CoRoT pictures transiting exoplanets
CCoRoT pictures transiting exoplanets
Claire Moutou , and Magali Deleuil Canada-France-Hawaii Telescope Corporation, 65-1238 Mamalahoa Highway, Kamuela HI 96743,[email protected] Laboratoire d’Astrophysique de Marseille, 38 rue Fr´ed´eric Joliot-Curie, 13376 Marseille cedex 13,France, [email protected]
Abstract
The detection and characterization of exoplanets have made huge progresses since thefirst discoveries in the late nineties. In particular, the independent measurement of themass and radius of planets, by combining the transit and radial-velociy techniques,allowed exploring their density and hence, their internal structure. With CoRoT (2007-2012), the pioneering CNES space-based mission in this investigation, about thirty newplanets were characterized. CoRoT has enhanced the diversity of giant exoplanets anddiscovered the first telluric exoplanet. Following CoRoT, the NASA Kepler missionhas extended our knowledge to small-size planets, multiple systems and planets orbit-ing binaries. Exploring these new worlds will continue with the NASA / TESS (2017)and ESA / PLATO (2024) missions.
Keywords:
Astrophysics - Space mission - Planetary systems - Observations -Internal structure of planets
1. Introduction
The search for exoplanets around solar-type stars has started in the 1990’s withthe direct imaging method and the radial velocity method. While the first one useshigh-contrast and high-resolution imaging of nearby stars to search for the dim light ofa sub-stellar companion, the second method uses indirect measurements of stars andsearches for velocity wobbling due to the gravitational pull by an invisible planet. Bothmethods had their first results in 1995: the first detected brown dwarf companion Gl229 B (about 50 Jupiter masses) by direct imaging [1] and the first detected Jupiter-like exoplanet 51 Pegasi b, orbiting a solar-type star [2]. The main di ff erence betweenthese two companions is probably their distance to their parent star: while the orbitalperiod of Gl 229 B is larger than 10,000 years, the one of 51 Pegasi b is 4.5 days.This exemplifies the di ff erent biases of the two methods, imaging being sensitive toplanets at long orbital period whereas radial velocity is performant for systems at shortorbital period. In the meantime, the Hubble Space Telescope was discovering thatyoung protostars were being formed in opaque, dusty disk structures, another hint thatplanetary formation could be a universal phenomenon. Preprint submitted to Elsevier September 3, 2018 a r X i v : . [ a s t r o - ph . E P ] O c t hese discoveries triggered a new field of research and extreme enthusiasm in theastrophysical community. A third method is quickly proposed and experimented onthe first detected radial-velocity planets: it consists in searching for a slight dimmingof the star’s light that would be due to the crossing of a planet on the stellar disk – atransit. This method works for planets whose orbit is perfectly aligned with the line ofsight between the observer and the parent star. To discover transiting planets, it is thusnecessary to observe for a long time a very large number of stars. The method is, asthe radial-velocity method, biased towards detecting short-period planets.A few months after the discovery of 51 Peg b, a team of French astronomers andCNES engineers proposed the spaced-based CoRoT mission [3], aimed at observinglarge numbers of stars with extreme photometric precision and long time series -as onlypossible from space-, in search for exoplanet transits. While CoRoT was designed,built, and integrated, a few exoplanetary transits were being detected from ground-based observatories. The first of them is HD 209458 b [4], a short-period Jupiter-likeplanet as 51 Pegasi b. The independent detections of HD 209458 b with the radialvelocity and the transit methods have been a strong confirmation that these giant planetsin very close distance to their stars actually existed. The combination of both methodsalso allowed to measure the planet’s mass (from the radial-velocity amplitude) and itsradius (from the transit depth), hence the bulk density of the planet. While the densityof HD 209458 b is within the range of the Solar System’s largest gaseous planets,the planet radius could not be reproduced by models of internal structure: the giantexoplanet appeared largely inflated, for its star’s composition and age. When CoRoTwas then launched in 2006, there were 15 known transiting exoplanets; all were giant,gaseous planets, whose transits (1-3% deep, 2-3 hour long) were easily detected fromground-based telescopes of small size (typically less than 1m).In this article, we will review the findings of the CoRoT mission in its search forexoplanets. After describing the mission concept and instrument, we will review theprocess that starts with candidates and leads to exoplanets. Finally, we will emphasizethe learnings from the CoRoT exoplanets and have a look towards other exoplanetsurveys and future missions in this field.
2. The CoRoT satellite
Searching for exoplanets with the transit method requires both very good photomet-ric precision and a very large number of observed stars. The duration of the observingrun sets the domain of orbital periods that is within reach, as the transit occurs onceper orbital period (note that the secondary transit, occurring when the planet passes be-hind the star with respect to the observer, is only rarely detected due to a much smalleramplitude and additional constraint on the geometry). In addition to transit detection,precise stellar photometry from space also allows to probe the stellar interiors withasteroseismology, another main science objective of CoRoT (on a smaller number ofmuch brighter stars) [3]. In the domain of transiting exoplanets, a space-based missionallows to detect transiting planets of smaller size and / or of longer orbital period thanground-based photometric surveys.The CoRoT design can be summarized by [5]:2 a PROTEUS platform ensuring a polar orbit at 900km altitude and 6-month con-tinuous access to a given field of view • a 27-cm mirror in an o ff -axis telescope • a focal plane divided in two, with an asteroseismology channel and an exoplanetchannel; a field of view of 1.4 ◦ x2.8 ◦ for the exoplanet channel and 2.3 arcsecpixel size • an operation plan adapted to the satellite orbit, with a succession of short runs(15-30 days) and long runs (80 or 150 days) • two main fields of observation, located in two opposite regions in the Milky Waytowards the Monoceros and Aquila constellations and close to the Galactic plane.These locations were optimized for stellar density, considering other observingconstraints, such as the presence of bright nearby stars of scientific interest forthe asteroseismology channel • a 7.10 − photometric precision on a R =
15 star in a 2h timescale • about 6,000 stars observed simultaneously in the exoplanet channel, in the mag-nitude range of V =
11 to 16.5 • • for the 30% brightest stars, three-color light curves generated by a low-resolutionprism. Only the white light curve is provided for all other starsLaunched on December 26, 2006 by a Soyouz rocket from Baikonour, the CoRoTsatellite has provided astronomical data from February 2007 (after a run of commis-sioning) to November 2012. Its lifetime was originally granted to be 3 years, and wasextended twice. Observations were discontinued due to electronic failures, probablydue to high-energy particle bombardment. The satellite was stopped in June 2014 afterprogramming its slow decay to Earth.A picture of the satellite during its integration in Cannes is shown on Figure 1.An archive of the CoRoT data is now fully public and can be accessed at http: // idoc-corotn2-public.ias.u-psud.fr.
3. Candidates or planets?
In total, 163,664 stars were observed during the lifetime of CoRoT for a total of169,967 light curves. Light curves were processed onboard and sent to Earth whereadditional data reduction got rid of instrumental e ff ects. Several algorithms for transitdetection were then applied to the light curves, generating several lists of exoplanetcandidates.A planet candidate is not necessary a planet. There are several other astrophysicalscenarios than mimic the planetary transits, involving multiple eclipsing stars ratherthan planets: grazing binaries, giant-dwarf binaries, eclipsing binary diluted by a third3 igure 1: The CoRoT satellite in its integration hall at Thales Alenia Space in Cannes, in 2005.Table 1: Stars, observing runs, light curves, candidates and planets counts from the CoRoT Exoplanet pro-gram. Direction Runs Light curves Candidates PlanetsAquila 13 85489 348 17Monoceros 13 89215 251 16body. If the detected transit is very shallow, a possible contamination is a giant transit-ing planet orbiting a star diluted by another star. In such a case, the planet interpretationis correct, but the inferred radius of the planet would be largely incorrect.As CoRoT provides lists of transiting candidates, these are then submitted to seriesof tests to evaluate their origin. For instance, the three-colored light curves, whenavailable, can be used to eliminate candidates showing di ff erent transit depth in thedi ff erent colors (signature of a blended stellar scenario). The distance between transitsin a light curve giving the orbital period, it is possible to evaluate the expected transitduration from this period and the stellar radius; if too discrepant, the chances are highthat the transit is due to a diluted eclipsing binary with a giant primary star. For eachrun, about 50 candidates are detected, and about half of them are further consideredas compatible with a planetary signature. On these ones, it is then required to performadditional observations to unveil their nature and complete their characterization. Table1 gives a summary of the number of CoRoT observing runs, observed stars, detectedcandidates and final planet yield in both Galactic directions.Complementary observations of CoRoT candidates include time-critical ground-4ased photometry, adaptive optics, and high-resolution spectroscopy with a high radial-velocity precision instrument. The first observation provides us with a confirmationthat the transit does not occur on a nearby star (less than 30 arc sec away); it is re-quired by the fact that CoRoT pixels and CoRoT photometric apertures are, respec-tively, 2.3 arcsec and ∼
30 arcsec wide, and the observed star can be contaminatedby fainter neighbors. Adaptive optics follows up on this objective, putting constraintson fainter neighbors at shorter distances. Finally, a radial velocity monitoring of thetarget identifies the grazing binaries, the multi-component spectroscopic binaries, andthe planets with a measurable mass. The magnitude range of CoRoT targets, definedto have an appropriate stellar sample size due to geometrical restrictions of the tran-sit alignment, severely restricts the capability of spectroscopic follow-up observations,especially for stars fainter than 14. This is even more critical even for very shallowtransits that are potentially due to very low-mass planets. Nevertheless, more than 120nights of telescope time with high-precision radial-velocity spectrographs were dedi-cated to complementary observations of CoRoT candidates, including HARPS at theEuropean Southern Observatory, SOPHIE at Observatoire de Haute Provence, HIRESat Keck Observatory and FIES on the Nordic Optical Telescope. The outputs of com-plementary observations are among the following: • another neighbor star is identified in the vicinity of the main target, which showsdeep, stellar eclipses at the expected time. The transit detected by CoRoT is theresult of a multiple-system dilution and usually not considered further. • the spectrum shows two or more components, or an amplitude of the radial-velocity curve consistent with a companion in the stellar mass regime and at thecorrect period. This candidate planet is rejected. • the spectrum shows a single component, and the radial-velocity curve reveals amass of the transiting companion which is compatible with a planetary nature.Then a full characterization of the planetary system is undertaken. • the spectrum reveals a fast rotating star, or a star hotter than ∼ • the radial-velocity series has su ffi cient precision but no signature is revealed atthe ephemeris of the CoRoT transit signals. Unless a complex, peculiar systemis identified, the system’s nature remains unclear and an upper limit for the massof a possible planetary candidate is estimated.In the remaining cases where complementary observations bring no definite answer,we have to rely on statistical comparison of data with a set of di ff erent astrophysicalmodels. The CoRoT team has developed such a tool using Bayesian statistics, PASTIS[6], which has played a large role in some planet identifications, such as CoRoT-16 b,CoRoT-22 b and CoRoT-24 b. 5 igure 3: The period and depth of CoRoT planetary candidates (courtesy A. Santerne). The size of the symbols indicatesthe apparent luminosity of the parent star (small meaning faint). diluted binaries as for the solved cases (see Table 1). With a more conservative approach, one can only saythat 6% of the detected candidates turn out to be confirmed planets, with a mass measurement throughthe radial-velocity detection of the planet signature. The number of confirmed planets is mainly limitedby performances of complementary observations, in relation with the faint luminosity of the target stars.Estimating the planet occurrence of the CoRoT mission requires as inputs the number of relevantobserved stars, the final number of confirmed planets, the detection yield of the instrument, and cor-rections for the transit probability. The calculation has been done for hot Jupiters where the detectionyield of CoRoT is optimal (assumed 100%). The result is that 1 ± ± ± ± Figure 2: The transit depth versus the orbital period of CoRoT candidates. The symbol size is proportionalto the star brightness. (illustration by A. Santerne)
4. CoRoT main discoveries
As for other subsamples of exoplanets, the CoRoT discoveries show an extreme di-versity of parameters, especially when compared to planets in the Solar System. First,as CoRoT measures the planetary radii, it is directly observed that some planets havelarge radius anomalies. These anomalies can go in both directions: either positive(radius inflation) or negative (radius contraction) and are related to the theories of for-mation and evolution of a solar-composition gaseous planet made of hydrogen andhelium, with no core. Several planets characterized by CoRoT and its follow-up instru-ments have a radius larger by more than 10%: CoRoT-1 b, 2 b, 5 b, 11 b, 12 b, 19 b[7], and 26 b. This inflation is thought to be due to a small fraction of additional energyreceived by the planet and dissipated in its interior [8]. Several mechanisms have beenproposed for these non-radiative processes, for instance invoking tidal interactions withthe close-by star. If most exoplanet inflated radii can be explained with a very smallamount of additional energy (about 1%), one of them, CoRoT-2 b, requires much morebecause of its large mass. Other mechanisms have been proposed, invoking dynamicalevents during the past evolution of this system [9].Other planets have a negative radius anomaly; their radius is smaller than expectedwith respect to their mass, expected composition and age. This anomaly sometimesrequires huge amounts of heavy elements, probably condensed in a central core. Thedense planets discovered by CoRoT are: CoRoT-8 b, 10 b, 13 b, 20 b. CoRoT-20 b [10]is an extreme case, requiring about 500 Earth masses in heavy elements and questioningthe formation mechanism of planets with such heavy elements enrichment.CoRoT-9 b [11] has attracted interest, among CoRoT discoveries, because of itsrelatively long orbital period (95 days). It is a giant planet and its radius is not inflated6ith respect to the expect H-He composition, illustrating the fact that radius inflationin other transiting giant planets was indeed due to excess energy income from theirproximity to the star.CoRoT also discovered the first transiting planet in the super-Earth regime, CoRoT-7 b [12]. The planet is qualified as Super-Earth because it has a density of 6.61 g / cm ,close to the density of Earth [13], and a radius of ∼ ff erence between both hemispheresreaching 2400K [15]. The composition of CoRoT-7 b is likely dominated by silicates,that are eroded into a thin atmosphere of rocky vapor due to the large surface tempera-tures [15]. Today, thanks to Kepler and ground-based surveys around M stars, a smallnumber of similar planets have been discovered.CoRoT has found another multiple planetary system composed of two transitingplanets or 3.7 and 5 Earth radii and 5.1 and 11.8 day periods, respectively [16]. Bothare compatible with Neptune-like planets, with a significant gaseous envelope. Theremay be a third, massive planet in the same system, at an orbital distance at least 30times larger. Except for CoRoT-7 and CoRoT-24 systems, all other planets discoveredby CoRoT seem to be single companions of their parent star. CoRoT has found noplanet orbiting binary stars.As a summary, the mass-radius and the period-eccentricity diagrams of CoRoTexoplanets are shown on Figure 3. Superimposed to CoRoT exoplanet parameters areexoplanet properties measured with other surveys until January 2015. The diversityobserved with CoRoT has been confirmed by other observing campaigns, and evenfurther extended.
5. Future is bright
CoRoT is not the only transit-search survey in this past decade. Simultaneouslyto the preparation and launch of CoRoT, several ground-based surveys have started, tosearch for exoplanets. These surveys (eg, Super-WASP, HAT, TrES, Qatar), if biasedtowards giant exoplanets with short periods, were extremely productive and largelycontributed to our current knowledge of these planetary companions (Figure 3).In March 2009, NASA launched the Kepler satellite, a more ambitious, large-scalemission for exoplanet transit searches. The outcome of the Kepler mission has shownthe ubiquity of small-size planets and the existence of compact aligned multiple sys-tems [17, 18, 19]. Kepler stopped observations in May 2013 and the telescope resumedobservations in a lower-performance mode, under the name K2, in June 2014.7 igure 3: The mass and radius of the CoRoT planets (left) and the orbital eccentricity as a function of period(right). Open diamonds show all transiting planets known at the beginning of 2015, red filled circles showCoRoT planets, and small green dots show non-transiting radial-velocity planets.
The limitations of the pioneer space transit surveys undertaken by CoRoT and Ke-pler are related to the magnitude range of the target stars. For both missions, the transitcandidates orbiting the most numerous faint stars could not be followed up with thehighest accuracy radial velocity spectrographs, and only an upper limit of their massis (or will be) available. This limitation will not exist anymore for the next space mis-sions TESS (NASA, launch in 2017) and PLATO (ESA, launch in 2024). Focusingon the brightest stars in a much larger part of the sky, TESS and PLATO will havethe capacity of unveiling the low-mass planets in the habitable zones of their stars(cool stars for TESS, solar-type stars for PLATO), and to characterize these stars to-tally through intense complementary observing campaigns. Getting the planet masseswith a few % accuracy in the domain of Earth masses is challenging, but possible ifthe parent star is bright enough. New facilities will be added in the next years to theexisting high-precision radial-velocity instruments (VLT / ESPRESSO, CFHT / SPIRou,Calar-Alto / CARMENES, Subaru / IRD...), and their contribution to TESS and PLATOcomplementary observations will be highly required. The field of research pioneeredby CoRoT has thus an inestimable future. The next step after identifying and char-acterizing other terrestrial worlds in the solar neighbourhood will be to explore theiratmospheric properties, in particular in a quest for biosignatures.
Acknowledgements
This is an invited contribution to the special issue ”Invited contributions of 2014 geo-science laureates of the French Academy of Sciences ”. It has been reviewed by8ranc¸oise Combes and editor Vincent Courtillot. We are thankful to the technical teamsthat have been in charge of CoRoT at CNES and in partner institutes from design tooperations. We want to acknowledge the role of the science team, for its high motiva-tion for about ten years. CoRoT would not have been such an adventure without thefar-reaching involvement of A. Baglin, J. Schneider, A. L´eger and D. Rouan particu-larly.
References [1] Oppenheimer B.R., Kulkarni S.R., Matthews K., Nakajima T. Infrared Spec-trum of the Cool Brown Dwarf Gl 229B. Science1995;270:1478–9.[2] Mayor M., Queloz D. A Jupiter-mass companion to a solar-type star.Nature1995;378:355–9.[3] Baglin A. COROT: A minisat for pionnier science, asteroseismology and planetsfinding. Advances in Space Research2003;31:345–9.[4] Charbonneau D., Brown T.M., Latham D.W., Mayor M. Detection of PlanetaryTransits Across a Sun-like Star. ApJL2000;529:L45–8. doi: . arXiv:astro-ph/9911436arXiv:astro-ph/9911436