mean motion resonance(MMR), although the detailed TTV modeling is beyondthe scope of this paper.K2-146 also exhibits the deepest transit among oursampled stars, making it a very unique target for atmo-spheric characterizations and TTV modeling by transitfollow-ups from the ground and space. However, thepredicted transit times are now highly uncertain due tothe TTV combined with the long time interval after the K2 observation, and it would be required to cover a longbaseline around predicted transits. Fortunately, K2-146is supposed to be observed by K2 again in the Campaignfield 16, by which we can refine the ephemeris and pos-sibly put a constraint on the object inducing the TTV.K2-146 is very faint in the optical ( m V = 16 . m H = 11 . ∼ . M ⊕ and the corresponding RV semi-amplitude induced bythis planet is ∼ . − .5.3. K2-122
K2-122 is a quite metal-rich early M dwarf ([Fe / H] =0 . ± . R p =1 . R ⊕ , P = 2 .
22 days). Pope et al. (2016) reportedthis system to be a candidate planet-host, which waslater validated by Dressing et al. (2017b). In additionto an independent validation by AO imaging and highresolution spectroscopy, we attempted a measurementof the planet mass. As shown in Figure 2, however, RVsmeasured by FIES and HARPS-N show a small vari-ation. Assuming a circular orbit, we fit the observedRV datasets, for which we find the RV semi-amplitudeof K = − . ± . − . This is consistent with anon-detection, but the 1 σ upper limit of K translates to ≈ . M ⊕ for K2-122b’s mass, suggesting that its com-position may be somewhat similar to that of the Earth.Future monitoring with a greater number of RV pointswould allow for a more robust mass measurement.5.4. K2-123
The detection of a transiting mini-Neptune ( R p =2 . R ⊕ ) was reported around K2-123 by Pope et al.(2016), and Dressing et al. (2017b) later validated thisplanet. We have presented our own observations anddata analysis including the precise RV measurement(Figure 2), and independently validated K2-123b as agenuine planet in a 31 − day orbit.The relatively large orbital distance ( a = 0 .
164 AU)0
Hirano et al. translates to K2-123b’s equilibrium temperature of 325K on the assumption that its Bond albedo is 0.3 ( ∼ Earth’s albedo). Thus, the planet is near the potentialhabitable zone, making it an attractive target for furthercharacterizations. Given the moderate transit depth ( ∼ . K2-147
K2-147 is a metal-rich M dwarf, orbited by a super-Earth with the ultra-short period (USP; ∼
23 hours).No detection has so far been reported for this planet.According to exoplanet.eu , K2-147b is the seventh val-idated USP planet ( P < R p as a func-tion of the orbital period P . We will later discuss thedependence of planetary sizes on insolation flux fromhost stars. 5.6. EPIC 220187552
The transit-like signal was first detected for this targetwith a period of 17 .
09 days and we measured its depthand duration as 0 . .
64 hours. As shown inFigures 3 and 4, however, EPIC 220187552 is comprisedof at least two stars separated by ∼ . ′′
3. The transitcurve is also V-shaped, and the preliminary light-curvefitting preferred a grazing transit. We thus concludethat either of the two stars seen in Figure 3 has an eclips-ing stellar companion (a late M dwarf or a brown dwarf),which is responsible for the relative Doppler shift in thecross-correlation profile (Figure 4). Indeed, as we de-scribed in Section 3.2, multiple spectra were obtained forthis target by Subaru/HDS with the I cell but the RVanalysis did not converge, which is most likely becausethe observed spectra (with the I cell) for RV measure-ment are different in shape from the template (withoutthe I cell), which complicates the fitting procedure.In Figure 4, the two line positions in the cross-correlation profile are separated by ∆RV = 18 km s − .The template spectrum for EPIC 220187552 was takenat JD = 2457676 . φ ∼ .
19 when folded by the period ofEPIC 220187552.01. This phase implies that the leftline (RV ∼
19 km s − ) in the cross-correlation profilecorresponds to the star with a companion (i.e., EB) andright one (RV ∼
37 km s − ) corresponds to the other http://exoplanet.eu/catalog star. Assuming a circular orbit ( e = 0) and the orbitalinclination of 90 ◦ for the EB, we can roughly estimatethe secondary-to-primary mass ratio q via∆RV = 212 . (cid:18) M /M ⊙ P/ day (cid:19) q (1 + q ) sin φ (km s − ) , (2)where M is the mass of the primary star. When weadopt M = 0 . M ⊙ , we obtain ∼ . M ⊙ for the massof the secondary. This would be easily confirmed bytaking additional spectra for the absolute RV measure-ment. EPIC 220187552 provides a good testing bench,where high resolution imaging and/or high dispersionspectroscopy become powerful tools to identify and char-acterize hierarchical triple systems.5.7. EPIC 220194953 and K2-148
As we have seen in Section 4.2.1, K2-148 turnedout to host three planets, whose radii we estimate as1 . R ⊕ , 1 . R ⊕ , and 1 . R ⊕ for the innermost ( P =4 .
38 days), middle ( P = 6 .
92 days), and outermost( P = 9 .
76 days) planets, respectively. In order tosee if EPIC 220194953 and K2-148 are bound to eachother (common proper-motion stars), we checked theproper motions of the two stars and found ( µ α , µ δ ) =( − . ± . − , − . ± . − ) and( − . ± . − , − . ± . − ), for EPIC220194953 and K2-148, respectively (Smart et al. 2013),indicating that the two stars share the same proper mo-tion within the errorbars. The almost identical RV val-ues (Figure 4), along with the same distance (Table 3)to the stars, all imply that EPIC 220194953 and K2-148are bound to each other. The separation of 9 . ′′ ∼ K2-149
K2-149 is a slightly metal-rich early M dwarf, havinga super-Earth ( R p = 1 . R ⊕ ) in a 11-day orbit. The RVmeasurement by Subaru/HDS shows no significant RVvariation, supporting the planetary nature of K2-149b.5.9. K2-150
The validated super-Earth K2-150b is similar to K2-149b in terms of its period ( P = 11 days) and size( R p = 2 . R ⊕ ), except that it is orbiting a cooler hoststar ( T eff = 3499 K). Two absolute RVs were measured alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 ∼ . K2-151
K2-151 is a metal-poor M dwarf hosting a transit-ing small planet with P = 3 .
84 days. The size ofK2-151b ( R p = 1 . R ⊕ ) suggests that it is likely arocky planet. The relative brightness of the host starallowed us to observe the follow-up transits from theground, enabling a considerable improvement in thetransit ephemeris (Section 4.2.2). We also measuredrough RVs, which completely ruled out the EB sce-nario. K2-151 is also a good target for future preciseRV measurements in the near infrared; with m J = 10 . K2-152
The transiting mini-Neptune K2-152 is orbiting thehost M dwarf every 33 days. Assuming the Bond albedoof A B = 0 .
3, we estimate the equilibrium temperatureof K2-152b as T eq = 331 K, putting this planet near thehabitable zone. The host star’s brightness ( m V = 13 . m J = 10 .
96 mag) and moderate transit depth( ∼ . ∼ . M ⊕ ,corresponding to the RV semi-amplitude of K ∼ . − . 5.12. K2-153
We did not obtain multiple spectra for K2-153, whichdoes not allow us to rule out completely the grazing EBscenario. Our HDS spectrum for K2-153, however, wastaken at JD = 2457920 .
857 corresponding to φ ∼ . vespa validation. K2-153 is a slightlymetal-poor, early-to-mid M dwarf orbited by a super-Earth ( R p = 2 . R ⊕ ) with P = 7 . K2-154
We identified and validated two transiting mini-Neptunes ( R p = 2 . R ⊕ and 2 . R ⊕ ) around K2-154, aslightly metal-rich early M dwarf. The orbital periodsare 3.68 and 7.95 days for K2-154b and c, respectively, Table 6 . Revised Spectroscopic Parameters Based on
SpecMatch-Emp
System T eff (K) [Fe/H] (dex) R s ( R ⊙ )K2-3 3799 ± − . ± .
12 0 . ± . ± − . ± .
12 0 . ± . ± − . ± .
12 0 . ± . ±
70 0 . ± .
12 0 . ± . ± − . ± .
12 0 . ± . ± − . ± .
12 0 . ± . ± − . ± .
12 0 . ± . ± − . ± .
12 0 . ± . whose ratio is somewhat close to the 2:1 resonance. Wesearched for TTVs for this system, but found no clearevidence as shown in Figure 12. A longer-term transitfollow-ups with a better T c precision would be required. DISCUSSIONAll together, we have validated 16 planets around 12of the low-mass stars observed by K2 , based on high-resolution imaging and optical spectroscopy. Since thenumber of planets around M dwarfs has been increasingrapidly, thanks to K2 and other projects, it is temptingto investigate the entire ensemble of M-dwarf planets,seeking patterns among their properties. We focus hereon a search for any relationships between planet size,the stellar insolation (the flux received by the planet),and the stellar metallicity. This is because insolationand metallicity are strongly suspected of playing an im-portant role in the formation and evolution of plan-ets, and some possible correlations with planetary ra-dius have already been discussed in the literature (e.g.,Owen & Wu 2013; Buchhave et al. 2014; Dawson et al.2015; Lundkvist et al. 2016).To this end, we created a list of transiting planetsaround M dwarfs based on information in the NASAExoplanet Archive , exoplanet.eu, and exoplanets.org .We restricted our sample to confirmed or validated plan-ets around dwarf stars with T eff ≤ σ . Forthe sake of homogeneity, we adopted the stellar pa-rameters of Mann et al. (2013b,a, 2016a,b, 2017b,a) https://exoplanetarchive.ipac.caltech.edu http://exoplanets.org Hirano et al. N u m be r o f P l ane t s R p [R Earth ]mid-to-late Mearly M
Figure 17 . Histogram of planet radius, for the validated andwell-characterized transiting planets around M dwarfs. Thenumber counts for mid-to-late M dwarfs are shown abovethose for early M dwarfs. for a majority of the
Kepler and K2 stars in oursample, since those were derived based on the same(or similar) observing and reduction schemes. Wealso used the SpecMatch-Emp code to derive ourown versions of the stellar parameters (Table 6),for cases in which high-resolution spectra were avail-able on the ExoFOP website . As noted byYee et al. (2017), the M-dwarf parameters derived bythe SpecMatch-Emp code were calibrated using thesample of Mann et al. (2015), faciliating comparisons.For the other systems, for which high-resolution spec-tra were not available, we adopted the stellar pa-rameters from the literature (Rojas-Ayala et al. 2012;Biddle et al. 2014; Torres et al. 2015; Hartman et al.2015; Berta-Thompson et al. 2015; Hirano et al. 2016a;Dressing et al. 2017a; Martinez et al. 2017; Gillon et al.2017; Dittmann et al. 2017), although no metallicity val-ues were reported by Martinez et al. (2017). Planetradii were estimated based on the revised stellar radiiand the values of R p /R s reported in the literature or bythe Kepler team.We split the sample into (1) planets around earlyM dwarfs (3500-4000 K) and (2) mid-to-late M dwarfs( < . − . R ⊕ ) are found around the later-typestars, in spite of the smaller number of such stars in https://exofop.ipac.caltech.edu -1 early M hostsK2-33bHATS-6bKepler-45b R p [ R E a r t h ] insolation [S Earth ] Figure 18 . Stellar insolation fluxes vs. radii of planetsaround early M dwarfs (3500 K < T eff ≤ K2 (blue squares), and planets from the Kepler pri-mary mission and other surveys (black triangles). The cyanrectangle area is the “hot-super-Earth desert” described byLundkvist et al. (2016). See the text for the upper boundaryof R p (green solid line). our sample. Although no completeness correction hasbeen applied, it is interesting that Figure 17 showsthat both types of stars have deficit of planets with R p = 1 . − . R ⊕ , relative to somewhat smaller orlarger planets. This is consistent with the findings ofFulton et al. (2017) and Van Eylen et al. (2017), basedmainly on solar-type stars, that planets with sizes be-tween 1.5-2 R ⊕ are rarer than somewhat smaller orlarger planets. This paucity has been interpreted as theoutcome of photoevaporation on a population of plan-ets with rocky cores ( ≈ . R ⊕ ) with differing massesof gaseous envelopes and different levels of irradiation(Owen & Wu 2017), or as the outcome of the erosion ofplanetary envelopes by internal heat from cooling rockycores (Ginzburg et al. 2017). The same sort of deficitseen in Figure 17 suggests that the same processes seemto be taking place around M dwarfs.6.1. Insolation Dependence
Figures 18 and 19 display the planet radius as a func-tion of stellar insolation S . In these figures, red circlesrepresent our newly validated planets, blue squares areother K2 planets, and black triangles are planets dis-covered during the primary Kepler prime mission or byground-based surveys. Looking at Figures 18 and 19,we note that an important contribution of K2 has beenthe discovery of relatively large planets ( R p & . R ⊕ ),which were not frequently detected during the Kepler primary mission.Figures 18 and 19 show a lack of larger planets alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 -1 mid-to-late M hosts R p [ R E a r t h ] insolation [S Earth ] Figure 19 . Stellar insolation fluxes vs. radii of planetsaround mid-to-late M dwarfs ( T eff ≤ ( R p & R ⊕ ) in the close proximity of M stars. Thedeficit of close-in planets ( P . R p for each inso-lation bin with its width being 0 . S space .We then estimated the maximum radius for each binby finding the 97 % upper limit of this cumulative dis-tribution. Finally, these upper limits were fitted with alinear function in the log S − R p space. We restricted thisanalysis to close-in planets ( P .
10 days) and excludedhot Jupiters ( R p > R ⊕ ) since they seem to form a dif-ferent population from their smaller counterparts (e.g.,Mazeh et al. 2016).The green line in Figure 18 represents this esti-mated boundary line. The moderate slope of the line( R p /R ⊕ = ( − . ± .
47) log
S/S ⊕ + (8 . ± . R p & R ⊕ ) are missingin the proximity of the host stars. Owen & Wu (2013)showed that close-in low-mass planets are likely to suf-fer significant envelope evaporation due to the X-rayand extreme ultraviolet (EUV) radiation from the hoststar. On the other hand, theoretical works have shownthat the gravitational potential of hot Jupiters is sodeep that the XUV radiation from host stars cannot sig-nificantly strip their envelopes (e.g., Murray-Clay et al. The bin size was set to 0.1 in log S , and thus each bin isoverlapping with the neighboring bins ≈
11 Myr. This suggeststhat K2-33b is actively evaporating, and that its radiuswill shrink significantly over the next 100 Myr. Notethat we did not exclude K2-33b from the analysis todraw the boundary.The cyan rectangles in Figures 18 and 19 depict the“hot-super-Earth” desert discussed by Lundkvist et al.(2016), for close-in planets around solar-type stars (i.e.,2 . R ⊕ < R p < . R ⊕ and S > S ⊕ ). Evidentlythis rectangle is not a good description of the “desert”seen around M dwarfs. Instead, for M dwarfs, the“desert” seems to extend towards much lower insola-tion. Also interesting is that the observed “desert” isshifted toward lower insolation for the mid-to-late Mstars. In Figure 19, we draw a similar upper bound-ary of R p for the mid-to-late M sample by the pur-ple dashed line. The derived slope of this boundary( R p /R ⊕ = ( − . ± .
34) log
S/S ⊕ +(7 . ± . σ . To makethis easier to see, the same green line that was drawn inFigure 18 is also drawn in Figure 19.This result can be understood in the framework ofOwen & Wu (2013), which implies that plotting theplanet radius against the current bolometric insolationis not the most direct way to seek evidence for photoe-vaporation. Envelope evaporation is caused specificallyby X-ray and EUV irradiation from the star, and not bythe bolometric flux. This is especially so for M dwarfsbecause they emit a higher fraction of X-rays relative tothe bolometric flux than solar-type stars. Thus planetsaround M dwarfs should have been eroded more effi-ciently, relative to planets around solar-type stars withthe same level of bolometric insolation. This was shownin Figure 7 of Owen & Wu (2013), wherein the lack oflarge planets extends to smaller bolometric fluxes forlater-type stars. Owen & Wu (2013) also showed thatwhen R p is plotted against the empirically estimated X-ray exposure, the maximum planet size at a given X-rayexposure is approximately the same for all types of hoststars. Although we do not attempt here to reproducethis type of plot, a comparison between Figures 18 and19 does suggest a similar pattern. We note that thispattern is also compatible with the scenario in whichphotoevaporation is responsible for the radius gap (Fig-ure 17), and favors photoevaporation over planetary in-ternal heat as the explanation (Ginzburg et al. 2017),because in the latter case it should be the bolometric4 Hirano et al. luminosity (not the XUV luminosity) that is relevant toatmospheric loss.Another possible mechanism that could lead toa deficiency of close-in planets with large sizes ishigh-eccentricity migration (e.g., Rasio & Ford 1996;Nagasawa & Ida 2011) coupled with the disruption ofplanetary envelopes in the vicinity of the Roche limit(Matsakos & K¨onigl 2016; Giacalone et al. 2017). SinceNeptune-sized planets are often observed to have lowermean densities than Jovian or Earth-sized planets, theirplanetary envelopes should be relatively easy to strip.Mulders et al. (2015b) and Lee & Chiang (2017) sug-gested that the decline of planet occurrence rate of allsizes at shortest orbital distances (
P <
10 days) could bethe result of disk truncation at these orbital distances.Several mechanisms that truncate the planet popula-tions around different types of stars are discussed in theliterature (e.g., Plavchan & Bilinski 2013; Mulders et al.2015b), including tidal halting of migrating planets. Thelack of planets of all sizes at higher insolation level inFigures 18 and 19 may also be consistent this interpre-tation. In this picture, the disk truncation likely hap-pens at ≈ P & R p & R ⊕ ) seemto occur within a relatively narrow range of periods.However, given that the occurrence rate of planets with R p > R ⊕ is known to dwindle dramatically and long-period planets are more affected by detection biases as-sociated with the transit geometry and short span of the K2 monitoring, it is premature to draw any conclusionson those outer planets. Compared to planetary systemsaround solar-type stars, little is known on the formationand evolution of M-dwarf planets, but measurements ofeccentricity for close-in planets and other orbital param-eters (e.g., the stellar obliquity) would help to test allthese hypotheses for M-dwarf planets.6.2. Metallicity Dependence
Stellar metallicity is also known to be related to planetsize in exoplanetary systems (see, e.g., Buchhave et al.2014). It is well known that the occurrence rate of R p [ R E a r t h ] [Fe/H] Figure 20 . Host stars’ metallicities from spectroscopy vs.radii of the planets around early M dwarfs (3500 K < T eff ≤ y − scale is logarithmic. giant planets around solar-type stars depends sensi-tively on [Fe/H] (e.g., Johnson et al. 2010). The occur-rence of Earth and Neptune-sized planets were reportedto be less dependent on metallicity (e.g., Sousa et al.2008; Mayor et al. 2011), although some recent stud-ies have shown that such planets are at least some-what more frequent around metal-rich solar-type stars(e.g., Wang & Fischer 2015). In particular, there isgrowing evidence that small close-in planets ( P < P crit ≈ . R p and[Fe/H] for M-dwarf planets, based on our new mea-surements and the parameters available in the litera-ture. Previously, Schlaufman & Laughlin (2010) founda hint that planet-hosting M dwarfs are preferentiallyfound in the region of the ( m V − m K s ) − M K s dia-gram where one expects metal-rich stars to be located.Rojas-Ayala et al. (2012) also investigated the metal-licity of eight planet-hosting M dwarfs. They foundthat M-dwarf planets appear to be hosted by system-atically metal-rich stars, and that Jovian planet hostsare more metal rich than Neptune-sized planet hosts.Mann et al. (2012), however, found no significant differ-ence in g − r color, a metallicity indicator, between theplanet-candidate cool hosts and other cool stars. Theyascribed the apparently high metallicity of cool planet-host stars reported in the literature to contamination ofthe sample by misidentified giant stars. alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 R p [ R E a r t h ] [Fe/H] Figure 21 . Host stars’ metallicities from spectroscopy vs.radii of the planets around mid-to-late M dwarfs ( T eff ≤ Figures 20 and 21 show the radii of confirmed and vali-dated transiting planets as a function of stellar metallic-ity, for early-M hosts (3500-4000 K) and mid-to-late Mhosts ( < & R ⊕ ) haveonly been found around metal-rich stars ([Fe / H] & . R p . R ⊕ ) around super-solar metallicity stars. How-ever, it must be remembered that these results have notbeen corrected for survey sensitivity. Transit surveyshave a strong bias favoring the detection of short-periodplanets; there may be larger-radius planets that havebeen missed due to their longer periods. It is most sig-nificant that there are no detections of super-Neptuneplanets around metal-poor M dwarfs (the upper left re-gion in both figures), since such large planets are easierto detect than smaller planets.Based on RV mass measurements for small plan-ets around solar-type stars, it has been demonstratedthat the observed maximum planet mass increases withmetallicity (Courcol et al. 2016; Petigura et al. 2017a).A similar trend is seen for planet radius in Figures20 and 21. To compare the previous finding with thedistribution of M-dwarf planets, we draw in Figures20 and 21 the upper envelope by the green solid linecorresponding to Equation (1) of Courcol et al. (2016),where the planet mass is converted into radius assum- ing R p /R ⊕ ∝ ( M p /M ⊕ ) . (Chen & Kipping 2017); allthe planets except hot Jupiters are below this line. Al-though the number of systems plotted is much smallerthan in previous works for solar-type stars, the upperenvelopes of planet radius seem to be pushed towardslower values for coolest stars.Dawson et al. (2015) advanced an explanation for thepaucity of gaseous planets around metal-poor stars.They argued that metal-rich stars possessed protoplan-etary disks with a higher surface density of solids, whichled to more rapid formation of rocky cores with a crit-ical mass ( > M ⊕ ) for gas accretion. If the formationtimescale of critical-mass cores is longer than the disklifetime, gaseous planets are unlikely to form. Althoughtheir argument focused on planets around solar-typestars, Figures 20 and 21 suggest that a similar argumentmight apply to low-mass stars.To be more quantitative, we computed the Pearson’scorrelation coefficient r between R p and [Fe/H]. Wefound r = 0 .
332 and 0 .
689 for early M and mid-to-lateM stars, respectively, corresponding to the p − values of0 . . / H] − R p distribu-tion, we performed a Monte Carlo simulation in which17 systems (the number of mid-to-late M systems) arerandomly selected from the 57 early M dwarfs, and wecomputed the probability that the correlation coefficient r for the subset of 17 systems is higher than 0.689 (theobserved r for the mid-to-late M stars). We found thatits probability is 0 . / H] and R p .Since the envelopes of close-in planets may have beenevaporated (at least to some degree) by X-ray and EUVradiation from the star, we also tried to compute the cor-relation coefficients after removing planets for which theinsolation exceeds 100 times the Earth’s insolation, ap-proximately the minimum value for which Figure 18 sug-gests that shrinkage takes place. We obtained a slightlyhigher correlation coefficient ( r = 0 . p = 0 . R p − [Fe/H] correlation is especially strong forcoolest M dwarfs ( T eff ≤ Hirano et al.
Another relevant factor that affects the [Fe / H] − R p relation is the correlation between the planet period andits host star’s metallicity. Mulders et al. (2016) andDong et al. (2017) have recently shown that stars withclose-in rocky planets ( P <
10 days) are preferentiallyseen around metal-rich stars, and thus the [Fe / H] − R p correlation could be in part affected by the [Fe / H] − P correlation. In order to examine such a correlation forM-dwarf planets, we split the whole sample (both earlyM and mid-to-late M samples) into inner planets ( P <
P > / H] = − . ± . / H] = − . ± . ≈ . σ difference). More planets are needed toconfirm the [Fe / H] − P correlation.Following Buchhave et al. (2014), we also computedthe mean metallicity for our samples. We found theweighted mean metallicity to be [Fe / H] = − . ± . . ± .
017 for mid-to-lateM dwarfs. Schlaufman & Laughlin (2010) noted thatthe mean metallicity of M dwarfs in the solar neigh-borhood is [Fe / H] ≈ − .
17. Therefore, our result alsoindicates that the confirmed/validated planet-hostingM dwarfs have systematically high metallicities, Thedifference in the mean metallicities was also seen byRojas-Ayala et al. (2012), but here we have extendedtheir argument down to lower-mass stars and have useda larger number of well-characterized systems. We note,however, that unknown selection effects and/or differ-ent methodologies for metallicity measurements mayhave introduced biases in the mean metallicities in thetwo samples. Homogeneous measurements for volume-limited samples would be required to draw a firm con-clusion.There is no obvious reason why transit surveys shouldhave a detection bias favoring high stellar metallicity,but there might be some effects. For instance, since Mdwarfs with higher metallicity are more luminous thanlower-metallicity counterparts for a given temperature,it may be somewhat easier to detect planet candidatesand conduct follow-up observations for high-metallicitystars, leading to the validation the transiting planets,as we have done in the present paper. Given that wehave included a variety of transiting planets detected bymany space-based and ground-based surveys, it is notstraightforward to account for any detection biases as-sociated with stellar metallicity. We leave this for futurework. CONCLUSIONS As a part of our K2 follow-up program (e.g.,Sanchis-Ojeda et al. 2015), we have detected tens ofplanet candidates around M dwarfs in K2 campaignfields 5–10, and conducted follow-up observations forcandidate planets around M dwarfs. We have validated16 transiting planets around 12 low-mass stars, out ofwhich 12 are newly validated planets. All the vali-dated planets are relatively small in size (Earth-sized tomini-Neptunes), with periods ranging from 0.96 to 33days. We have also identified a hierarchical triple sys-tem (EPIC 220187552) based on AO imaging and highresolution spectroscopy.We also reviewed the relationships between planetsize, insolation, and metallicity that are emerging fromthe growing sample of M-dwarf planets. The planet-radius distribution suggested the same “gap” at around1.5-2 R ⊕ that was found by Fulton et al. (2017) for alarger sample of mainly solar-type stars. We saw an in-dication of the “desert” of very hot planets larger thanabout 2 R ⊕ , although for the coolest M stars the desertbegins at significantly lower insolation levels than forsolar-type stars. We also confirmed that planets largerthan about 3 R ⊕ are preferentially seen around metal-rich stars ([Fe / H] > the Transiting Exoplanet Survey Satellite ( TESS ; Ricker et al. 2015) will be launched and startthe transit survey in the near future, which would makeit more straightforward to deal with selection biases andextract the true distributions of stellar and planetaryparameters with a larger number of sampled stars. Tocorroborate our findings, homogeneous characterizationsof the systems with and without planets are essential.Some of the new M-dwarf planets offer excellentprospects for further characterization, including Dopplermass measurement with optical or near-infrared spec-troscopy (e.g., Kotani et al. 2014). As discussed above,the sizes of M-dwarf planets show some qualitativetrends similar to those around solar-type stars, but theyalso exhibit quantitatively different dependences on stel-lar insolation and metallicity. Perhaps the mass-radiusrelation for M-dwarf planets will also be seen to bedifferent from that of planets around solar-type stars(Weiss & Marcy 2014). Measurements of orbital eccen-tricity and stellar obliquity could also provide helpfulclues to the processes of planet formation and evolutionaround low-mass stars.This paper is based on data collected at Subaru Tele-scope, which is operated by the National Astronomi- alidation of M-dwarf Planets in K2 Campaign Fields 5 – 10 a ) with theNordic Optical Telescope (NOT), operated on the is-land of La Palma jointly by Denmark, Finland, Iceland,Norway, and Sweden, in the Spanish Observatorio delRoque de los Muchachos (ORM) of the Instituto de As-trof´ısica de Canarias (IAC); b ) with the Italian Telesco-pio Nazionale Galileo (TNG) operated on the island ofLa Palma by the Fundaci´on Galileo Galilei of the INAF(Istituto Nazionale di Astrofisica) at the Spanish Ob-servatorio del Roque de los Muchachos of the Institutode Astrofisica de Canaria. The data analysis was inpart carried out on common use data analysis computersystem at the Astronomy Data Center, ADC, of the Na-tional Astronomical Observatory of Japan. We thankAkito Tajitsu, Joanna Bulger, and Ji Hoon Kim, thesupport astronomers at Subaru, and Jun Hashimoto,Shoya Kamiaka, Yohei Koizumi, and Shota Sasaki fortheir helps to carry out the Subaru observations. Wealso thank Santos Pedraz for carrying out the CAFOSobservations at the Calar Alto observatory. We arevery grateful to the NOT and TNG staff members fortheir unique and superb support during the observa-tions. T.H. is grateful to Samuel Yee for providing in-structions to install SpecMatch-Emp . We are thankfulto Christophe Lovis, who provided the numerical maskfor the spectral cross-correlation analysis. The discus- sions with Eric Gaidos, Hiroyuki Kurokawa, Jose Ca-ballero, Alexis Klutsch, and Kento Masuda were veryfruitful. This work was supported by Japan Society forPromotion of Science (JSPS) KAKENHI Grant Num-ber JP16K17660. D. G. gratefully acknowledges the fi-nancial support of the
Programma Giovani Ricercatori– Rita Levi Montalcini – Rientro dei Cervelli (2012) awarded by the Italian Ministry of Education, Universi-ties and Research (MIUR). The research leading to theseresults has received funding from the European UnionSeventh Framework Programme (FP7/2013-2016) un-der grant agreement No. 312430 (OPTICON). D.M. ac-knowledge financial support from the Universidad Com-plutense de Madrid (UCM), the Spanish Ministry ofEconomy and Competitiveness (MINECO) from projectAYA2016-79425-C3-1-P. I.R. acknowledges support bythe Spanish Ministry of Economy and Competitive-ness (MINECO) and the Fondo Europeo de Desar-rollo Regional (FEDER) through grant ESP2016-80435-C2-1-R, as well as the support of the Generalitat deCatalunya/CERCA programme. We acknowledge thevery significant cultural role and reverence that the sum-mit of Mauna Kea has always had within the indigenouspeople in Hawai’i.
Software:
IRAF (Tody 1986, 1993), ACORNSpipeline (Brandt et al. 2013), SpecMatch-Emp(Yee et al. 2017), EVEREST (Luger et al. 2016,2017), PHOENIX (Allard et al. 2013), vespa (Morton2015b), pyfits (Barrett et al. 2012)REFERENCES
Ahn, C. P., Alexandroff, R., Allende Prieto, C., et al. 2012,ApJS, 203, 21Allard, F., Homeier, D., Freytag, B., et al. 2013, Memorie dellaSocieta Astronomica Italiana Supplementi, 24, 128Alonso-Floriano, F. J. 2015, PhD thesis, Departamento deAstrof´ısica y Ciencias de la Atm´osfera, Facultad de CienciasF´ısicas, Universidad Complutense de Madrid, 28040, Madrid,SpainAlonso-Floriano, F. J., Morales, J. C., Caballero, J. A., et al.2015, A&A, 577, A128Ballard, S., & Johnson, J. A. 2016, ApJ, 816, 66Barrett, P., Hsu, J. C., Hanley, C., et al. 2012, PyFITS: PythonFITS Module, Astrophysics Source Code Library, , ,ascl:1207.009Berta-Thompson, Z. K., Irwin, J., Charbonneau, D., et al. 2015,Nature, 527, 204Biddle, L. I., Pearson, K. A., Crossfield, I. J. M., et al. 2014,MNRAS, 443, 1810Bonfils, X., Delfosse, X., Udry, S., et al. 2013, A&A, 549, A109Brandt, T. D., McElwain, M. W., Turner, E. L., et al. 2013,ApJ, 764, 183Buchhave, L. A., Bakos, G. ´A., Hartman, J. D., et al. 2010, ApJ,720, 1118Buchhave, L. A., Latham, D. W., Johansen, A., et al. 2012,Nature, 486, 375 Buchhave, L. A., Bizzarro, M., Latham, D. W., et al. 2014,Nature, 509, 593Butler, R. P., Marcy, G. W., Williams, E., et al. 1996, PASP,108, 500Chen, J., & Kipping, D. 2017, ApJ, 834, 17Ciardi, D. R., Crossfield, I. J. M., Feinstein, A. D., et al. 2017,ArXiv e-prints, arXiv:1709.10398Claret, A., Hauschildt, P. H., & Witte, S. 2013, A&A, 552, A16Clough, S. A., Shephard, M. W., Mlawer, E. J., et al. 2005,J. Quant. Spec. Radiat. Transf., 91, 233Cosentino, R., Lovis, C., Pepe, F., et al. 2012, in Proc. SPIE,Vol. 8446, Ground-based and Airborne Instrumentation forAstronomy IV, 84461VCourcol, B., Bouchy, F., & Deleuil, M. 2016, MNRAS, 461, 1841Crossfield, I. J. M., Ciardi, D. R., Petigura, E. A., et al. 2016,ApJS, 226, 7Dai, F., Winn, J. N., Yu, L., & Albrecht, S. 2017, AJ, 153, 40Dawson, R. I., Chiang, E., & Lee, E. J. 2015, MNRAS, 453, 1471D´ıaz, R. F., Almenara, J. M., Santerne, A., et al. 2014, MNRAS,441, 983Dittmann, J. A., Irwin, J. M., Charbonneau, D., et al. 2017,Nature, 544, 333Dong, S., Xie, J.-W., Zhou, J.-L., Zheng, Z., & Luo, A. 2017,ArXiv e-prints, arXiv:1706.07807Dressing, C. D., & Charbonneau, D. 2013, ApJ, 767, 95—. 2015, ApJ, 807, 45 Hirano et al.
Dressing, C. D., Newton, E. R., Schlieder, J. E., et al. 2017a,ApJ, 836, 167Dressing, C. D., Vanderburg, A., Schlieder, J. E., et al. 2017b,AJ, 154, 207Endl, M., & Cochran, W. D. 2016, PASP, 128, 094502Evans, T. M., Aigrain, S., Gibson, N., et al. 2015, MNRAS, 451,680Frandsen, S., & Lindberg, B. 1999, in Astrophysics with theNOT, ed. H. Karttunen & V. Piirola, 71Fridlund, M., Gaidos, E., Barrag´an, O., et al. 2017, A&A, 604,A16Fukui, A., Livingston, J., Narita, N., et al. 2016, AJ, 152, 171Fukui, A., Narita, N., Tristram, P. J., et al. 2011, PASJ, 63, 287Fulton, B. J., Petigura, E. A., Howard, A. W., et al. 2017, AJ,154, 109Gandolfi, D., Parviainen, H., Fridlund, M., et al. 2013, A&A,557, A74Gandolfi, D., Barrag´an, O., Hatzes, A. P., et al. 2017, AJ, 154,123Giacalone, S., Matsakos, T., & K¨onigl, A. 2017, AJ, 154, 192Gillon, M., Triaud, A. H. M. J., Demory, B.-O., et al. 2017,Nature, 542, 456Ginzburg, S., Schlichting, H. E., & Sari, R. 2017, ArXiv e-prints,arXiv:1708.01621Girardi, L., Groenewegen, M. A. T., Hatziminaoglou, E., & daCosta, L. 2005, A&A, 436, 895Guenther, E. W., Barrag´an, O., Dai, F., et al. 2017, A&A, 608,A93Hartman, J. D., Bayliss, D., Brahm, R., et al. 2015, AJ, 149, 166Hayano, Y., Takami, H., Oya, S., et al. 2010, in Proc. SPIE, Vol.7736, Adaptive Optics Systems II, 77360NHirano, T., Masuda, K., Sato, B., et al. 2015, ApJ, 799, 9Hirano, T., Sanchis-Ojeda, R., Takeda, Y., et al. 2014, ApJ, 783,9Hirano, T., Fukui, A., Mann, A. W., et al. 2016a, ApJ, 820, 41Hirano, T., Nowak, G., Kuzuhara, M., et al. 2016b, ApJ, 825, 53Howard, A. W., Marcy, G. W., Bryson, S. T., et al. 2012, ApJS,201, 15Howell, S. B., Sobeck, C., Haas, M., et al. 2014, PASP, 126, 398Johnson, J. A., Aller, K. M., Howard, A. W., & Crepp, J. R.2010, PASP, 122, 905Kirkpatrick, J. D., Henry, T. J., & McCarthy, Jr., D. W. 1991,ApJS, 77, 417Kobayashi, N., Tokunaga, A. T., Terada, H., et al. 2000, inProc. SPIE, Vol. 4008, Optical and IR TelescopeInstrumentation and Detectors, ed. M. Iye & A. F. Moorwood,1056–1066Kotani, T., Tamura, M., Suto, H., et al. 2014, in Society ofPhoto-Optical Instrumentation Engineers (SPIE) ConferenceSeries, Vol. 9147, Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, 14Kov´acs, G., Zucker, S., & Mazeh, T. 2002, A&A, 391, 369Lee, E. J., & Chiang, E. 2017, ApJ, 842, 40Luger, R., Agol, E., Kruse, E., et al. 2016, AJ, 152, 100Luger, R., Kruse, E., Foreman-Mackey, D., Agol, E., &Saunders, N. 2017, ArXiv e-prints, arXiv:1702.05488Lundkvist, M. S., Kjeldsen, H., Albrecht, S., et al. 2016, NatureCommunications, 7, 11201Mandel, K., & Agol, E. 2002, ApJ, 580, L171Mann, A. W., Brewer, J. M., Gaidos, E., L´epine, S., & Hilton,E. J. 2013a, AJ, 145, 52Mann, A. W., Feiden, G. A., Gaidos, E., Boyajian, T., & vonBraun, K. 2015, ApJ, 804, 64Mann, A. W., Gaidos, E., & Ansdell, M. 2013b, ApJ, 779, 188Mann, A. W., Gaidos, E., L´epine, S., & Hilton, E. J. 2012, ApJ,753, 90Mann, A. W., Gaidos, E., Mace, G. N., et al. 2016a, ApJ, 818, 46 Mann, A. W., Newton, E. R., Rizzuto, A. C., et al. 2016b, AJ,152, 61Mann, A. W., Dupuy, T., Muirhead, P. S., et al. 2017a, AJ, 153,267Mann, A. W., Gaidos, E., Vanderburg, A., et al. 2017b, AJ, 153,64Mann, A. W., Vanderburg, A., Rizzuto, A. C., et al. 2018, AJ,155, 4Marois, C., Lafreni`ere, D., Doyon, R., Macintosh, B., & Nadeau,D. 2006, ApJ, 641, 556Martinez, A. O., Crossfield, I. J. M., Schlieder, J. E., et al. 2017,ApJ, 837, 72Matsakos, T., & K¨onigl, A. 2016, ApJ, 820, L8Mayor, M., Marmier, M., Lovis, C., et al. 2011, ArXiv e-prints,arXiv:1109.2497Mazeh, T., Holczer, T., & Faigler, S. 2016, A&A, 589, A75Montet, B. T., Morton, T. D., Foreman-Mackey, D., et al. 2015,ApJ, 809, 25Morton, T. D. 2015a, isochrones: Stellar model grid package,Astrophysics Source Code Library, , , ascl:1503.010—. 2015b, VESPA: False positive probabilities calculator,Astrophysics Source Code Library, , , ascl:1503.011Morton, T. D., & Swift, J. 2014, ApJ, 791, 10Mulders, G. D., Pascucci, I., & Apai, D. 2015a, ApJ, 798, 112—. 2015b, ApJ, 814, 130Mulders, G. D., Pascucci, I., Apai, D., Frasca, A., &Molenda- ˙Zakowicz, J. 2016, AJ, 152, 187Murray-Clay, R. A., Chiang, E. I., & Murray, N. 2009, ApJ, 693,23Nagasawa, M., & Ida, S. 2011, ApJ, 742, 72Nagayama, T., Nagashima, C., Nakajima, Y., et al. 2003, inSociety of Photo-Optical Instrumentation Engineers (SPIE)Conference Series, Vol. 4841, Instrument Design andPerformance for Optical/Infrared Ground-based Telescopes,ed. M. Iye & A. F. M. Moorwood, 459–464Narita, N., Nagayama, T., Suenaga, T., et al. 2013, PASJ, 65, 27Narita, N., Fukui, A., Kusakabe, N., et al. 2015, Journal ofAstronomical Telescopes, Instruments, and Systems, 1, 045001Noguchi, K., Aoki, W., Kawanomoto, S., et al. 2002, PASJ, 54,855Ofir, A. 2014, A&A, 561, A138Ohta, Y., Taruya, A., & Suto, Y. 2009, ApJ, 690, 1Owen, J. E., & Wu, Y. 2013, ApJ, 775, 105—. 2017, ApJ, 847, 29Petigura, E. A., Sinukoff, E., Lopez, E. D., et al. 2017a, AJ, 153,142Petigura, E. A., Marcy, G. W., Winn, J. N., et al. 2017b, ArXive-prints, arXiv:1712.04042Plavchan, P., & Bilinski, C. 2013, ApJ, 769, 86Pont, F., Zucker, S., & Queloz, D. 2006, MNRAS, 373, 231Pope, B. J. S., Parviainen, H., & Aigrain, S. 2016, MNRAS, 461,3399Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery,B. P. 1992, Numerical recipes in C. The art of scientificcomputing, ed. Press, W. H., Teukolsky, S. A., Vetterling,W. T., & Flannery, B. P.Quirrenbach, A., Amado, P. J., Caballero, J. A., et al. 2014, inSociety of Photo-Optical Instrumentation Engineers (SPIE)Conference Series, Vol. 9147, Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series, 1Rappaport, S., Sanchis-Ojeda, R., Rogers, L. A., Levine, A., &Winn, J. N. 2013, ApJ, 773, L15Rasio, F. A., & Ford, E. B. 1996, Science, 274, 954Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, Journalof Astronomical Telescopes, Instruments, and Systems, 1,014003 alidation of M-dwarf Planets in K2 Campaign Fields 5 – 1029