Table Of Content

DraftversionFebruary5,2008 PreprinttypesetusingLATEXstyleemulateapjv.11/12/01 LUPUS-TR-3B: A LOW-MASS TRANSITING HOT JUPITER IN THE GALACTIC PLANE? David T F Weldrake1, Daniel D R Bayliss2, Penny D Sackett2, Brandon W Tingley3, Michaël Gillon4,5 and Johny Setiawan1 Draft version February 5, 2008 ABSTRACT We present a strong case for a transiting Hot Jupiter planet identified during a single-field transit survey towards the Lupus Galactic plane. The object, Lupus-TR-3b, transits a V=17.4 K1V host star every 3.91405d. Spectroscopy and stellar colors indicate a host star with effective temperature 5000 ± 8 150K,with astellarmassandradiusof0.87±0.04M⊙ and0.82± 0.05R⊙, respectively. Limb-darkened 0 transit fitting yields a companion radius of 0.89±0.07 R and an orbital inclination of 88.3+1.3 deg. 0 J −0.8 Magellan6.5mMIKEradialvelocitymeasurementsreveala2.4σ K=114±25m/ssinusoidalvariationin 2 phase with the transit ephemeris. The resulting mass is 0.81 ± 0.18M and density 1.4 ± 0.4g/cm3. Y- J n bandPANICimagedeconvolutionrevealaV>21redneighbor0.4′′awaywhich,althoughhighlyunlikely, a wecannotconclusivelyruleoutasablendedbinarywithcurrentdata. However,blendsimulationsshow J that only the most unusual binary system can reproduce our observations. This object is very likely 6 a planet, detected from a highly efficient observational strategy. Lupus-TR-3b constitutes the faintest 1 ground-based detection to date, and one of the lowest mass Hot Jupiters known. ] Subject headings: planetary systems: individual (Lupus-TR-3b) - techniques: photometric - techniques: h radial velocities p o- 1. introduction project for increased efficiency, and PANIC imagery in- r cluding overheads), and 1 hour of ANU 2.3m time (DBS Transiting extrasolar planets provide unique oppor- st tunities to study the structure of exoplanets. Tran- spectroscopy)for a targetat V=17.4. We detail the likely a nature of this object and the overall system parameters. sit photometry, coupled with radial velocities, allows [ direct measurement of the exoplanetary radius, total 2. the photometric survey 2 mass and density. Currently ∼30 transiting planets are v known6, the majority identified in wide-field transit sur- We conducteda 52′×52′ fieldtransitsurveycenteredat 6 RA=15h30m36.3s, DEC=−42◦53′53.0′′, b=11◦ l=331.5◦ veys and confirmed by radial velocities. Due to their 4 towards Lupus, using the Wide Field Imager (WFI) on easy detection and transit-related bias, nearly all the 7 the Australian National University (ANU) 1m telescope currently known eclipsing exoplanets are large (∼ R ), 1 J at Siding Spring Observatory (SSO). Seven of the eight the exception being GJ 436b (Gillon et al. 2007a), and 1. in close orbits. Transits identified against bright stars WFICCDswereoperational,yieldinganeffective0.66deg2 1 (V612) permit investigation into planetary atmosphere field. Weobservedwithasinglebroad-bandV+Rfilterfor 7 26and27nightsinJune2005andJune2006,respectively. (Vidal-Madjar et al. 2004; Tinetti et al. 2007) and in- 0 Survey motivation is threefold: a transit searchin its own frared emission (Charbonneau et al. 2005; Deming et al. : right, a control field for the earlier globular cluster tran- v 2005). Deeper surveys using more numerous fainter stars i can detect large numbers of planets for statistical studies sit surveys of Weldrake et al. (2005, 2007), and a feasibil- X ity study for the massive 5.7 deg2 SkyMapper transiting of exoplanet populations, and additional system planets r planet survey (Bayliss & Sackett 2007). a through large aperture photometry. The stellar light curves consist of 1783 five-minute ex- Transit surveys uncover many candidates, most of posures in V+R with a mean seeing of 2.2′′. A total of whichareastrophysicalfalse-positiveswhichmustundergo 110,372 light curves were produced via differential imag- strictfollow-upprocedures(O’Donovan et al.2006,2007). ing techniques (Alard & Lupton 1998; Wozniak 2000), of Distinguishing hierarchical triples and blended binaries which16,134havesuitablephotometricquality(totallight (Torres et al. 2004; Mandushev et al. 2005) requires large curveuncertainty≤0.025mag)forthemaintransitsearch. time investmentandcarefulanalysiswith high-resolution, Correlated,systematic errorswereremovedfrom the light large-aperture spectroscopy and photometry. curvesusing the SYSREMalgorithm(Tamuz et al.2005). WepresentthedetectionofLupus-TR-3b,astrongcase From a tandem application of both the BLS foratransiting0.81±0.18M HotJupiterplanetidentified J (Kovács et al. 2002) and the Weldrake & Sackett (2005) from a deep, 0.66deg2 transitsurvey towardsLupus. This transit detection algorithms, six candidates were identi- result required 12.8 hours of Magellan 6.5m time spread fied. These were subjected to vigorous tests to determine over 5 nights (MIKE spectroscopy shared with another theirnature;Lupus-TR-3wasquicklyclassifiedasaprime 1 MaxPlanckInstitut fu¨rAstronomie,Ko¨nigstuhl17,D-69117,Heidelberg,Germany. Email: [email protected] 2 ResearchSchoolofAstronomyandAstrophysics,MountStromloObservatory,Cotter Road,WestonCreek,ACT2611,Australia 3 Instituted’AstronomyetAstrophysique,UniversitéLibrédeBruxelles,Belgium 4 GenevaObservatory,51ChemindesMaillettes,1290Versoix,Switzerland 5 Institutd’AstrophysiqueetdeGéophysique, UniversitéDeLiége,4000Liége,Belgium 6 http://exoplanet.eu/ 1 2 Weldrake et al target. A full analysis of the other candidates will form FEROS spectra on the MPG/ESO 2.2m, taken over four the subjectofaseparatepaper. Fourfulltransitsandtwo nights, ruled out a stellar binary at a few km/s level. half-transits (in egress phase) of Lupus-TR-3b were ob- TheDBSspectrumwascorrectedforE(B−V)=0.182mag served during our 53 nights. The photometry was used to reddening (Schlegel et al. 1998) and compared to MILES determine the ephemeris, best-fitting model, and system (Saćhez-Bláquez et al. 2006) empirical spectra. The re- parameters for comparison with the spectroscopic results. sults indicate that the target is a K1V star with effective temperature 5000 ± 150K and log(g)=4.5 ± 0.5. With these constraints and the measured stellar V−I, Yi et al. (2003) isochrones indicate a mass of 0.87 ± 0.04M⊙ and radius of 0.82 ± 0.05R⊙, assuming solar metallicity. As a check,weexaminedBertelli et al.(1994)isochrones,which produced a consistent mass of 0.83 ± 0.03M⊙. The Yi isochrone stellar mass and radius were used to derive R p and M . From the assumed isochrone luminosity, the dis- p tance is estimated at ∼2 Kpc. The stellar parameters are summarized in Table 2. Parameter Value R.A.(J2000.0) 15h30m18.7s DEC.(J2000.0) −42◦58′41.5” V 17.4±0.2 V-I 0.859±0.008 SpectralType K1V Teff 5000±150K Mass 0.87±0.04M⊙ Radius 0.82±0.05R⊙ Dist ∼2Kpc Table2: StellarparametersforLupus-TR-3. 4. radial velocity measurements Fig. 1.— Differential, phase-wrapped Lupus-TR-3 photome- The photometry and RV measurements for Lupus- try as a function of transit phase after application of SYSREM TR-3b are available upon request. We obtained high- (Tamuzetal. 2005). Data is binned using a 9-point binning, with the best-fitting model shown (middle and bottom) as a solid line. resolution spectroscopic observations using the MIKE Thebottom panel iscenteredonthetimeofanti-transit. spectrograph on the Magellan II (Clay) telescope at Las The transit photometry is shown in Fig. 1, covering Campanas Observatory in June 2007. MIKE had suc- the full range in phase and zoomed on the transit itself. cessfully confirmed the transiting planet orbiting OGLE- Also shown in Fig. 1 is the best-fitting model assuming a TR-113b (Konacki et al. 2004), a somewhat brighter star circular orbit, a V+R quadratic limb-darkening law, and (I=14.4) of a similar spectral type to Lupus-TR-3. Using R∗=0.82 ± 0.05R⊙ (see next section). All system param- MIKE in its standard configuration and a 0.35′′ slit, fif- eters derived from the photometric data and the best-fit teen usable spectra were obtained over four nights, with modelaregiveninTable1. Thetransithasobserveddura- resolutions of R=47,000 to 48,000, signal-to-noise of 5-10 tion0.109+0.003 daysandmaximumδF/F=0.013± 0.001. per pixel, and wavelength ranges of 5000-9500˚A. Expo- −0.007 The rms residuals from the model are 7.48×10−3 mag. sure times varied between 1400-2400s. In addition, spectra were taken of three bright radial velocity standards Parameter Value selected to match the spectral type and class of Lupus- Period 3.91405±0.00004days a 0.0464±0.0007AU TR-3. The targetand one of the standards were observed Tcenter 2453887.0818 ±0.0013HJD at airmass near unity each night. The seeing during the Rp 0.89±0.07RJ run varied from 0.7′′ to 1.05′′ and there is no correlation i 88.3+−10..38 deg betweenradialvelocity andseeing,minimizing the chance K 114±25m/s of any variable contaminating spectra. Mρ P 01..841±±0.04.1g8/cMmJ3 These data were reduced using standard IRAF7 tasks e 0(fixed) and the mtools package (for correcting the tilted slits in MIKE spectra). ThAr arcs, taken before and after each Table1: Lupus-TR-3bparameters. exposure, were used to wavelength calibrate the spectra. 3. characterizing lupus-tr-3 Radial velocities (RV) for each echelle order were com- V and I magnitudes were obtained for all stars in the puted by cross-correlating the Lupus-TR-3 spectra with survey field using the ANU 1m telescope. In addition, one long exposure of K5V HD 219509 that consistently spectroscopy was obtained with the DBS on the ANU produced the most stable, low-scatter results for the tar- 2.3m telescope to determine the stellar type (R=200, getandHD126053,anotherstandardobservedatapprox- per-pixel S/N=13), reduced using the ”onedspec” pack- imately the same times and airmass as Lupus-TR-3. age of IRAF. Wavelength calibration was performed us- Thefinalresults,showninFig. 2,use18ordersselected ing CuHe arcs taken throughout the same night. Five forhighsignal-to-noise(S/N) andlowtelluric contamina- 7 IRAFisdistributedbytheNationalOpticalAstronomyObservatories,whichareoperatedbytheAssociationofUniversitiesforResearchin Astronomy,Inc.,undercooperativeagreementwiththeNationalScienceFoundation. Lupus-TR-3b 3 tion;thescatterintheseorderswasusedasanestimateof ions (B and C) were detected inside the 2′′ PSF of the the uncertainity in the radial velocity. Only the period of target on the ANU 1m, one 4.1 mag fainter than Lupus- thetransitwasfixedinfittingtheradialvelocitymeasure- TR-3located 1.7′′away(B),andtheother5.6magfainter ments. Assuming a circular orbit, the best fit sinusoidal at2.2′′distance(C).StarCistoofainttobeabinarymas- variation, detected at 2.4σ comparedto the null hypothe- querading as a planet. sis, has an amplitude of K=114 ± 25 m/s and a phase consistent with the transit ephemeris, to within 0.13d, 5.2. Y-Band Imaging wellwithinfituncertainties. Uncertaintieswerecalculated Wealsoobtainedhigh-spatialresolutionimagesin0.68′′ from the covariance matrix of the fitted parameters. seeing taken in the Y-band (1.035 microns) with PANIC (0′′.125 pixels) on MagellanI, to further rule out the pos- siblity of a blend and to check the DECPHOT deconvolution. We obtained 18 frames with exposure times of 120s each, and the best quality frames were combined to make a deep image of the field (Fig. 3). Fig. 2.—RVforLupus-TR-3(lowerpoints)andthebrightstan- dardHD126053(upper,offsetpoints)asafunctionofHeliocentric Julian Date. Error bars are determined from the actual scatter in theechelleordersused. Largeropencirclesareuncertainty-weighted nightlyaverages. The RV signal is small and its detection significance rather low, which is not surprising given the faintness of Lupus-TR-3. Detecting the signalrequiredanexceptional wavelengthsolutionwithanrms of0.0011pixels,rejection of echelle orders susceptible to terrestrial contaminants, observationsatnearlyconstantandverylowairmass,and high-signal-to-noiseRVstandardsofsimilarspectraltype. We note that the scatter in the points taken on any given Fig. 3.—Y-band40×45arcsec2 PANICimageryforLupus-TR-3 night can be taken as a proxy for the reliability of our (left of center). Three blended neighbors are revealed, the two de- measurements. This scatter over 2-3 hour time frames is tected with DECPHOT on our ANU 1m images (B and C) and a third(D).Insetisthepost-DECPHOTresultforthetarget,reveal- consistent with the uncertainties displayed on individual ingafourthblendedcompanion,starE. points in Fig. 2, which were, in turn, calculated from the actual scatter across individual echelle orders for that This image confirms the presence of the two neighbors spectrum. Theresultingplanetarymassis0.81±0.18M , foundusingDECPHOTonour1mimagesatthesamesep- J assuming a 0.87 ± 0.04M⊙ host star. Fixing the phase to arations,withY-bandmagnitudedifferencesof2.5forstar match the transit data produced a nearly identical plan- B and 4.2 for star C. In addition, one other close neigh- etary mass (of even greater significance), as did different bor(D)wasdetected4.3magnitudesfainterintheY-band choices for the echelle orders used in the analysis. than Lupus-TR-3 at a separation of 1.8′′. Star D is too faint in V+R to harbor a feasible confusing eclipse signal, 5. ruling out blended binaries and in any event we have already shown that the transit is strongly associated with the position of the target. 5.1. Image Deconvolution with DECPHOT WealsoperformedaDECPHOTanalysisofthePANIC In principle, a transit signal can be caused by a faint image, seen superimposed on Fig. 3, which revealed an- eclipsingbinarysystemthatislocatedcloseenoughtothe other star, star E, 0.4′′ away from the target. This could targettobeblendedwithinthetarget’spointspreadfunc- constitute a blended binary. Star E is not seen in our de- tion(PSF),producingthe observedtransitandusuallyno convolved V+R images, indicating a V magnitude fainter detectable radialvelocityperiodicity. Onbrightertargets, than 21, contributing ≤0.036 magnitudes to the target linebisectoranalysiscanbeusedtodetermineifablendis brightness. It is 2.8 magnitudes fainter than Lupus-TR-3 present. Ourdata showno signofin-phasebisectorvaria- in the Y-band, indicating a red spectral type. If star E tions, but given the low signal-to-noise of the spectra, the is a binary that mimics the planetary hypothesis, it must constraint on blended binaries is inconclusive. harbor eclipses with a depth of at least 36%. Planned We therefore also tested for blends by performing observations will test this possibility. a DECPHOT analysis (Magain et al. 2007; Gillon et al. DECPHOT was then used to analyze 73 images from 2007b) over one night in order to effectively “de-blend” one night containing a well-sampled transit. Light curves our ANU 1m images of Lupus-TR-3. Two faint compan- were created for the target and star B, to determine with 4 Weldrake et al which star, and in what location, the transit signal was rare small companion close to the hydrogenburning limit produced. The result is shown in Fig. 4. The transit sig- (ie: OGLE-TR-122b,Pont et al.2005)canalsoberejected nalclearlyoccursatthetargetlocationandwiththesame as it would have to be of very comparable brightness to depth and duration. This rules out star B as a binary, Lupus-TR-3toproducetheobservedtransitdepth(sucha with C and D being ruled out by their angular distance system would have similar undiluted transit depth). Such and faintness. Only star E remains as a blend candidate. a star is not observed in our DECPHOT analysis. In summary, in the unlikely event that the eclipse in Lupus-TR-3 is caused by a blend, only star E can be responsible. Howeveritsrednessandfaintnessargueagainst theA+Mbinaryconfigurationsmostcapableofreproduc- ingourobservations. Ourdatacannotruleoutthedistant possibility that star E is an usual binary system or a stel- larcompaniontoLupus-TR-3itself. Theotherthreeclose neighbors are ruled out using DECPHOT. 6. conclusions We announce the detection of Lupus-TR-3b, a strong candidate for a transiting Hot Jupiter orbiting a V=17.4 Fig. 4.—Post-DECPHOTin-transitphotometryforLupus-TR-3 K1V star with a period of 3.91405d. Our observations (bottom) and star B (top). The signal is strongly associated with are consistent with a planet of radius R = 0.89 ± 0.07 thethetargetlocation,rulingoutstarBasabinary. StarCisruled p R (assuming a Jovian equatorial radius of 71,492 kms). out duetoitslargerangular distance, andDistoo fainttoharbor J anyeclipse. MIKE radial velocities reveal a 2.4σ low-amplitude variation in phase with the transit ephemeris, yielding a plan- 5.3. Modeling Possible Binary Configurations etary mass of 0.81 ± 0.18M and thus density of 1.4 ± J A large number of stellar binaries were modeled to ex- 0.4g/cm3, equivalent to that of Jupiter. amine if star E could be a blended binary. The tested Deep imaging, image deconvolution techniques and ex- binary models spanned 21 spectral types between A7 and tensive binary modeling argue strongly against a blended M3 eclipsing at 41 different impact parameters. Our well- or triple configuration for the system that can reproduce determinedtransitmorphologyplacesstrictconstraintson the photometric observations. With currentdata,we can- the nature of any prospective blend, as Lupus-TR-3 dis- not rule out the unlikely possibility that the nearby faint, plays a large flat-bottomed transit, which depends most red star E is a blended binary responsible for the transit, significantly on the ratio of radiibetween the components butthis istestablewithminimalamountsofplannedhigh and the impact parameter of the transit and less impor- resolution imaging. The low-mass, Hot Jupiter Lupus- tantly on the orbital eccentricity and total system mass. TR-3b is an example of the extraordinary discoveries The best fit blend scenario (A7 primary and M3 sec- thatawaitif innovativecomputationalmethods andtime- ondary with a 0.134 mag central eclipse, hence 4.7 mags efficientfollow-upobservationsarebroughttobearoncan- fainterthanthetarget)canberuledoutbasedonthetran- didates selected from wide and deep transit surveys. sit duration. An eccentricity of >0.56 would be required to reduce the circular orbit duration of 0.21d to the ob- The authors wish to thank Grant Kennedy and Karen served0.1devent. Non-zeroeccentricitiesarenotobserved Lewis for assistance during the original observing runs; inanybinarysystemswithsuchshortperiods(Abt2006). Ken Freeman for the ANU 2.3m spectrum; Andre Mu¨ller InanycasetherednessandV+RfaintnessofstarEmakes for the FEROS spectra; Nidia Morrell, Miguel Roth, and it unlikely to contain an A-type primary. Chris Burns for the Magellan Y-band images of Lupus- Binary systems with larger primaries can also mimic TR-3, and the anonymous referee. We acknowledge fi- the observations, albeit less ably, and can also be ruled nancial support from the Access to Major Research Fa- out based on the more extreme eccentricities required for cilities Programme which is a component of the Interna- larger, earlier spectral types primaries. Also, hotter and tional Science Linkages Programme established under the larger stars increase the probability that ellipsoidal varia- Australian Government’s innovation statement, Backing tions will be observed in our data. A blended star with a Australia’s Ability. REFERENCES Alard,C.&Lupton, R.H.1998,ApJ,503,325 Gillon,M.,Magain,P.,Chantry, V.,Letawe, G., Sohy, S.,Courbin, Alonso,R.,etal.2004,ApJ,613,L153 F.,Pont, F.,&Moutou, C.2007a, TransitingExtrapolar Planets Abt,H.A.2006,ApJ,651,1151 Workshop, 366,113 Bayliss, D. D. R., & Sackett, P. D. 2007, Transiting Extrapolar Gillon,M.,etal.2007b, A&A,472,L13 PlanetsWorkshop,366,320 Konacki, M., Torres, Sasselov, D. D. Pietrzyn´ski, G., Udalski, A., Bertelli, G., Bressan, A., Chiosi, C., Fagotto, F., & Nasi, E. 1994, Jha,S.,Ruiz,M.,Gieren,W.,Minniti,D., 2003, ApJ,609,L37 A&AS,106,275 Kovaćs,G.,Zucker,S.,&Mazeh,T.2002, A&A,391,369 Charbonneau, D.,etal.2005,ApJ,626,523 Magain, P.,Courbin, F., Gillon,M.,Sohy, S., Letawe, G., Chantry, Claret,A.2000,A&A,363,1081 V.,&Letawe, Y.2007,A&A,461,373 Deming, D., Seager, S., Richardson, L. J., & Harrington, J. 2005, Mandushev,G.,etal.2005, ApJ,621,1061 Nature,434,740 O’Donovan,F.T.,etal.2006,ApJ,644,1237 O’Donovan,F.T.,etal.2007,ApJ,662,658 Lupus-TR-3b 5 Pont, F., Melo, C. H. F., Bouchy, F., Udry, S., Queloz, D., Mayor, Torres,G.,Konacki, M.,Sasselov, D.D.,&Jha,S. 2004, ApJ,614, M.,&Santos,N.C.2005,A&A,433,L21 979 Saćhez-Bla´quez,P.;Peletier,R.F.;Jiménez-Vicente,J.;Cardiel,N.; Vidal-Madjar,A.,etal.2004, ApJ,604,L69 Cenarro,A.J.;Falcoń-Barroso,J.;Gorgas,J.;Selam,S.;Vazdekis, Weldrake, D. T., Sackett, P. D, Bridges, T. J, & . 2007, ArXiv e- A. 2006, MNRAS,371,703 prints,710,arXiv:0710.3461 Schlegel, D.J.,Finkbeiner,D.P.,&Davis,M.1998,ApJ,500,525 Weldrake,D.T.F.,&Sackett, P.D.2005,ApJ,620,1033 Seager, S.,&Mallén-Ornelas,G.2003, ApJ,585,1038 Weldrake,D.T.F.,Sackett,P.D.,Bridges,T.J.,&Freeman,K.C. Southworth, J., Wheatley, P. J., & Sams, G. 2007, ArXiv e-prints, 2005, ApJ,620,1043 704,arXiv:0704.1570 Wozniak,P.R.2000,ActaAstronomica,50,421 Tamuz,O.,Mazeh,T.,&Zucker,S.2005, MNRAS,356,1466 Yi,S.K.,Kim,Y.-C.,&Demarque,P.2003,ApJS,144,259 Tinetti, G.,Vidal-Madjar, A., Liang, M.-C., Beaulieu, J.-P., Yung, Y., Carey, S., Barber, R. J., Tennyson, J., Ribas, I., Allard, N., Ballester,G.E.,Sing,D.K.,Selsis,F.2007,Nature,448,169