K2-66b and K2-106b: Two extremely hot sub-Neptune-size planets with high densities
Evan Sinukoff, Andrew W. Howard, Erik A. Petigura, Benjamin J. Fulton, Ian J. M. Crossfield, Howard Isaacson, Erica Gonzales, Justin R. Crepp, John M. Brewer, Lea Hirsch, Lauren M. Weiss, David R. Ciardi, Joshua E. Schlieder, Bjoern Benneke, Jessie L. Christiansen, Courtney D. Dressing, Brad M. S. Hansen, Heather A. Knutson, Molly Kosiarek, John H. Livingston, Thomas P. Greene, Leslie A. Rogers, Sebastien Lepine
DDraft version May 11, 2017
Preprint typeset using L A TEX style emulateapj v. 01/23/15
K2-66B AND K2-106B: TWO EXTREMELY HOT SUB-NEPTUNE-SIZE PLANETS WITH HIGH DENSITIES
Evan Sinukoff
Andrew W. Howard , Erik A. Petigura , Benjamin J. Fulton , Ian J. M.Crossfield , Howard Isaacson , Erica Gonzales , Justin R. Crepp , John M. Brewer , Lea Hirsch , LaurenM. Weiss , David R. Ciardi , Joshua E. Schlieder , Bjoern Benneke , Jessie L. Christiansen , Courtney D.Dressing , Brad M. S. Hansen , Heather A. Knutson , Molly Kosiarek , John H. Livingston , Thomas P.Greene Leslie A. Rogers Sébastien Lépine Draft version May 11, 2017
ABSTRACTWe report precise mass and density measurements of two extremely hot sub-Neptune-size planetsfrom the K2 mission using radial velocities, K2 photometry, and adaptive optics imaging. K2-66harbors a close-in sub-Neptune-sized ( . +0 . − . R ⊕ ) planet (K2-66b) with a mass of . ± . M ⊕ .Because the star is evolving up the sub-giant branch, K2-66b receives a high level of irradiation,roughly twice the main sequence value. K2-66b may reside within the so-called “photoevaporationdesert”, a domain of planet size and incident flux that is almost completely devoid of planets. Itsmass and radius imply that K2-66b has, at most, a meager envelope fraction (< 5%) and perhaps noenvelope at all, making it one of the largest planets without a significant envelope. K2-106 hosts anultra-short-period planet ( P = 13.7 hrs) that is one of the hottest sub-Neptune-size planets discoveredto date. Its radius ( . +0 . − . R ⊕ ) and mass ( . ± . M ⊕ ) are consistent with a rocky composition,as are all other small ultra-short-period planets with well-measured masses. K2-106 also hosts alarger, longer-period planet ( R p = . +0 . − . R ⊕ , P = 13.3 days) with a mass less than . M ⊕ at99.7% confidence. K2-66b and K2-106b probe planetary physics in extreme radiation environments.Their high densities reflect the challenge of retaining a substantial gas envelope in such extremeenvironments. INTRODUCTION Institute for Astronomy, University of Hawai‘i at M¯anoa,Honolulu, HI 96822, USA Cahill Center for Astrophysics, California Institute of Tech-nology, 1216 East California Boulevard, Pasadena, CA 91125,USA Division of Geological and Planetary Sciences, California In-stitute of Technology, 1255 East California Blvd, Pasadena, CA91125, USA Department of Astronomy & Astrophysics, University of Cal-ifornia Santa Cruz, 1156 High St., Santa Cruz, CA, USA Astronomy Department, University of California, Berkeley,CA, USA Department of Physics, University of Notre Dame, 225Nieuwland Science Hall, Notre Dame, IN, USA Department of Astronomy, Yale University and 260 WhitneyAvenue, New Haven, CT 06511, USA Institut de Recherche sur les Exoplanètes, Dèpartement dePhysique, Universitè de Montrèal, C.P. 6128, Succ. Centre-ville,Montréal, QC H3C 3J7, Canada IPAC-NExScI, Mail Code 100-22, Caltech, 1200 E. Califor-nia Blvd., Pasadena, CA 91125, USA Exoplanets and Stellar Astrophysics Laboratory, NASAGoddard Space Flight Center, Greenbelt, MD 20771, USA Department of Physics & Astronomy and Institute of Geo-physics & Planetary Physics, University of California Los Ange-les, Los Angeles, CA 90095, USA Department of Astronomy, The University of Tokyo, 7-3-1Bunkyo-ku, Tokyo 113-0033, Japan NASA Ames Research Center, Space Science and Astrobi-ology Division, M.S. 245-6, Moffett Field, CA 94035, USA Department of Astronomy & Astrophysics, University ofChicago, 5640 South Ellis Avenue, Chicago, IL 60637, USA Department of Physics and Astronomy, Georgia State Uni-versity, GA, USA NSERC Postgraduate Research Fellow Hubble Fellow NSF Graduate Research Fellow NASA Sagan Fellow
Approximately one third of Sun-like stars host planetsbetween the size of Earth and Neptune (“sub-Neptunes”)with orbital periods P < 100 days (Howard et al. 2012;Fressin et al. 2013; Petigura et al. 2013; Burke et al.2015). Most sub-Neptunes detected to date were discov-ered by the prime Kepler mission (2009–2013). While
Kepler provided a detailed measure of the distribu-tion of planet radii, only a few tens of stars host-ing sub-Neptunes were bright enough for secure mass-measurements by current generation precision radial ve-locity (RV) facilities (e.g. Marcy et al. 2014). Many otherplanets have masses measured from transit timing varia-tions (TTVs, Holman & Murray 2005; Agol et al. 2005),a technique that is limited to compact, multiplanet sys-tems (e.g. Carter et al. 2012; Hadden & Lithwick 2014).Mass and radius measurements yield planet densities,which can be used to infer bulk compositions and probeplanet formation histories. From the dozens of sub-Neptunes with measured densities, bulk compositionaltrends have become apparent. Most notably, the ma-jority of planets smaller than ≈ R ⊕ have primarilyrocky compositions, whereas most larger planets havelower densities, consistent with the presence of extendedenvelopes of H/He and other low-density volatiles (Weiss& Marcy 2014; Marcy et al. 2014; Rogers 2015; Dressinget al. 2015).This overall trend in bulk compositions likely has atemperature dependence, which has yet to be fully ex-plored. The gaseous envelopes of planets at extreme tem-peratures are subjected to photoevaporation by the in-cident radiation from their host stars (e.g. Owen & Wu2013; Lopez & Fortney 2014). Probing planets at ex-treme temperatures is crucial to understand these sculpt-ing effects and the formation histories of planets close a r X i v : . [ a s t r o - ph . E P ] M a y Sinukoffto their host stars. If these planets did form as mini-Neptunes and/or giant planets, studying the masses andcompositions of their remnants provides insight into thenature of the cores of such planets, specifically the mech-anisms that formed them, put them so close to their hoststars, and removed their surrounding envelopes.Recent studies of planet occurrence as a function ofradius and temperature have shed light on the formationand evolution of sub-Neptunes. The prime
Kepler mis-sion revealed that the occurrence of 2–4 R ⊕ planets dropssignificantly at very short orbital periods ( P < 10 days,Howard et al. 2012; Fressin et al. 2013). Moreover, froma study of Kepler planets and planet candidates, includ-ing 157 with astroseismically characterized host stars,Lundkvist et al. (2016) reported a complete absence ofplanets with radii 2.2–3.8 R ⊕ and incident fluxes S inc > S ⊕ . Evolutionary models have explained this gap asa “photoevaporation desert”, because planets in this sizeand temperature regime have their envelopes stripped byphotoevaporation (Owen & Wu 2013; Lopez & Fortney2013). Alternatively, smaller planet cores might form toolate and/or too close to the star to accrete much gas andgrow in size (Lee & Chiang 2016).Another rare sub-class of small planets are those withorbital periods P < ∼
1% ofSun-like stars (Sanchis-Ojeda et al. 2014). While it is un-clear how USPs form and how they end up so close to thestar, there are several observational clues: Systems withUSPs commonly host additional planets, which mighthave played a role in their formation and/or migrationhistories. Moreover, Sanchis-Ojeda et al. (2014) mea-sured a sharp decrease in the occurrence of USPs largerthan ∼ R ⊕ , and a complete lack of USPs > R ⊕ .Lopez (2016) showed that the observed dearth of USPs R p = 2–4 R ⊕ suggests that they formed with water-poorH/He envelopes that were subsequently lost via photoe-vaporation.Bulk density measurements of these two rare types ofsub-Neptunes can reveal whether they are bare cores, orcontain a significant amount of volatiles. Unfortunately,there have been few opportunities to study their com-positions. The few of them discovered in the prime Ke-pler field orbit stars too faint for spectroscopic follow-up.However, in 2014, NASA’s K2 mission began a new chap-ter in the search for planets orbiting bright stars. The Kepler spacecraft has been collecting precise photometryof numerous fields along the ecliptic plane, each for nearlythree continuous months (Howell et al. 2014). With10,000–20,000 stars per campaign, hundreds of transit-ing planet candidates have been discovered (Vanderburget al. 2015; Pope et al. 2016; Barros et al. 2016; Adamset al. 2016a), many of which have been statistically vali-dated or confirmed as planets (Sinukoff et al. 2016; Cross-field et al. 2016). This includes several USPs aroundbright stars amenable to Doppler spectroscopy, includingWASP-47e (Becker et al. 2015; Dai et al. 2015; Sinukoffet al. 2017) and HD 3167b (Vanderburg et al. 2016). K2 also provides an opportunity to probe the compositionsof planets in and at the boundaries of the photoevapora-tion desert.Here we report the first mass and density measure-ments of a planet in the photoevaporation desert as wellas the mass and density of a USP planet in a multi- planet system. K2-66 (EPIC 206153219) is a G1 subgiantstar in K2 Campaign 3 (C3), which hosts a transitingsub-Neptune in the photoevaporation desert. K2-106(EPIC 220674823) is a G-star in K2 Campaign 8 (C8)with two transiting sub-Neptunes, including a USP sub-Neptune (K2-106b). We note that K2-66b was first re-ported as a planet candidate by Vanderburg et al. (2015)and statistically validated by Crossfield et al. (2016).Both K2-106 planets were first reported and statisticallyvalidated by Adams et al. (2016b) as part of the Short-Period Planets Group effort (SuPerPiG).In §2 we describe the methods by which we generatestellar light curves from raw K2 photometry and sum-marize our adaptive optics imaging and Doppler obser-vations. §3 explains our analysis of the resulting lightcurves, AO images, and RV time-series to precisely char-acterize the host stars and determine planet masses andradii. In §4, we present our results, discuss possibleplanet compositions, and place these planets in contextwith other sub-Neptunes. Concluding statements areprovided in §5. OBSERVATIONS
K2 Photometry
NASA’s
Kepler
Telescope collected nearly continuousphotometry of K2-66 from 2014 November 15 – 2015 Jan-uary 23 UT (69 days) as part of K2 Campaign 3. K2-106was observed from 2016 January 04 – 2016 March 23 UT(80 days) as part of K2 Campaign 8. We generated stel-lar light curves from the respective target pixel files usingthe same procedures detailed in Sinukoff et al. (2016) andCrossfield et al. (2016). The same Gaussian process wasused to model and subtract the spacecraft motion from K2 pixel data. We use the same K2-66 light curve pre-sented in Crossfield et al. (2016), so we do not display itin this work. Adaptive Optics Imaging
We observed K2-106 on 2016 August 24 UT with thehigh-contrast adaptive optics (AO) system on the Keck-II telescope using the NIRC2 imaging instrument (PI:Keith Matthews). The images were obtained in the nar-row camera mode using a 3-point dither pattern withnods of (cid:48)(cid:48) in each cardinal direction to remove back-ground light. The K s filter was used for all observations.Conditions were foggy and the star was at airmass 1.2with seeing of . (cid:48)(cid:48) during the observations. Crossfieldet al. (2016) presented NIRC2 adaptive optics imagingof K2-66 obtained by our group, which we do not showhere. The star was found to be single. Moreover, Adamset al. (2016b) presented similar NIRC2 observations ofK2-106, finding no evidence of secondary sources. Radial Velocity Measurements
RV measurements of K2-66 and K2-106 were made us-ing HIRES (Vogt et al. 1994) at the W. M. Keck Obser-vatory. We collected 38 RV measurements of K2-66 from2015 September 20 UT to 2017 January 07 UT and 35RV measurements of K2-106 from 2016 August 12 UTto 2017 January 22 UT. Observations and data reduc-tion followed the usual methods of the California PlanetSearch (CPS; Howard et al. 2010). An iodine cell wasused for each observation as a wavelength calibrator and2-66 and K2-106 3
TABLE 1K2-66 Relative radial velocities, Keck-HIRES
BJD RV [m s − ] Unc. [m s − ] a S HK b a Uncertainties estimated from the dispersion in the radialvelocity measured from 718 chunks. These uncertaintiesdo not include “jitter” which is incorporated as a free pa-rameter during the RV modeling ( σ jit , Table 3). b For three observations, the S HK measurement failed dueto a combination of poor seeing, scattered light, and over-lapping orders at blue wavelengths. These measurementsare listed as N/A. point spread function (PSF) reference. The “C2” decker( . (cid:48)(cid:48) × (cid:48)(cid:48) slit) provided spectral resolution R ≈ ≈
100 and typicallylasted 20 min. K2-106 exposures proceeded until SNR ≈
125 ( ∼
25 min). For each star, a single iodine-free expo-sure was taken at roughly twice the SNR using the “B3”decker ( . (cid:48)(cid:48) × (cid:48)(cid:48) slit). The standard CPS Dopplerpipeline was used to measure RVs (Marcy & Butler 1992;Valenti et al. 1995; Butler et al. 1996; Howard et al.2009). RV measurements are listed in Tables 1 and 2for K2-66 and K2-106, respectively. ANALYSIS
Here we describe the methods used to characterizeplanet host stars and to model our K2 light curves andRV time series. Measured stellar parameters, light curvemodel parameters, and RV model parameters are listed TABLE 2K2-106 Relative radial velocities, Keck-HIRES
BJD RV [m s − ] Unc. [m s − ] a S HK a Uncertainties estimated from the dispersion in the ra-dial velocity measured from 718 chunks. These uncer-tainties do not include “jitter” which is incorporated as afree parameter during the RV modeling ( σ jit , Table 4). in Tables 3 and 4 for K2-66 and K2-106, respectively. Stellar characterization
From the iodine-free HIRES spectra, we measured theeffective temperature ( T eff ), surface gravity ( log g ), andmetallicity ( [ Fe/H ] ) of K2-66 and K2-106, using the up-dated “Spectroscopy Made Easy” (SME) analysis tool de-scribed in Brewer et al. (2016). Previous comparison ofSME results with astroseismic results demonstrated log g values accurate to 0.05 dex (Brewer et al. 2015). Stellarmasses and radii were estimated using the isochrones Python package (Morton 2015), which fit our T eff , log g ,and [ Fe/H ] measurements to a grid of models from theDartmouth Stellar Evolution Database (Dotter et al.2008). Posteriors were sampled using the emcee MarkovChain Monte Carlo (MCMC) package (Foreman-Mackeyet al. 2013). The adopted uncertainties on stellar massand radius correspond to 68.3% (1 σ ) confidence inter-vals of the resulting posterior distributions. For K2-66,we measure a mass M (cid:63) = . ± . M (cid:12) and radius R (cid:63) = . ± . R (cid:12) . These are consistent with the val-ues M (cid:63) = 1.16 ± M (cid:12) , and R (cid:63) = 1.71 ± R (cid:12) reported by Crossfield et al. (2016), who used the Spec-Match algorithm (Petigura 2015) instead of SME. ForK2-106, we measure a mass of . ± . M (cid:12) and radiusof . ± . R (cid:12) . Adams et al. (2016b) measured M (cid:63) =0.93 ± M (cid:12) , which is consistent with our measure- Sinukoffment, but they estimated R (cid:63) = 0.83 ± R (cid:12) , which issmaller than our measurement at the ∼ σ level (seediscussion in §4.2.2).To test for spectroscopic blends, we used the algorithmof Kolbl et al. (2015) to search for multiple sets of stel-lar lines. For both K2-66 and K2-106, we ruled out thepossibility of companions in the . (cid:48)(cid:48) × (cid:48)(cid:48) HIRES slitwith T eff = 3400–6100 K, down to 1% contrast in V andR bands, and ∆ RV > km s − .The magnetic activity of each star was assessed bymeasuring S HK indices using the Ca II H & K spectrallines (Isaacson & Fischer 2010). The S HK measurementsare listed in Tables 1 and 2 for K2-66 and K2-106 respec-tively. The median S HK values from all spectra are . and . . The measured T eff and S HK were convertedinto log R (cid:48) HK values, a metric of the Ca II flux relativeto the photospheric continuum (Middelkoop 1982; Noyeset al. 1984). We measure median log R (cid:48) HK values of − . and − . dex, consistent with magnetically quiet starsfrom the California Planet Search (Isaacson & Fischer2010). For comparison, the Sun ranges from log R (cid:48) HK = − . dex to − . dex over a typical magnetic cycle(Meunier et al. 2010).Our NIRC2 images were processed using a standardflat-field, background subtraction, and image stackingtechniques (e.g Crepp et al. 2012). Figure 1(a) displaysthe final reduced image and angular scale. Both raw andstacked images were examined for companion sources.A speckle to the right of the host star was ruled outas a companion as stacked images in the J-band filtershowed it moving as a function of wavelength. Figure1(b) shows the sensitivity to nearby companions. Con-trast levels reach ∆ K = 7.7 for separations beyond . (cid:48)(cid:48) .Adams et al. (2016b) achieve similar contrast limits fromK-band observations of K2-106, also with Keck/NIRC2AO. Light curve analysis
We fit transit models to the detrended K2-106 lightcurve using the same MCMC analysis described in Cross-field et al. (2016). In brief, our code employs the MarkovChain Monte Carlo (MCMC) package emcee (Foreman-Mackey et al. 2013) and model light curves are generatedusing the Python package
BATMAN (Kreidberg 2015). Themodel parameters are: time of conjunction ( T conj ), or-bital period, eccentricity, inclination, and longitude ofperiastron ( P e , i , and ω ), scaled semimajor axis ( a/R (cid:63) ),ratio of planet radius to stellar radius ( R p /R (cid:63) ), a sin-gle multiplicative offset for the absolute flux level, andquadratic limb-darkening coefficients ( u and u ). Thedetrended K2-106 light curve and fitted transit modelsfor planets b and c are shown in Figure 2 RV Analysis
Methodology
To analyze the RV time-series of K2-66 and K2-106, weused the RV fitting package
RadVel (B. Fulton & E. Pe-tigura, in prep.), which is publicly available on GitHub . RadVel is written in object-oriented Python. It uses afast Kepler equation solver written in C and the affine-invariant sampler (Goodman & Weare 2010) of the emcee https://github.com/California-Planet-Search/radvelhttp://radvel.readthedocs.io/en/master/index.html EPIC 2206748232016 − − ∆ t = 144 secondsNIRC2, K s (a) σ C on t r as t ( ∆ m ) Angular Separation (arcseconds)
EPIC 220674823 (b)
Fig. 1.—
Keck/NIRC2 K s -band adaptive optics imaging of K2-106. (a) Reduced image, showing no evidence of secondary stars. (b) σ contrast limits. package (Foreman-Mackey et al. 2013). RadVel is eas-ily adaptable to a variety of maximum-likelihood fittingand MCMC applications. The standard version allowsfor modeling of multi-planet, multi-instrument RV time-series, and assumes no interaction between planets (e.g.Sinukoff et al. 2017).We adopt the same likelihood function for RV modeling2-66 and K2-106 5
Hours From Mid-Transit N o r m a l i z e d F l u x b Hours From Mid-Transit c
BJD_TBD - 2454833 N o r m a l i z e d F l u x Fig. 2.—
Top : Calibrated K2 photometry for K2-106. Vertical ticks indicate the locations of each planets’ transits. Bottom: Phase-foldedphotometry and best-fit light curves for each of the two planets. as Howard et al. (2014): ln L = − (cid:88) i ( v i − v m ( t i )) (cid:16) σ i + σ (cid:17) + ln (cid:114) π (cid:16) σ i + σ (cid:17) (cid:35) , (1)where v i and σ i are the i th RV measurement and corre-sponding uncertainty, and v m ( t i ) is the Keplerian modelvelocity at time t i . The same RV model parameters areused as MCMC step parameters. Before starting theMCMC exploration, we first use the minimization tech-nique of Powell (1964) to find the maximum-likelihoodmodel. Fifty parallel MCMC chains (“walkers”) are theninitialized by perturbing each of the free parameters fromthe maximum likelihood values by as much as 3%. Aninitial round of MCMC exploration continues until theGelman-Rubin (GR) statistic (Gelman & Rubin 1992)drops below 1.10, at which point the chains are reset.Following this burn-in phase, the remaining chains arekept and the MCMC run proceeds until the GR < T z statistic (Ford 2006) exceeds 1000 for all freeparameters. This ensures that the chains are well-mixedand converged.The adopted basis for our RV model for both K2-66and K2-106 is: { P , T conj , K , γ }, where P is orbitalperiod, T conj is the time of conjunction, K is the RVsemi-amplitude and γ is a constant RV offset. For K2-106, we fit for P , T conj , and K of both planets. Welock the orbital periods and phases at the photometri-cally measured values in Tables 3 and 4. Since the or-bital ephemeris is tightly constrained from photometry,it made no difference whether we fixed the ephemerisor assigned Gaussian priors according to uncertaintieson P and T conj . When testing non-circular orbits, weinclude two additional model parameters, √ e cos ω and √ e sin ω , where e is the orbital eccentricity and ω is thelongitude of periapsis of the star’s orbit. This parameter-ization mitigates the Lucy-Sweeney bias toward non-zeroeccentricity (Lucy & Sweeney 1971; Eastman et al. 2013).We also search for additional bodies at orbital periodsbeyond the duration of RV observations by testing RV models that include a constant acceleration term, d v/ d t (i.e. a linear trend in the RV time series). To assesswhether the addition of eccentricity and constant accel-eration parameters are warranted, we use the BayesianInformation Criterion (BIC). When comparing models,we lock the RV jitter at the values in Tables 3 and 4.In §3.3.4, we discuss our search for additional planetsin these two systems. We found no conclusive evidencefor additional planets. K2-66
After testing several different RV model parameteri-zations for K2-66, we adopt a circular orbit (sinusoidal)model with zero acceleration ( d v/ d t ≡ ). The adoptedRV parameters for K2-66 are listed in Table 3, includ-ing K =7 . ± . m s − . The maximum likelihood RV fitis shown in Figure 3. When the orbital eccentricity isallowed to float, the MCMC fit yields e =0.10 +0 . − . , anda planet mass consistent with the circular orbit model.The change in the BIC is ∆ BIC = BIC ecc − BIC circ = 1.0,which indicates that the fit does not improve enough tojustify the additional free parameters (Kass & Raftery1995). Similarly, introducing d v/ d t as a free parameteryields ∆ BIC = BIC d v/ d t − BIC d v/ d t ≡ = − σ of the adopted value. K2-106
The adopted RV model for K2-106 is the sum of twosinusoids (two circular orbits), with d v/ d t ≡ . The fit-ted RV parameters for K2-106 are listed in Table 4 andthe adopted RV fit is displayed in Figure 4. Overall, thechoice of model did not significantly affect the planetmass measurements — all of the RV models yieldedplanet mass constraints consistent with the adopted val-ues. For planet b, we measure K = . ± . m s − ,for a . σ detection. For planet c, we measure K = . ± . m s − , which is not a reliable detection. From theposterior distribution, we place an upper limit, K < . m s − ( M p < . M ⊕ ) at 99.7% confidence. Due to itsproximity to the host star, the orbit of K2-106b has likelybeen circularized by tidal interactions with the star: We Sinukoffcompute a circularization timescale of ≈ Q = 100 estimated for terrestrial planets inthe Solar System (Goldreich & Soter 1966; Henning et al.2009; Lainey 2016). Nevertheless, we tested a fit to theRV time series in which the eccentricity of planet b wasallowed to float. The MCMC fit yielded e = . +0 . − . ,and a planet mass consistent with the best circular orbitmodel. Moreover, the eccentric model is not statisticallyfavored ( ∆ BIC = 0.1). When the eccentricity of planet cwas allowed to float, the preferred eccentricity was 0.75and the MCMC chains did not converge. Any orbit e (cid:38) d v/ d t as a free parameter, but found this additionalmodel complexity was not statistically warranted ( ∆ BIC= 0.2). Finally, since planet c was not significantly de-tected, we also tried fitting for planet b alone but themeasured mass changes by < σ .There are several possible reasons why we do not de-tect the RV signal of planet c. One possibility is that K is sufficiently small that more data are needed to se-curely detect the planet. Alternatively, stellar activityon the timescale of the planet’s orbital period (13 days)could partially wash out the planet signal. However, our log R (cid:48) HK measurement of − . indicates a magneticallyquiet star. Finally, the star might host additional planetsnot included in our RV model. Search for Additional Planets
We conducted a search for additional planets inboth systems using the planet search algorithm de-scribed in Howard & Fulton (2016), which utilizes atwo-dimensional Keplerian Lomb-Scargle periodogram(2DKLS, O’Toole et al. 2009). The periodogram val-ues represent the difference in χ between an N -planetmodel ( χ N ) and an N +1 planet model ( χ N +1 ) for eachorbital period value. When searching for the first planetin a given system we compare χ for a 1-planet model to χ for a flat line. Figure 5 shows the periodograms for N = 0 and N = 1. We estimate an empirical false alarmprobability (eFAP) for any peaks in the 2DKLS peri-odogram by fitting a log-linear function to a histogramof periodogram values.For K2-66, we find no evidence of additional planet sig-nals in the RV time series. In the N = 0 case, the tallestpeak in the periodogram occurs at 5.1 days, correspond-ing to the known transiting planet K2-66b. For N = 1,which tests the 2-planet hypothesis, the tallest peak is at P = 4.0 days and has eFAP > ≈ σ . Therefore, evenif there is an additional planet at P ≈ N = 0 has a global maximum at the orbitalperiod of K2-106b (0.57 days). The N = 1 periodogramdoes not have any significant peaks — the tallest is at P = 35 days with eFAP > < σ when a3-planet RV model is tested with an initial period guessof 35 days for the third Keplerian. Thus, even if there isan additional planet at P ≈
35 days, it has a negligibleeffect on our mass measurement for K2-106b. RESULTS & DISCUSSION
No Significant Dilution
Our RV detections of K2-66b and K2-106b confirmthat they are bonafide planets. To verify that theplanet radius measurements are accurate, we investigatedthe possibility that the photometric aperture contains ablend of multiple stars. Blends would dilute the transitdepth, causing the planet radius to be underestimated(Ciardi et al. 2015). Figure 6 shows blend constraintsfrom the spectroscopic analysis, AO images, and RVmeasurements. Together, these rule out the presenceof companions that would significantly alter the mea-sured planet radii. Contrasts in the NIRC2-AO band-pass were converted to the Kepler bandpass and to com-panion masses using riJHK photometric calibrations ofKraus & Hillenbrand (2007). A blend with Kepler-bandcontrast ∆ K p (cid:46) ∼
100 AU of K2-66 or K2-106 would have been detectedas a linear trend in the RV time-series and would havebeen detected inside ∼ ∼
10 AU. We note that the plot-ted constraints from RV observations use Equation 1 ofWinn et al. (2010), and conservatively assume dv/dt val-ues equal to the 3- σ upper limits obtained when dv/dt is included as a free model parameter. The only con-ceivable problematic blend that would be undetected isa companion near apastron of a highly eccentric orbit(hence low dv/dt ), at an orbital phase of low projectedseparation (hence undetected in AO images) and witha spectrum similar to that of the primary star (henceundetected spectral lines). However, such a scenario ishighly improbable and we conclude that the likelihood ofa problematic blend is negligibly low. Planetary Bulk Compositions
The derived planet properties for K2-66 and K2-106are listed in Tables 3 and 4 respectively. Figure 7(a)shows the masses and radii of K2-66b and K2-106b alongwith all other planets smaller than 4 R ⊕ , whose massesand radii are each known to better than 50% precision .Here we discuss possible planet bulk compositions. K2-66
For K2-66b, we measure a radius R p = . +0 . − . R ⊕ ,and a mass M p = . ± . M ⊕ , corresponding to bulkdensity ρ p = . ± . g cm − . It is one of the most mas-sive planets between 2 and 3 R ⊕ , and likely has a massiveheavy-element core. The compositions of planets in thisregion of the mass radius diagram are not uniquely de-termined and could be a range of different admixturesof various chemical species including iron, rock, waterand H/He (Rogers & Seager 2010; Valencia et al. 2013). NASA Exoplanet Archive, UT 08 February 2017,http://exoplanetarchive.ipac.caltech.edu R V [ m s - ] a) HIRES
TDB - 245483325025 R e s i d u a l s b) R V [ m s - ] c) P b = 5.07 days K b = 7.4 ± -1 e b = 0.00 Fig. 3.—
Single-planet RV model of K2-66, assuming a circular orbit and adopting the ephemeris from transit fits. a) The RV time-series.Open black circles indicate Keck/HIRES data. The solid blue line corresponds to the most likely model. Note that the orbital parameterslisted in Table 3 are the median values of the posterior distributions. Error bars for each independent dataset include an RV jitter termlisted in Table 3, which are added in quadrature to the measurement uncertainties. b) Residuals to the maximum-likelihood fit. c) TheRV time-series phase folded at the orbital period of K2-66b.
To assess possible compositions, we considered a coupleof different two-layer planet models and in each case weconstrained the mass fraction of each layer.First, we assumed an Earth-composition core (33%iron, 67% rock) surrounded by a solar-composition H/Heenvelope. We used the work of Lopez & Fortney (2014),who started with a sample of 1–20 M ⊕ cores surroundedby H/He envelopes that are 0.1–50% of the total planetmass and recorded the evolution of planet radius and en-velope mass over a range of incident fluxes. Their modelsconsist of planet radii ( R p ) computed over a 4-D grid ofplanet core mass ( M core ), planet envelope mass ( M env ),age, and incident stellar flux ( S inc ), i.e. R p = R p ( M core , M env , age, S inc ). Following Petigura et al. (2017), we in-terpolated this grid to convert our measured M p , R p , S inc , and age into a core mass (envelope mass). Wegenerated probability distributions for core mass frac-tion (CMF) by randomly sampling the posteriors of M p , R p , and S inc , assuming an age of 5 Gyr. Varying the agebetween 3–8 Gyr had negligible effect, which is explainedby the fact that at Gyr ages, there is little dependenceon age as the heating/cooling budget is close to a steadystate value. From the resulting probability distribution, we constrain CMF > M core > M ⊕ at 99.7%confidence (3 σ ). One potential limitation of our methodis that the Lopez & Fortney (2014) models assume theplanet incident flux is constant. However, the luminosityof K2-66 has increased by a factor of ∼ σ lower limit on the CMF changes bya negligible amount, from 0.96 to 0.95. We conclude thatif the planet consists of a H/He envelope atop an Earth-composition core, the envelope is <
5% of the planet’smass and the core is > M ⊕ . If the iron mass fractionis larger (smaller) than that of Earth, then the planetwould need a more (less) extended H/He atmosphere tomaintain the same radius.We also considered a composition of rock (Mg SiO )and water ice. We randomly drew 100,000 planet massesand radii from the posterior distributions, and convertedthem into a rock-mass-fraction (RMF) using Equation 7of Fortney et al. (2007). From the resulting distributionof RMFs, we conclude that if the planet is indeed a mix-ture of rock and water ice, then RMF >
81% at 68.3% Sinukoff R V [ m s - ] a) HIRES
TDB - 245483325025 R e s i d u a l s b) R V [ m s - ] c) P b = 0.57 days K b = 7.1 ± -1 e b = 0.00 R V [ m s - ] d) P c = 13.34 days K c = 1.6 ± -1 e c = 0.00 Fig. 4.—
Two-planet RV model of K2-106, assuming circular orbits and adopting the ephemerides from transit fits. Details are same asFigure 3, with panels c and d showing the phase-folded light curves for planets b and c, after subtracting the signal of the other planet.We do not make a statistically significant measurement of the mass of planet c. confidence (1 σ ). Moreover, the total mass of rock M rock > M ⊕ at 68.3% confidence and the planet is denserthan pure rock at 39% confidence. K2-106
For the USP planet K2-106b, we measure radius, mass,and density R p = . +0 . − . R ⊕ , M p = . ± . M ⊕ , and ρ p = . +4 . − . g cm − . These are consistent with anEarth-like composition. Assuming the planet is a mix-ture of iron and rock, we used Equation 8 of Fortneyet al. (2007) to convert our mass and radius posteriorsinto an iron mass fraction (IMF) probability distribu-tion. The median IMF is 19% with a 1 σ upper limit of33%, consistent with an Earth-like composition. Withan extremely high incident flux of ± S ⊕ , andequilibrium temperature of ± K, K2-106b is thehottest sub-Neptune with a measured density. At suchclose proximity to the star, any volatiles would likelyhave been lost by photoevaporation, leaving a bare ∼ M ⊕ core.The measured radii of planets b and c are larger thanthose reported by Adams et al. (2016b) at the ∼ σ and ∼ σ level respectively. Adams et al. (2016b) mea-sure R p = 1.46 ± R ⊕ for planet b and R p = 2.53 ± R ⊕ for planet c. Adopting their measured radiusfor planet b with our measured mass yields an iron massfraction, IMF = 0.8 ± T eff - R (cid:63) relations of Boyajian et al.(2012), which they used to convert their spectroscopicallymeasured T eff (5590 ±
51 K) into a radius. Equation 8 ofBoyajian et al. (2012) was reported as being a third-orderpolynomial fit to a sample of 33 K–M-dwarfs with pre-cisely measured radii and T eff . Equation 9 was reportedas a second polynomial fit that extends to hotter temper-atures by including the Sun. However, these equationsseem to have been mistakenly swapped — the polynomialcoefficients in Equation 8 belong in Equation 9 and vice-versa. This can be seen by computing R (cid:63) (5778 K) = 1.00and 0.86 R (cid:12) for Equations 8 and 9 respectively. The twoequations diverge as T eff exceeds ∼ R (cid:63) = 0.83 R (cid:12) but would have com-puted R (cid:63) = 0.91 R (cid:12) if they had used Equation 8, whichis consistent with our measurement. Although Equation2-66 and K2-106 9 χ − χ K2-66b
10 100 1000
Period [days] χ − χ (a) χ − χ K2-106b
Period [days] χ − χ
35 days(eFAP > 90%) (b)
Fig. 5.—
Two-dimensional Keplerian Lomb-Scargle periodograms of the measured RV time series of a) K2-66 and b) K2-106. Valueson the vertical axis represent the difference in χ between an N -planet model ( χ N ) and an N +1 planet model ( χ N +1 ) at each period.The tallest peaks in the N = 0 cases (top panels) correspond to the periods of known transiting planets, as labeled. For the N = 1 cases(bottom panels), empirical false alarm probabilities (eFAPs) for the tallest peaks are > T eff (cid:38) R (cid:63) and T eff become significantly age-dependent because ofmain sequence evolution. We encourage the authors ofany studies who have used Equations 8 and 9 of Boya-jian et al. (2012) to verify their results. T. Boyajian hasconfirmed the error and is working to publish an erratum.We note that the T eff and log g measured by Adamset al. (2016b) are higher than our measurements. Ourspectroscopic parameters for K2-106 are derived fromSME, which has been well-validated by asteroseismicallycharacterized stars (Brewer et al. 2015). Nevertheless,even if we run the isochrones Python package assum-ing the T eff , log g , and [ Fe/H ] values from Adams et al.(2016b), we measure stellar parameters M (cid:63) = 0.96 M (cid:12) and R (cid:63) = 0.90 M (cid:12) , which are within our measurementerrors. Photoevaporation Desert
The radius and temperature of K2-66b and K2-106bconstitute the extremes of planet parameter space. Fig-ure 8 shows the radius and incident flux of confirmedplanets from the NASA Exoplanet Archive (NEA). K2-106b ranks among the hottest sub-Neptunes found todate. There is a clear absence of very hot planets largerthan ∼ R ⊕ . Another noticeable feature is that hottergiant planets tend to have larger radii— the reason forwhich is highly debated (see Ginzburg & Sari 2015, andreferences therein). It would be interesting to see if anytrends exist for the larger sub-Neptunes of similar tem-perature. K2-66b occupies the region of parameter space NASA Exoplanet Archive, UT 15 February 2017,http://exoplanetarchive.ipac.caltech.edu found to be completely devoid of planets by Lundkvistet al. (2016) (2.2 R ⊕ < R p < 3.8 R ⊕ , S inc > 650 S ⊕ ),hereafter referred to as the “L16 desert”.We find that seven other planets fall within the L16desert. To assess the reliability of these seven mea-surements, we examined constraints on the host stel-lar parameters from spectroscopic and imaging obser-vations. None of them were asteroseismically char-acterized by Lundkvist et al. (2016). According tothe Exoplanet Follow-up Observing Program (ExoFOP)database , five of these stars (K2-100, Kepler-480,Kepler-536, Kepler-656, and Kepler-1270) have proper-ties constrained from spectroscopy and AO imaging. Oneof these five stars, Kepler-536, has a stellar companionat . (cid:48)(cid:48) separation. The planet in this system wouldbe much larger than 4 R ⊕ if it orbits the companionstar rather than the primary (Law et al. 2014; Furlanet al. 2017) so we deem this measurement unreliable.We consider the planet parameters for the other four sys-tems to be reliable and confirm that planets remain inthe L16 desert when spectroscopic stellar parameters areadopted. For K2-100, we adopt the stellar and planetparameters reported in (Mann et al. 2017). The star is alate F dwarf in the 800 Myr Praesepe Cluster. For Kepler480, Kepler-656, and Kepler-1270, we had previously ob-tained HIRES spectra and used the SpecMatch algorithm(Petigura 2015) to derive T eff , log g , and [ Fe/H ] . Wecomputed stellar masses and radii using the isochrones package (see §3.1). We find that Kepler-480 is an F8dwarf, Kepler-1270 is a K1 subgiant, and K2-656 is ahigh-metallicity G dwarf ( [ Fe/H ] = 0.23 ± https://exofop.ipac.caltech.edu/cfop.php Separation (AU) ∆ K p
10% planet radius error20% planet radius error M a ss [ M fl ] S p e c t r o s c o p y A O I m a g i n g R V AllowedCompanion (a)
Separation (AU) ∆ K p
10% planet radius error
20% planet radius error M a ss [ M fl ] S p e c t r o s c o p y A O I m a g i n g R V AllowedCompanion (b)
Fig. 6.—
Constraints on the presence of other stars in the pho-tometric aperture for (a)
K2-66 and (b)
K2-106, which would di-lute the measure transit depth. The vertical axes show compan-ion brightness contrast and companion mass plotted against or-bital separation. NIRC2 AO imaging excludes companions in thehatched blue region, assuming distances of 400 pc and 250 pc toK2-66 and K2-106, respectively. The dashed red line shows thelimits of our search for secondary lines in the HIRES spectrum.Companions in the hatched green region would induce a linear RVtrend larger than the 3- σ upper limit determined from the RV time-series, assuming a circular, edge-on orbit. The horizontal dottedlines represent companion contrasts at which the dilution of the ob-served transit depths of K2-66b and K2-106b would cause planetradii to be overestimated by 10% and 20%. Together, AO imagingand spectroscopy, and RVs rule out companions that would causesystematic errors of >
10% in planet radius with high confidence(see §4.1 for discussion) points and labeled in Figure 8.We examine whether the five planets in the L16 desertshare common properties that can be linked to their ori-gins. First, we note that none of them are USPs — theyhave orbital periods of 1.3–6.0 days. Moreover, four ofthe five host stars have luminosities
L > L (cid:12) . Basedon these two observations, we speculate that planets inthe L16 desert are 2–4 R ⊕ cores of larger planets thatwere stripped of their gaseous envelopes by means ofphotoevaporation. Such 2–4 R ⊕ cores would have highersurface gravities and orbit further from the star than thesmaller cores of USPs. Therefore, the removal of theirenvelopes by photoevaporation would require stars that TABLE 3K2-66 system parameters
Parameter Value Units
Stellar Parameters V . ± . mag T eff ± K log g . ± . dex [ Fe/H ] − . ± . dex v sin i . ± . km s − M (cid:63) . ± . M (cid:12) R (cid:63) . ± . R (cid:12) Planet b
Transit Model P . ± . days T conj . ± . BJD R p /R (cid:63) . +0 . − . — R (cid:63) /a . +0 . − . — u . ± . — u . ± . — b . ± . — i . +2 . − . deg T . +0 . − . hrs ρ (cid:63), circ . +0 . − . g cm − RV Model (circular orbit assumed) K . ± . m s − Derived Planet Parameters a . ± . au S inc ± S ⊕ T eq ± K R p . +0 . − . R ⊕ M p . ± . M ⊕ ρ p . ± . g cm − Other γ − . ± . m s − σ jit . ± . m s − Note . — S inc = incident flux, T conj = timeof conjunction. T eq = equilibrium temperature,assuming albedo = 0.3 are systematically more luminous than USP hosts, con-sistent with observations. Mass measurements of otherplanets in the L16 desert are needed to test the hypoth-esis that they are cores surrounded by little to no gas.Given that K2-66 is a subgiant star, we consider theevolution of the planet’s irradiance since the star left themain sequence. According to Dartmouth stellar evolu-tion models, a star with mass M (cid:63) = 1.1 M (cid:12) and [ Fe/H ] = 0.05 dex would have had a radius R (cid:63) ≈ R (cid:12) dur-ing its main sequence lifetime and have luminosity L (cid:63) ≈ L (cid:12) . Its current luminosity is ≈ L (cid:12) , mean-ing that the planet incident flux has increased twofold,from ≈
420 to 840 S ⊕ since the main sequence era. Thiswould have boosted the rate of photoevaporation of low-density volatiles in the planet’s envelope. Alternatively,EPIC 206153219 might have formed in a gas-poor disk,preventing it from accumulating much H/He.If K2-66b was stripped of its envelope as the star be-came a subgiant, then the rapid post-main sequence evo-lution explains the lack of known planets similar in sizeand density. Perhaps we are catching a glimpse of aplanet from a population that quickly spirals into theirhost stars as they evolve off the main sequence (e.g.2-66 and K2-106 11 TABLE 4K2-106 system parameters
Parameter Value Units
Stellar Parameters V . ± . mag T eff ± K log g . ± . dex [ Fe/H ] 0 . ± . dex v sin i < . km s − M (cid:63) . ± . M (cid:12) R (cid:63) . ± . R (cid:12) Planet b
Transit Model P . ± . days T conj . ± . BJD R p /R (cid:63) . +0 . − . — R (cid:63) /a . +0 . − . — u . ± . — u . ± . — b . ± . — i . +7 . − . deg T . +0 . − . hrs ρ (cid:63), circ . +0 . − . g cm − RV Model (circular orbit assumed) K . ± . m s − Derived Planet Parameters a . ± . au S inc ± S ⊕ T eq ± K R p . +0 . − . R ⊕ M p . ± . M ⊕ ρ p . +4 . − . g cm − Planet c
Transit Model P . ± . days T conj . ± . BJD R p /R (cid:63) . +0 . − . — R (cid:63) /a . +0 . − . — u . ± . — u . ± . — b . ± . — i . +0 . − . deg T . ± . hrs ρ (cid:63), circ . +0 . − . g cm − RV Model (circular orbit assumed) K . ± . m s − Derived Planet Parameters a . ± . au S inc ± S ⊕ T eq ± K R p . +0 . − . R ⊕ M p . ± . M ⊕ ρ p . ± . g cm − Other γ − . ± . m s − σ jit . ± . m s − Note . — S inc = incident flux, T conj = timeof conjunction. T eq = equilibrium temperature,assuming albedo = 0.3 Mass [Earth Masses] R a d i u s [ E a r t h R a d ii ] i r o n r o c k w a t e r V E U N K2-66bK2-106b (a)
Mass [Earth Masses] R a d i u s [ E a r t h R a d ii ] r o c k i r o n w a t e r K2-106b WASP-47e55 Cnc eK-10bK-78 bCoRoT-7b (b)
Fig. 7.— (a)
Masses and radii of all confirmed planets whosemass and radius are measured to better than 50% (2 σ ) precision(blue triangles). Solar System planets are represented as blacksquares. Red circles indicate our measurements of K2-66b and K2-106b. Dark red squares represent other USP measurements fromthe literature. Green curves show the expected planet mass-radiuscurves for 100% iron, 100% rock (Mg SiO ), and 100% water (ice)compositions according to models by Fortney et al. (2007). (b) Azoomed in look of the top panel. The five well-characterized USPsall have masses and radii consistent with mostly rocky compositionsand little to no gaseous envelopes.
KELT-8b, Fulton et al. 2015). To test this scenario, wecomputed an inspiral time, t inspiral ≈
370 Gyr for K2-66b using Equation 1 of Lai (2012) assuming a nominalreduced tidal quality factor Q (cid:48) (cid:63) = 10 . We conclude thatthe planet is not on the verge of spiraling into its hoststar. Ultra-short-period Planets
Only five other USPs have measured masses and den-sities: 55 Cnc (Fischer et al. 2008; Dawson & Fabrycky2010; Nelson et al. 2014; Demory et al. 2016), CoRoT-7b(Léger et al. 2009; Bruntt et al. 2010; Haywood et al.2014), Kepler-10b (Batalha et al. 2011; Esteves et al.2015), Kepler-78b (Howard et al. 2013; Pepe et al. 2013;Grunblatt et al. 2015), and WASP-47e (Becker et al.2 Sinukoff
Incident Flux Relative to Earth P l a n e t R a d i u s [ E a r t h R a d ii ] K-480bK-1270b K-656bK2-100bK2-66bK2-106b
L16 desert
Fig. 8.—
Radii and incident fluxes of all confirmed planets fromthe NASA Exoplanet Archive. K2-66b and K2-106b are shown inred. The black dashed box encloses the region of parameter spacefound by citetLundkvist16 to completely lack planets, which werefer to as the L16 desert. K2-66b, as well as three other planets(blue) occupy the L16 desert and have host stars characterized byboth spectroscopic and AO observations. Four of these five planetshave host stars with super-solar luminosities. K2-106b is one of thehottest sub-Neptunes found to date. ∼ R ⊕ and ∼ M ⊕ . Perhapsthese planets constitute an upper size and mass limit tothe cores of the larger planets from which they form. If allUSPs have similar rocky compositions, then the observedabsence of USPs > R ⊕ naturally translates to an up-per mass limit. Some sub-Neptune-size planets with P> > M ⊕ (e.g. K2-66b), but thereare no such examples of USPs. More well-characterizedUSPs are needed to reveal their core mass distribution.We note that the three well-characterized USPs with ∼ M ⊕ cores (K2-106b, 55 Cnc e, WASP-47e) havehost stars with super-solar metallicities, whereas two ofthe three well-characterized USPs with masses (cid:46) M ⊕ (Kepler-78b and Kepler-10b) have host stars with sub-solar metallicities. With only six data points, a correla-tion cannot be claimed, but this motivates a more com-plete analysis of all USPs beyond the scope of this study.USPs are unlikely to be remnants of hot-Jupiters.While earlier studies argued that USPs could be the left-over cores of hot-Jupiters that experienced Roche lobeoverflow (RLO, e.g. Valsecchi et al. 2014), simulations byValsecchi et al. (2015) and Jackson et al. (2016) suggestthat RLO of planets with cores (cid:46) M ⊕ would tend toexpand their orbits to P > [ Fe/H ] distribution of USP hoststars is inconsistent with that of hot-Jupiter host stars,and consistent with that of stars hosting hot planets ofNeptune-size or smaller. This suggests the that the ma- jority of USPs are not remnants of hot-Jupiters but couldbe remnants of Neptune- or sub-Neptune-size planets.Five of the six well-characterized USPs have knownplanetary companions. The single exception is Kepler-78b, which orbits an active star, hampering the abilityto detect planets with longer orbital periods. The num-ber of detected companions to USPs is consistent witha 50–100% occurrence rate of additional planets P < 45days, depending on the assumed distribution of mutualinclinations and assuming 100% detection completeness(Adams et al. 2016b).It remains unclear how USPs settle so close to theirhost stars, but the multiplicity of these systems ( P < ∼ M ⊕ USPs withclose neighbors. CONCLUSION
We have measured the masses and densities of twoextremely hot sub-Neptunes, K2-66b and K2-106b.We have characterized their stellar hosts using high-resolution spectroscopy and adaptive optics imaging.The radius of K2-66b, R p = . +0 . − . R ⊕ measured from K2 photometry and mass, M p = . ± . M ⊕ measuredfrom Keck-HIRES RVs are consistent with a mostly rockycomposition and little to no low-density volatiles, mak-ing it one of the densest planets of its size. It is one ofthe few known planets in the “photoevaporation desert"( R p = 2.2–3.8 R ⊕ , S inc ≥ S ⊕ ), and the first suchplanet with a measured mass. These planets tend to or-bit stars more luminous than the Sun, which suggeststhat they might have systematically higher densities dueto increased photoevaporation. The measured radius, R p = . +0 . − . R ⊕ and mass, M p = . ± . M ⊕ ofK2-106b indicate an Earth-like composition, similar tothe four other USPs with measured densities. It is thehottest sub-Neptune with a measured mass, and could2-66 and K2-106 13 TABLE 5Ultra-short-period planets with measured masses.
Name M (cid:63) R (cid:63) [ Fe/H ] P R p M p ρ p N pl References( M (cid:12) ) ( R (cid:12) ) (dex) (days) ( R ⊕ ) ( M ⊕ ) (g cm − )55 Cnc e 0.905 ± ± ± ± ± +0 . − . ± ± ± ± ± ± ± ± − . ± +0 . − . +1 . − . ± ± ± ± +0 . − . +0 . − . +3 . − . ± ± ± ± ± ± . ± .
05 0 . ± .
03 0 . ± . . +0 . − . . ± . . +4 . − . Note . — V05: Valenti & Fischer (2005), V11: von Braun et al. (2011), D16: Demory et al. (2016), L09: Léger et al. (2009), B10: Brunttet al. (2010), H14: Haywood et al. (2014), B11: Batalha et al. (2011), E15: Esteves et al. (2015), S13: Sanchis-Ojeda et al. (2013), H13:Howard et al. (2013), P13: Pepe et al. (2013), B15: Becker et al. (2015), S17: Sinukoff et al. (2017). be the stripped core of a more massive planet. K2-66band K2-106b join the rare class of planets larger than 1.5 R ⊕ with mostly rocky compositions.We thank the many observers who contributed to themeasurements reported here. We gratefully acknowledgethe efforts and dedication of the Keck Observatory staff.We thank Tabetha Boyajian for helpful discussions. Thispaper includes data collected by the K2 mission. Fund-ing for the K2 mission is provided by the NASA Sci-ence Mission directorate. E. S. is supported by a post-graduate scholarship from the Natural Sciences and Engi-neering Research Council of Canada. E. A. P. acknowl-edges support by NASA through a Hubble Fellowshipgrant awarded by the Space Telescope Science Institute,which is operated by the Association of Universities forResearch in Astronomy, Inc., for NASA, under contractNAS 5-26555. B. J. F. was supported by the NationalScience Foundation Graduate Research Fellowship undergrant No. 2014184874. A. W. H. acknowledges support for our K2 team through a NASA Astrophysics DataAnalysis Program grant. A. W. H. and I. J. M. C. ac-knowledge support from the K2 Guest Observer Pro-gram. L. M. W. acknowledges the Trottier Family Foun-dation for their generous support. J. R. C. acknowledgessupport from the Kepler Participating Scientist program(NNX14AB85G). This work was performed [in part] un-der contract with the Jet Propulsion Laboratory (JPL)funded by NASA through the Sagan Fellowship Programexecuted by the NASA Exoplanet Science Institute. Thisresearch has made use of the NASA Exoplanet Archive,which is operated by the California Institute of Technol-ogy, under contract with the National Aeronautics andSpace Administration under the Exoplanet ExplorationProgram. Finally, the authors extend special thanks tothose of Hawai‘ian ancestry on whose sacred mountainof Maunakea we are privileged to be guests. Withouttheir generous hospitality, the Keck observations pre-sented herein would not have been possible.
Facilities:
Kepler, Keck-HIRES.
REFERENCESAdams, E. R., Jackson, B., & Endl, M. 2016a, AJ, 152, 47Adams, E. R., Jackson, B., Endl, M., et al. 2016b, ArXiv e-prints,arXiv:1611.00397Agol, E., Steffen, J., Sari, R., & Clarkson, W. 2005, MNRAS,359, 567Barros, S. C. C., Demangeon, O., & Deleuil, M. 2016, A&A, 594,A100Batalha, N. M., Borucki, W. J., Bryson, S. T., et al. 2011, ApJ,729, 27Becker, J. C., Vanderburg, A., Adams, F. C., Rappaport, S. A., &Schwengeler, H. M. 2015, ApJ, 812, L18Boyajian, T. S., von Braun, K., van Belle, G., et al. 2012, ApJ,757, 112Brewer, J. M., Fischer, D. A., Basu, S., Valenti, J. A., &Piskunov, N. 2015, ApJ, 805, 126Brewer, J. M., Fischer, D. A., Valenti, J. A., & Piskunov, N.2016, ApJS, 225, 32Bruntt, H., Deleuil, M., Fridlund, M., et al. 2010, A&A, 519, A51Burke, C. J., Christiansen, J. L., Mullally, F., et al. 2015, ApJ,809, 8Butler, R. P., Marcy, G. W., Williams, E., et al. 1996, PASP, 108,500Carter, J. A., Agol, E., Chaplin, W. J., et al. 2012, Science, 337,556Ciardi, D. R., Beichman, C. A., Horch, E. P., & Howell, S. B.2015, ApJ, 805, 16Crepp, J. R., Johnson, J. A., Howard, A. W., et al. 2012, ApJ,761, 39Crossfield, I. J. M., Ciardi, D. R., Petigura, E. A., et al. 2016,ApJS, 226, 7Dai, F., Winn, J. N., Arriagada, P., et al. 2015, ApJ, 813, L9 Dawson, R. I., & Fabrycky, D. C. 2010, ApJ, 722, 937Demory, B.-O., Gillon, M., Madhusudhan, N., & Queloz, D. 2016,MNRAS, 455, 2018Dotter, A., Chaboyer, B., Jevremović, D., et al. 2008, ApJS, 178,89Dressing, C. D., Charbonneau, D., Dumusque, X., et al. 2015,ApJ, 800, 135Eastman, J., Gaudi, B. S., & Agol, E. 2013, PASP, 125, 83Esteves, L. J., De Mooij, E. J. W., & Jayawardhana, R. 2015,ApJ, 804, 150Fischer, D. A., Marcy, G. W., Butler, R. P., et al. 2008, ApJ, 675,790Ford, E. B. 2006, ApJ, 642, 505Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J.2013, PASP, 125, 306Fortney, J. J., Marley, M. S., & Barnes, J. W. 2007, ApJ, 659,1661Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ, 766, 81Fulton, B. J., Collins, K. A., Gaudi, B. S., et al. 2015, ApJ, 810,30Furlan, E., Ciardi, D. R., Everett, M. E., et al. 2017, AJ, 153, 71Gelman, A., & Rubin, D. B. 1992, Statist. Sci., 7, 457Ginzburg, S., & Sari, R. 2015, ApJ, 803, 111Goldreich, P., & Soter, S. 1966, Icarus, 5, 375Goodman, J., & Weare, J. 2010, Communications in AppliedMathematics and Computational Science, 5, 65Grunblatt, S. K., Howard, A. W., & Haywood, R. D. 2015,arXiv:1501.00369, arXiv:1501.00369Hadden, S., & Lithwick, Y. 2014, ApJ, 787, 80Hansen, B. M. S., & Zink, J. 2015, MNRAS, 450, 45054 Sinukoff