Astronomy&Astrophysicsmanuscriptno.pont February2,2008 (DOI:willbeinsertedbyhandlater) Doppler follow-up of OGLE planetary transit candidates in ⋆ Carina F.Pont1,F.Bouchy2,3,C.Melo4,N.C.Santos5,M.Mayor1,D.Queloz1 andS.Udry1 1 ObservatoiredeGene`ve,51ch.desMaillettes,1290Sauverny,Switzerland 5 2 Laboratoired’AstrophysiquedeMarseille,TraverseduSiphon,13013Marseille,France 0 3 ObservatoiredeHauteProvence,04870StMichell’Observatoire,France 0 4 ESO,Casilla19001,Santiago19,Chile 2 5 LisbonObservatory,TapadadaAjuda,1349-018Lisboa,Portugal n a Received;accepted J 7 Abstract. Wepresenttheresultsofourhigh-resolutionspectroscopicfollow-upof42planetarytransitcandidatesinCarina 2 fromtheOGLEsurvey.Thisfollow-uphasalreadyallowedthediscoveryofthreenewtransitingexoplanets,OGLE-TR-111, 113and132,presentedinearlierLetters(Bouchyetal.2004;Pontetal.2004).Hereweanalysethedatafortheremaining39 1 candidates.Theradialvelocitydatashowthatmostofthemareeclipsingbinaries,inveryvariedconfigurations.Preciseradial v velocityorbitswerederivedfor15binaries,revealing9transitsofsmallstars(generallyM-dwarfs)infrontofF-Gdwarfs,1 5 grazingequal-mass eclipsingbinary, 4tripleand 1quadruple systems. Aremaining14systemsappear binary, but theexact 1 orbitisuncertainorwasnotdetermined.2objectsdonotshowanyradialvelocityvariationsinphasewiththetransitsignal, 6 and6donotpossessspectrallinesstrongenough forareliablecross-correlationfunctiontobemeasured. Amongtheselast 1 0 twocategories,upto6objectsaresuspectedfalsepositivesofthephotometrictransitdetection.Finally2objectsareunsolved 5 casesthatdeservefurtherobservations. 0 Ourstudyillustratesthewidevarietyofcasesthatcanmimicphotometricplanetarytransits,andtheimportanceofspectroscopic / follow-up. Multi-fiber capacities and an optimized follow-up strategy, which we present here, can help deal with the high h numberofcandidatesthatarelikelytoturnupinthenearfuture. p Animportantby-productofthisstudyisthedeterminationofexactmassesandradiiforsixverylow-massstars,includingtwo - o attheveryedgeofthestellardomain,OGLE-TR-106(M =0.116 0.021M )andOGLE-TR-122(M =0.089 0.007M ). r Theradiusoftheseobjectsisconsistentwiththeoreticalexpectation±s.Twofur⊙therobjects,OGLE-TR-123andOG±LE-TR-12⊙9, t s may harbour transitingcompanions near thebrown-dwarf/stellar limit (M 0.07M ),whose confirmation requiresfurther a high-resolutionspectroscopicmonitoring. ≃ ⊙ : v Notransitingmassiveplanets(M=2 10M)weredetected,confirmingtherarityofsuchsystemsatshortperiodasindicated − J i byDopplersurveys.Nolight(M < 0.5M),large(R > R)planetswerefoundeither,indicatingthat”hotSaturns”generally X J J havesmallerradiithanhotJupiters.ThreeshortperiodbinarieswithaM-dwarfcompanionshowdefiniteorbitaleccentricities, r withperiodsrangingfrom5.3to9.2days.Thisconfirmstheoreticalindicationsthatorbitalcircularisationinclosebinariesis a lessefficientforsmallercompanionmasses. WealsodiscusstheimplicationsofourresultsforthestatisticalinterpretationoftheOGLEplanetarytransitsurveyinCarinain termsofplanetfrequencyanddetectionefficiency.Wefindthattheactualtransitdetectionthresholdisconsiderablyhigherthan expectedfromsimpleestimates,andverystronglyfavoursthedetectionofplanetswithperiodsshorterthanabout2days.The apparentcontraditionbetweentheresultsoftheOGLEtransitsurveyandDopplersurveyscanberesolvedwhenthisdetection biasistakenintoaccount. Keywords.techniques:radialvelocities-instrumentation:spectrographs-stars:binaries-stars:planetarysystems 1. Introduction have been detected around other stars, the vast majority of them by radial velocity surveys. In the wake of the first Since the groundbreaking discovery of 51 Pegasi exoplanet detection, Gillilandetal. (2000) have monitored (Mayor&Queloz 1995), more that one hundred planets several thousand stars in the globular cluster 47 Tuc with the HST in search of photometric planetary transits, but none Sendoffprintrequeststo:e-mail:[email protected] were found. A results that is now attributed to the paucity of ⋆ Based on observations collected with the UVES and FLAMES short-period planets around metal-poor stars (Sackettetal. spectrographs at the VLT/UT2 Kueyen telescope (Paranal 2005). Observatory,ESO,Chile,Programme72.C-0191) 2 F.Pontetal.:Follow-upofOGLEtransitsinCarina Transiting planets are especially valuable in the study of 2004; Chabrieretal. 2004; LecavelierdesEtangsetal. 2004; exoplanets,becausethe presenceof a transitallowsthe deter- Sasselov2003;Mazehetal.2005). mination of the exact mass and radius – and therefore mean In Paper I, we showed that most of the OGLE transiting density–oftheplanet.Theseareobviouslyveryimportantpa- candidatesareinfacteclipsingbinaries.Thesebinariescanbe rameters in the physical understanding of exoplanet structure interestingintheirownright,especiallysinceinmanyofthem and evolution. Data from transits are thus an important com- the eclisping bodyis a very small star. Because the mass and plementto radialvelocityplanetsearcheswhichprovideonly radius of the transiting body can be obtained from the transit anestimateofM siniinadditiontoorbitalparameters. andradialvelocitydata,animportantby-productoftheradial pl ThefirstexoplanettransitwasdetectedaroundHD209458 velocityfollow-upistoprovideconstraintsonthemass-radius by Charbonneauetal. (2000) and Henryetal. (2000), the relation for low-mass stars (see Paper I). These data augment planet having been previously discovered in radial veloc- thosefrombrightereclipsingbinariesandfrominterferometric ity by Mazehetal. (2000). This planet has turned out to studiesornearbyM-dwarfs. have a radius much larger than Jupiter, R = 1.35 The present paper exposes the results of our spectro- ± 0.06 R (Brownetal. 2001), and to be undergoing signifi- scopic follow-up of OGLE transiting candidates in Carina J cantevaporation(Charbonneauetal.2002;Brownetal.2002; (Candidatesnumber60-132describedinUdalskietal.2002b, Vidal-Madjaretal.2003,2004). 2003).Section2presentsthespectroscopicdataandreduction, In recent years, increasing telescope automation and the Section 3 explains the tools used in the analysis of the light capacity to process large amounts of CCD photometric data curveandspectroscopicdata,Section4presentstheresolution have make possible several ambitious ground-based searches ofallcasesbycategory,Section5includesindividualnoteson for planetary transits: e.g. STARE (Brown&Charbonneau someobjects,andSection6discussessomeimplicationsofthis 2000), WASP (Kaneetal. 2001), PLANET (Sackettetal. study,notablyonthestellarmass-radiusrelationforlow-mass 2004). Up to now, the most successful of these searches starsandthestatisticalimplicationsoftheplanetdetections. has been the OGLE survey, which monitored 3 fields in the direction of the Galactic bulge and 3 fields in Carina 2. Observationsandreductions during 2001/2002, annoucing 137 possible transiting can- didates (Udalskietal. 2002a,c,b, 2003). After these candi- The observations were acquired in 8 half-nights (32 hours) dates were announced, several spectroscopic follow-up pro- on FLAMES on 13 to 21 March 2004 (Prog. 72.C-0191). grammes were initiated (Konackietal. 2003b; Dreizleretal. FLAMES is a muli-fiber link which allows to feed the high- 2002;Gallardoetal.2005),includingourownfollow-upof17 resolution spectrographUVES with up to 7 targets on a field candidatesintheGalacticbulgefield(Bouchyetal.2005,here- of view of 25 arcmin diameter, in addition to a simultaneous afterPaperI).Itwasfoundthatthevastmajorityoftransiting thoriumcalibration.Inapreviousrunonthisinstrument(Paper candidateswereeclipsingbinaries,aspredictedbythesimula- I), we have shown that FLAMES was able to measure radial tionsofBrown(2003). velocitieswithanaccuracyofabout30 ms−1onstarsdownto In this context, Konackietal. (2003a) announced the first the 16thmagnitudein I. Some trialswith HARPS andUVES exoplanet discovered by photometric transit surveys, OGLE- inslit mode(seePaperI)ledtothe conclusionthatFLAMES TR-56b (Bouchyetal. 2005, confirmed by). This planet was wasaveryefficientinstrumentfortheradialvelocityfollow-up foundtoorbitwiththeunexpectedlyshortperiodof1.2-days, ofOGLEtransitcandidates. muchshorterthantheobservedpile-upofperiodsabove3days inradialvelocitysurveys. 2.1.Targetselectionandobservingstrategy In March 2004, using the FLAMES/UVES multi-fiber spectrograph on the VLT, we acquired high-resolution spec- Forthe73OGLEtransitcandidatesintheCarinafield(OGLE- troscopic information for 42 of the most promising candi- TR-60toTR-132),themostpromisingcandidatesin termsof dates in the OGLE Carina fields, up to 8 spectra per ob- planetarytransitswere selected accordingto threemain crite- ject. Three planet were discovered in this way: OGLE-TR- ria: 111b(Pontetal.2004),OGLE-TR-113bandOGLE-TR-132b 1. theradiusoftheeclipsingbodyindicatedbythedepthand (Bouchyetal. 2004).Two ofthemareotherinstancesofvery durationofthetransit. short-periodplanets,revealingthatthecaseofOGLE-TR-56b 2. theshapeofthetransit.AU-shape(flat-bottom)transitin- was not uncommon. OGLE-TR-113b was subsequently con- dicatesacentraltransit,whileaV-shapeindicatesagrazing firmed by independent radial velocities (Konackietal. 2004) transit,thereforeaprobableeclipsingbinary. and a high-accuracy photometric transit curve was obtained 3. theamplitudeofthesineanddouble-sinemodulationsseen for OGLE-TR-132 by Moutouetal. (2004). One more object inthelightcurve.Sirko&Paczinski(2003)haveanalysed in the OGLE survey, OGLE-TR-10, has gradually emerged theOGLEcandidatelightcurvesandshownhowtheinflu- as a solid transiting planet detection (Konackietal. 2003b; enceofamassivecompanioncouldbedetectedbymodu- Bouchyetal.2005;Konackietal.2005). lationsofthelightcurveoutsidethetransits. The only other transiting planet discovered to date by transit searches is TrES-1 (Alonsoetal. 2004). The total of Using these threecriteria andother occasionalindications known transiting planets now stand at seven, and have al- suchasthepresenceofananti-transitsecondarysignal,anini- ready spurred several interesting studies (e.g. Burrowsetal. tial list was built with 14 first-priority and 16 second-priority F.Pontetal.:Follow-upofOGLEtransitsinCarina 3 objects. 43 candidateswere consideredalmost certainly bina- 2.2.Radialvelocities riesfromtheseconsiderationsalone. ThespectraobtainedfromFLAMES/UVESwereextractedus- In order to optimize the use of telescope time, we used a ing the standard ESO-pipeline with bias, flat-field and back- real-time replacement strategy throughout the observing run: groundcorrection.Wavelengthcalibrationwasperformedwith assoonasanobjectwasdetectedasabinary(i.e.photometric theThArspectra.Theradialvelocitieswereobtainedbycross- transitsignalnotcausedbyaplanetbutbyaneclipsingbinary), correlationwithanumericaltemplateconstructedfromtheSun the object was marked for possible re-allocationof the corre- spectrumatlas.TheThArspectrumwasusedinordertocom- sponding FLAMES fiber to another target. The objects were putetheinstrumentaldriftbycross-correlationwithaThorium also markedif no signal was detected in the cross-correlation template (see Paper I for details). Radial velocity uncertain- function(indicatingafast-rotatingstar,anearly-typestar ora ties were computed as in Paper I. The most precise of our heavilyblendedsystem).Inthefollowingnight,thefibercorre- measurements (σ < 50 ms 1) are not photon-noise limited spondingtotheobjectsmarkedwasre-allocatedtoanewtarget, − and we added quadraticallyan uncertainty of 30 ms 1 in or- if allowed by the distribution of the targetsin the sky. In that − der to take into the accountsystematic errorsprobablydueto way, we could observe and characterise 42 objects with four wavelengthcalibrationerrors,fiber-to-fibercontamination,and fieldsof 7 fibersandmakea highlyefficientuse of the multi- residualcosmicrays.ThisvaluewasajustedontheO-Cresid- plexfacility.Wecouldobserveallfirst-prioritytargets,and11 ualsofthenon-rotatingstarwithoutsignificantradialvelocity ofthe16second-prioritytargets(OGLE-TR-70,71,73,74and variationsOGLE-TR-131(Bouchyetal.2004). 115 were left unobserved because of their inconvenientposi- Inthecaseofdouble-linedandtriple-linesspectra,thera- tioninthefield). dial velocities of each component were calculated with stan- As reminded in the introduction, most OGLE candidates dard techniques for spectroscopic binaries. Iterative solutions areeclipsingbinariesratherthanplanets.Weusedthreecriteria wereoftennecessarytodisentanglethecomponentswhenthe toidentifyeclipsingbinariesfromthe spectroscopicmeasure- multiple line systems are blended with eachother. We note ments:(i)presenceofmorethanonesetoflinesinthespectrum thattheradialvelocityuncertaintieswerecalculatedassuming (double-or triple-lined binaries). (ii) rotationalbroadeningof single-linedspectraandinthecaseofblendedspectratheymay thelinescorrespondingtotidalsynchronisation.(iii)largevari- beunderestimated. ationsoftheradialvelocity.Figure1showsexamplesofspec- Our radial velocity measurements and cross-correlation tral cross-correlation function (CCF) illustrating the different function parameters are listed in Table 8 and analysed in cases. Section4. 2.3.Rotationvelocities Rotationalvelocitieswerecomputedfromtheobservedcross- correlation functionusing rotationallybroadenedline profiles convolved with a Gaussian instrumental profile of width 4.0 kms 1 (found suitable for our instrumental setup in Paper I). − TheprofileswerefittedtotheCCFsimultaneouslywiththera- dialvelocitytodeterminetheprojectedrotationvelocityvsini of the target objects. A quadratic limb-darkening with coef- ficients u1+u2 = 0.6 was assumed (The computations of Barbanetal.2003,findthatsuchacoefficientisasuitableap- proximationforawiderangeofspectraltypesinwavelengths correspondingtotheVfilter). Fig.1. Example of cross-correlation functions from our pro- 2.4.Stellarspectroscopicparameters gramme:foranunrotatingstar(left),forarotatingstarsynchro- For the slowly-rotatingstars in our sample, the stellar param- nised with the eclipsing companion(middle) and for a triple- eters (temperatures,gravities and metallicities) were obtained lined spectroscopic binary (right). In the last case, the third from an analysis of a set of Fe and Fe lines, followingthe componentisthewidedepressionbetweenthetwodeeperdips. procedureusedinSantosetal.(2004).Theprecisionofthede- rived atmospheric parameters is limited by the relatively low Notethatcriteria(i)and(ii)requireonlyonespectroscopic S/Nofthecombinedspectra(30-50),togetherwithsomepos- measurement.Criteria(ii)isbasedonthefactthatbinarieswith siblecontaminationcomingfromtheThArspectrum. periods smaller than about 10 days are expected to be tidally Forthestarsinoursamplerotatingwithvsini 20 kms 1, − ≥ locked to their companions in synchronous rotation. The in- ormultiple-linedspectra,themethoddescribedinSantosetal. creased rotation velocity is observed as a broadening of the (2004)isnotapplicable,becausethemeasurementofindivid- correlationdipinthecross-correlationfunction.SeeSection3 ualequivalentwidthsisnotaccurateenoughduetolineblend- formoredetails. ing. The majority of the objects in our sample are actually in 4 F.Pontetal.:Follow-upofOGLEtransitsinCarina this case. When necessary for the resolution of the case, we have determined very rough estimates of the spectral type by visualcomparisonof the observedspectra with a grid of syn- theticspectra. 3. Analysis Adetaileddescriptionofouranalysistoolsforphotometricand spectroscopic data of transiting candidate is given in Paper I. We briefly sketch the analysis ”toolbox” below and refer to PaperIforamoredetaileddescription. 3.1.Synchronisedrotationofeclipsingbinaries For close binaries,with rotationperiodsof the orderof a few days,weexpecttherotationaxistobealignedwiththeorbital axisandthesystemtobetidallylocked(e.g.Levato1976;Hut 1981).Forknownclosebinaries,thealignmentoftheaxesand the tidal locking is fast. In that case, P = P and the rot transit rotationvelocityisdirectlyrelatedtotheradiusoftheprimary. Fig.2.Rotationvelocityoftheprimaryvsperiodforoursam- In Paper I, we could indeed verify this hypothesis in all the ple. The lines show the expected rotation velocity for tidally target observed, and we shall use it here both to estimate the lockedsystemswithradiiof1 R and2 R fortheprimary.For primary radius R and as an indication that the companion is objects at vsini=5 kms 1(dotte⊙d line), t⊙he value given is an massivewhenwehaveonlyonespectroscopicmeasurement. − upperlimit. Uncertaintiesareindicatedonlywhenlargerthan Figure2shows,fortheCarinasample,therotationvelocity the symbols (an uncertainty of 10% is used for values based of the primarytargetas a functionof the periodof the transit on single spectra). Black dots show objects with a precise ra- signal (for transiting objects, sini 1 and therefore V rot ≃ ≃ dial velocity orbits, triangles the detected planets. The three vsini). Most objectshave a rotationvelocitycompatiblewith opencirclesisolatedinthelowerleftareOGLE-TR-97(triple synchronisationforreasonablevaluesoftheprimaryradius(1- system), 124 and 131 (suspected false transit detections). For 2 R ),indicatingthateclipsingbinariesdominatethesample. ⊙ all other objects the rotation velocity is compatible with syn- TheonlyexceptionsarethethreeplanethostsOGLE-TR-111, chronousrotation. 113and132,andtheobjectsOGLE-TR-97,124and131.The first of these has a triple-lined spectrum and the last two are suspectedfalsetransitdetections(seeSect.4.4). total mass (m + M) – via the orbital period and semi-major 3.2.Revisedperiod axisforaKeplerianorbit–andthelimbdarkeningcoefficients. Theparameterq ismainlyconstrainedbythetransitdepth,b R Theradialvelocitydata,obtainedtwoyearsafterthephotetric bythetransitshapeandthefactorR(m+M) 1/3 bythetransit − data, generally allows a significantimprovementof the deter- duration. mination of the orbital periods. Some systems turn out to be The light curves were fitted by non-linear least square grazing, equal-mass eclipsing binaries where both the eclipse fitting with analytic transit curves computed according to andanti-eclipse were visible in the lightcurve.In these cases Mandel&Agol (2002), using a quadratic limb darkening the real orbital period of the system is double the period in model with u1+u2=0.3. Note that this is different from the Udalskietal.(2002b,2003). coefficientsusedforthedeterminationoftherotationalveloc- ContrarilytoPaperI,therewerenocaseswheretheradial ity,becausethewavelengthsaredifferent.TheOGLEdatawas velocity indicated a completely differentperiod than the pho- obtainedwith an I filter while the spectra are centeredon the tometry. This reflects the fact that much less candidates with visible.Thefittedparameterswereq ,V /Randb,whereV is onlytwoorthreemeasuredtransitswereacceptedintheOGLE R T T thetransversalorbitalvelocityatthetimeofthetransit.Theun- CarinasamplethanintheGalacticbulgesample. certaintiesontheseparameterswereestimatedusingamethod thattakesintoaccountthecovarianceofthephotometricresid- 3.3.Analysisofthetransitshape uals(seePaperIfordetails). Thedepth,widthandgeneralshapeofthetransitsignaldepend onacombinationofphysicalvariables,mainlytheradiusratio 3.4.Synthesisofthespectroscopicandphotometric qR (noted r in Paper I), the primary radius R and the impact constraints parameterb (or,equivalentlyforcircularorbits,theanglei of the normalof the orbitalplane with the line-of-sight)and the Forconvenience,thesixrelationsusedtoinferthesystempa- orbital eccentricity. It is also more weakly dependent on the rametersfromtheobservablesarerepeatedbelowfromPaperI: F.Pontetal.:Follow-upofOGLEtransitsinCarina 5 whichimpliesthattherotationvelocityderivedfromthebroad- eningofthecross-correlationfunctionisdirectlyrelatedtothe vsini P = 50.6 R (1) star’sradius(Equ.1).ContrarilytoPaperI,formostoftheob- sini · · jects in this study we have no precise spectroscopic estimate m K = 214 P−1/3 (2) of the temperature, because of the rotational line broadening · (m+M)2/3 · and thelow signal-to-noiseof the spectra.When necessaryto b R i = acos( · ) (3) constrain the solution, the existence of a signal in the cross- a correlationfunctionwasusedtogiveabroadupperlimittothe a VT = 2π· P (4) temperature,Teff < 7000K.Thisimprovesthesolutionform byeliminatingveryhighvaluesof M incaseswhenRislarge q = r/R (5) R (namelyforOGLE-TR-78and125). (logT ,logg,[Fe/H],R) = f(M,age,Z) (6) eff Theresultsfortheradialvelocityorbitandtherotationve- witha=4.20 P2/3 (m+M)1/3 locity are givenin Table 1. Theparametersobtainedfromthe · · combination of the light curve and spectroscopy are given in where vsini is the projectedrotationvelocityin kms 1, i Table7.Twodifferentmethodswereusedtocombinethecon- − theorbitalinclination, Pthe periodin days,R,r,M andm the straints (1)-(6):if the impact parameter was low, the solution radii and masses of the eclipsed and eclipsing bodiesin solar wasobtainedbyχ2 minimisationasinPaperI.Forhigherval- units,K theprimaryradial-velocitysemi-amplitudein kms 1, uesoftheimpactparameter,theradiusratioispartlydegenerate − andatheorbitalsemi-majoraxisinunitsof R .Thefunction withtheorbitalangle.Inthesecases(OGLE-TR-72,105,121), f in the last equation is the relation given by⊙the theoretical wefirstobtainedanestimateoftheprimarymassusingtheap- stellarevolutionmodelsofGirardietal.(2002). proximate relation M √R, then fitted the light curve with ∼ Weassumedthatallorbitswerecircular(e = 0)unlessthe VT/Rfixedandonlytheradiusratioandimpactparametersas radialvelocitydataclearlyindicatedotherwise.Veryclosebi- freeparameters.Obviously,in thatcase some valuesobtained naries are expected to be circularized on a timescale shorter canhavehigheruncertainties,asareindicatedinTable7with thantypicalstellarages.However,thecircularisationtimescale columns. increases with decreasing companion mass (Zahn 1989), so For OGLE-TR-105,the radial velocity data show that the thatsomeofthelow-masscompanionsinoursamplemaystill periodneedstobedoubled,andthatbothtransitandanti-transit haveeccentricorbits.Fourobjectsshowdefiniteindicationsof werevisibleinthephotometricdata. e , 0 in the velocitycurve.Foreccentricorbits, Equ.(2),(3) OGLE-TR-122isthesmalleststellarobjectinoursample, and (4) above must be slightly modified. The distribution of withaplanet-likeradius,andassuchishighlyinteresting.Itis eccentricitiesarediscussedinSection6.2. analysedinmoredetailsinaseparateLetter(Meloetal.2005). 4. Results 4.1.2. Eclipsingbinarieswithtentativeorbits The analysis of the 42 objects measured in spectroscopy is OGLE-TR-123 was measured only three times and the solu- presented below, divided for convenience according to cate- tionproposedinTable1isonlytentative.Ourthreespectrafor gories reflecting the treatment of the data and the nature of thisobjectsshowawideCCFsignal,withvsini 34.5 kms 1, the systems: single-lined spectroscopic binaires (eclipsing bi- ≃ − andsmallvelocityvariationstothelevelofafew kms 1.The narieswithsmallcompanion),multiple-linedspectroscopicbi- − rotation is compatible with synchronisation, and would indi- naries (grazing eclipsing binaries, multiple systems, line-of- cate R 1.2 R . A circular orbitfitted on the velocitypoints sight blends), transiting planets, suspected false transit detec- with the∼epoch⊙of the transit signal gives K = 12.08 kms 1, tions,andobjectswithoutCCFsignal. − V =0.66 kms 1andP=1.8038days.Thesearetentativeval- 0 − uesbecausethenumberoffreeparametersisequaltothenum- 4.1.Single-linedspectroscopicbinaries berofpoints.IftheperiodisfixedtotheOGLEvalueaswell, theorbitfityieldsK =11.32 0.6 kms 1andV =0.77 0.40 − 0 4.1.1. Eclipsingbinarieswithresolvedorbits ± ± kms 1.Thisimpliesr = 0.097 0.006R andm 0.070M − ± ⊙ ≃ ⊙ Objects OGLE-TR-72, 78, 105, 106, 120, 121, 122, 125 and forthetransitingbody,makingOGLE-TR-123anextremelyin- 130 show single-lined spectra with radial velocity variations terestingcandidateoftransitingbrowndwarforlow-massstar compatiblewithorbitalmotionattheperiodgivenbythepho- near the Hydrogen-burning limit (see Fig. 10). However, the tometrictransits.Theyareeclipsingbinaries,withafainter,un- transitdurationis longerthan wouldbepredictedbythissce- seencompanioncausingtheobservedphotometricsignal.The nario. More radial velocity data would be needed to confirm radialvelocitydatawerefittedwithKeplerianorbits.Figure3 theperiodandexcludemorecomplexblendscenarios. shows the resulting radial velocity orbits. The spectroscopic OGLE-TR-129 has a wide and very shallow signal in the and light-curvedata were then combined as described briefly CCF,andtheradialvelocitieshavelargeuncertainties.Thepro- above and more fully in Paper I to determine the mass and jected rotation velocity is vsini 23 kms 1, which implies − ∼ radius of both components. The target stars are assumed to R 2.6 R in case of synchronised rotation. The spectrum ≃ ⊙ be tidally locked to their companion in synchronousrotation, indicates a high temperature, T 6500, compatible with eff ≥ 6 F.Pontetal.:Follow-upofOGLEtransitsinCarina Fig.3.Radialvelocitydataandorbitsforsingle-linedspectroscopicbinaries.Theorbitalperiodsandepochsareconstrainedin combinationwith the photometricsignal. Thecorrespondingparametersare givenin Table 1. Themeasurementsuncertainties aresmallerthanthesymbols.ThesolutionsforOGLE-TR-123andOGLE-TR-129aretentative(seeText). Fig.4.RadialvelocitydataandtentativeorbitalsolutionforOGLE-TR-123andOGLE-TR-129.Theorbitalperiodsandepochs areconstrainedin combinationwiththe photometricsignal.Thecorrespondingparametersaregivenin Table1. Themeasure- mentsuncertaintiesareplottedonlywhenlargerthanthesymbols. this radius value. The radial velocity data phased on the pe- PossiblevaluesforKwouldimplym=0.07 0.13M forthe − ⊙ riod of the photometric transit signal are compatible with a transitingbody.AtransitingM-dwarfscenarioisnotcoherent, markedly excentric orbit. There is some degeneracy between however,withthetransitshapeanddepthiftheprimaryissyn- the excentricityand K, and orbitswith lower eccentricityand chronised: the amplitude of the radial velocity variation indi- lowervaluesofKarealsomarginallycompatiblewiththedata. cates m 0.2 M , but the transit depth indicates r/R 0.2, ≤ ⊙ ∼ F.Pontetal.:Follow-upofOGLEtransitsinCarina 7 thereforewithasynchronisedprimaryr 0.5R forthesec- thatcasetheperiodhastobedoubledandthesecondaryradius ∼ ⊙ ondary. Such a mass-radius relation for a M-dwarf is not ex- iscomparabletotheprimaryradius,sothatthevalueofrgiven pected. Moreover,the transitis clearly V-shaped,indicatinga inthetableisreallyalowerlimit. grazingtransitwith b 1, whichwouldimplyan evenlarger ∼ secondary.ThepositionofOGLE-TR-129inFig.11indicates Name N vsini R r P that it lies near the transit detection threshold, with S = 20. OGLE d [ kms 1] [R ] [R ] [days] − Comparisonwithothersimilarcases(seeSection4.4)suggest ⊙ ⊙ OGLE-TR-63 3 74 1.6 0.15 1.06698 that it may be a false positive. In that case, the radial veloc- OGLE-TR-94 1 28 1.7 >0.2 3.09222 ityvariationmaybeunrelatedtothephotometrictransitsignal. OGLE-TR-98 1 18 2.2 0.34 6.39800 Alternatively,theradialvelocityvariationsmaybeexplaindby OGLE-TR-99 1 51 1.1 0.19 1.10280 ablendofmorethanonesetoflines.Theradialvelocitiesare OGLE-TR-126 1 18 1.8 0.25 5.11080 notpreciseenoughtodistinguishaKeplerianorbitfromother Table2.Rotationvelocityandradiusestimatesforprimaryand typesofvariations.Insummary,apossiblescenarioforOGLE- secondarycomponentsofsingle-linedsuspectedbinarieswith- TR-129isthatofa0.07 0.015M transitingM-dwarfwitha out orbits. N: number of spectra; vsini: projected rotational − ⊙ hot,fast-rotatingprimarythatisnotyetsynchronised,butother velocity of the primary; R and r: radius of the eclipsed and scenarioscannotbeentirelyexcluded. eclipsingbodiesrespectively. Theproposedorbitsforthesetwo objectsaredisplayedin Fig.4. 4.1.3. Probableeclipsingbinarieswithoutresolved 4.2.Double-linedortriple-linedspectroscopicbinaries orbits Such systems range across a wide variety of cases: grazing OGLE-TR-63 was measured three times in spectroscopy. Its eclipsing binaries, triple and quadruple systems. The resolu- rotation velocity is compatible with orbital synchronisation. tionofthecasesinthissectionnecessitatesthefullarsenalof Because the rotational broadening is very large (vsini 74 spectroscopicbinariesanalysis,thatwewillnotrepeatindetail ∼ kms−1),theradialvelocitieshavehighuncertainties.Moreover, here. The CCF componentsare often blended with eachother theperiodnear1daycausesanunfavourablephasesampling. andtheirseparationrequiresaglobal,iterativetreatmentofall As a consequence,the radial velocitydata is compatiblewith themeasurements.Whenanorbitcouldbedetermined,theor- bothasmallM-dwarftransitingcompanionandwithaconstant bital solution is given in Table 3 and illustrated in Figure 5. velocity.Thiscaseremainsunsolved. EstimatesofmassesandradiiaregiveninTable 7.Othercases OGLE-TR-94, 98, 99 and 126 were measured only once arepresentedinTable4.Theradiusofprobablysynchronised inspectroscopy.Theirrotationallinebroadeningiscompatible componentisestimatedfromtherotationvelocity. withtidalsynchronisationattheperiodoftheobservedtransit signal (see Table 1 and Fig. 2), so that these objects are very 4.2.1. Grazingequal-masseclipsingbinaries likelytobeeclipsingbinaries.Notethatevenifbychancethe highrotationvelocitywerenotduetoorbitalsynchronisation,it OGLE-TR-64showstwodipsinthecross-correlationfunction wouldmakethedetectionofaplanetorbitalmotionverychal- varying in anti-phase. The mass of both components can be lenging anyway, because the high width of the spectral lines determinedpreciselyfromtheorbitintheusualwayofdouble- significantlyincreasestheuncertaintyonthederivedradialve- linedspectroscopicbinaries.Bothradiicanbedeterminedfrom locities.Theresultingradialvelocityuncertaintieswouldbetoo the rotationvelocitiesderivedfromthe line broadeningin the hightorevealaplanetaryorbitwithourFLAMES/VLTobser- CCF.Thesinifactorcanbecalculatedfromtheaposteriorifit vationalsetupandreasonableexposuretimes.Moreover,ifthe ofthelightcurvewithbothradiifixed.Theperiodisthedouble rapid rotation were not due to synchronisation,it would most ofthatgivenbyUdalskietal.(2002b),becauseboththeeclipse probablyindicateanearly-typeprimary.Inthatcase,thelarge andanti-eclipseareseeninthelightcurve. radiusoftheprimarywouldimplyasecondaryradiustoolarge OGLE-TR-69and110,withonlyoneortwomeasurements for a planet. Therefore, we do not expect to miss any planet showingtwosetsoflineswithalargeradialvelocitydifference detectionbyrejectingthesefastrotators. androtationalbroadeningcompatiblewithorbitalsynchronisa- For these objects, a tentative value of the primary radius tion, are also probablegrazingeclipsingbinaries. The masses can be obtained from the rotation velocity and Equ. (1), and ofthecomponentscouldbeestimatedfromthevelocitiesusing thesecondaryradiuscanbeestimatedthroughtheradiusratio the periodand epochof the photometrictransits, but the time obtained by a fit of the lightcurve. No mass estimate can be intervalbetween the two sets of measurementsis too large to derived.Table 2 givesvsini, R andr for these objects.These providereliablevalues. valuesaretentativeestimatesandnouncertaintiesarederived. InthecaseofOGLE-TR-94,thelightcurveindicatesagraz- 4.2.2. Triplesystemswithfainteclipsingcompanion ingtransit,sothattheradiusratioisdegeneratewiththeimpact parameter. Only a lower limit can be assigned to r. It is also OGLE-TR-76and85are triplesystemswithtwo setsoflines possiblethatboththeeclipseandanti-eclipseareseen,andin inthespectrum,aneclipsingbinaryblendedwithathirdbody. 8 F.Pontetal.:Follow-upofOGLEtransitsinCarina Name P Ttr(OGLE) POGLE Tp w K V0 e vsini [days] [-2452000] [days] [-2452000] [deg] [kms 1] [kms 1] [kms 1] − − − 72 6.8581 77.399 (6.854) - - 27.4 19.9 0 10.9 0.2 ± 78 5.3187 328.812 (5.32038) 327.849 348 27.08 0.19 -10.47 0.12 0.117 0.007 17.5 0.8 ± ± ± ± 105 6.1161 324.380 (3.0581) - - 67.73 0.93 54.23 0.61 0.0 0.01 17 2.5 ± ± ± ± 106 2.5359 324.783 (2.53585) - - 18.27 0.38 -5.60 0.27 0.0 0.02 25.5 2.0 ± ± ± ± 120 9.1662 331.498 (9.16590) 331.268 70 32.99 0.08 34.072 0.06 0.361 0.002 9.6 0.4 ± ± ± ± 121 3.2321 325.689 (3.2321) - - 39.00 0.12 8.100 0.086 0.0 0.006 20.8 2.4 ± ± ± ± 122 7.2695 342.283 (7.26867) 335.152 101 9.887 0.065 -0.252 0.058 0.231 0.006 5.7 0.6 ± ± ± ± 123 1.8039 324.979 (1.8038) - - 12.08 0.66 0 34 3 ± 125 5.3039 343.825 (5.30382) - - 18.83 0.25 27.60 0.18 0.0 0.01 18.7 1.7 ± ± ± ± 129 5.7339 327.368 (5.74073) 327.5: 36: 16: 5: 0.6: 23 3 ± 130 4.83103 327.281 (4.83027) - - 38.482 0.037 4.516 0.026 0.0 0.001 11.9 0.4 ± ± ± ± Table1.Orbitalparametersforsingle-linedsystems.Columns2-9:Parametersoftheorbitalsolution.P:revisedperiod,T : tr(OGLE) epochofthetransit(fixed),P :originalOGLEperiod,T :epochofperiastron,w:omegaangle,,K:orbitalsemi-amplitude, OGLE p V :systemicvelocity,e:eccentricity.Column10:projectedrotationvelocityfromthecross-correlationfunction. 0 Inthescasesthe CCF showsonesetof lineswith orbitalmo- Name Comp N vsini R POGLE tion,broadenedbysynchronousrotation,andtheotherwithout [kms−1] R [days] ⊙ radialvelocitychange.Thereforethebodycausingtheeclipse OGLE-TR-69 a 2 15.6 1.44 2 2.33708 × is an unseen third body in orbit around the first. The second b 11.5 1.06 OGLE-TR-81 a 1 21: 1.4 3.21650 bodyseenintheCCFiseithergravitationallyboundtothetwo b 1 <5 - others in a triple system, or an unrelated star along the same OGLE-TR-93 a 1 45: 2.0 2.20674 line-of-sight. The treatment of these cases is the same as for b <5 - single-lined spectroscopic binaries, except that the lightcurve OGLE-TR-95 a 1 65:: 1.8 1.39358 also containsan unknowncontaminationfromthe thirdbody, b 12 - so that the radius ratio cannot be determined from the transit OGLE-TR-96 a 1 11: 1.4 2 3.20820 signal.OGLE-TR-81,93and95,withonemeasurementonly, b 9: 1.2 × showonebroadandonenarrowcomponentintheCCFandare c 71: - probablysimilarsystems. OGLE-TR-97 a 1 10.7 - 2 0.56765 × b 9.4 - c <5 - OGLE-TR-110 a 1 11.7 1.32 2 2.84857 4.2.3. Triplesystemswithequal-massgrazingbinary × b 11.4 1.28 Table 4. Number of measurements and rotation velocity for OGLE-TR-65 and 114 show three sets of lines in the CCF. double-linedandtriple-linedsystemswithoutorbitalsolution. These targets are equal-mass eclipsing binaries in triple sys- Column5indicatestheimpliedradiusincaseofsynchronised tems. OGLE-TR-96, with only one measurement, appears to rotation.Uncertaintiesonthevsiniareoftheorderof10per- be a similar system with two synchronisedcomponentsand a cent.Whenindicated,theorbitalperiodistwicetheOGLEpe- widercomponent(afast-rotatingF-dwarf). riod because the system contains an equal-mass eclipsing bi- nary. 4.2.4. Quadruplesystem OGLE-TR-112 is a truly involved case with three dips in the 4.2.5. Probablebinarieswithambiguityinthe CCF, all varyingin radial velocity on short timescales. It is a configuration quadruplysystem, with three componentsvisible in the spec- For some double-lined objects with only one spectroscopic tra.Twoofthemdescribeanexcentricorbitaroundeachother measurement,the broadeningof some of the spectral compo- with a period unrelated to the photometricsignal (P 10.63 ≃ nentsarecompatiblewithsynchronousrotationwithamassive days).Thethirdhasanorbitwiththeperiodofthetransitsig- eclipsingcompanion,butthereisnotenoughinformationtode- nal,revealingthatitiseclipsedbyafourth,unseencompanion. terminetheexactparametersof thesystem.Table 4gives,for Therefore OGLE-TR-112 is a system consiting in two close theseobjects,therotationvelocitiesindicatedbytheCCFcom- binaries. ponents,andthecorrespondingradiusiftherotationvelocityis Note that for all triple systems, the radial velocity of the compatiblewiththeperiodofthephotometrictransitsignal,or thirdcomponentisnearenoughtothesystemicvelocityofthe withdoublethatperiodforsuspectedgrazingbinaries. eclipsing binary for the systems to be gravitationally bound ForOGLE-TR-69,81,93,96and110,therotationveloci- multiplesystemsratherthanline-of-sightcontamination. tiesallowsatentativescenariotobe proposed.OGLE-TR-69, F.Pontetal.:Follow-upofOGLEtransitsinCarina 9 Name Comp P Ttr(OGLE) POGLE K V0 vsini [days] [-2452000] [days] [kms 1] [kms 1] [kms 1] − − − OGLE-TR-64 a 5.434691 78.569 (2.71740) 62.43 0.95 1.17 0.75 12 1 ± − ± ± b 96.46 1.02 9: ± OGLE-TR-65 a 1.720390 76.319 (0.86013) 114.53 0.84 2.18 0.70 43.6 1.9 ± − ± ± b 119.09 0.97 43.9 1.3 ± ± c 6: <5 − OGLE-TR-76 a 2.12725 323.545 (2.12678) 74.1 2.4 11.8 1.8 29: ± ± c 11: 20:: − OGLE-TR-85 a 2.11481 324.440 (2.11460) 48.62 0.75 4.48 0.30 31 3 ± ± ± c 1: 60: − OGLE-TR-112 a 3.8754 327.532 (3.87900) 29.8 2.5 3.3 1.8 40:: ± − ± c 10.6 - - 80.9 1.2 11.32 0.04 <5 ∼ ± − ± d 90.6 1.2 <5 ± OGLE-TR-114 a 3.4218 323.249 (1.71213) 80 3 5 2 10.8 1.4 ± ± ± b 80 3 10.6 1.3 ± ± c 7(drifting) <5 Table 3.Orbitalelementsandrotationvelocitiesfordouble-linedandtriple-linedsystems. Column2:systemcomponents.By conventiontheeclipsingcomponentsarealways”a”and”b”.Columns3-7:Parametersoftheorbitalsolution.P:revisedperiod, T :epochofthetransits(fixed),P :originalOGLEperiod,K:orbitalsemi-amplitude,V :systemicvelocity.Theexcentricity tr OGLE 0 wasfixedtozeroinallcasesexceptforthesecondsystem(”cd”)ofOGLE-TR-112,wheree = 0.55 0.02.Column8:rotation ± velocityfromthecross-correlationfunction. Fig.5. Radial velocity data and resulting orbits for double-lined and triple-lined spectroscopic binaries. The transit epoch is derivedfromthephotometricdata.ThecorrespondingparametersaregiveninTable3.Thedifferentcomponentsidentifiedinthe spectraareidentifiedwithdifferentsymbols.InallplotstheblackdotsindicatetheobjectundergoingtheeclipseatT .Error tr(OGLE) barsare comparableto the size of the symbolsor smaller. Theradialvelocityofcomponentswitoutsignificantradialvelocity changeswasfixedtoaconstantvaluestoallowabettersolutionfortheothercomponents.NotethatOGLE-TR-112isplottedin dateinsteadofphasebecauseofthetwodifferentperiods. 96and110appeartobegrazingequal-massbinarieswithdou- photometry. More measurements would be needed to deter- bletheOGLEperiod(Sect.4.2.1).OGLE-TR-81and93appear mine the nature of this system. However, since the presence to be spectroscopicbinariesblendedwith a non-rotatingthird of three sets of lines in its spectrum show that it is a blended body(Sect.4.2.2). system, it loses its interest in the context of planetary transit search,whichwasthemainobjectiveofourstudy. ForOGLE-TR-97,severalscenariosarepossibleandnone isclearlyfavoured.Detailsaregiveninthecommentsonindi- vidualobjectsinSection5.TheCCFshowsthreecleardips,but none of the calculated rotation velocities are compatible with synchronisationwiththe(veryshort)periodindicatedwiththe 10 F.Pontetal.:Follow-upofOGLEtransitsinCarina 4.3.Transitingplanets gives K = 1.5 0.9 kms 1. The temperature of the target − − ± isT 7000Kaccordingtoourestimation,T =7580 370K Transiting exoplanets were securely detected around three of ∼ ± according to Gallardoetal. (2005). Synchronisation with the the objects: OGLE-TR-111, OGLE-TR-113 and OGLE-TR- transiting companion would imply R 0.4 R , which is in- 132. A detailed account has been previously published in ≃ ⊙ compatiblewiththetemperatureobserved.Evenifbotheclise Bouchyetal. (2004) and Pontetal. (2004). The orbitsare in- and anti-eclipse were seen, the primary radius would only be cludedinFigure6forcompleteness. R 0.8 R , still much too low to match the spectral type. ∼ ⊙ Thereforethetargetisanon-synchronised,fast-rotatingF-star, whichleavesthreescenarioopen: – atransitingplanet – ablendwithabackgroundeclipsingbinary – afalsepositiveofthetransitdetection Thethirdpossibilityis discussedfurtherattheend ofthis Section. OGLE-TR-124: This object exhibits a linear drift in the radial velocity across eight nights of measurements, V˙ = r 73 15ms 1day 1.Themeasuredrotationalvelocity,vsini= − − − ± 6.3 0.1 kms 1, is not compatible with tidal synchronisation. − ± That the radial velocity drift is indeed a long-term drift and notpartofashort-periodorbitwasverifiedwithasinglelater measurementonthisobjectswiththeESOHARPSspectrom- eteron JD = 2453151.53withV = 8.982 13 kms 1.The Fig.6.Radialvelocitydataandorbitforthethreedetectedtran- r − ± − residuals around a linear relation (90 ms 1, 29ms 1without siting planets. Adapted from Pontetal. (2004); Bouchyetal. − − thediscrepantpointatJD=53083)putanupperlimitofaround (2004);Moutouetal.(2004). 0.1M onaputativeplanet. J OGLE-TR-131: This object shows no velocity variations, 4.4.Starswithoutshort-periodradialvelocityorbit with residuals of 60 ms 1. Ajusting an orbital solution gives − Two objects,measured8 timeseach,with veryhigh-accuracy K=16 17 ms 1, which implies m 0.2M for a possible − J ± ≤ radialvelocitydata,shownoradialvelocityvariationinphase planet. The shape of the transit is quite remote from a typi- with the transit signals to the level of less than 50 ms 1(see calcentral-transitshape,andthelightcurvefityieldsb > 0.79 − Fig.7).OGLE-TR-124showsaconstantdriftoftheradialve- fortheimpactparameter. locity throughout the observation period, and OGLE-TR-131 showsradialvelocityresidualsconsistentwiththenoisearound a constant velocity. A third object, OGLE-TR-109 shows no ForbothOGLE-TR-124andOGLE-TR-131,thebestfitto significantradialvelocityvariationseither,butwithlargerun- the light curve is given by a grazing transit with a large im- certaintiesdueto a large rotationalbroadening.These objects pactparameter.Thereforetheexplanationintermsoftransiting are discussed in some detail here. They are interesting in the planetis muchless likely than in terms ofblend with a back- contextofplanetsearchbecausetheabsenceofdetectedshort- groundeclipsingbinarycontributingafewpercentofthelight periodradialvelocityorbitiscompatiblewiththeexplanation and not seen in the CCF, or in terms of false positive of the ofthephotometricdataintermsofplanetarytransitiftheplanet transitdetectionprocedure. massissmallerthansomeupperlimit.Spectroscopicinforma- We note that all three objects in this section belong to a tion for these objects is given in Table 5. For the reasons ex- category where the existence of the transit signal itself is not posedbelowhowever,weestimatethattheseobjectsareprob- beyonddoubt–seeSection6.4andFig.11,wherewetrytoes- ablyfalsepositivesofthetransitdetectionprocedure. timatetheactualdetectionthresholdofthisOGLEtransitsur- veyinCarina.ThepositionofOGLE-TR-109,124and131in Figure11showsthattheystandoutbelowtheotherconfirmed OGLE-TR-109:ThespectrallinesofOGLE-TR-109show candidatesintermsofsignificanceofthetransitdetection.The a large rotational broadening, indicating vsini= 35.4 1.8 position of these objects without detected velocity variations ± kms 1. The radial velocity uncertainties are correspondingly nearorbelowthedetectabilitythresholdisunlikelytobeaco- − higher,and our 8 measurementsshow no significantvariation incidence.Someorallofthemmaysimplybefalsepositiveof withintheseuncertainties.Theradialvelocitydataisdisplayed thetransitdetectionprocedure,asuspicionreinforcedbytheir in Fig. 7. Fitting a circular orbit at the period of the transit non-squareshapeforOGLE-TR-124andOGLE-TR-131.