arXiv:1407.1057v3 [astro-ph.SR] 11 May 2015 · 2018. 8. 25. · photometric aperture down to about mag = 10. The F555W and F775W lters on WFC3/UVIS are ideally suited to observe the

Accepted version March 5 2015 by ApJPreprint typeset using LATEX style emulateapj v 5211

REVISION OF EARTH-SIZED KEPLER PLANET CANDIDATE PROPERTIES WITH HIGH RESOLUTIONIMAGING BY HUBBLE SPACE TELESCOPE

Kimberly M S Cartier12 Ronald L Gilliland12 Jason T Wright12 and David R Ciardi3

Accepted version March 5 2015 by ApJ

ABSTRACT

We present the results of our Hubble Space Telescope program and describe how our analysis methodswere used to re-evaluate the habitability of some of the most interesting Kepler planet candidatesOur program observed 22 Kepler Object of Interest (KOI) host stars several of which were foundto be multiple star systems unresolved by Kepler We use our high-resolution imaging to spatiallyresolve the stellar multiplicity of Kepler-296 KOI-2626 and KOI-3049 and develop a conversion tothe Kepler photometry (Kp) from the F555W and F775W filters on WFC3UVIS The binary systemKepler-296 (5 planets) has a projected separation of 0primeprime217 (80 AU) KOI-2626 (1 planet candidate) isa triple star system with a projected separation of 0primeprime201 (70 AU) between the primary and secondarycomponents and 0primeprime161 (55 AU) between the primary and tertiary and the binary system KOI-3049 (1planet candidate) has a projected separation of 0primeprime464 (225 AU) We use our measured photometryto fit the separated stellar components to the latest Victoria-Regina Stellar Models with syntheticphotometry to conclude that the systems are coeval The components of the three systems range frommid-K dwarf to mid-M dwarf spectral typesWe solved for the planetary properties of each systemanalytically and via an MCMC algorithm using our independent stellar parameters The planetsrange from sim 16 Roplus to sim 42 Roplus mostly Super Earths and mini-Neptunes As a result of the stellarmultiplicity some planets previously in the Habitable Zone are in fact not and other planets maybe habitable depending on their assumed stellar hostSubject headings planetary systems - stars fundamental parameters - stars individual (KIC 6263593

KIC 11497958 KIC 11768142) - techniques photometric

1 INTRODUCTION

Since its advent the Kepler mission has increased thenumber of candidate exoplanets by thousands confirmedhundreds of planets and has pushed the boundaries oftransiting exoplanets to smaller radii and longer orbitalperiods than previously detected (Batalha et al 2013Borucki et al 2010 2011 Burke et al 2014 Fressin etal 2013 Howard et al 2012 Lissauer et al 2014) The2013 release of the first 16 quarters of Kepler data hasincreased the number of known transiting exoplanet can-didates of all radii but has been especially fruitful for thesmallest candidates (with a fractional increase of 201known planets smaller than 2Roplus) and for the longest or-bital periods (with a fractional increase of 124 for orbitslonger than 50 days Batalha et al 2013 Borucki et al2011) More recently Rowe et al (2014) has nearly dou-bled the total number of validated exoplanets throughcareful elimination of false-positive detections in multi-planet systems Nearly40 of Kepler planet candidateshave been found to reside in multiple planet systems(Batalha et al 2013 Rowe et al 2014) and recent sur-veys show that the vast majority of multiple transiting

kms648psuedu1 Department of Astronomy amp Astrophysics The Pennsylva-

nia State University 525 Davey Lab University Park PA 168022 Center for Exoplanets and Habitable Worlds The Pennsyl-

vania State University University Park PA 168023 NASA Exoplanet Science Institute California Institute of

Technology Pasadena CA USA Based on observations with the NASAESA Hubble Space

Telescope obtained at the Space Telescope Science Institutewhich is operated by AURA Inc under NASA contract NAS5-26555

system detections are true multiple planet systems (Lis-sauer et al 2014 Rowe et al 2014) Howard et al (2012)showed that the planet occurrence rate increases from Fto K dwarfs and followup studies by Dressing amp Char-bonneau (2013) and Kopparapu (2013) showed that thistrend continues increasing towards M dwarfs New esti-mates of ηoplus have made use of these more robust dataarriving at a conservative prediction that between 6-15of Sun-like stars have an Earth-size planet in the Habit-able Zone (HZ Kasting et al 1993 Petigura et al 2013Silburt et al 2015) though utilization of state-of-the-artHabitable Zone calculations will likely reduce this num-ber (Kopparapu et al 2013)

While the majority (gt 2000) of the Kepler planet can-didates reside in apparently single-star systems this per-centage is likely due to a selection effect that avoids bi-nary targets (Kratter amp Perets 2012) Accounting forthe frequency of binary stars the occurrence of plan-ets in multiple star systems could be as high as 50(Kaib et al 2013) Nearly all of the Kepler targets havebeen imaged by the United Kingdom Infrared Telescope(UKIRT) or other ground based telescopes that providesim 1primeprime seeingbut only 3051 of planet candidate hostshave been followed up with speckle interferometry adap-tive optics imaging or other high-resolution imaging ca-pable of resolving tightly bound systems This impliesthat a significant fraction of Kepler targets may in factbe close-in binary or higher multiple star systems that re-main unresolved Recent advancements in ground-basedadaptive optics (AO) particularly at the Keck Obser-

1 httpcfopipaccaltecheduhome

2 Cartier et al

vatory have accelerated high-resolution imaging of Ke-pler Objects of Interest (KOIs) especially those with thesmallest planets at the coolest temperatures The identi-fication of any diluting sources in the aperture allows forimproved precision when determining planet habitabilityand can also reveal previously unresolved stellar compan-ions Gilliland amp Rajan (2011) and Gilliland et al (2015)have shown that the sharp and stable point spread func-tion (PSF) of the WFC3 camera on Hubble Space Tele-scope is ideal for detailed photometric study of Keplertargets and for the identification of field stars in the HSTphotometric aperture down to about ∆mag = 10 TheF555W and F775W filters on WFC3UVIS are ideallysuited to observe the majority of Kepler targets

Our HST Guest Observing Snapshot Program GO-12893 observed 22 targets before May 1 2014 six ofwhich were found to be multiple star systems unresolvedby Kepler Gilliland et al (2015) discusses the overar-ching scientific goals and conclusions of the observingprogram including program parameters and basic im-age analysis stellar companion detections and detectioncompleteness comparison to other high-resolution imag-ing and tests for physical association of detected stel-lar companions Gilliland et al (2015) presents analysisthat directly supports the methods in this paper andserves as a companion paper to this work Here weperform multiple-star isochrone fitting using the latestrelease of the Victoria-Regina Stellar Models (Vanden-Berg et al 2014b Casagrande amp VandenBerg 2014) forthree Kepler targets of particular interest KIC 11497958(KOI-1422 hereafter Kepler-296) KIC 11768142 (here-after KOI-2626) and KIC 6263593 (hereafter KOI-3049)We discuss the parameters of GO-12893 and our imageanalysis in Section 2 including our use of the DrizzlePacsoftware and our conversion of our HST photometry tothe Kepler photometric bandpass In Section 3 we dis-cuss the importance of our three targets and detail ourcharacterization of the stellar components in each multi-star system including the use of our empirically derivedPSF to calculate the photometry of our systems fittingto the Victoria-Regina isochrones and examination oftheir suitability for our targets Section 4 presents ourre-evaluation of the planetary habitability For the pur-poses of this paper we define a ldquohabitable planetrdquo to be aplanet that falls between the moist greenhouse limit andthe maximum greenhouse limit as defined by Kopparapuet al (2013) Finally we discuss our results in contextof previous and future work in Section 5 and summarizeour findings in Section 6

2 OBSERVATIONS AND IMAGE ANALYSIS

The 158 targets proposed for observation were selectedfrom the 2013 data release of Kepler planet candidates byBatalha et al (2013) prioritized by smaller candidate ra-dius and cooler equilibrium temperature The remainingranked targets were then sorted between ground-basedAO and HST observations based on the quality of obser-vations for the fainter targets where HST would providecomparable or better data in half an orbit than a fullnight of ground-based AO observation on Lick or Palo-mar systems This resulted in the selected HST targetshaving the shallowest transit signatures which thus re-quire the deepest imaging The targets have a nominalupper limit of Rp lt 25Roplus (Batalha et al 2013)though

Fig 1mdash AstroDrizzled image of KIC 4139816 in the F775Wfilter showing a 1primeprime0 scale bar and orientation The image is ap-proximately 2primeprime0 on a side Units are log10 of eminuss The FWHMof the PSF is 0primeprime0777

our revision of the stellar parameters indicates that someof the planets are actually larger than this limit Of the158 proposed targets 22 were observed before May 2014and are included in our analysis Any observations col-lected after May 2014 will be analyzed using the tech-niques presented in this section but are not included inthis paper Our image analysis utilized the latest im-age registration and drizzling software from STScI Driz-zlePac (Gonzaga et al 2012) and our own PSF definitionand subtraction

21 HST High Resolution Imaging

Our HST program provided high resolution imaging inthe F555W (λ sim 0531microm) and F775W (λ sim 0765microm)filters of the WFC3UVIS camera to support the analysisof faint KOIs In particular the parameters of our ob-servations allowed us to examine the properties of faintstellar hosts of small and cool planet candidates Atthe faint magnitudes of typical Kepler stars our WFC3imaging provides resolution that is competitive with cur-rent ground-based AO and has the advantage of usingtwo well calibrated optical filters well matched to theKepler bandpass

The observations made by HST closely resemble thosemade by Gilliland amp Rajan (2011) though we only usedobservations in F555W and F775W since the faintest Ke-pler targets could still be probed in these bandpassesObservations planned for each of the 158 SNAP targetswere identical in form In each filter we took 5 observa-tions of each target 4 observations with exposure timesto reach 90 of full well depth in the brightest pixel andan additional observation at an exposure time equal to50 more than the sum of the unsaturated exposures tobring up the wings of the PSF The saturated exposureyielded a ∆-mag of sim 9 outside 2primeprime and helped with thesignal-to-noise anywhere outside the inner 0primeprime1

22 AstroDrizzle

The ldquodrizzlerdquo process formally known as variable-pixellinear reconstruction was developed to align and com-bine multiple under-sampled dithered images from HST

Revision of Kepler Planet Candidates with HST 3

into a single image with improved resolution reduction incorrelated noise and superior cosmic ray removal whencompared to images combined using a lower quality shift-and-add method (Gonzaga et al 2012) AstroDrizzle re-placed MultiDrizzle in the HST data pipeline in June2012 and is a significant improvement over the previ-ous MultiDrizzle software as it directly utilizes the FITSheaders for the instrument exposure time etc insteadof through user input AstroDrizzle also provides morefreedom in regard to the parameters for the image combi-nation leading to faster more compact and target spe-cific drizzled products (Frutcher et al 2010) Using As-troDrizzle we were able to adjust the parameters used increating the median image the shape of the kernel usedin the final drizzled image and the linear drop in pixelsize when creating the final drizzled image all of whichallowed us to create products with sharper and smootherPSFs than previous MultiDrizzle or STScI pipeline prod-ucts

We processed each target in our sample in the samemanner in order to best compare the final productsThe 5 images in each filter were first registered usingthe tweakreg task in DrizzlePac which performed fine-alignment of the images via additional sources found us-ing a daofind-like algorithm This fine-alignment wasnecessary to fully realize the high resolution of our obser-vations to create accurate PSFs out of the drizzled prod-ucts After registering the images they were combinedthrough astrodrizzle which first drizzled each sepa-rate image created a median image and split the medianimage back into the separate exposures to convolve eachexposure with the instrumental PSF and reconstruct itafter the instrumental effects were removed These recon-structed images were then corrected for cosmic ray con-tamination and finally drizzled together with the finalastrodrizzle product scaled to 0primeprime03333pixel Lastlywe centered the target on a pixel to within plusmn001 pix byutilizing the astrodrizzle output world coordinate sys-tem rotation matrix to transform the desired shift of thecentroid of the star in pixel-space to a shift in RADEC-space The drizzling and centering process was iteratedas often as necessary to center the target on a pixel tothe desired accuracy which aided in constructing an ac-curate PSF

Fig 1 shows the final drizzled product in the F775Wband for KIC 4139816 a typical single star from our sam-ple The HST pipeline product for this target showed arough PSF near the center of the target and further ex-amination showed that the pipeline had incorrectly clas-sified pixels in the saturated exposure Manual adjust-ment of the data quality flags allowed us to correct theissue in our AstroDrizzled product leading to a smootherand sharper PSF than the pipeline product

23 KpminusHST Photometric Conversion

Converting the Kepler photometric system to the HSTsystem served two purposes the first to provide a checkon the quality of our images and analysis and the sec-ond to calculate the dilution of the transit depths due toadditional stars in the Kepler photometric aperture Wecalculated photometry from the AstroDrizzle productsby summing the flux within a square aperture equivalentin area to a 20primeprimeradius aperture centered on the targetWe then used the published encircled energy of 99 rel-

TABLE 1Derived WFC3 photometry and Kp magnitudes from the

Kepler Input Catalogue used to derive Eq 1

KIC ID Obs Date Kp F555W F775W

2853029 2013-08-12 15679 16017 150064139816 2013-04-12 15954 16604 151414813563 2012-11-12 14254 14602 135105358241 2013-02-04 15386 15656 149025942949 2012-10-29 15699 16154 149906026438 2013-05-22 15549 16075 148276149553 2013-06-12 15886 17004 148126263593 2013-02-14 15037 15524 142756435936 2013-08-18 15849 16846 147967455287 2013-10-04 15847 16720 148378150320 2013-09-02 15791 16303 149858890150 2013-08-16 15987 16853 149698973129 2013-07-07 15056 15329 144559838468 2012-10-28 13852 14108 1332410004738 2014-01-07 14279 14563 1370410118816 2012-10-27 15233 16000 1422610600955 2013-02-10 14872 15135 1425311305996 2013-03-31 14807 15519 1385011497958 2013-04-06 15921 16807 1480511768142 2013-07-31 15931 17056 1489512256520 2013-07-28 14477 14805 1395712470844 2013-03-19 15339 15636 1469512557548 2013-02-06 15692 16349 14936

Note mdash HST photometry is for blended stellar componentsin KIC 6263593 11497958 and 11768142 systems KIC 12557548data are from Croll et al (2014) Observation Date is the same forall exposures of the same target

ative to an infinite aperture along with published zeropoints2 to obtain F555W and F775W magnitudes for thetargets Errors on the magnitudes are estimated to be003 in both filters

We then compared the published values for Kp fromthe Kepler Input Catalogue to F555W and F775Wfor the 22 observed targets and one from Croll et al(2014) that had identical observations (Table 1) Basedon a plot of Kpminus F555W vsF555W minus F775W we ob-served that the transformation between Kp F555W andF775W would follow a linear relation Fitting of a linearmodel to the data produced the correlation shown in Fig2 whose form follows

Kp = 0236 + 0406times F555W + 0594times F775W (1)

The fitted errors for this relation are 0019 mag for theF555W and F775W coefficients and 0027 mag for theintercept with an RMS scatter about the fit of 0042showing that our simple linear modeling works well forthis sample The error on the derived Kp magnitudedepends on the F555W minus F775W color as

σKp =radic

00192 (F555W minus F775W)2 + 00272 (2)

leading to slightly higher errors in Kp for redder targetsin HST

3 EVALUATION OF KEPLER-296 KOI-2626 AND KOI-3049STELLAR PARAMETERS

Our program observed three systems of particular in-terest Kepler-296 KOI-2626 and KOI-3049 Kepler-296 was first published as a multiple planet system by

2 wwwstscieduhstwfc3phot_zp_lbn

4 Cartier et al

Fig 2mdash Plot of Kp minus F555W vsF555W minus F775W (black pointsTable 1) with the best fit linear model (Eq 1) plotted in red Thetightness of the fit validates our echoice of a linear model to fit theconversion The errors on fit and points are in the text

Fig 3mdash Drizzled image of Kepler-296 in the F775W filter show-ing a 1primeprime0 scale bar and orientation The fainter component B isto the left Scale and units as in Fig 1 The FWHM of the PSFis 0primeprime1719 for blended system

Borucki et al (2011) and it has since been confirmed asa five planet system The stellar properties for this sys-tem were significantly updated by Muirhead et al (2012)Dressing amp Charbonneau (2013) and Mann et al (2013)and as a result of these studies it was found that Kepler-296 contained at least three potentially habitable plan-ets However Lissauer et al (2014) showed using KeckAO and these HST images that Kepler-296 is actuallya tight binary star system that appeared blended in theKepler CCDs KOI-2626 was first published in Batalhaet al (2013) and examination by Dressing amp Charbon-neau showed that the single planet candidate in the sys-tem was potentially habitable though Mann et al (2013)

Fig 4mdash Drizzled image of KOI-2626 in the F775W filter showinga 1primeprime0 scale bar and orientation Component B is lowest in theimage with component C to the left Scale and units as in Fig 1The FWHM of the PSF is 0primeprime3870 for blended system

Fig 5mdash Drizzled image of KOI-3049 in the F775W filter showinga 1primeprime0 scale bar and orientation The fainter component B istowards the top Scale and units as in Fig 1 The FWHM of thePSF is 05563primeprime for blended system

disputed this finding Later Keck AO observations3 re-vealed KOI-2626 to be a tight triple star system and thisrealization challenged all previous arguments about hab-itability It was noted in July 2013 on the Kepler Com-munity Follow-up Observing Program (CFOP) that LickAO detected a secondary star in their image 0primeprime5 awayfrom KOI-3049 4(1 planet candidate) but no confirma-tion of association has been published to date The stel-lar multiplicity of each system has profound impacts onthe habitability of their planets which we re-evaluatedin this study

Figures 3 4 and 5 show the AstroDrizzle combinedimages of Kepler-296 KOI-2626 and KOI-3049 respec-tively and display the tight apparent multiplicity of the

3 httpscfopipaccaltecheduedit_obsnotesphpid=2626lsquolsquociardi

4 httpscfopipaccaltecheduedit_obsnotesphpid=3049lsquolsquohirsch

systems We performed PSF fitting for each system asdescribed in Gilliland et al (2015) to photometricallyseparate the components in the HST filters

To ensure that the multiple components are not ran-dom superpositions of stars at different distances wethen attempted to fit the components of each system to asingle isochrone to prove that the systemsrsquo are most likelybound and therefore that the stars are the same age (co-eval) We then determined the probability that a randomstar in the field would produce a false isochrone match tothe same precision while not being physically associatedwith the target star This determines the probability ofthe isochrone fits for our target systems indicating boundsystems over randomly superimposed stars on the CCDThe PSF definition and the false association probabilityare outlined here and described in detail in Gilliland etal (2015)

31 PSF Definition and Photometry Used

We adopted the global PSF solution of Gilliland et al(2015) in each HST filter in order to separate the stellarcomponents of each of the three systems This globalPSF was empirically generated from our observations ofapparently single stars and is a function of target colorHST focus (which changes by small amounts from ther-mal stresses) and sub-pixel centering of the target Weextracted the necessary parameters for the PSF from thedrizzled image of each system of interest and iteration ofthe PSF fitting returned the separation and orientationsof the components of the systems and their fractionalcontributions in each HST bandpass Lastly combiningthe fractional contributions in the HST filters with theKpminusHST conversion in Eq 1 returned the fractionalcontribution of light from each component in Kp whichis directly relevant to the planetary parameters inferredfrom the Kepler transit depth

Application of this algorithm for Kepler-296 shows thatcomponent A contributes 809 of the light in the Keplerbandpass while component B contributes 191 (Lis-sauer et al 2014) Estimated uncertainties for these per-centages are 3 We found that component B is offsetfrom the brighter component A by 0primeprime217 plusmn 0primeprime004 at aposition angle of 2173plusmn 08 north through east

We used the same aforementioned global PSF and fit-ting algorithm for KOI-2626 using the appropriate colorfocus and offset values We inspected the drizzled imageminus the PSF fit for both F555W and F775W and foundno evidence for yet further components in the KOI-2626system For KOI-2626 component A contributes 545in the Kepler bandpass component B contributes 310and component C contributes 145 Estimated errorsfor these fractions are 6 We found that component Bis separated from component A by 0primeprime201plusmn0primeprime008 at a po-sition angle of 2127plusmn16 and component C is separatedfrom component A by 0primeprime161plusmn 0primeprime008 at 1816plusmn 16

Fitting of the global PSF for KOI-3049 using the cor-responding color and focus values for this system showedthat component A contributes 623 in the Kepler band-pass and component B contributes 377 with estimatederrors of 2 We found that component B is separatedfrom component A by 0primeprime464 plusmn 0primeprime004 at a position an-gle of 1969 plusmn 08 The estimated error for this systemis lower than for either Kepler-296 or KOI-2626 as thecomponents of the system are both brighter and more

Fig 6mdash Keck Kprime image of KOI-2626 showing a 0primeprime5 scale barComponent A is highest in the image with component B to thelower right and C to the lower left

widely separated and thus the PSF fitting was able tomore distinctly separate the components

In addition to the derived WFC3-based magnitudesand colors for the individual components of Kepler-296KOI-2626 and KOI-3049 we also utilized the SDSS-based magnitudes (Fukugita et al 1996) available in theKepler Input Catalogue (KIC) (Brown et al 2011) aswell as the 2MASS near-IR photometry available for theblended components We found that the SDSS g and rband photometry was redundant for our late-type starsgiven our WFC3 photometry and the SDSS z band wasunreliable at the apparent magnitudes examined here(Brown et al 2011) We therefore chose to include theblended photometry for the SDSS i band adopting thetransformation to standard SDSS photometry as detailedin Pinsonneault et al (2012) As 2MASS J minusK is rela-tively constant for a large span of early M dwarfs wechose to utilize iminus J for the blended components inthe fitting Keck-AO data for KOI-2626 from NIRC-2 (Fig 6) allowed PSF fitting to derive photometry forthe individual components of that system in the Ks bandwhich were used to replace the blended i minus J color inthe isochrone fits Our derived WFC3-based photome-try the blended iminus J colors and the Ks band photome-try for KOI-2626 used in the isochrone fitting are listedin Table 2 for Kepler-296 KOI-2626 and KOI-3049 Wechose to use the ∆mag in F775W between components ineach system as the longer wavelength of that filter shouldbe more reliable for our late-type stars than the F555Wphotometry

32 Reddening Corrections

As we did not assume a distance (and therefore a red-dening) value a priori for any of our systems we al-lowed for adjustment of E(BminusV ) in order to find thebest isochrone fit We used the extinction laws for J iand Ks bands from Pinsonneault et al (2012) which are

AJ = 0282timesAV

Ai = 0672timesAV

AKs = 0117timesAV

6 Cartier et al

TABLE 2Observed Photometry

Kepler-296 Photometry

Star F555W F775W Ks Kp F555W-F775W iminus J F775W-Ks

A 16997 15040 ndash 16076 plusmn 0045 1957 ndash ndashB 18874 16396 ndash 17641 plusmn 0053 2478 ndash ndash

A + B 16820 14766 ndash 15845 plusmn 0047 2053 1807 ndashB minus A ndash 1356 ndash ndash ndash ndash ndash

KOI-2626 Photometry

A 17643 15598 13400 16669 plusmn 0047 2045 ndash 2198B 18406 16107 13838 17280 plusmn 0051 2299 ndash 2269C 19289 16900 14520 18109 plusmn 0052 2389 ndash 2380

A+B+C 17057 14886 12634 16010 plusmn 0049 2172 1807 2252B minus A ndash 0509 0438 ndash ndash ndash ndashC minus A ndash 1302 1120 ndash ndash ndash ndash

KOI-3049 Photometry

Note mdash Kp magnitudes and errors derived from Eq 1 and 2

where Aband is the extinction in the desired band andAV = 31 times E(BminusV ) is the extinction in the Vband We calculated the extinction laws for F555Wand F775W with the HST Exposure Time Calculatorfor WFC3UVIS 5 to be

AF555W = 311times E(BminusV )AF775W = 198times E(BminusV )

33 Fitting Using Victoria-Regina Isochrones

Based on the derived WFC3 photometry for the com-ponents of Kepler-296 KOI-2626 and KOI-3049 we an-ticipated that Kepler-296A would match the temperatureof an early M dwarf with Kepler-296B a slightly laterM dwarf (Lepine et al 2013) We also predicted KOI-2626A to be a slightly later M dwarf than Kepler-296AKOI-2626B between Kepler-296A and Kepler-296B andKOI-2626C slightly later than Kepler-296B We expectedboth KOI-3049A and KOI-3049B to be earlier types thanKepler-296A falling near late-Kearly-M dwarfs (Boya-jian et al 2012) Dressing amp Charbonneau (2013) arguethat the Dartmouth Stellar Evolution Database (DSED)(Dotter et al 2008) provides the most state-of-the-artrepresentation of the evolution of M dwarfs and thuswould provide reliable solutions for Kepler-296 KOI-2626 and KOI-3049 Feiden et al (2011) also demon-strated the reliability of the Dartmouth isochrones in fit-ting for late-type stars

We have found that the DSED isochrones systemati-cally underestimate the temperatures masses and radiifor M dwarfs when optical bandpasses are relied upon forthe fitting The latest release of the DSED isochrones in2012 utilizes the BT-Settl model atmosphere line listsand physics of Allard et al (2011) The Dartmouth Stel-lar Evolution Program generated their synthetic photom-etry using the PHOENIX atmospheric code (Hauschildtet al 1999ab) and inputted DSED boundary condi-tions from their isochrone grids Thus while the DSED

5 httpetcstscieduetcinputwfc3uvisimaging

isochrones did not use the exact model atmosphere gridsreleased by Allard et al (2011) the synthetic photom-etry included in the latest DSED release is still subjectto the same strengths and weaknesses as the BT-Settlatmospheres Examination of Fig 2 of Allard et al(2011) and Fig 9 of Mann et al (2013) shows that whilethe synthetic spectra for M dwarfs are remarkably accu-rate for infrared wavelengths the molecular line lists forM dwarfs are incomplete in the optical and thus do notadequately represent the M dwarf spectral energy distri-bution in this wavelength range These regions of thesynthetic spectra are often masked out when attemptingto use the BT-Settl atmospheric spectra to fit to observedM dwarf spectra As BT-Settl appears to overestimatethe SED of M dwarfs in the optical inclusion of opti-cal photometry when attempting to fit using BT-Settlphotometry should always predict more optical flux thanappears for a given stellar temperature so would skewthe fitting towards cooler temperatures This is consis-tent with our comparison with Dressing amp Charbonneau(2013) (see sect5 for more information) The synthetic pho-tometry included in DSED predicts that below a certaintemperature all M dwarfs have the same color in opti-cal bandpasses which does not match our full observa-tional sample (Gilliland et al 2015) The newest releaseof the Victoria-Regina (VR) Stellar Models (VandenBerget al 2014ab Casagrande amp VandenBerg 2014) uses theMARCS model atmospheres that demonstrate increas-ingly red colors for decreasing stellar brightness a muchmore accurate representation of observed M dwarfs inthe solar neighborhood and our full target sample

The discrepancy in photometry tabulated in DSEDand VR can be traced back to the differences betweenthe latest PHOENIX (Allard et al 2011) and MARCS(Casagrande amp VandenBerg 2014) model atmosphere in-puts and physics To solve for the emergent intensity as afunction of wavelength MARCS uses a spherical 1D lo-cal thermodynamic equilibrium (LTE) atmosphere whileBT-Settl uses a spherically symmetric LTE 2D solution

with non-LTE physics for specific species The most sig-nificant difference between these two atmospheric modelsare the molecular lines and opacities included in their cal-culations as well as the inclusion of dust opacities cloudformation condensation and sedimentation BT-Settlincludes all of the aforementioned advanced atmosphericcalculations while MARCS contains limited ionic andmolecular opacities and no dust opacity or high-orderatmospheric physics As these details are most impor-tant for M dwarfs in the infrared it logically follows thatBT-Settl more accurately models stellar photometry inthat range while the missing optical molecular bands inthe PHOENIX models leads to inaccuracies in opticalbandpasses (Allard et al 2011 Mann et al 2013)

Fig 7 shows solar sub-solar and super-solar metal-licity 5 Gyr isochrones from the VR and DSED mod-els with stars from the RECONS project (Henry et al1999 2006 Cantrell et al 2013 Jao et al 2014) within 5pc of the Sun overplotted From this we can see thatthe stellar models are indistinguishable for stars withF555W minus F775W colors bluer than sim 1 Stars with col-ors redder than 1 follow the VR models more closely thanthe Dartmouth models The deviation becomes great-est for colors redder than 25 where the RECONS datashow a continual reddening of color with decrease in mag-nitude which Dartmouth models do not show Initialanalysis using the Dartmouth isochrones yielded stellartemperatures that were significantly hotter than previousstudies suggested (Dressing amp Charbonneau 2013 Muir-head et al 2012) and the lack of consistency with thosecalculations remained troubling until the limitations ofDartmouth models for cool stars in optical bandpasseswere realized We therefore used the synthetic photome-try available for the VR isochrones for F555W F775Wi J and Ks bands to perform our fitting

It has been noted in the past that stars in the solarneighborhood have a sub-solar average [FeH] metallic-ity (Hinkel et al 2014) Therefore the RECONS starsshould fall between the [FeH] = 0 and [FeH] = -05isochrones in Fig 7 The recently released Hypatia Cat-alog (Hinkel et al 2014) which compiles spectroscopicabundance data from 84 literature sources for 50 ele-ments across 3058 stars within 150 pc of the Sun chal-lenges this conclusion After re-normalizing the raw spec-troscopic data of their catalog stars to the same solarabundances they find that the mean [FeH] for thin-disk stars in the solar neighborhood is +00643 and hasa median value of +008 As the Hypatia Catalog indi-cates that solar neighborhood stars are actually slightlysuper-solar in metallicity the location of the RECONSstars in relation to the VR isochrones in Fig 7 appearsconsistent

Using the data and codes provided by VandenBerg etal (2014a) and the interpolation methods described inAppendix A of Casagrande amp VandenBerg (2014) wegenerated ten 5 Gyr isochrones assuming a helium frac-tion of 027 [αFe] = 00 and spanning the metallicityrange [FeH] = minus05rarr +04 in steps of 01 dex We thenlinearly interpolated the generated isochrones halfwaybetween the given points and added calculations of LLand RR from the quantities provided The resultingisochrones contained synthetic photometry for F555WF775W i J and Ks bandpasses as well as fundamentalstellar parameters The final isochrones used spanned a

Fig 7mdash Comparison of 5 Gyr isochrones from the Victoria-Regina Stellar Models (black) and the Dartmouth Stellar EvolutionDatabase (red) Numbers in legend indicate the isochrone value of[FeH] Crosses are stars within 5 pc of the sun from the RECONSproject with absolute photometry

range of 012 MM 12The Kepler light curves for Kepler-296 KOI-2626 and

KOI-3049 all show low amplitude long period variations(sim weeks) which are characteristic of older stars As M-dwarfs evolve little over the course of their very longlives we have adopted an age for all systems of 5 Gyradjustment of this age showed insignificant impact on theresults Assuming these are systems of late-type main se-quence stars we further restricted our isochrone fittingonly to stars with MM le 10 Lastly we requiredthat the brightest component of each system be the mostmassive with the dimmer component(s) being less mas-sive If the systems are truly bound then each componentis at the same distance from us meaning that the appar-ent magnitudes correlate with the effective temperaturesand therefore with the mass

To fit both stellar components of Kepler-296 and KOI-3049 to an isochrone we performed a minimum-χ2 fit-ting between the observed and synthetic photometry de-scribed above We chose to minimize the quadraturesum of the differences for the color of component A thecolor of component B the magnitude difference of B-Ain F775W and the blended iminus J color given as

χ2binary = (∆(F555W minus F775W)AσA)2 (5)

+ (∆(F555W minus F775W)BσB)2

+ (∆ F775WBminusAσBminusA)2

+ (∆(iminus J)A+BσA+B)2

where ∆(F555W minus F775W) are the color differences be-tween the observed colors and the tabulated values in thesynthetic VR isochrones ∆F775WBminusA is the observeddifference in magnitude between components B and Ain the F775W band minus the same quantity from theisochrones and ∆(i minus J)A+B is the i minus J color for theobserved blended A+B photometry minus the blendedisochrone values for A+B The σ values represent the

8 Cartier et al

uncertainties in the measured photometry and were setto 003 mag for Kepler-296 and 002 mag for KOI-3049for colors within the same photometric system and 008for cross-system colors (ie for i minus J )

For the three components of KOI-2626 we performeda similar minimum-χ2 fitting including Ks band pho-tometry in place of i minus J and adding appropriate termsfor component C given as

χ2triple = (∆(F555W minus F775W)AσA)2 (6)

+ (∆(F555W minus F775W)CσC)2

+ (∆(F775W minusKs)AσA)2

+ (∆(F775W minusKs)BσB)2

+ (∆(F775W minusKs)CσC)2

+ (∆ F775WCminusAσCminusA)2

+ (∆ KsBminusAσBminusA)2

+ (∆ KsCminusAσCminusA)2

Terms in Eq 6 are the same as Eq 5 with the ad-dition of ∆(F555W minus F775W) for the C component∆F775WCminusA for the observed difference in magnitudebetween components C and A in the F775W band mi-nus the same quantity from the isochrones and similarquantities for F775W-Ks colors and ∆Ks magnitudes ofall components The σ values in Eq 6 were set to 005mag for all terms except any involving component Cwhich were set to 008 The σrsquos were increased to ac-count for the larger uncertainty in the PSF fitting andthus the contributions of each component to the totalmagnitude When fitting the observed photometry tothe isochrones we used the reduced χ2 metrics whereχ2

binary was reduced by a factor of (1 minus dof) = 3 andχ2

triple was reduced by a factor of (1minus dof) = 9In the fitting of Kepler-296 and KOI-3049 for each pri-

mary mass value (MA) the secondary mass value (MB)that produced the minimum χ2 as per Eq 5 was se-lected assuming MB lt MA The overall best isochronematch was the combination of A and B masses that pro-duced the global minimum χ2

binary This two-level fit-ting was performed for the three binary permutationsof components of KOI-2626 as well to determine thateach binary permutation of the system (A-B A-C andB-C) could also be coeval to ensure that the photom-etry was producing consistent results between combina-tions of components and to provide initial values for themasses of each component in the triple-star fitting Toperform the three-component fitting we took the initialestimates for the masses of each component and searcheda range of surrounding masses for the best fit with thesize of the range dependent on the reliability of the pho-tometry for that component For each mass in the rangeof component A Eq 6 was minimized for every combi-nation of B and C masses The overall combination of AB and C that produced the global minimum of χ2

triple

was adopted as the best fitIn order to test the systematic uncertainties in using

the VR isochrones to determine the stellar mass radiusand bolometric luminosity of our three target systems

we applied an offset to the solar metallicity VR modelin order to match the RECONS stars in Fig 7 Wethen fit the isochrones with the offset to Kepler-296 ac-cording to the method described above to test how theslight offset in metallicity affects the determination ofthe stellar parameters We first fit the solar metallicityisochrone to the Kepler-296 photometry as is then didthe same by applying a shift in F555W-F775W color tomatch RECONS colors and finally by applying a shiftin F775W magnitude to match the RECONS magni-tudes This yielded two measurements of the system-atic uncertainty when fitting for mass radius and lu-minosity We find that the VR models required a shiftof ∆F775W = minus05 or ∆(F555W minus F775W) = +02 inorder to best match the RECONS sampleWe note thatthe chosen shift in color matches the colors of the coolerstars in the sample while being slightly too red to prop-erly match the hotter stars The shift in magnitude didnot affect the fit at all since the search range to matchthe magnitudes of the Kepler-296 components was largerthan the model shift and so the fitting algorithm stillselected the minimum χ2 fit To calculate the system-atic uncertainty of our isochrone fitting we averaged thedifferences between the best fit stellar parameters andthe color-shifted best fit stellar parameters for the pri-mary and secondary stars in Kepler-296 We find that∆M = minus0081M ∆R = minus0071R ∆L = minus0014Land ∆Teff = minus15455K From this we conclude that thesystematic uncertainties when fitting for stellar mass ra-dius and luminosity are small but not insignificant con-tributions to the total error budget

Lacking spectroscopic determinations for metallicityfor Kepler-296 KOI-2626 or KOI-3049 we fit each sys-tem to isochrones of each metallicity in our range atE(BminusV )= 0 to find the best fitting metallicity and thenincreased the reddening to determine whether that wouldprovide a better fit In all cases E(BminusV )=0 providedthe best fits Table 3 provides the minimum χ2 for eachsystem at each metallicity for E(BminusV )=0 Kepler-296and KOI-2626 both show a clear best fit for [FeH] =+03 and +01 respectively While KOI-3049 has a bestfit for [FeH] = minus04 all metallicities tested show ap-proximately the same goodness of fit suggesting the in-dependence of the goodness-of-fit with regard to metal-licity for that system and an even weaker assertion aboutthe true metallicity of KOI-3049 For the evaluation ofplanetary habitability stellar parameters from the bestfit metallicity (highlighted in bold in Table 3) were cho-sen As the best fit χ2 for Kepler-296 is significantlybelow 1 we are likely overestimating our errors for thatsystem

34 False Association Odds

In addition to showing that the suspected companionstars for Kepler-296 KOI-2626 and KOI-3049 are co-eval we performed a Bayesian-like odds ratio analysis onthe three systems to determine the probability that theisochrone fitting described in sect33 could have produced agood match for all components without the stars beingphysically associated (Gilliland et al 2015) For the com-ponents of Kepler-296 the odds ratio associatedrandomwas 410161 for KOI-2626 the ratio was 283291 for theprimary and secondary companions and 92811 for the

TABLE 3Values of the min χ2 for changing values of metallicity

for Kepler-296 KOI-2626 and KOI-3049

[FeH] Kepler-296 KOI-2626 KOI-3049

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

TABLE 4Best fit stellar parameters for the components of

Kepler-296

Parameter Kepler-296A Kepler-296B

MM 0626 plusmn 0082 0453 plusmn 0082Teff [K] 3821 plusmn 160 3434 plusmn 156RR 0595 plusmn 0072 0429 plusmn 0072

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

F555W minus F775W 1952 2490F775WBminusA 1356

Note mdash Tabulated values were calculated for E(BminusV ) = 000[FeH] = +03 age = 5 Gyr and were matched to the observedvalues in Table 2 χ2

min = 0218

primary and tertiary companions for KOI-3049 the ratiowas 192371 From this we conclude that isochrone fit-ting utilizing the photometry of these three cases wouldbe very unlikely to produce a good fit if the stars wererandom superpositions and not truly associated

35 Kepler-296 Best-fit Stellar Parameters

Using the procedures described in sect33 and sect32 wefound that the best fit for the stellar components ofKepler-296 occurred for [FeH] = +03 with MAM =0626 plusmn 0082 and MBM = 0453 plusmn 0082 The tab-ulated temperatures that correspond to these masses inthe VR isochrones are TA = 3821 plusmn 160 K and TB =3434plusmn156 K These roughly correspond to spectral typesM00V and M30V respectively based on the Lepineet al (2013) spectroscopic catalogue of the brightestK and M dwarfs in the northern sky which providedranges and average temperature for each spectral sub-type The stellar radii are RAR = 0595 plusmn 0072 andRBR = 0429plusmn0072 as calculated from the tabulatedvalues of Teff and stellar luminosity from the isochronesErrors on all of these values are δX =

radic1σ2

iso + ∆(X)2where 1σiso are the 1σ errors above the minimum re-duced χ2 value of 0218 from the isochrone fitting and∆(X) are the systematic uncertainties in the isochronefitting as described in sect33 Fig 8 shows the variationof χ2 (calculated as in Eq 5) with the best-fit massesof the primary and secondary component of Kepler-296indicated The 1σiso errors were calculated by findingthe two points along the χ2 curves in Fig 8 that cor-responded to values of χ2

min + 157 accounting for 4

KOI-2626

Parameter KOI-2626A KOI-2626B KOI-2626C

MM 0501 plusmn 0086 0436 plusmn 0086 0329 plusmn 0085Teff [K] 3649 plusmn 166 3523 plusmn 160 3391 plusmn 158RR 0478 plusmn 0075 0415 plusmn 0077 0321 plusmn 0076

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

F775W minusKs 2221 2321 2435F775WBminusA 0518F775WCminusA 1321Ks BminusA 0420Ks CminusA 1107

min = 0860

degrees of freedom in the fit (Press et al 1986) Theoptimal stellar parameters and their errors are tabulatedin Table 4

We calculated the distance to Kepler-296 by applyingthe distance modulus formula to the observed and ab-solute magnitudes of each component in each HST filterthen averaging the four estimates The absolute magni-tudes from the isochrone match combined with the ap-parent magnitudes from our HST imaging implies a dis-tance to Kepler-296 of 360plusmn 20 pc At this distance theempirically measured separation of 0primeprime217plusmn 0primeprime004 trans-lates to a physical separation of 80plusmn5 AU and an orbitalperiod of 660 plusmn 60 years The true values of both theseparation and period are likely larger due to projectioneffects foreshortening the true separation and orbital pe-riod

36 KOI-2626 Best-fit Stellar Parameters

The best fit for KOI-2626 occurred for [FeH] = +01with MAM = 0501plusmn 0086 MBM = 0436plusmn 0086and MCM = 0329 plusmn 0085 The tabulated tem-peratures that correspond to these masses in the VRisochrones are TA = 3649 plusmn 166 K TB = 3523 plusmn 160 Kand TC = 3391 plusmn 158 K These temperatures trans-late roughly to M10V M20V and M25V respectivelybased on Lepine et al (2013) The stellar radii areRAR = 0478 plusmn 0075 RBR = 0415 plusmn 0077 andRCR = 0321plusmn0076 as calculated from the tabulatedvalues of Teff and stellar luminosity from the isochronesThese parameters are tabulated in Table 5 Curves show-ing the variation of χ2 (calculated as in Eq 6) as a func-tion of stellar mass similar to Fig 8 were created andused to determine the best fit and 1σiso points The listederrors are calculated as in sect35 with 1σiso =χ2

min + 128above the minimum χ2 value of 0860 accounting for the10 degrees of freedom in the fitting (Press et al 1986)

The absolute magnitudes from the isochrone matchcombined with the apparent magnitudes from our HSTimaging implies a distance to KOI-2626 of 340 plusmn 35 pcAt this distance the empirically measured separation of0primeprime203 between components A and B translates to a phys-ical separation of 70plusmn 7 AU and for the measured sepa-ration of components A and C of 0primeprime161 we calculated a

10 Cartier et al

Fig 8mdash Left variation of χ2 from Eq 5 for MM for component A of Kepler-296 Right same as left panel for component B ofKepler-296 Black curve shows the variation of χ2 red dashed line shows mass of components for the minimum χ2

KOI-3049

Parameter KOI-3049A KOI-3049B

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

Note mdash Tabulated values were calculated for E(BminusV ) = 0[FeH] = -04 age = 5 Gyr and were matched to the observedvalues in Table 2 χ2

min = 0907

physical separation of 55plusmn 6 AU Again the real valuesare likely larger due to projection effects

The best fit for the components of KOI-3049 occurredfor [FeH] = minus04 We find that MAM = 0607plusmn0081and MBM = 0557 plusmn 0081 The tabulated tem-peratures that correspond to these masses in the VRisochrones are TA = 4529plusmn163 K and TB = 4274plusmn159 KThese effective temperatures match approximately toK40V and K55V respectively based on the spectraltypes tabulated in Boyajian et al (2012) as the tem-peratures are outside the range provided by Lepine etal (2013) We find the stellar radii to be RAR =0588plusmn 0071 and RBR = 0536plusmn 0071 The optimalstellar parameters and their errors are tabulated in Ta-ble 6 Curves showing the variation of χ2 (calculated asin Eq 5) as a function of stellar mass similar to Fig 8were created and used to determine the best fit and 1σpoints The listed errors are determined as in sect35 with1σiso calculated using the minimum χ2 value of 0907

The absolute magnitudes from the isochrone matchcombined with the apparent magnitudes from our HST

imaging implies a distance to KOI-3049 of 485 plusmn 20 pcAt this distance the empirically measured separationof 0primeprime464 plusmn 0primeprime004 translates to a physical separation of225 plusmn 10 AU and an orbital period of 3150 plusmn 205 yearsAgain the true values are likely larger due to projectioneffects

38 Isochrone Fit Discussion

To compare the best-fit stellar properties of Kepler-296 KOI-2626 and KOI-3049 we plotted each compo-nent atop their respective best fit isochrones in Fig 9The observed photometry tabulated in Table 2 was con-verted to absolute photometry using the distances de-rived from the respective isochrone fits From Fig 9we note that our initial guesses at the relative magni-tudes of the components of all three systems were cor-rect and that Kepler-296 and KOI-3049 are very likelybound binary systems based on their close fits to theVR isochrones The only star that falls somewhat off ofthe isochrone is KOI-2626 B which appears to be slightlyredder than the isochrone fit would suggest However asKOI-2626 B still fits the isochrone within its 1σ error oncolor we still report with high confidence that KOI-2626is a bound triple star system

4 PLANETARY HABITABILITY

The multiplicity of Kepler-296 KOI-2626 and KOI-3049 have interesting implications on the habitabilityof the planets in each system Dressing amp Charbon-neau (2013) determined that the planets Kepler-296 d(the third planet in the system) and KOI-262601 (theonly detected planet candidate in the system) were hab-itable given the systemsrsquo previously assumed single-starproperties Mann et al (2013) re-evaluated the temper-atures of these stars using stellar temperatures derivedfrom mid-resolution spectra and found that those twoplanets were actually interior to their respective Habit-able Zones However neither of those studies accountedfor the multiplicity of those systems and thus their HZanalyses are inaccurate for these targets Knowing now

Fig 9mdash Absolute photometry of stellar components of Kepler-296 KOI-2626 and KOI-3049 plotted over their respective bestfit 5 Gyr isochrones Kepler-296 components are in red circlesplotted over an [FeH] = +03 isochrone (red solid line) KOI-2626 components are in blue squares plotted over an [FeH] =+01 isochrone (blue dashed) KOI-3049 components are in greentriangles plotted over an [FeH] = -04 isochrone (green dotted)Error bars are 1σ Spectral types are from Lepine et al (2013)for types later than K60 and from Boyajian et al (2012) for typesearlier than K60

that Kepler-296 KOI-2626 and KOI-3049 are multiple-star systems we recalculated the planetary parameters ofall detected planets around each potential stellar host us-ing the best-fit stellar parameters in order to re-evaluatethe planetary habitability

Circumbinary and circum-triple planetary orbits werenot tested for habitability as the wide physical separa-tions of the systems coupled with the short transit pe-riods preclude planetary orbits around multiple starsOur projected separations of the stellar components ofKepler-296 KOI-2626 and KOI-3049 indicate that theyare either close or moderately separated systems butas we cannot correct for projection effects the systemscould be more widely separated While circum-primaryorbits reduce the likelihood of the additional stellar com-ponent(s) interacting catastrophically with the planetaryorbits we tested the habitability of each planet assumingan orbit around each stellar component separately as wecurrently lack data indicating which stars host which (orany) planets in these systems

The existence of other bright stars in the Kepler pho-tometric aperture (in this case due to the stellar multi-plicity of the systems) required that the recorded transitdepth be corrected for the light dilution from the addi-tional star(s) To account for the transit dilution wescaled the blended transit depth observed by Kepler bythe photometric contribution of the star of interest as

∆Ftrue = ∆FMASTdilution (7)

where ∆FMAST is the transit depth as measured by Ke-pler and dilution is the fraction of the blended light in

TABLE 7Transit Parameters for Kepler-296 KOI-2626 and

KOI-3049Components

Planeta ∆FMASTb ∆Ftrue

c Period b

[ppm] [ppm] [days]

Kepler-296 Ac 14230 plusmn 281 17677 plusmn 349 5842Kepler-296 Ad 15670 plusmn 412 19466 plusmn 512 19850Kepler-296 Ab 8200 plusmn 363 10186 plusmn 451 10864Kepler-296 Af 9790 plusmn 608 12161 plusmn 755 63338Kepler-296 Ae 7870 plusmn 458 9776 plusmn 568 34142

Kepler-296 Bc 14230 plusmn 281 72974 plusmn 1439 5842Kepler-296 Bd 15670 plusmn 412 80359 plusmn 2115 19850Kepler-296 Bb 8200 plusmn 363 42051 plusmn 1861 10864Kepler-296 Bf 9790 plusmn 608 50205 plusmn 3118 63338Kepler-296 Be 7870 plusmn 458 40359 plusmn 2346 34142

KOI-2626 A01 8180 plusmn 473 15064 plusmn 871 38098KOI-2626 B01 8180 plusmn 473 26908 plusmn 1555 38098KOI-2626 C01 8180 plusmn 473 53464 plusmn 3090 38098

KOI-3049 A01 5400 plusmn 320 8668 plusmn 513 22477KOI-3049 B01 5400 plusmn 320 14324 plusmn 848 22477

aldquoKepler-296 Acrdquo etc indicates the solution for planet c aroundcomponent A of Kepler-296bFrom MASTcCorrected for dilution from the stellar companion via Eq 7

the Kepler aperture that is contributed by the individualstellar components The dilutions to the transit depthwere calculated using the PSF fitting (sect31) coupled withthe KpminusHST conversion (sect23) and are listed in sect31As each star is smaller and cooler than the raw Keplerphotometry indicates (as Kepler only shows the blendedsystem) the relative drop in the stellar flux due to thetransit is actually larger than was measured which inturn increases the ratio of RpRlowast The input transit pa-rameters used in the habitability calculations are foundin Table 7 The errors listed for ∆Ftrue were calculatedusing the detection SN and the archive-listed transitdepth in parts per million

41 Calculation of Planetary Parameters

Using the transit parameters listed in Table 7 we cal-culated the planet radius the semi-major axis the equi-librium temperature and incident stellar flux of eachplanet around each of its potential host stars usingthe equations listed in Seager amp Mallen-Ornelas (2003)Planetary masses and bulk densities were calculated us-ing the formalisms of Weiss amp Marcy (2014) and Lissaueret al (2011) These formalisms do not take into accountstellar limb darkening instead assuming a uniform stellardisk We provide these results as a first order calculationand provide the results of limb darkened model fits to thefull folded time series in the next subsection

The planetary radius was directly calculated from thestellar radius and the transit depth using the equationsof Seager amp Mallen-Ornelas (2003) as

Rp = R

radic∆Ftrue (8)

where ∆Ftrue is the dilution-corrected transit depthfrom Eq 7 and R is the stellar radius The plane-tary orbital semi-major axis was calculated from the KICtransit period and the best-fit stellar mass using

ap = aoplus

Poplus

12 Cartier et al

where Pp is the planetary orbital period and M is thestellar mass The semi-major axis calculated in Eq 9was combined with the best-fit stellar effective tempera-ture and radius to get the planetary equilibrium temper-ature via

Teq = Teff(1minusA)14

radicR

2 ap(10)

where A is the assumed Bond albedo of 03 and ap is theplanetary semi-major axis as calculated in Eq 9 Thisequilibrium temperature does not account for any po-tential greenhouse effects which would warm the surfaceand are unavoidable if there is any liquid water on thesurface Next the stellar flux incident on the planet wascalculated relative to the flux received at Earth by

)2(TlowastT

where ap is the planetary semi-major axis R is the stel-lar radius Tlowast is the stellar temperature and T = 5779 Kis the adopted value of solar effective temperature

Lastly the mass and density of the planets were cal-culated using the empirical relations of Weiss amp Marcy(2014) for planets less than 4 Earth-radii given as

ρp = 243 + 339

Roplus

)gcm3 (12)

for RpRoplus lt 15 and

Moplus= 269

Roplus

gcm3 (13)

for 15 le RpRoplus lt 4 The relation of Lissauer et al(2011) was used for planets with RpRoplus ge 4 as

Roplus

Moplus (14)

which fits exoplanet observations for planets smaller thanSaturn Conversion between mass and density was doneusing

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

We used the formalism of Kopparapu et al (2013) todetermine the habitability of the planets Using Eq 2from that paper we calculated the locations of the moistgreenhouse limit (inner) and the maximum greenhouselimit (outer) for each of our component stars and com-pared the limits to the calculated effective stellar fluxincident on the planets from Eq 11 If a planet fallsbetween the moist and maximum greenhouse limits weconsidered it to be habitable The moist and maxi-mum greenhouse limits were chosen to be conservativelocations of the Habitable Zone though for stars withTeff 5000 K the moist greenhouse limit is indistinguish-able from the runaway greenhouse limit

The projected separations of the stellar components inboth systems range from sim 50minus225AU while the orbitalperiods of the planets as measured by Kepler are on the

order of weeks The wide separations of the componentsof each system greatly reduce the chances that the stellarcomponents produce overlapping Habitable Zones like inclose (ie lt 50AU) multi-star systems (Kaltenegger ampHaghighipour 2013) Furthermore censuses of the popu-lations of protoplanetary disks in wide (amp 40AU) binarysystems show that the influence of a binary companionreduces the lifetime of the disk by a few Myr which de-creases the likelihood of planet formation (Kraus et al2012) As these systems successfully completed planetformation the protoplanetary disk was likely only af-fected minimally by the stellar companion(s) furthersuggesting independent Habitable Zones

42 Transit Light Curve Fitting

The above evaluation of planet habitability in each sys-tem is accurate to first order but the equations in sect41do not account for stellar limb darkening orbital eccen-tricity inclination or impact parameter These exclu-sions affect our calculation of the planetary radius andmass and thus could potentially change our conclusionsabout planetary habitability We adopted a more robustmethod of transit analysis by fitting a transit model us-ing an MCMC algorithm to iteratively solve for the bestfitting transit model Attempts at using publicly avail-able MCMC transit fitting software including the Tran-sit Analysis Package (TAP Gazak et al 2012) EXO-FAST (Eastman et al 2013) and PyKE packages (Stillamp Barclay 2012) illuminated limitations in dealing withlow mass and low stellar temperature cases We foundthat the transit identifying function autokep built in toTAP was unable to identify the transits of these systemswithout first stitching together light curves from all ofthe quarters folding them on their linear ephemeridesand binning the phase-folded light curve using PyKEpackages The EXOFAST transit fitter attempted firstthrough the TAP GUI and then use of the functiondirectly showed that their stellar mass-radius relation(Torres et al 2010) was unable to handle stellar massesbelow 06 M and that their limb-darkening interpolationfunctions were unsupported for stellar temperatures be-low 3500 K While tests using EXOFAST showed that thetransit solutions for M gt 06M Teff gt 3500K transitswere reliable the mass and temperature limits imposedby the program during execution were unsuitable for thestellar solutions in this study

We modified both the EXOFAST code itself and theinput transit light curves We applied an adaptive bin-ning algorithm to the input transit light curves to ensurethat the transit itself was properly sampled This prop-erly preserved the shape and depth of the transits whilereducing computation time with broader bins outside oftransit We took the mean time of all the data pointswithin a bin as the bin time value rather than the binmidpoint to account for any clumps or gradients withina bin and aid in accurate reproduction of transit shapeWe used Poisson statistics to calculate the uncertaintyin the mean flux value of each bin this led to smalleruncertainties in the out-of-transit points and larger un-certainties within the transit which allowed EXOFASTto properly weight each binned flux value Finally afterbinning the light curves for each planet in our samplewe applied the stellar dilution corrections directly to thelight curves themselves using Eq 7 as before This pro-

duced a separate light curve for each possible planetstarpermutation EXOFAST was then used in a mode thatintegrates the Mandel amp Agol (2002) light curve modelover a long cadence period (294 minutes) a smoothingto the data that applies even when binning within tran-sits to shorter intervals

Within the EXOFAST package itself we overrode thebuilt-in stellar mass-radius relation from Torres et al(2010) since the function was unreliable when extrap-olated to stellar masses below 06 M As we wantedto enforce our isochrone solutions for the stellar massand radius we imposed those solutions as prior valuesand calculated the prior widths from our uncertaintiesin the stellar mass and radius solutions We then addeda penalty to the χ2calculation within EXOFAST for de-viating from the desired stellar mass and radius Theuncertainties in the stellar mass and radius from theisochrone fitting are then accuratly propagated throughEXOFAST into the posterior distributions and result-ing uncertainties for the planetary values We utilizedthe online limb darkening applet from Eastman et al(2013) to calculate stellar limb darkening priors for ourtransit fitting to support calculation of limb darkeningcoefficients for stellar temperatures below 3500K Theonline limb darkening utility interpolates the quadraticlimb darkening tables of Claret amp Bloemen (2011) givena bandpass effective temperature surface gravity andstellar metallically We calculated the quadratic limbdarkening separately and imposed those values as addi-tional priors with small prior widths In addition to pri-ors on the stellar properties the planetary orbital periodand transit center time we included a prior restrictionon the orbital eccentricity to downweight high eccentric-ity solutions that are unphysical and skew the posteriordistributions of all related variables

We applied these modifications to EXOFAST and theinput transit light curves and then fit transit modelsto the light curves for each possible permutation ofplanet and star as done previously with the analyticsolutions Before accepting the EXOFAST solution asldquogoodrdquo we assured that the reduced χ2 of the transit fitwas sim 1 that the best fit stellar parameters indicated byEXOFAST (especially the stellar effective temperature)matched our isochrone solutions within 1σ and that thecalculated RP Rlowast matched the value calculated analyt-ically in Eq 8 As the MCMC fitting did not accountfor the observed HST photometry which constrained ourstellar solutions these checks ensured that the MCMCalgorithm did not diverge from the isochrone fits or indi-cate a solution that was not consistent with observations

43 Implications on Habitability

Table 8 lists the calculated planetary parameters foreach planet around each potential stellar host for boththe analytic method and the EXOFAST method Thetabulated EXOFAST solutions are the median values andthe 68 confidence intervals on the posterior MCMCdistributions We find planetary radii that range from157Roplus to 423Roplus and are larger than those listed inthe Mikulski Archive for Space Telescopes6 (MAST)due to the dilution corrections Regardless of the hoststar around which the planets orbit all planets around

6 httparchivestsciedu

Kepler-296 and the single planets around KOI-2626 andKOI-3049 are super-Earthsmini-Neptunes Our calcu-lated values of planetary radius are larger than those tab-ulated in Dressing amp Charbonneau (2013) and Muirheadet al (2012) for Kepler-296 c Kepler-296 d and Kepler-296 b and larger than the radii recorded in MAST forall planets in the Kepler-296 system due to our inclusionof the transit depth dilution Our planetary radius forKOI-262601 is also larger than those recorded in MASTand Dressing amp Charbonneau (2013) and our radius forKOI-304901 is larger than the MAST value for the samereason

Upon comparison of the analytic and EXOFAST solu-tions we note that the planetary radius (rather RpRlowastin the calculation) and the effective stellar flux are mildlydependent on the inclusion of limb darkening and con-sequently the planetary mass and equilibrium tempera-tures are also mildly dependent on the inclusion of higherorder calculations As expected planets that fall in theHZ according to the analytic solutions are still habitablewith the EXOFAST calculations either falling directlywithin the HZ or within 1σ of the inner edge of the HZ

Figure 10 displays a subset of planets that fall in ornear the Habitable Zones of their potential host staraccording to the EXOFAST solutions and helps high-light the differences between our calculations and thoseof of Dressing amp Charbonneau (2013) and Muirhead etal (2012) Both Dressing amp Charbonneau and Muirheadet al determined that Kepler-296 d was in the HabitableZone of the assumed single star Using our stellar solu-tions for Kepler-296 Kepler-296 d is not habitable aroundeither star and in fact falls significantly interior to theHabitable Zone of either star The outermost planet inthe system (Kepler-296 f) now falls comfortably withinthe Habitable Zones of both the primary and the sec-ondary stars Kepler-296 e also falls just barely interiorto the Habitable Zone of the secondary but the uncer-tainty on the effective stellar flux at that planet makes itanother likely habitable candidate Neither Dressing ampCharbonneau nor Muirhead et al reported on the statusof Kepler-296 f or Kepler-296 e due to the timing of thetwo studies

The multiplicity of KOI-2626 also changes our under-standing of the habitability of its single planet Dressingamp Charbonneau report that KOI-262601 falls within theHabitable Zone of the assumed single star but our resultsshow that this is only possible around the tertiary starThe uncertainty in the effective stellar flux indicates thatKOI-262601 may also be habitable around the primaryand secondary stars despite its location interior to theHZ

Lastly we find that the multiplicity of KOI-3049 doesnot improve its planetrsquos chances of habitability Evenwith the stellar dilution to the transit depth accountedfor KOI-304901 remains well interior to the HabitableZone around both the primary and secondary compo-nents as it also did for the initial single-star analysis

5 DISCUSSIONS AND FUTURE WORK

Dressing amp Charbonneau (2013) report a temperaturefor the blended Kepler-296 of 3424 plusmn 50 K while Muir-head et al (2012) report a temperature of 3517 K basedon spectral index matching Our best-fit isochrone tem-peratures for both components A and B are warmer than

14 Cartier et al

TABLE 8Analytic and EXOFAST Solutions for Kepler-296 KOI-2626 and KOI-3049 Planets

Planeta Rp aP Mp ρp Teq Seff HZb

[Roplus] [AU] [Moplus] [gcm3] [K] [S0]

Kepler-296 Ac 275 plusmn 033 0054 69 18 5586 plusmn 410 2292 plusmn 673 no335 plusmn 021 0054 83 12 6060 plusmn 320 2263 plusmn 220 no

Kepler-296 Ad 288 plusmn 035 0123 72 17 3715 plusmn 273 449 plusmn 132 no269 plusmn 021 0123 68 19 4030 plusmn 215 426 plusmn 098 no

Kepler-296 Ab 209 plusmn 026 0082 53 32 4542 plusmn 333 1002 plusmn 294 no215 plusmn 021 0082 55 30 4950 plusmn 255 1007 plusmn 458 no

Kepler-296 Af 228 plusmn 028 0266 58 27 2524 plusmn 185 095 plusmn 028 maybe208 plusmn 021 0266 53 32 2740 plusmn 150 088 plusmn 046 yes

Kepler-296 Ae 204 plusmn 025 0176 52 34 3101 plusmn 228 218 plusmn 064 no186 plusmn 017 0176 48 41 3370 plusmn 175 204 plusmn 062 no

Kepler-296 Bc 403 plusmn 068 0049 177 15 4503 plusmn 429 968 plusmn 369 no378 plusmn 045 0049 93 09 4970 plusmn 270 999 plusmn 148 no

Kepler-296 Bd 423 plusmn 071 0110 195 14 2995 plusmn 286 189 plusmn 072 no400 plusmn 045 0110 174 15 3310 plusmn 215 198 plusmn 071 no

Kepler-296 Bb 306 plusmn 052 0074 76 15 3661 plusmn 349 423 plusmn 161 no291 plusmn 063 0074 73 16 3950 plusmn 330 382 plusmn 112 no

Kepler-296 Bf 335 plusmn 057 0239 83 12 2034 plusmn 194 040 plusmn 015 yes278 plusmn 040 0240 70 18 2140 plusmn 165 034 plusmn 031 yes

Kepler-296 Be 300 plusmn 051 0158 75 15 2500 plusmn 237 092 plusmn 035 maybe272 plusmn 038 0158 68 19 2730 plusmn 175 091 plusmn 048 maybe

KOI-2626 A01 204 plusmn 033 0176 52 34 2656 plusmn 242 117 plusmn 043 maybe186 plusmn 025 0176 48 41 2890 plusmn 200 113 plusmn 058 maybe

KOI-2626 B01 237 plusmn 044 0168 60 25 2446 plusmn 252 084 plusmn 035 yes247 plusmn 035 0176 62 23 2780 plusmn 185 099 plusmn 053 maybe

KOI-2626 C01 258 plusmn 062 0153 65 21 2169 plusmn 276 052 plusmn 027 yes265 plusmn 028 0150 66 20 2520 plusmn 130 068 plusmn 037 yes

KOI-3049 A01 190 plusmn 024 0132 49 39 4221 plusmn 298 747 plusmn 211 no157 plusmn 010 0132 41 58 4610 plusmn 205 757 plusmn 117 no

KOI-3049 B01 223 plusmn 030 0128 57 28 3861 plusmn 294 523 plusmn 160 no197 plusmn 017 0128 51 36 4360 plusmn 220 588 plusmn 110 no

Note mdash The first row for each planet contains the analytic planet solution and the second row for each planet contains the EXOFASTplanet solution The HZ determination is italicized for the EXOFAST solution and bolded for any HZ planets

aThe notation ldquoKepler-296 Acrdquo etc indicates the solution for planet c around component A of Kepler-296bHZ indicates falling between the moist greenhouse inner limit and max greenhouse outer limit ldquomayberdquo indicates falling within 1σ of

the HZ

the Dressing amp Charbonneau values However our tem-peratures do straddle the blended temperature of Muir-head et al (2012) as expected Mann et al (2013) reportTeff = 3622 K for Kepler-296 which also falls between ourtemperatures of the individual components as expectedLikewise for KOI-2626 Dressing amp Charbonneau (2013)adopt a value of Teff = 3482 K which falls between ourvalues for components B and C while Mann et al (2013)report Teff = 3637 K which falls between our solutionsfor components A and B That our solutions agree withblended temperature estimates derived using two differ-ent methods suggests that the VR isochrones provideda logical solution for both Kepler-296 and KOI-2626Muirhead et al (2012) did not include the KOI-2626system in their studies and none of the aforementionedreports included KOI-3049

Our initial analysis attempted to follow the procedureoutlined in earlier sections of this paper but utilizing theDSED isochrones in place of the VR isochrones Thiswas initially an attempt to best compare to the stud-ies of Dressing amp Charbonneau (2013) and Muirhead etal (2012) the former of which also fit to Dartmouthisochrones and the latter which produced consistent re-sults using spectroscopic methods Our first results fromusing the Dartmouth isochrones indicated temperatures

for all components that were much hotter than the tem-peratures reported by both studies (and later reported byMann et al (2013) as well) Investigating the cause ofthis difference we attempted first to replicate the resultsof Dressing amp Charbonneau (2013) regarding the tem-perature of Kepler-296 using the same seven bands thatwere used in that study (grizJHK) We were able tomatch the Dressing amp Charbonneau (2013) Teff to within100 K and found that the inclusion on the SDSS g bandphotometry skewed the isochrone fitting to significantlycooler temperatures Dropping the g band photometryproduced a warmer midpoint between A and B tempera-tures and a large drop of χ2 while exclusion of any otherband made little difference on the temperature midpointor χ2 Knowing a priori the late spectral types of thetargets we observe that the inclusion of g band photom-etry may bias some of the isochrone solutions of Dress-ing amp Charbonneau Photometry in the g band is alsoobservationally suspect in the KIC at those faint mag-nitudes (Brown et al 2011) The photometric issues arethen coupled with the uncertainties of the Dartmouthisochrones for late-type stars as discussed in sect33 Wealso note that our analysis is limited to the use of opti-cal and near-optical bandpasses which are not the mostreliable wavelength ranges for cooler stars To mitigate

Fig 10mdash Stellar effective temperature versus effective incidentstellar flux from EXOFAST in solar units for planets in and nearthe Habitable Zones of their respective stars Red circles indicateKepler-296 A gold squares indicate Kepler-296 B and blue trian-gles indicate KOI-2626 Moist and max greenhouse curves are cal-culated using formalism of Kopparapu et al (2013) Any planetsnot shown fall significantly interior to the Habitable Zone Planetlabels as in Table 7

this we relied more heavily on our NIR bandpass overour optical bandpass when fitting our photometry to theVR isochrones Inclusion of infrared bands for these tar-gets will likely affect the temperatures derived from theisochrone fitting and reduce the differences between VRand Dartmouth isochrones

Habitable planets in the canonical sense must not onlyhave the capability for liquid water on the surface butalso have a solid surface on which that water can exist Inshort the planets must be rocky and not gaseous Usingradial velocity measurements coupled with Doppler spec-troscopy high-resolution imaging and asteroseismologyMarcy et al (2014) measured the radii and masses for65 planet candidates and concluded that only planetswith radii less than sim 15Roplus are compatible with purelyrocky compositions Planets larger than that must havea larger fraction of low-density material eg H Heand H2O Our updated planet radii from EXOFASTindicate that none of our potentially habitable plan-ets (Kepler-296 Af Kepler-296 Bf Kepler-296 Be KOI-2626 A01 KOI-2626 B01 and KOI-2626 C01) are smallenough to have purely rocky compositions according toMarcy et al (2014) and thus are not habitable in thecanonical sense KOI-3049 A01 however is within 1σof the purely rocky composition limit and so may stillbe a rocky planet We cannot exclude the possibil-ity of a very massive yet rocky planet like Kepler-10c(Dumusque et al 2014) as we lack radial velocity mea-surements needed to calculate the planetary masses anddensities directly Even if Kepler-296 Af Kepler-296 BfKepler-296 Be KOI-2626 A01 KOI-2626 B01 and KOI-2626 C01 remain too large to be rocky the possibility ofhabitable exomoons would remain

6 CONCLUSION

Using the results of our HST GOSNAP program GO-12893 we derived HST-based photometry for the hostsof some of the most interesting Kepler planet candi-dates and created a conversion between the broad-bandKp and our two filters from HST We utilized the em-pirical PSF from Gilliland et al (2015) for Kepler-296KOI-2626 and KOI-3049 three Kepler targets that wererecently discovered to be tight multi-star systems withsmall and cool planets Based on the goodness of the bi-nary isochrone fitting we determined that componentsA and B in Kepler-296 are almost certainly a bound co-eval system consisting of two early-M dwarfs Based onthe updated stellar properties from the Victoria-ReginaStellar Model isochrone matches we found that the sys-tem still contains a potentially habitable planet aroundits primary star and two potentially habitable planetsaround its secondary star with all other combinationsof star-planet producing too-hot planets Likewise wefound that KOI-2626 is likely a bound coeval triple starsystem containing three early- to mid-M dwarfs with asingle planet that is potentially habitable around any ofthe stellar components Lastly while KOI-3049 is likelyalso a bound binary K dwarf system its single planetis not habitable around either stellar component Whilethe sizes of Kepler-296 Af Kepler-296 Bf Kepler-296 BeKOI-2626 A01 KOI-2626 B01 and KOI-2626 C01 in-dicate that those planets are most likely gaseous KOI-3049 A01 likely has a mostly rocky compositions basedon the work of Marcy et al (2014) though it is wellinterior to the HZ of its star The six potentially habit-able planets have densities more consistent with a highergaseous fraction and are not likely habitable in the canon-ical sense

KMSC performed analyses found in sect2 sect3 and sect4and discussion in sect1 sect5 and sect6 RLG contributedanalysis to sect31 and sect34 as well as overall guidance anddirection for this work and the companion paper Gillilandet al (2015) JTW contributed to sect1 sect6 and valuablediscussion and advice regarding isochrone use DRCcontributed Keck AO K-band data to sect36 and provideddiscussion on KOI-2626 KMSC and RLG have beenpartially supported through grant HST-GO-1289301-Afrom STScI We thank Don VandenBerg for permittinguse of the latest Victoria-Regina Stellar Models beforepublication We also thank Sharon X Wang for discus-sion on error analysis for our isochrone fitting

Some of the data presented in this paper were obtainedfrom the Mikulski Archive for Space Telescopes (MAST)STScI is operated by the Association of Universitiesfor Research in Astronomy Inc under NASA contractNAS5-26555 Support for MAST for non-HST data isprovided by the NASA Office of Space Science via grantNNX13AC07G and by other grants and contracts Thispaper makes use of data collected by the Kepler missionFunding for the Kepler mission is provided by the NASAScience Mission directorate Some of the data presentedherein were obtained at the WM Keck Observatorywhich is operated as a scientific partnership amongthe California Institute of Technology the Universityof California and the National Aeronautics and SpaceAdministration The Observatory was made possibleby the generous financial support of the WM Keck

16 Cartier et al

Foundation The Center for Exoplanets and HabitableWorlds is supported by the Pennsylvania State Univer-sity the Eberly College of Science and the PennsylvaniaSpace Grant ConsortiumWe gratefully acknowledge the

use of SOANASA ADS NASA and STScI resources

Facilities HST (WFC3) Kepler

REFERENCES

Allard F Homeier D amp Freytag B 2011 16th CambridgeWorkshop on Cool Stars Stellar Systems and the Sun 448 91

Batalha N M Rowe J F Bryson S T et al 2013 ApJS204 24

Borucki W J Koch D Basri G et al 2010 Science 327 977Borucki W J Koch D G Basri G et al 2011 ApJ 736 19Boyajian T S von Braun K van Belle G et al 2012 ApJ

757 112Brown T M Latham D W Everett M E amp Esquerdo G A

2011 AJ 142 112Burke C J Bryson S T Mullally F et al 2014 ApJS 210 19Cantrell J R Henry T J amp White R J 2013 AJ 146 99Casagrande L amp VandenBerg D A 2014 MNRAS 444 392Claret A amp Bloemen S 2011 AampA 529 AA75Croll B Rappaport S DeVore J et al 2014 ApJ 786 100Dotter A Chaboyer B Jevremovic D et al 2008 ApJS 178

89Dressing C D amp Charbonneau D 2013 ApJ 767 95Dumusque X Bonomo A S Haywood R D et al 2014 ApJ

789 154Eastman J Gaudi B S amp Agol E 2013 PASP 125 83Feiden G A Chaboyer B amp Dotter A 2011 ApJ 740 L25Fressin F Torres G Charbonneau D et al 2013 ApJ 766 81Fruchter AS Hack W Dencheva N Droettboom M

Greenfield P 2010 STSCI Calibration Workshop ProceedingsBaltimore MD STScI 376

Fukugita M Ichikawa T Gunn J E et al 1996 AJ 111 1748Gazak J Z Johnson J A Tonry J et al 2012 Advances in

Astronomy 2012Gilliland R L amp Rajan A 2011 Instrument Science Report

WFC3 2011-03 (Baltimore MD STScI)Gilliland R L Cartier K M S Adams E R et al 2015 AJ

149 24Gonzaga S Hack W Fruchter A amp Mack J 2012 The

DrizzlePac Handbook Baltimore STScIHauschildt P H Allard F amp Baron E 1999 ApJ 512 377Hauschildt P H Allard F Ferguson J Baron E amp

Alexander D R 1999 ApJ 525 871Henry T J Franz O G Wasserman L H et al 1999 ApJ

512 864Henry T J Jao W-C Subasavage J P et al 2006 AJ 132

2360Hinkel N R Timmes F X Young P A Pagano M D amp

Turnbull M C 2014 AJ 148 54Howard A W Marcy G W Bryson S T et al 2012 ApJS

201 15

Jao W-C Henry T J Subasavage J P et al 2014 AJ 14721

Kaib N A Raymond S N amp Duncan M 2013 Nature 493381

Kaltenegger L amp Haghighipour N 2013 ApJ 777 165Kasting J F Whitmire D P amp Reynolds R T 1993 Icarus

101 108Kopparapu R K 2013 ApJ 767 L8Kopparapu R K Ramirez R Kasting J F et al 2013 ApJ

765 131Kratter K M amp Perets H B 2012 ApJ 753 91Kraus A L Ireland M J Hillenbrand L A amp Martinache F

2012 ApJ 745 19Lepine S Hilton E J Mann A W et al 2013 AJ 145 102Lissauer J J Marcy G W Bryson S T et al 2014 ApJ

784 44Lissauer J J Ragozzine D Fabrycky D C et al 2011 ApJS

197 8Mandel K amp Agol E 2002 ApJ 580 L171Mann A W Gaidos E amp Ansdell M 2013 ApJ 779 188Marcy G W Isaacson H Howard A W et al 2014 ApJS

210 20

Muirhead P S Hamren K Schlawin E et al 2012 ApJ 750L37

Petigura E A Howard A W amp Marcy G W 2013Proceedings of the National Academy of Science 110 19273

Pinsonneault M H An D Molenda-Zakowicz J et al 2012ApJS 199 30

Press W H Flannery B P amp Teukolsky S A 1986Cambridge University Press 1986

Rowe J F Bryson S T Marcy G W et al 2014 ApJ 78445

Seager S amp Mallen-Ornelas G 2003 ApJ 585 1038Silburt A Gaidos E amp Wu Y 2015 ApJ 799 180Still M amp Barclay T 2012 Astrophysics Source Code Library

8004Torres G Andersen J amp Gimenez A 2010 AampA Rev 18 67VandenBerg D A Bergbusch P A amp Dowler P D 2014

Astrophysics Source Code Library 4010VandenBerg D A Bergbusch P A Ferguson J W amp

Edvardsson B 2014 ApJ 794 72Weiss L M amp Marcy G W 2014 ApJ 783 LL6

ABSTRACT

1 Introduction

2 Observations and Image Analysis


22 AstroDrizzle

23 Kp-HST Photometric Conversion

3 Evaluation of Kepler-296 KOI-2626 and KOI-3049 Stellar Parameters









4 Planetary Habitability




5 Discussions and Future Work

6 Conclusion

2 Cartier et al

vatory have accelerated high-resolution imaging of Ke-pler Objects of Interest (KOIs) especially those with thesmallest planets at the coolest temperatures The identi-fication of any diluting sources in the aperture allows forimproved precision when determining planet habitabilityand can also reveal previously unresolved stellar compan-ions Gilliland amp Rajan (2011) and Gilliland et al (2015)have shown that the sharp and stable point spread func-tion (PSF) of the WFC3 camera on Hubble Space Tele-scope is ideal for detailed photometric study of Keplertargets and for the identification of field stars in the HSTphotometric aperture down to about ∆mag = 10 TheF555W and F775W filters on WFC3UVIS are ideallysuited to observe the majority of Kepler targets

Our HST Guest Observing Snapshot Program GO-12893 observed 22 targets before May 1 2014 six ofwhich were found to be multiple star systems unresolvedby Kepler Gilliland et al (2015) discusses the overar-ching scientific goals and conclusions of the observingprogram including program parameters and basic im-age analysis stellar companion detections and detectioncompleteness comparison to other high-resolution imag-ing and tests for physical association of detected stel-lar companions Gilliland et al (2015) presents analysisthat directly supports the methods in this paper andserves as a companion paper to this work Here weperform multiple-star isochrone fitting using the latestrelease of the Victoria-Regina Stellar Models (Vanden-Berg et al 2014b Casagrande amp VandenBerg 2014) forthree Kepler targets of particular interest KIC 11497958(KOI-1422 hereafter Kepler-296) KIC 11768142 (here-after KOI-2626) and KIC 6263593 (hereafter KOI-3049)We discuss the parameters of GO-12893 and our imageanalysis in Section 2 including our use of the DrizzlePacsoftware and our conversion of our HST photometry tothe Kepler photometric bandpass In Section 3 we dis-cuss the importance of our three targets and detail ourcharacterization of the stellar components in each multi-star system including the use of our empirically derivedPSF to calculate the photometry of our systems fittingto the Victoria-Regina isochrones and examination oftheir suitability for our targets Section 4 presents ourre-evaluation of the planetary habitability For the pur-poses of this paper we define a ldquohabitable planetrdquo to be aplanet that falls between the moist greenhouse limit andthe maximum greenhouse limit as defined by Kopparapuet al (2013) Finally we discuss our results in contextof previous and future work in Section 5 and summarizeour findings in Section 6

2 OBSERVATIONS AND IMAGE ANALYSIS

The 158 targets proposed for observation were selectedfrom the 2013 data release of Kepler planet candidates byBatalha et al (2013) prioritized by smaller candidate ra-dius and cooler equilibrium temperature The remainingranked targets were then sorted between ground-basedAO and HST observations based on the quality of obser-vations for the fainter targets where HST would providecomparable or better data in half an orbit than a fullnight of ground-based AO observation on Lick or Palo-mar systems This resulted in the selected HST targetshaving the shallowest transit signatures which thus re-quire the deepest imaging The targets have a nominalupper limit of Rp lt 25Roplus (Batalha et al 2013)though

Fig 1mdash AstroDrizzled image of KIC 4139816 in the F775Wfilter showing a 1primeprime0 scale bar and orientation The image is ap-proximately 2primeprime0 on a side Units are log10 of eminuss The FWHMof the PSF is 0primeprime0777

our revision of the stellar parameters indicates that someof the planets are actually larger than this limit Of the158 proposed targets 22 were observed before May 2014and are included in our analysis Any observations col-lected after May 2014 will be analyzed using the tech-niques presented in this section but are not included inthis paper Our image analysis utilized the latest im-age registration and drizzling software from STScI Driz-zlePac (Gonzaga et al 2012) and our own PSF definitionand subtraction

Our HST program provided high resolution imaging inthe F555W (λ sim 0531microm) and F775W (λ sim 0765microm)filters of the WFC3UVIS camera to support the analysisof faint KOIs In particular the parameters of our ob-servations allowed us to examine the properties of faintstellar hosts of small and cool planet candidates Atthe faint magnitudes of typical Kepler stars our WFC3imaging provides resolution that is competitive with cur-rent ground-based AO and has the advantage of usingtwo well calibrated optical filters well matched to theKepler bandpass

The observations made by HST closely resemble thosemade by Gilliland amp Rajan (2011) though we only usedobservations in F555W and F775W since the faintest Ke-pler targets could still be probed in these bandpassesObservations planned for each of the 158 SNAP targetswere identical in form In each filter we took 5 observa-tions of each target 4 observations with exposure timesto reach 90 of full well depth in the brightest pixel andan additional observation at an exposure time equal to50 more than the sum of the unsaturated exposures tobring up the wings of the PSF The saturated exposureyielded a ∆-mag of sim 9 outside 2primeprime and helped with thesignal-to-noise anywhere outside the inner 0primeprime1

22 AstroDrizzle

The ldquodrizzlerdquo process formally known as variable-pixellinear reconstruction was developed to align and com-bine multiple under-sampled dithered images from HST

2853029 2013-08-12 15679 16017 150064139816 2013-04-12 15954 16604 151414813563 2012-11-12 14254 14602 135105358241 2013-02-04 15386 15656 149025942949 2012-10-29 15699 16154 149906026438 2013-05-22 15549 16075 148276149553 2013-06-12 15886 17004 148126263593 2013-02-14 15037 15524 142756435936 2013-08-18 15849 16846 147967455287 2013-10-04 15847 16720 148378150320 2013-09-02 15791 16303 149858890150 2013-08-16 15987 16853 149698973129 2013-07-07 15056 15329 144559838468 2012-10-28 13852 14108 1332410004738 2014-01-07 14279 14563 1370410118816 2012-10-27 15233 16000 1422610600955 2013-02-10 14872 15135 1425311305996 2013-03-31 14807 15519 1385011497958 2013-04-06 15921 16807 1480511768142 2013-07-31 15931 17056 1489512256520 2013-07-28 14477 14805 1395712470844 2013-03-19 15339 15636 1469512557548 2013-02-06 15692 16349 14936

σKp =radic

00192 (F555W minus F775W)2 + 00272 (2)

4 Cartier et al

AJ = 0282timesAV

Ai = 0672timesAV

AKs = 0117timesAV

6 Cartier et al

KOI-2626 Photometry

KOI-3049 Photometry

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

2853029 2013-08-12 15679 16017 150064139816 2013-04-12 15954 16604 151414813563 2012-11-12 14254 14602 135105358241 2013-02-04 15386 15656 149025942949 2012-10-29 15699 16154 149906026438 2013-05-22 15549 16075 148276149553 2013-06-12 15886 17004 148126263593 2013-02-14 15037 15524 142756435936 2013-08-18 15849 16846 147967455287 2013-10-04 15847 16720 148378150320 2013-09-02 15791 16303 149858890150 2013-08-16 15987 16853 149698973129 2013-07-07 15056 15329 144559838468 2012-10-28 13852 14108 1332410004738 2014-01-07 14279 14563 1370410118816 2012-10-27 15233 16000 1422610600955 2013-02-10 14872 15135 1425311305996 2013-03-31 14807 15519 1385011497958 2013-04-06 15921 16807 1480511768142 2013-07-31 15931 17056 1489512256520 2013-07-28 14477 14805 1395712470844 2013-03-19 15339 15636 1469512557548 2013-02-06 15692 16349 14936

σKp =radic

00192 (F555W minus F775W)2 + 00272 (2)

4 Cartier et al

AJ = 0282timesAV

Ai = 0672timesAV

AKs = 0117timesAV

6 Cartier et al

KOI-2626 Photometry

KOI-3049 Photometry

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

4 Cartier et al

AJ = 0282timesAV

Ai = 0672timesAV

AKs = 0117timesAV

6 Cartier et al

KOI-2626 Photometry

KOI-3049 Photometry

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

AJ = 0282timesAV

Ai = 0672timesAV

AKs = 0117timesAV

6 Cartier et al

KOI-2626 Photometry

KOI-3049 Photometry

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

6 Cartier et al

KOI-2626 Photometry

KOI-3049 Photometry

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

8 Cartier et al

triple

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

-05 3187 1610 0936-04 3187 1491 0908-03 6227 1313 1056-02 7531 1191 1179-01 8365 1139 108600 6246 0941 0943

+01 3207 0860 1049+02 0704 1258 1073+03 0218 2123 1039+04 1568 3987 1041

Kepler-296

Distance [pc] 359 358F555W 9218 11111F775W 7266 8621

min = 0218

radic1σ2

KOI-2626

Distance [pc] 337 342 333F555W 10007 10697 11690F775W 7953 8472 9274

Ks 5732 6151 6839F555W minus F775W 2054 2225 2416

min = 0860

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

10 Cartier et al

KOI-3049

Distance [pc] 485 484F555W 7567 8222F775W 6381 6858

min = 0907

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

KOI-3049Components

c Period b

[ppm] [ppm] [days]

Rp = R

radic∆Ftrue (8)

ap = aoplus

Poplus

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

12 Cartier et al

radicR

2 ap(10)

)2(TlowastT

ρp = 243 + 339

Roplus

)gcm3 (12)

Moplus= 269

Roplus

gcm3 (13)

Roplus

Moplus (14)

ρpρoplus

=MpMoplus

(RpRoplus)3 (15)

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

14 Cartier et al

the HZ

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

6 CONCLUSION

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

16 Cartier et al

REFERENCES

201 15

210 20

ABSTRACT

1 Introduction


22 AstroDrizzle
















6 Conclusion

Documents

arXiv:1407.1057v3 [astro-ph.SR] 11 May 2015 · 2018. 8. 25. · photometric aperture down to about mag = 10. The F555W and F775W lters on WFC3/UVIS are ideally suited to observe the