The K2-ESPRINT Project VI: K2-105 b, a Hot-Neptune around a Metal-rich G-dwarf

1.2 MB·

by  Norio Narita

Publ.Astron.Soc.Japan(2014)00(0),1–11 1
doi:10.1093/pasj/xxx000
The K2-ESPRINT Project VI: K2-105 b,
a Hot-Neptune around a Metal-rich G-dwarf
Norio NARITA1,2,3, Teruyuki HIRANO,4, Akihiko FUKUI5, Yasunori HORI2,3,
7 Fei DAI6, Liang YU6, John LIVINGSTON1, Tsuguru RYU3,7, Grzegorz
1 NOWAK8,9, Masayuki KUZUHARA2,3,4, Yoichi TAKEDA3, Simon ALBRECHT10,
0
2 Tomoyuki KUDO11, Nobuhiko KUSAKABE2,3, Enric PALLE8, Ignasi RIBAS12,
n Motohide TAMURA1,2,3, Vincent VAN EYLEN13 and Joshua N. WINN14
a
J 1DepartmentofAstronomy,TheUniversityofTokyo,7-3-1Hongo,Bunkyo-ku,Tokyo

5 113-0033,Japan
2AstrobiologyCenter,NINS,2-21-1Osawa,Mitaka,Tokyo181-8588,Japan
]
P 3NationalAstronomicalObservatoryofJapan,NINS,2-21-1Osawa,Mitaka,Tokyo181-8588,
E Japan
h. 4DepartmentofEarthandPlanetary Sciences,TokyoInstituteofTechnology,2-12-1
p Ookayama,Meguro-ku,Tokyo152-8551,Japan
o- 5OkayamaAstrophysicalObservatory,NationalAstronomicalObservatoryofJapan,NINS,
r Asakuchi,Okayama719-0232,Japan
t
s 6DepartmentofPhysics,andKavliInstituteforAstrophysicsandSpaceResearch,
a
[ MassachusettsInstituteofTechnology,Cambridge,MA02139,USA
7SOKENDAI(TheGraduateUniversityforAdvancedStudies),2-21-1Osawa,Mitaka,Tokyo
1
181-8588,Japan
v
4 8InstitutodeAstrof´ısicadeCanarias(IAC),38205LaLaguna,Tenerife,Spain
9 9DepartamentodeAstrof´ısica,UniversidaddeLaLaguna(ULL),38206LaLaguna,Tenerife,
2
Spain
1
0 10StellarAstrophysicsCentre,DepartmentofPhysicsandAstronomy,AarhusUniversity,Ny
. Munkegade120,DK-8000AarhusC,Denmark
1
0 11SubaruTelescope,NationalAstronomicalObservatoryofJapan,650NorthAohokuPlace,
7 Hilo,HI96720,USA
1
12InstitutdeCie`nciesdel’Espai(CSIC-IEEC),CarrerdeCanMagrans,CampusUAB,08193
:
v Bellaterra,Spain
i
X 13LeidenObservatory,LeidenUniversity,2333CALeiden,Netherlands
r 14DepartmentofAstrophysicalSciences,PrincetonUniversity,Princeton,NJ08544
a
E-mail:[email protected]
∗
Received2016December2;Accepted2016December29
Abstract
WereportontheconfirmationthatthecandidatetransitsobservedforthestarEPIC211525389
areduetoashort-periodNeptune-sizedplanet. Thehoststar,locatedinK2campaignfield5,
isametal-rich([Fe/H]=0.26±0.05)G-dwarf(Teff=5430±70Kandlogg=4.48±0.09),based
onobservationswiththeHighDispersionSpectrograph(HDS)ontheSubaru8.2mtelescope.
High-spatial resolution AO imaging with HiCIAO on the Subaru telescope excludes faint com-
panionsnearthehoststar,andthefalsepositiveprobability ofthistargetisfoundtobe<10−6
(cid:13)c 2014.AstronomicalSocietyofJapan.
2 PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0
using the open source vespa code. A joint analysis of transit light curves from K2 and addi-
tional ground-based multi-color transitphotometry with MuSCAT on theOkayama1.88m tele-
scopegivestheorbitalperiodofP =8.266902±0.000070daysandconsistenttransitdepthsof
R /R ∼0.035 or (R /R )2 ∼0.0012. The transit depth corresponds to a planetary radius of
p ⋆ p ⋆
R =3.59+0.44 R , indicating that EPIC 211525389 b is a short-period Neptune-sized planet.
p −0.39 ⊕
Radial velocities of the host star, obtained with the Subaru HDS, lead to a 3σ upper limit of
90M (0.00027M ) on the mass of EPIC 211525389 b, confirming its planetary nature. We
⊕ ⊙
expectthisplanet,newlynamedK2-105b,tobethesubjectoffuturestudiestocharacterizeits
mass,atmosphere,spin-orbit(mis)alignment,aswellasinvestigatethepossibilityofadditional
planetsinthesystem.
Keywords:planetsandsatellites: individual(K2-105b=EPIC211525389b)—stars: individual(TYC
807-1019-1=EPIC211525389)—techniques: spectroscopic—techniques: highangularresolution—
techniques:photometric—techniques:radialvelocities
1 Introduction inghotNeptunearoundametal-richG-dwarfEPIC211525389
(TYC 807-1019-1) in K2 campaign field 5. The host star is
Transiting planets are especially valuable targets in exoplanet
studies due to the fact that both their radius and mass can relatively bright (mKp =11.69 mag) with colors of B−V =
0.79 and V −J = 1.38, and is located at the distance of
be determined. Thanks to previous ground-based and space-
∼230 pc according to GAIA parallax (see §3.1). This target
based transit surveys, thousands of transiting exoplanets have
was identified as an interesting candidate planet host by the
been discovered. The K2mission (Howell et al. 2014) is cur-
international collaboration ESPRINT, Equipo de Seguimiento
rently continuing the legacy of Kepler in providing dozens of
de Planetas Rocosos Intepretando sus Transitos, (Sanchis-
interesting transiting exoplanet candidates in each of its suc-
Ojeda et al. 2015; Van Eylen et al. 2016a; Van Eylen et al.
cessive ∼ 80 day observing campaigns in the ecliptic plane.
2016b;Hiranoetal.2016a;Hiranoetal.2016b;Daietal.2016).
Since 2014, K2 has discovered more than 100 new transiting
Althoughthestarwasalsoreportedtobeacandidateplanethost
exoplanets by several follow-up teams (e.g., Crossfield et al.
by Pope et al. (2016) and Barros et al. (2016), we have con-
2015; Sanchis-Ojeda et al. 2015; Lillo-Box et al. 2016; Mann
firmedthe planetary nature of this object for the firsttime via
etal.2016;Crossfieldetal.2016).
highdispersionspectroscopy,high-contrastAOimaging,addi-
The large number of confirmed transiting exoplanets pro-
tionalground-basedtransitphotometry,andradialvelocity(RV)
vide us a unique opportunity to investigate their distribution
measurements.TheplanetisnewlynamedasK2-105b.
in parameter space, such as the Period-Radius (P-R) relation,
the Period-Mass (P-M) relation, and the Mass-Radius (M-R)
relation. While the M-R relation is useful to investigate the
compositionandexistenceofvolatile-richatmosphere(Zeng& Therestofthispaperisorganizedasfollows. Observations
Sasselov 2013), P-R and P-M relations are suggested to pro- of K2, high dispersion spectroscopy, AO imaging, and ad-
vide possible insights into planet formation and the migration ditional transit photometry as well as our reduction methods
ofshort-periodplanets (Beauge´ &Nesvorny´ 2013; Adibekyan are described in §2.1–2.4. We present stellar parameters of
etal.2013;Helledetal.2016;Mazehetal.2016). EPIC211525389(hereafterK2-105)fromhighdispersionspec-
In this regard, short-period Neptune-sized planets are es- troscopy in §3.1, a contrast curve around the host star from
pecially interesting, since such planets occupy a region of AO imaging in §3.2, a joint transit analysis with light curves
parameter space which corresponds to the proposed dearth from K2 and ground-based multi-color transit photometry in
of short-period super-Neptune/sub-Jovian planets. Moreover, §3.3, and RVs and a corresponding upper limit on the mass
since transiting exoplanets allow us to investigate their atmo- of EPIC 211525389 b (hereafter K2-105 b) in §3.4. We con-
spheres via transmission spectroscopy, to measure spin-orbit firmtheplanetarynatureofK2-105bbasedonthemassupper
(mis)alignmentsviatheRossiter-McLaughlineffectordoppler limit and a statistical analysis using the vespa code (Morton
tomography,andtoprobepossiblepresenceofouterplanetsvia 2012; Morton 2015) in §4.1. We report an improved transit
transit timing variations, they will incubate further follow-up ephemeris and a hint of a possible transit timing variation for
sciencecasesandprovideadditionalcluestouncoverformation K2-105bin§4.2. Wediscusstheimportanceofadiscoveryof
andmigrationmechanismsofshort-periodplanets. anewshort-periodhot-Neptunefromatheoreticalpointofview
In this paper, we report the confirmation of a new transit- in§4.3.Finally,wesummarizeourfindingsin§5.
PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0 3
!"" "
non-linearitycorrection,biassubtraction,flatfielding,scattered
light subtraction, aperture extraction, and wavelength calibra-
!"""#
tionusingTh-Arlines,resultinginacalibratedone-dimensional
!"""" spectrum. WealsotookHDSspectrawith thesamesetupand
x with the iodine cell to monitor RVs of K2-105 on 2015 Nov
e flu "!$$$# 26-28,2016Feb2,and2016Oct12-14(UT).
v
ati "!$$$"
el
R 2.3 Subaru8.2mTelescope/HICIAO&AO188
"!$$%#
High contrast, high spatial resolution images of K2-105 were
"!$$%" takenintheHbandwithHiCIAO(Tamuraetal.2006)incom-
bination with AO188 (188 element curvature sensor adaptive
&&)" &&(" &&%" &’""
opticssystem: Hayanoetal.2008)ontheSubarutelescopeon
BJD - 2454900
2015December30(UT).ThefieldofviewofHiCIAOisabout
Fig.1.K2lightcurveofK2-105processedbytheESPRINTpipeline. Red 20′′×20′′. Weusedthetargetitselfasanaturalguidestarfor
barsindicatepositionsoftransits.
AO188,producingthetypicalAO-correctedfull-width-at-half-
maximum (FWHM) of ∼0′.′06. All of our observations were
2 Observations andReductions carried out in siderial tracking mode. We acquired 10 object
frameswithanindividualexposuretimeof5s,usinga9.74%-
2.1 K2PhotometrywiththeESPRINTpipeline
transmittance neutral density (ND) filterthat avoids saturation
K2-105wasobservedinK2campaignfield5from2015April
ofthetargetstar. Thoseunsaturated frameswereusedtocali-
27 to July 10. We obtained the K2 data from the Mikulski
bratethecontrast limit aroundthe target. Inaddition, weper-
Archive for Space Telescopes (MAST). We found small vari-
formedtheobservations without theNDfilterandtook60ob-
ability in the raw light curve with the amplitude of ∼ 0.4 %
ject frames with 15 s exposure, enabling us to search for the
and the period of ∼ 24 days, possibly related with the stel-
faintsourcesaroundthetarget. Thetotalexposuretimewas15
lar rotation. We processed it with the ESPRINT pipeline
min(50s)without(with)theNDfilter.
(Sanchis-Ojeda et al. 2015) to create a detrended light curve
TheACORNSpipeline(seeBrandtetal.2013)wasusedto
(see Figure 1). In brief, we identified the candidate planet
reduce the HiCIAO data as follows. First we remove a char-
EPIC211525389 bwithaBox-Least-Squaresroutine(Kova´cs
acteristicstripebiaspattern(Suzukietal.2010), andthenbad
et al. 2002; Jenkins et al. 2010) using the optimal frequency
pixel and flat-field correction are performed. In order to cor-
sampling described by Ofir (2014). The transit-like dimming
rectHiCIAO’sfielddistortion,wecompareHiCIAOdataofthe
occurredevery∼8.2672478dayswithadepthof∼0.1%. No
M5globularclustertakenduringthesamerunwiththearchival
odd-even differencewas observed inthedimming andno sig-
M5images taken byAdvanced CameraforSurveys(ACS)on
nificant evidence of secondary eclipses were seen, suggesting
Hubble Space Telescope, basedon thesameway as explained
thatthesignalislikelytobecausedbyatransitingplanet. We
inBrandtetal.2013. Finally,theplatescaleiscorrectedtobe
thusaddedthisobjectasoneofourfollow-uptargets.
9.5maspixel 1.
−
2.2 Subaru8.2mTelescope/HDS
2.4 Okayama188cmTelescope/MuSCAT
We observed K2-105 with the High Dispersion Spectrograph
Weobtainedsimultaneousmulti-bandtransitphotometryofthe
(HDS: Noguchi et al. 2002) on the Subaru 8.2m telescope lo-
targeton2016February10(UT)usingMuSCAT(Naritaetal.
catedatthesummitofMaunaKea,Hawaii. Weemployed the
2015b) onthe188cmtelescope attheOkayamaAstrophysical
std-I2a setup covering 4940A˚ – 7590A˚ with the Image Slicer
ObservatoryinJapan.Theskyconditionduringourobservation
#2 (R ∼ 80,000: Tajitsu et al. 2012). We took a template wasfreeofcloudsandmoonlight(2-day-oldmoon),butslightly
spectrum without an iodine cell on 2015 November 28 (UT). hazyduetoyellowdust. MuSCAThasthecapabilityofsimul-
Thetotal exposure timewas 1,500 sand thetypical signal-to- taneousthree-bandimagingwithSloanGen2filters(g2′,r2′,and
noiseratioaround6000A˚ was∼60. ThestandardIRAF1 pro- zs,2)andthreeCCDcameras,eachhaving6.′1×6.′1FOV,with
ceduresforHDSwereapplied,includingoverscansubtraction, apixelscaleofabout0.′′358. Theexposuretimewassetto60
s for g2′ and zs,2 bands, and 20 s for r2′ band. We defocused
1The Image Reduction and Analysis Facility (IRAF) is distributed by the telescope such that the FWHM of the stellar point spread
the National Optical Astronomy Observatory, which is operated by the
AssociationofUniversitiesforResearchinAstronomy(AURA)underaco- function(PSF)waskeptaround24(g2′),29(r2′),and32(zs,2)
operativeagreementwiththeNationalScienceFoundation. pixels, respectively. The observations were conducted during
4 PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0
JD2457428.95-2457429.20. Table1.StellarParametersofK2-105
Parameter Value
The observed images are dark-subtracted, flat-fielded, and
(StellarParameters)a
correctedfornon-linearity, separatelyforeachCCD.Aperture
RA(J2000.0) 08:21:40.871
photometry is performed for the target and two brighter com-
Dec(J2000.0) +13:29:51.08
parisonstarsinthefieldofview(TYC807-1069-1hereafterC1,
andTYC807-1165-1hereafterC2)usingacustomizedpipeline mKp [mag] 11.687
(Fukuietal.2011).Theapertureradiusforeachbandischosen mg′ [mag]b 12.244±0.001
as24(g2′),26(r2′),and28(zs,2)pixelsrespectivelysothatthe mr′ [mag]b 11.656±0.001
apparentroot-mean-square(RMS)forafractionallightcurveof mi′ [mag]b 11.484±0.001
C1/C2isminimized.Wecheckforpossiblesystematicvariabil- mz′ [mag]b 11.419±0.015
ityofthetargetandcomparisonstarsbymakingthefractional mJ [mag] 10.541±0.02
light curves of each combination. We find that the fractional mH [mag] 10.173±0.03
lightcurvesofC1/C2inr2′ andzs,2bandssmoothlychangeina mKs [mag] 10.091±0.02
linearmannerwiththedeviationfromalinearapproximationof B−V [mag] 0.79±0.05
∼0.1%. Ontheotherhand,thefractionallightcurveofC1/C2 V −J [mag] 1.38±0.05
ing2′ bandshowsastrangesystematicvariationwiththeampli- (SpectroscopicParameters)
tude of∼0.4%. Although wesuspect thesystematic variation Teff [K] 5434±35(stat.)±40(sys.)
iscausedbystrong2nd-orderextinctionofEarth’satmosphere, logg[dex] 4.477±0.085
wedecidenottouseg2′ bandlightcurveinthesubsequentanal- [Fe/H][dex] 0.26±0.05
ysis, since we cannot correct the variation with the observed ξ[kms−1] 0.21±0.44
data. ThetotalfluxofC1+C2 is usedasacomparison fluxto vsini[kms−1] 1.76±0.86
the target. We also find that the peak count and total flux of (DerivedParameters)
thetargetsuddenlydroppedafterJD2457429.17, eventhough M⋆[M ] 1.01±0.07
⊙
the FWHM did not change, suggesting the sky transparency R⋆[R ] 0.95+00..1110
changedsignificantlyatthattime. Wethusconfineusabledata ρ⋆[ρ ⊙] 1.19−+00..4342
toaroundJD2457428.95-2457429.17inthesubsequentanal- ρ⋆[g⊙cm−3] 1.68−+00..6425
ysistoavoidsystematicerrors. Distance[pc]c 220−±30
Distance[pc]d 233+29
23
−
Age[Gyr] ≥0.6
3 Analysesand Results
aBasedontheEPIC,SDSS,UCAC4,and2MASSCatalogs.bBasedontheSDSS
PSFmagnitude. cBasedonthe2MASSapparentmagnitudeandtheestimated
3.1 SpectroscopicParameters
absolutemagnitudesforthestellarparameters. dBasedontheparallaxreported
Weperform aline-by-line analysis forthe HDS spectrum fol- byGAIADataRelease1.
lowingthemethoddescribedinTakedaetal.(2002)andTakeda
etal.(2005).BymeasuringtheequivalentwidthsofFeIandFe ingATLAS9model(aplane-parallelstellaratmospheremodel
IIlinesbetween5000A˚ and7400A˚,weestimatethestellaref- inLTE;Kurucz1993)assumingtheabovederivedatmospheric
fectivetemperatureTeff,thesurfacegravitylogg,themetallic- parameters,andconvolvethemodelspectrumwiththerotation
ity[Fe/H],andthemicroturbulentvelocityξfromtheexcitation plusmacroturbulence kernel(Gray2005) andtheinstrumental
andionization equilibria. Basedontheestimatedatmospheric profile of Subaru/HDS. Taking account of the intrinsic uncer-
parameters, we estimate the mass, radius, and density of the tainty in the macroturbulent velocity (Hirano et al. 2012), we
host star using the empirical relations derived by Torres et al. estimatevsinitobe1.76±0.86kms−1.
(2010)fromdetachedbinaries.Notethattheempiricalrelations The distance of the host star is estimated as 220±30 pc
haveuncertainties of6%and3%forthemassandradius,re- bycomparingtheabsolutemagnitudesbasedontheDartmouth
spectively, and these uncertainties aretaken into account. We isochrones (Dotter et al. 2008) for the above stellar parame-
also take into account the fact that the effective temperature ters with the apparent magnitudes in JHKs bands from the
derived from the excitation/ionization may have a systematic 2MASS point source catalog (Skrutskie et al. 2006). In ad-
error of about 40 K (see details in Bruntt et al. 2010; Hirano dition, recently GAIA Data Release 1 (Gaia Collaboration
etal.2014). Theestimatedmassandradiusareingoodagree- 2016; Lindegren et al. 2016) reported the parallax of K2-105
ment with those based on the Yonsei-Yale stellar-evolutionary as4.288±0.467mas,correspondingtothedistanceof233+29
23
−
model(Yietal.2001),whichweusetosetalowerlimitonthe pc. These two estimates arein excellent agreement, implying
ageofthehoststarof0.6Gyr. Toderivethestellar rotational thespectroscopically-derivedstellarparametersarereasonable.
velocity vsini, we generate the stellar intrinsic spectrum us- Derived stellar parameters and their errorsaresummarized
PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0 5
The combined saturated image is convolved with the FWHM
ofthecombinedunsaturatedimage,andthestandarddeviations
ofcountsinannulisegmentedfromthecenterofthetargetare
calculated. Aperturephotometryforthecombinedunsaturated
image is done tocompute thefluxcount of thetarget perunit
of time. By comparing the flux of the target to the standard
deviations of counts in annuli, we create a 5σ contrast curve
asafunctionoftheseparationinarcsec. Thecombinedimage
ofthesaturatedframesandthe5σ contrastcurveareshownin
Figures2and3. Consequently,wedonotfindanyevidenceof
contaminants whichcouldmimicthetransitsignalaroundK2-
105atthelevelshowninthecontrastcurve.
3.3 JointTransitLightCurveAnalysis
Fig.2.Acombined imagearound K2-105. Theimageisgiveninthelog
Wesimultaneously fitK2and MuSCAT transitlight curves as
color scale. The central region around K2-105 is saturated. Thisfigure
showsa5”×5”portion.NorthisupandEastisleft. follows.
TheK2lightcurveshowninFigure1isseparatedintonine
transitsegments,eachcontainingafull-transit(namely,before,
100 0 during, and after a transit). Note that there is another transit
EPIC211525389
atthebeginning ofK2observation, butweexcludethistransit
10-1 2 sincethedatabeforethetransitarenotavailable. Eachsegment
t includes the data within ∼ 9 hours from the apparent transit
mi 4
center. Wecomputethestandarddeviationoftheout-of-transit
ast li10-2 6 mag K2 data, excluding data in the transit segments and apparent
tr H outliers (with the excursion larger than 0.001 from the unity).
on10-3 ∆Weadoptthestandarddeviationoftheout-of-transitdataasan
c 8
σ estimateoftheuncertaintyforeachK2fluxvalue.ForK2transit
5
10-4 10 light curves, we adopt a linear function in time as a baseline
model,
10-50 1 2 3 4 5 6 7 812 Fbase(t)=k0+ktt,
separation [arcsec]
Fig.3.A5σcontrastlimitcurvearoundK2-105. where k0 is the normalization factor and kt is the coefficient
for the time. Other parameters for the K2 transit light curves
in Table 1. As aresult, we find that the host star K2-105 is a arethemid-transittimeforeachtransitsegment(Tc(E),where
metal-richG-dwarf. This resultis, however, inconsistent with E=0–8),andtheplanet-to-starradiusratioRp/R⋆fortheK2
Huber et al. (2016), who reported EPIC 211525389 to be a bandpass(Kp).
For the MuSCAT transit light curves, we adopt a novel
metal-poor giant based on reduced proper motion and colors.
parametrization for thebaseline model to take account forthe
Such discrepancies are not uncommon, given the difficulty of
2nd-order extinction introduced byFukui etal.(2016b). As a
metallicitydeterminationbasedonlyonbroadbandphotometry.
brief introduction, the parametrization uses the apparent mag-
Thespectroscopicdeterminationismorereliable.
nitudeofthecomparisonstar(s)insteadoftheairmass,andthe
baselinefunctionisexpressedinmagnitudesasfollows,
3.2 ExcludingFaintContaminants
mt,base(t)=k0+ktt+kcmc(t),
Wecomputeoffsetsbetweenthecentralstar’scentroidsineach
frame obtained with and without the ND filter. The reduced wheremtandmc aretheapparentmagnitudeofthetargetand
images of saturated and unsaturated frames are then offset- comparisonstar(s),kcisthecoefficientfortheatmosphericex-
corrected,sky-level-subtracted, andcombinedtoproducefinal tinction. This parametrization allows us to correct both the
deep-integrationimages. Noadditionalpointsourcewhichcan airmass extinction and the 2nd-order extinction caused by the
mimic the observed transit signal is detected in the final im- different spectral types of comparison stars. See Fukui et al.
age. Hence we compute the detection limit for such sources. (2016b)forthemathematicalderivation ofthismethod. Other
6 PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0
Fig.4.Leftpanel:Thephase-folded,baseline-correctedK2transitlightcurve.BluedotsplottheK2data.Ablacksolidlinerepresentsthebest-fittransitmodel
thatisintegratedovertheK2cadence(∼29.4minutes).Residualsfromthebest-fitmodelareplottedwithaverticaloffsetof−0.01.Middleandrightpanels:
Sameastheleftpanel,butthebaseline-correctedMuSCATtransitlightcurvesforther2′ band(middle)andthezs,2band(right),respectively.Reddotsshow
theMuSCATdata.
parameters for the MuSCAT transit light curves are the mid- Table2.SummaryofpriorsfortheMCMCanalysis.
transittimefortheMuSCATobservation(Tc(34))andRp/R⋆ Parameter Prior Explanation
forMuSCATr2′ andzs,2bands. P [days] 8.2672478 Fixed
Finally, the common planetary transit parameters are the a/R⋆ 18.4±2.5 Addedinthepenaltyfunction
scaledsemi-majoraxisa/R⋆ andtheimpactparameterb. We w1,Kp [0.197,0.359] Uniformprior
placeanaprioriconstraintona/Rstobe18.4±2.5,basedon w1,r′ [0.180,0.329] Uniformprior
2
spectroscopicallyderivedM⋆,R⋆,andthesemi-majoraxises- w1,Kp [-0.028,0.191] Uniformprior
timatedbyKepler’sthirdlaw. Weassumeanorbitalperiodof w2,Kp 0.466±0.009 Gaussianprior
P =8.2672478 days, which wederivefromtheK2transits at w2,r′ 0.472±0.009 Gaussianprior
2
the time of identification of the candidate. Although we later w2,Kp 0.375±0.011 Gaussianprior
deriveanimprovedorbitalperiod,thedifferencehasnoimpact
onfitting results, sinceweallow allTc(E)tobefreeparame- ThepriorsfortheMCMCanalysisaresummarizedinTable2.
ters.
Before creating MCMC chains, we first optimize free pa-
To estimate the values of the freeparameters and their un- rameters for each light curve, using the AMOEBA algorithm
certainties,weuseacode(Naritaetal.2007)thatusesthean- (Pressetal.1992).Thepenaltyfunctionisgivenby
acluyrtvicemfoormdeul.laTghiveeannablyytOichttraanestiatlf.o(r2m0u0l9a)ifsoerqtuhievatlreanntsittolitghhatt χ2=XX(fobs,t−σ2fmodel,t)2
givenbyMandel&Agol(2002)whenusingthequadraticlimb- K2 t f,t
darkening law. We adopt a methodology for applying priors + (mobs,t−mmodel,t)2 +(a/R⋆−18.4)2,
X X σ2 2.52
on limb-darkening described in Fukui et al. (2016a). To re- MuSCAT t m,t
duceacorrelationbetween thequadraticlimb-darkening coef- wherefobs,tandσf,taretherelativefluxesofthetargetineach
ficients u1 and u2, and to appropriately estimate uncertainties K2 transit segment and their errors, and mobs,t and σm,t are
for other parameters, we use acombination formof the limb- the magnitude of the target in rs′ and zx,2 bands and their er-
darkeningparametrizationw1=u1cos(φ)−u2sin(φ)andw2= rors. The model functions fmodel,t and mmodel,t are combi-
u1sin(φ)+u2cos(φ),whichwasintroducedbyPa´l(2008).We nations of the baseline model and the analytic transit formula
adoptφ=40◦thatisrecommendedbyPa´l(2008)andHowarth mentioned above. Wenote that thetransitmodel is integrated
(2011).Wereferthetablesofquadraticlimb-darkeningparam- overtheK2cadence(∼29.4minutes)forfmodel,t.Inaddition,
eters by Claret et al. (2013) and compute allowed w1 and w2 thetime stamps of alldata areconverted tothe BJDTDB sys-
valuesforthestellarparameterspresentedinTable1. Weem- temusingthecodebyEastmanetal.(2010).Ifthereducedχ2is
ployuniformpriorsforw1between[0.197,0.359]forKpband, largerthanunity, werescalethephotometricerrorsofthedata
[0.180, 0.329] for MuSCAT r2′ band, and [-0.028, 0.191] for such that the reduced χ2 for each light curve becomes unity.
MuSCAT zs,2 band, respectively. For w2, we adopt Gaussian Wethen estimatetheleveloftime-correlatednoise(a.k.a. red
priorsas0.466±0.009forKpband,0.472±0.009forMuSCAT noise: Pontetal.2006)foreachlightcurve,bycalculatingthe
r2′ band,and0.375±0.011forMuSCATzs,2band,respectively. βfactorintroducedbyWinnetal.(2008). Theβfactorisused
PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0 7
to take into account the time-correlated noise and to properly Table3.PlanetaryParametersofK2-105b
compensatethepossibleunderestimateofderiveduncertainties Parameter Value
fromanalysesoftransitphotometry. Forthepurpose,wecom- (MCMCParameters)
putetheresidualsforeachlightcurveandaveragetheresiduals a/R⋆ 17.96+02..9314
intoM binsofN points. Wethencalculatetheactualstandard b 0.328+−0.249
0.225
deviationofthebinneddataσN,obsandtheidealstandarddevi- Rp/R⋆[Kpband] 0.03472−+00..0000103637
wathioenrewσit1hiosutthaensytatnimdaer-dcodrerveilaattieodnnoofistheeσrNe,siiddeualal=sf√oσrN1uqnbiMnMn−e1d, RRpp//RR⋆⋆[[rz2s′,2babnadn]d] 00..0033464541−+−+0000....00000000434548245811
data. To account for increased uncertainties due to the time- Tc(0)[BJD-2450000] 7147.9896−0+00..0000434711
correlated noise, we compute β =σN,obs/σN,ideal for various Tc(1)[BJD-2450000] 7156.25371−+00..0000220069
N.Ifβissignificantlyhigherthanunity,itimpliesthepresence Tc(2)[BJD-2450000] 7164.52415−+00..0000222214
ofthe time-correlatednoise. Consequently, wefindnosignif- Tc(3)[BJD-2450000] 7172.79050−+00..0000211939
−
icanttime-correlatednoiseinK2lightcurves,whileMuSCAT Tc(4)[BJD-2450000] 7181.06464±0.00244
light curves show significant time-correlated noise, especially Tc(5)[BJD-2450000] 7189.32677+00..0000220122
in the r2′ band. We adopt β = 1.56409 for the r2′ band and Tc(6)[BJD-2450000] 7197.59274−+00..0000228974
β=1.08181forthezs,2band,whicharethemedianvaluesof Tc(7)[BJD-2450000] 7205.86050−+00..0000210917
βforN=5–20binningcasesineachband,andfurtherrescale Tc(8)[BJD-2450000] 7214.12742−+00..0000339423
the errors of the light curves by multiplying them by their β Tc(34)[BJD-2450000] 7429.06529−+00..0000114527
factors. K[ms−1] 9.4±5.8(<−26.8∗)
(DerivedParameters)
Finally, we employ the Markov Chain Monte Carlo
(MCMC)code(Naritaetal.2013;Fukuietal.2016a)tocom- P [days] 8.266902±0.000070
putetheposteriordistributionsforthefreeparameters.Wecre- Tc(0)†[BJD-2450000] 7147.99107±0.00117
ate3chainsof12,000,000points,anddiscardthefirst2,000,000 Rp[R ]† 3.59+00..4349
pointsfromeachchainas“burn-in”.Thejumpsizesofparame- Rp[R⊕Jup]‡ 0.369−+00..003394
tersineachMCMCstepareadjustedsuchthatacceptanceratios i[◦] 88.95−+01..7037
become ∼23%, which is considered as an optimal acceptance T14[days] 0.14426+−00..0000222043
−
ratioforefficientconvergenceofMCMC(seee.g.,Ford2005). Mp[M ] 30±19(<90∗)
⊕
Table3presentsthemedianvaluesanduncertainties,which ∗Anupperlimitat99.865percentile(3σ)level.†Thisistheoriginforthetransit
are defined by the 15.87 and 84.13 percentile levels of the ephemeris.‡BasedonRp/R⋆inKpband.
merged posterior distributions. The baseline corrected tran-
sit light curves (in flux) are plotted in Figure 4. We find long-termradialvelocitydrift.
that Rp/R⋆ for all three bands are consistent with one an- To model the observed RVs, we adopt an RV model,
other and that the MuSCAT light curves areconsistent with a vmodel = −Ksinφ+γ, where K, φ, γ are the RV semi-
flat-bottomed transit. To derive the orbital ephemeris we fit amplitude, theorbitalphaserelativetothemid-transit,andthe
a linear model to the mid-transit times, yielding a period of offset RV relative to the template spectrum. We fix the or-
P = 8.2669016±0.0000581 days and a time of first transit bital period P to 8.2669016 days and the origin of the tran-
of Tc(0)=2457147.99107±0.00098 (BJDTDB), with χ2 of sitephemerisTc(0)to2457147.99107 (BJDTDB)asderivedin
11.499for8degreesoffreedom. Notethatinthisfitweadopt §3.3.WedonotconsidertheRossiter-McLaughlineffect,since
the larger-side uncertainty if uncertainties of respective mid- thereisnoRVdataduringatransit. Wealsoneglecttheeccen-
transittimesareasymmetric.Tobeconservative,werescalethe tricitye,sinceitisindeterminablewiththecurrentRVs.
uncertainties ofP andTc(0)byp11.499/8,andtherescaled We first optimize K and γ using the AMOEBA algorithm
uncertainties arepresented inTable3. This refinedephemeris
(Press et al. 1992), and create an MCMC chain of 500,000
willbeusefulforfuturetransitobservationsofK2-105b.
points starting from the optimal parameters. The acceptance
ratio is set to ∼23%. The phased RVs and the best-fit RV
model are shown in Figure 6. The median values and uncer-
3.4 Subaru/HDSRadialVelocitiesandaMass
tainties of the free parameters are presented in Table 3. We
UpperLimit
also present a 3σ upper limit (99.865 percentile level) of K
We employ the RV pipeline for the Subaru HDS described in inTable3. Consequently, theRVsemi-amplitude is 9.4±5.8
Satoetal.(2002)toextracttherelativeRVswithrespecttothe ms−1,whichcorrespondsto30±19M forthemassofK2-
⊕
templateiodine-freespectrum. ThederivedRVsarepresented 105 b. At this point, the current RVs are not sufficient to de-
inTable4andplottedinFigure5.Wedonotfindanysignificant terminetheRVsemi-amplitudeandthemassofK2-105bpre-
8 PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0
Table4.RadialvelocitiesofK2-105takenwithSubaru/HDS.
#"
BJD Value[ms 1] Error[ms 1]
TDB − −
2457352.95525 11.33 8.92
$"
1) 2457353.96869 5.08 7.25
-s
m 2457354.98800 4.45 7.14
"
y ( 2457420.93976 -23.06 8.42
t
ci 2457420.94741 -12.43 9.14
o $"
el 2457674.12841 7.07 7.38
v
al #" 2457675.13374 -6.44 7.45
di
a 2457676.11418 3.23 7.80
R
!"
lations,constraintsoncontaminantsfromhighresolutionimag-
!&" ’"" ’&" &"" &&" %"" %&"
BJD - 2457000 ing contrast curves, physical properties of the host star from
spectroscopically-derived parametersandbroadbandphotome-
Fig.5.Relative radial velocities (blue points) of K2-105 taken with the try,andcomparisonsoftheshapeofthephase-foldedK2light
SubaruHDS.
curvetoalargenumberofrealisticfalsepositivescenarios.
We input our results presented in the last section to vespa
and find the final FPP for this target to be extremely low (<
!"
10−6), which strongly indicates aplanetary nature fortheori-
-1m s) #" tghienfoaflstheepoobsisteivrveesdcetrnaanrsiiotssiagcncaolusn.tWedeftohrerbeyfovreesrpuale(oi.uet.ahllieor-f
city ( " aercclihpiscianlgtrbipinlearsieyss)t.ems,eclipsingbinaries,blendedbackground
o
el
al v #" Weconclude thatK2-105 bis abona fideplanet, basedon
di the mass constraint presented in §3.4 and the extremely low
a
R !" FPP.
"$% "$! "$" "$! "$% 4.2 TheNewTransitEphemerisandaHintofTransit
Phase TimingVariation
We have derived the new transit ephemeris for K2-105 b as
Fig.6.Radialvelocities(bluepoints)phasedbytheorbitalperiodgivenin
Table3andthebest-fitRVmodel(redline). P =8.266902±0.000070daysandTc(0)=2457147.99107±
0.00117 in BJDTDB. This is indeed the first reliable transit
ephemeris for K2-105 b, since Pope et al. (2016) and Barros
cisely.Nevertheless,wecanputaconstraintonK<26.8ms−1
et al. (2016), who reported EPIC211525389 b as acandidate
at the 3σ level, which corresponds to a mass upper limit of
planet, presented a transit ephemeris without any uncertainty.
90M or0.00027M .Thisupperlimitensuresthatthemassof
⊕ ⊙ Usingthetransitephemeriswehavederivedfromourobserva-
K2-105biswithin aplanetarymass,excluding thepossibility
tions, transits of K2-105 b in 2017 can be predicted with un-
ofaneclipsingbinaryscenarioforthissystem.
certainties of only about 10 minutes, which will facilitate the
schedulingoffuturetransitobservations.
4 Discussions We check the possible presence of transit timing variation
(TTV) for the transits of K2-105 b. Figure 7 plots residu-
4.1 ConfirmationofthePlanetaryNatureof
als of the observed mid-transit times from the current transit
K2-105b
ephemeris. While if only K2 transits are taken into account,
Tofurthervalidatetheplanetarynatureofthetransitsignal,we a linear fit to the mid-transit times gives P = 8.2675710±
usetheopen-sourcePythoncodevespa(Morton2012;Morton 0.0003569 days and Tc(0) = 2457147.98853 ±0.00164 in
2015), which employs a robust statistical framework to cal- BJDTDB,withχ2of7.884for7degree-of-freedom.Thetransit
culate the False Positive Probability (FPP) of the transit sig- fortheMuSCATrunoccurredabout30minearlierthanthepre-
nal. It does this by taking into account a variety of factors: diction fromtheK2-onlytransits, although thedifferenceisat
the size of the photometric aperture of K2, the source den- the2σlevel. Thediscrepancyisstatisticallynotsignificant,but
sity along the line of sight as determined from galaxy simu- itmaysuggestthatanadditional non-transiting planet existas
PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0 9
! 17
16
15
14
e) "! R)⊕ 1123
ut s( 11 Jupiter
u
min #! radi 109 Saturn
C ( etary 78 FGK---tttyyypppeee
- ! an 6 M-type
O Pl 5 EPIC211525389b
4 Neptune
3
2
$#!
1 Earth
0
! #! "! ! 1 10
Orbitalperiod(day)
Epoch Fig.8.Period-radius relation ofconfirmed transiting exoplanets withP ≤
50days around F-type (M⋆ = 1.04−1.4M⊙), G-type (M⋆ = 0.8−
Fig.7.Residualsoftheobservedmid-transittimesfromthecurrentbest-fit
1.04M⊙), K-type (M⋆ = 0.45−0.8M⊙), and M-type stars (M⋆ =
transitephemeris. Theredlinerepresentsthebest-fittransitephemerisfor
0.08−0.45M⊙)asof2016August;thedatacomefromhttp://exoplanet.eu.
thetransitsfromK2only.WenotethatthetransitephemerisfromK2hasan
Planetsforwhichtheradiusuncertaintyexceeds20%oftheirrepresentative
uncertaintyof∼17minutesattheepochof34.Thusthediscrepancyisstill
valuesareexcluded.
withinthe2σlevel.
isthecaseforK2-19b&c(e.g.,Armstrongetal.2015;Narita than30M⊕.K2-105borbitsatap=0.081±0.006AUarounda
G-dwarfwiththemassof1.01±0.07M .AccordingtoOwen
etal.2015a). Alternatively, stellaractivities suchasstarspots ⊙
& Wu (2013), K2-105 b can retain its atmosphere under an
mayplay arole inthe apparent discrepancy (seee.g., Oshagh
intense stellar X-rayand EUV irradiation during an estimated
et al. 2013). To confirm the presence of TTVs for K2-105 b,
stellarageofolderthan0.6Gyr,ifthecoremassisgreaterthan
furthertransitmonitoringisneeded.
∼6M .
⊕
IfK2-105bisagasdwarf,howdiditform? Therearetwo
4.3 K2-105bintheContextofPeriod-Radiusand possibleformationscenarios,namely,in-situgasaccretiononto
Mass-RadiusRelation a massive core (Ikoma & Hori 2012; Lee et al. 2014; Ormel
Occurrence rates of planets derived from RV surveys and the etal.2015)orinwardmigrationofaNeptune-likeplanet(e.g.,
Keplerindicatethatshort-periodNeptune-sizedplanetssuchas Bodenheimer &Lissauer2014). However, wecannotruleout
K2-105 b are only rarely found in planetary systems around both stories becauseof anunknown massof K2-105 b. Thus,
solar-typestars(e.g.,Howardetal.2012;Petiguraetal.2013). massdeterminationfromfollow-upRVobservationswillbein-
Figure8showsanorbitalperiod–radius distribution oftransit- dispensable for constraining the formation history and quanti-
ing planets with P ≤50days. We see a clear lack of planets fyingtheeffectofphoto-evaporation.
withradiiof∼3.5−10R aroundsolar-typestars,albeitinter- In addition, close-in Neptune-sized planets as represented
estingly, nohotJupiterw⊕ithradiusof>∼10R isseenaround byK2-105bwouldbesuggestiveofuncoveringhowtheSolar
⊕
starswithM⋆≤0.45M . Thesefeatures,alsopointedoutby Systemwasborn.ThereisnoK2-105b-likeplanetintheSolar
⊙
previous studies (e.g., Mazeh et al. 2016; Matsakos & Ko¨nigl System, instead the two ice giants orbit beyond ∼20AU. As
2016),mayreflectasizeboundarybetweenafailedcoreanda onepossibility, thismightbecausedbythepresenceofJupiter
gas giant, which corresponds to a critical core mass that trig- andSaturnorbitingwithintheorbitsofthetwoicegiants,act-
gersgasaccretioninarunawayfashion,ormasslossviaphoto- ingasabarrieragainstinwardmigratingcores. Long-termRV
evaporation. Thelattercasecanbeausefulindicatortoevaluate monitoringofK2-105toconstrainthepossibilityofoutergiant
theefficiencyofatmosphericescapeduetoastellarirradiation planetsshouldbehelpfulinunderstandingtheorbitalevolutions
orinjectionofhigh-energyparticles.Thus,thediscoveryofK2- ofNeptune-likeplanetsandtheSolarSystem. Therefore,long-
105bcanbeaninterestingbenchmarktodisentangletheorigin termRVmonitoringofthissystemwouldbealsoencouraged.
ofNeptune-sizedplanetsclosetocentralstars.
Figure 9 shows theoretical mass-radius relations for three
5 Summary
typesofplanetsandtransitingexoplanetswithknownmass.We
findthatK2-105bisnotabarerockyplanet but likely hasan We have confirmed the planetary nature of K2-105 b, using
atmosphere(<10%ofitstotalmass)ifitstotalmassissmaller transitphotometry fromtheK2mission, highdispersion spec-
10 PublicationsoftheAstronomicalSocietyofJapan,(2014),Vol.00,No.0
vital forunderstanding the formationand migration history of
thisplanetarysystem.
4.0
EPIC211525389b
)⊕ 3.5 6 Funding
R
(
us 3.0 This work was supported by Japan Society for Promotion
di
a of Science (JSPS) KAKENHI Grant Numbers JP25247026,
r
ary 2.5 JP16K17660, 25-8826, JP16K17671, and JP15H02063. This
net workwasalsosupportedbytheAstrobiologyCenterProjectof
Pla 2.0 H2O NationalInstitutesofNaturalSciences(NINS)(GrantNumbers
Fe
AB271009,AB281012andJY280092).I.R.acknowledgessup-
1.5 MgSiO3
portbytheSpanishMinistryofEconomyandCompetitiveness
1.0 (MINECO)throughgrantESP2014-57495-C2-2-R.
1 10
Planetarymass(M⊕)
Fig.9.Mass-radius relationofconfirmed transiting exoplanets asof2016 Acknowledgments
September; the data come from http://exoplanet.eu. Planets with uncer-
taintiesinmassandradiusover20%oftheirrepresentativevaluesarenot WeacknowledgeRobertoSanchis-Ojeda,whoestablishedtheESPRINT
shownhere.Theoreticalmodelsofiron,water,andsilicateplanetsarebased collaboration. WethanksupportsbyAkitoTajitsuandHikaruNagumo
onZeng&Sasselov(2013).WeadoptthepossiblerangeofK2-105b’smass forourSubaruHDSobservation,JunHashimotoforSubaruHiCIAOob-
derivedfromRVmeasurementsbytheSubaruHDS,Mp=30±19M⊕and servation, andTimothyBrandtforHiCIAO datareduction. Thispaper
itsradiusofRp=3.59+−00..4349R⊕. isbasedondatacollectedattheSubarutelescopeandOkayama188cm
telescope,whichareoperatedbytheNationalAstronomicalObservatory
of Japan. The data analysis was in part carried out on common use
troscopy and RVs from Subaru/HDS, high-contrast AO imag-
dataanalysiscomputersystemattheAstronomyDataCenter, ADC,of
ing from Subaru/HiCIAO, and ground-based transit photome- the National Astronomical Observatory of Japan. PyFITSand PyRAF
try from Okayama/MuSCAT. The host star K2-105 is located were useful for our data reductions. PyFITS and PyRAF are prod-
in K2 campaign field 5, and estimated to be a metal-rich G- ucts of the Space Telescope Science Institute, which is operated by
AURA for NASA. Our analysis is also based on observations made
dwarf.AlthoughfurtherRVmonitoringisrequiredtoprecisely
with the NASA/ESA Hubble Space Telescope, and obtained from the
determine the mass of K2-105 b, the Subaru HDS RVs put a
Hubble Legacy Archive, which is a collaboration between the Space
stringentconstraintonthemassofK2-105baslessthan90M TelescopeScienceInstitute,theSpaceTelescopeEuropeanCoordinating
⊕
or 0.00027M at the 3σ level, ensuring that the mass of K2- Facility (ST-ECF/ESA) and the Canadian Astronomy Data Centre
⊙
105biswellwithin theplanetary massrange. Ourjoint anal- (CADC/NRC/CSA).ThisworkhasmadeuseofdatafromtheEuropean
Space Agency (ESA) mission Gaia (http://www.cosmos.esa.int/gaia),
ysisofthetransitdatafromK2andMuSCATyieldsanorbital
processedbytheGaiaDataProcessingandAnalysisConsortium(DPAC,
periodofP =8.266902±0.000070daysandanoriginofmid-
http://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the
transittimeTc(0)=2457147.99107±0.00117inBJDTDB.The DPAChasbeenprovidedbynationalinstitutions,inparticulartheinstitu-
transitephemerisisaccurateenoughtopredicttransittimesof tionsparticipatingintheGaiaMultilateralAgreement. Weacknowledge
K2-105 b with uncertainties of less than 20 minutes for the theverysignificantculturalroleandreverencethatthesummitofMauna
next few years. We have found that the transit observed with KeahasalwayshadwithintheindigenouspeopleinHawai’i.
theOkayama/MuSCAT occurredabout30minutesearlierthan
thepredictionfromtheK2-onlytransits. Althoughthediscrep-
References
ancyfromthepredictionisstatisticallymarginalatthe2σlevel,
Adibekyan,V.Z.,etal.2013,A&A,560,A51
this may suggest that additional long period or non-transiting
Armstrong,D.J.,etal.2015,A&A,582,A33
planet(s)existinthesystem,whichincreasestheneedforfur-
Barros, S.C.C.,Demangeon, O.,&Deleuil, M.2016, ArXive-prints,
thertransitandRVmeasurementsofthissystem.
arXiv:1607.02339
ThetransitdepthofK2-105b,Rp/R⋆∼0.035,corresponds Beauge´,C.,&Nesvorny´,D.2013,ApJ,763,12
to a planetary radius of Rp =3.59+00..4349 R . Thus K2-105 b Bodenheimer,P.,&Lissauer,J.J.2014,ApJ,791,103
isashort-periodNeptune-sizedplan−et. As⊕K2-105bisatran- Brandt,T.D.,etal.2013,ApJ,764,183
siting planet around a relatively bright host star, it is a favor- Bruntt,H.,etal.2010,MNRAS,405,1907
Claret,A.,Hauschildt,P.H.,&Witte,S.2013,A&A,552,A16
ableandimportanttargetforcharacterizationofitsmassviaRV
Crossfield,I.J.M.,etal.2015,ApJ,804,10
measurements, its atmosphere via transmission spectroscopy,
Crossfield,I.J.M.,etal.2016,ApJS,226,7
spin-orbit(mis)alignmentviatheRossiter-McLaughlineffector
Dai,F.,etal.2016,ApJ,823,115
doppler tomography, andthepresenceof additional planet via Dotter, A., Chaboyer, B., Jevremovic´, D., Kostov, V., Baron, E., &
TTVs and/or RVtrends. Suchfurther characterization will be Ferguson,J.W.2008,ApJS,178,89