An Ultra-Hot Neptune in the Neptune desert
James S. Jenkins, Matías R. Díaz, Nicolás T. Kurtovic, Néstor Espinoza, Jose I. Vines, Pablo A. Peña Rojas, Rafael Brahm, Pascal Torres, Pía Cortés-Zuleta, Maritza G. Soto, Eric D. Lopez, George W. King, Peter J. Wheatley, Joshua N. Winn, David R. Ciardi, George Ricker, Roland Vanderspek, David W. Latham, Sara Seager, Jon M. Jenkins, Charles A. Beichman, Allyson Bieryla, Christopher J. Burke, Jessie L. Christiansen, Christopher E. Henze, Todd C. Klaus, Sean McCauliff, Mayuko Mori, Norio Narita, Taku Nishiumi, Motohide Tamura, Jerome Pitogo de Leon, Samuel N. Quinn, Jesus Noel Villaseñor, Michael Vezie, Jack J. Lissauer, Karen A. Collins, Kevin I. Collins, Giovanni Isopi, Franco Mallia, Andrea Ercolino, Cristobal Petrovich, Andrés Jordán, Jack S. Acton, David J. Armstrong, Daniel Bayliss, François Bouchy, Claudia Belardi, Edward M. Bryant, Matthew R. Burleigh, Juan Cabrera, Sarah L. Casewell, Alexander Chaushev, Benjamin F. Cooke, Philipp Eigmüller, Anders Erikson, Emma Foxell, Boris T. Gänsicke, Samuel Gill, Edward Gillen, Maximilian N. Günther, Michael R. Goad, Matthew J. Hooton, James A. G. Jackman, Tom Louden, James McCormac, Maximiliano Moyano, Louise D. Nielsen, Don Pollacco, Didier Queloz, Heike Rauer, Liam Raynard, Alexis M. S. Smith, Rosanna H. Tilbrook, Ruth Titz-Weider, Oliver Turner, Stéphane Udry, Simon. R. Walker, Christopher A. Watson, Richard G. West, Enric Palle, Carl Ziegler, Nicholas Law, Andrew W. Mann
DDraft version September 30, 2020
Typeset using L A TEX default style in AASTeX62
An Ultra-Hot Neptune in the Neptune desert
James S. Jenkins,
1, 2, ∗ Mat´ıas R. D´ıaz,
1, 2
Nicol´as T. Kurtovic, N´estor Espinoza, Jose I. Vines, Pablo A. Pe˜na Rojas, Rafael Brahm,
4, 5
Pascal Torres, P´ıa Cort´es-Zuleta, Maritza G. Soto, Eric D. Lopez, George W. King,
9, 10
Peter J. Wheatley,
9, 10
Joshua N. Winn, David R. Ciardi, George Ricker, Roland Vanderspek, David W. Latham, Sara Seager,
13, 16
Jon M. Jenkins, Charles A. Beichman, Allyson Bieryla, Christopher J. Burke, Jessie L. Christiansen, Christopher E. Henze, Todd C. Klaus, Sean McCauliff, Mayuko Mori, Norio Narita,
19, 20, 21, 22, 23
Taku Nishiumi, Motohide Tamura,
18, 21, 22
Jerome Pitogo de Leon, Samuel N. Quinn, Jesus Noel Villase˜nor, Michael Vezie, Jack J. Lissauer, Karen A. Collins, Kevin I. Collins, Giovanni Isopi, Franco Mallia, Andrea Ercolino, Cristobal Petrovich,
27, 28
Andr´es Jord´an,
4, 5
Jack S. Acton, David J. Armstrong,
9, 10
Daniel Bayliss, Franc¸ois Bouchy, Claudia Belardi, Edward M. Bryant,
9, 10
Matthew R. Burleigh, Juan Cabrera, Sarah L. Casewell, Alexander Chaushev, Benjamin F. Cooke,
9, 10
Philipp Eigm¨uller, Anders Erikson, Emma Foxell,
9, 10
Boris T. G¨ansicke, Samuel Gill,
9, 10
Edward Gillen, † Maximilian N. G¨unther, Michael R. Goad, Matthew J. Hooton, James A. G. Jackman,
9, 10
Tom Louden,
9, 10
James McCormac,
9, 10
Maximiliano Moyano, Louise D. Nielsen, Don Pollacco,
9, 10
Didier Queloz, Heike Rauer,
31, 32, 36
Liam Raynard, Alexis M. S. Smith, Rosanna H. Tilbrook, Ruth Titz-Weider, Oliver Turner, St´ephane Udry, Simon. R. Walker, Christopher A. Watson, Richard G. West,
9, 10
Enric Palle,
23, 37
Carl Ziegler, Nicholas Law, andAndrew W. Mann Departamento de Astronoma, Universidad de Chile, Camino El Observatorio 1515, Las Condes, Santiago, Chile Centro de Astrof´ısica y Tecnolog´ıas Afines (CATA), Casilla 36-D, Santiago, Chile Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA Facultad de Ingenier´ıa y Ciencias, Universidad Adolfo Ib´a˜nez, Av. Diagonal las Torres 2640, Pe˜nalol´en, Santiago, Chile Millennium Institute for Astrophysics, Av. Vicu˜na Mackenna 4860, 782-0436 Macul, Santiago, Chile Center of Astro-Engineering UC, Pontificia Universidad Cat´olica de Chile, Av. Vicu˜na Mackenna 4860, 7820436 Macul, Santiago, Chile School of Physics and Astronomy, Queen Mary University of London, 327 Mile End Road, London E1 4NS, UK NASA Goddard Space Flight Center, 8800 Greenbelt Rd, Greenbelt, MD 20771, USA Department of Physics, University of Warwick, Gibbet Hill Road, Coventry CV4 7AL, UK Centre for Exoplanets and Habitability, University of Warwick, Gibbet Hill Road, Coventry CV4 7AL, UK Department of Astrophysical Sciences, Princeton University, 4 Ivy Lane, Princeton, NJ 08544 USA NASA Exoplanet Science Institute/Caltech Pasadena, CA, USA Department of Physics and Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA02139, USA Royal Observatory of Belgium, Ringlaan 3, B-1180 Brussel, Belgium Center for Astrophysics | Harvard & Smithsonian, 60 Garden Street, Cambridge, MA 02138, USA Department of Earth and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA NASA Ames Research Center, Moffett Field, CA, 94035 Department of Astronomy, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Komaba Institute for Science, The University of Tokyo, 3-8-1 Komaba, Meguro, Tokyo 153-8902, Japan JST, PRESTO, 3-8-1 Komaba, Meguro, Tokyo 153-8902, Japan Astrobiology Center, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan Instituto de Astrof´ısica de Canarias (IAC), 38205 La Laguna, Tenerife, Spain Department of Physics, Kyoto Sangyo University, Motoyama, Kamigamo, Kita-ku, Kyoto, 603-8555 Japan George Mason University, 4400 University Drive, Fairfax, VA, 22030 USA Campo Catino Astronomical Observatory, Regione Lazio, Guarcino (FR), 03010 Italy Canadian Institute for Theoretical Astrophysics, University of Toronto, 60 St George Street, ON M5S 3H8, Canada Centre for Planetary Sciences, Department of Physical & Environmental Sciences, University of Toronto at Scarborough, Toronto,Ontario M1C 1A4, Canada Department of Physics and Astronomy, University of Leicester, University Road, Leicester, LE1 7RH, UK Observatoire de Gen`eve, Universit´e de Gen`eve, 51 Ch. des Maillettes, 1290 Sauverny, Switzerland Institute of Planetary Research, German Aerospace Center, Rutherfordstrasse 2, 12489 Berlin, Germany Center for Astronomy and Astrophysics, TU Berlin, Hardenbergstr. 36, D-10623 Berlin, Germany Astrophysics Group, Cavendish Laboratory, J.J. Thomson Avenue, Cambridge CB3 0HE, UK a r X i v : . [ a s t r o - ph . E P ] S e p Astrophysics Research Centre, School of Mathematics and Physics, Queen’s University Belfast, BT7 1NN Belfast, UK Instituto de Astronom´ıa, Universidad Cat´olica del Norte, Angamos 0610, 1270709, Antofagasta, Chile Institute of Geological Sciences, FU Berlin, Malteserstr. 74-100, D-12249 Berlin, Germany Departamento de Astrof´ısica, Universidad de La Laguna (ULL), 38206, La Laguna, Tenerife, Spain Dunlap Institute for Astronomy and Astrophysics, University of Toronto, 50 St. George Street, Toronto, Ontario M5S 3H4, Canada Department of Physics and Astronomy, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3255, USA
ABSTRACTAbout one out of 200 Sun-like stars has a planet with an orbital period shorter than one day: an ultra-short-period planet (Sanchis-Ojeda et al. 2014; Winn et al. 2018). All of the previously known ultra-short-period planets are either hot Jupiters, with sizes above 10 Earth radii ( R ⊕ ), or apparently rockyplanets smaller than 2 R ⊕ . Such lack of planets of intermediate size (the “hot Neptune desert”) hasbeen interpreted as the inability of low-mass planets to retain any hydrogen/helium (H/He) envelopein the face of strong stellar irradiation. Here, we report the discovery of an ultra-short-period planetwith a radius of 4.6 R ⊕ and a mass of 29 M ⊕ , firmly in the hot Neptune desert. Data from theTransiting Exoplanet Survey Satellite (Ricker et al. 2015) revealed transits of the bright Sun-like starLTT 9779 every 0.79 days. The planet’s mean density is similar to that of Neptune, and according tothermal evolution models, it has a H/He-rich envelope constituting 9.0 +2 . − . % of the total mass. Withan equilibrium temperature around 2000 K, it is unclear how this “ultra-hot Neptune” managed toretain such an envelope. Follow-up observations of the planet’s atmosphere to better understand itsorigin and physical nature will be facilitated by the star’s brightness ( V mag = 9 . MAIN MANUSCRIPTUsing high precision photometry from Sector 2 of the Transiting Exoplanet Survey Satellite (
TESS ) mission at acadence of two minutes, a candidate transiting planet was flagged for the star LTT 9779 (Jenkins et al. 2016). Thecandidate was released as a
TESS
Alert in October 2018, and assigned the TESS Object of Interest (TOI) tagnameTOI-193 (TIC183985250). The
TESS lightcurve was scrutinised prior to its public release. No transit depth variationswere apparent, no motion of the stellar image was detected during transits, and no secondary eclipses could be found.Data from the Gaia spacecraft (Gaia Collaboration et al. 2016, 2018) revealed only one background star within the
TESS photometric aperture, but it is 5 mag fainter than LTT 9779 and hence cannot be the source of the transit-likesignals, and no significant excess scatter was witnessed in the Gaia measurements. The lack of all these abnormalitiessupported the initial interpretation that the transit signals are due to a planet with an orbital period of 19 hours anda radius of 3.96 R ⊕ .We also observed four complete transits with ground-based facilities: three with the Las Cumbres Observatory (LCO)and one with the Next Generation Transit Survey (NGTS; Wheatley et al. 2018) telescopes. The LCO and NGTS datahave a similar precision to the TESS light curve and much better angular resolution. The observed transit depths werein agreement with the depth observed with
TESS . High-angular resolution imaging of LTT 9779 was performed withadaptive optics in the near-infrared using NIRC2 at the Keck Observatory, and with speckle imaging in the opticalusing HRCam on SOAR at the Cerro-Tololo Inter-American Observatory. No companions were detected within aradius of 3 (cid:48)(cid:48) down to a contrast level of 7.5 magnitudes, and no bright close binary was seen with a resolution of 0 . (cid:48)(cid:48) (see the SI). These observations sharply reduce the possibility that an unresolved background star is the source of thetransits. We also tested the probability of having background or foreground stars within a region of 0 . (cid:48)(cid:48) separation(AO limit) from the star, using a Besan¸con (Robin et al. 2003) model of the galaxy. The model indicates we canexpect over 2200 stars in a 1 square degree field around LTT 9779 providing a probability of only 0.0005% of havinga star down to a magnitude limit of 21 in V contaminating the lightcurves. If we consider only objects bright enoughto cause contamination of the transit depth that would significantly alter the planet properties, this probability dropseven more (see the Supplementary Information (SI) for more details). Furthermore, although there exists a 13.5%probability that LTT 9779 could be part of a binary system that passes within this separation limit, spectral analysisrules out all allowable masses whose contaminant light that would be required to push LTT 9779 b outside of theNeptune desert. ∗ Corresponding author E-mail: [email protected] † Winton Fellow
Final confirmation of the planet’s existence came from high-cadence radial-velocity observations with the HighAccuracy Radial-velocity Planet Searcher (HARPS; Pepe et al. 2000). A sinusoidal radial-velocity (RV) signal wasdetected with the
EMPEROR code (Pe˜na Rojas & Jenkins 2020) independently of the transit data, but with a matchingorbital period and phase. No other significant signals were detected, nor were any longer-term trends, ruling outadditional massive planets with orbital periods of a few years or less. Likewise, no transit-timing variations weredetected (see the SI).To determine the stellar properties, we combined the Gaia data with spectral information from HARPS, along withother spectra from the Tillinghast Reflector Echelle Spectrograph (TRES; F˝ur´esz et al. 2008) and the Network ofRobotic Echelle Spectrographs (NRES; Siverd et al. 2018) and compare the star’s observable properties to the outputsfrom theoretical stellar-evolutionary models (MIST and Y2). We also used our new
ARIADNE code to precisely calculatethe effective temperature and stellar radius (see SI for more information on these methods). The star was found to havea mass, radius, and age of 1.02 +0 . − . M (cid:12) , 0.949 ± R (cid:12) , and 2.0 +1 . − . Gyrs, respectively. The effective temperatureand surface gravity are consistent with a main-sequence star slightly cooler than the Sun. The spectra also revealedthe star to be approximately twice as metal-rich as the Sun ([Fe / H] = +0.25 ± juliet code (Espinoza et al. 2019) to perform a joint analysis of the transit and radial-velocitydata (Figure 1). The period, mass, and radius of the planet were found to be 0.792054 ± +0 . − . M ⊕ ,and 4.72 ± R ⊕ , respectively. The orbit is circular to within the limits allowed by the radial-velocity data (theposterior odds ratio is 49:1 in favor of a circular model over an eccentric model).LTT 9779 b sits in the hot Neptune desert (Mazeh et al. 2016) (Figure 2), providing an opportunity to studythe link between short-period gas giants and lower mass super-Earths. The planet’s mean density is similar to thatof Neptune, and the planet’s mass and radius are incompatible with either a pure rock or pure water composition(Figure 3), implying that it possesses a substantial H/He gaseous atmosphere. Using 1-D thermal evolution modelsfrom Lopez & Fortney (2014), assuming a silicate and iron core and a solar composition gaseous envelope, we find aplanet core mass of 27.9 +1 . − . M ⊕ , and an atmospheric mass fraction of 9.0 +2 . − . %. We also tested other planet structures,and even in the limiting case of a non-physical pure water-world, there still exists a significant H/He-rich envelope, atthe level of 2.2 +1 . − . %. When combined with the high equilibrium temperature for the planet of 1978 ±
19 K, this makesLTT 9779 b an excellent target for future transmission spectroscopy, secondary eclipse studies, and phase variationanalyses. All of the planetary model parameters are in Table 2.LTT 9779 b is the most highly irradiated Neptune-sized planet yet found. It is firmly in the region of parameterspace known as the ”evaporation desert” where observations have shown a clear absence of similarly sized planets(Sanchis-Ojeda et al. 2014; Lundkvist et al. 2016), and models of photo-evaporative atmospheric escape predict thatsuch low density gaseous atmospheres should be evaporated on short timescales (Lopez 2017; Owen & Wu 2017).As LTT 9779 b is a mature planet found in this desert, it is a particularly high priority target for transmissionspectroscopy at wavelengths that probe low density material escaping from planetary upper atmospheres such asLyman Alpha (Ehrenreich et al. 2015), FUV metal-lines (Vidal-Madjar et al. 2004), Ca and Fe lines (Casasayas-Barriset al. 2019), and the 1.083 µ m Helium line (Nortmann et al. 2018).An interesting comparison can be made between LTT 9779 b and NGTS-4 b (West et al. 2019), the most similar of allthe other known planets. NGTS-4 b is not as hot ( (cid:104) T eq (cid:105) a = 1650 ±
400 K) or short-period ( P = 1 . ± . , and orbits a metal-poor star ([ M/H ] = − . ± .
10 dex).These characteristics may be clues that the two planets formed differently: NGTS-4 b may have formed as a relativelysmall and dense world, whereas LTT 9779 b started life as a much larger and less dense planet (see Figure 4). Indeed,photoevaporation models posit that the bulk population of ultra-short period planets form by growing to around3 M ⊕ , through the accretion of various amounts of light elements from the proto-planetary disk. The intense radiationfrom the young star then evaporates these close-in planets over an interval on the order of 10 yrs, leaving behindsmall rocky planets with radii less than 1.5 R ⊕ (Owen & Wu 2017). The more massive population of planets canhold onto the bulk of their envelopes until the star becomes quiescent, leaving behind planets with radii 2 − R ⊕ .However, these planets are generally found to have orbital periods beyond one day, similar to NGTS-4 b, reaching outto 100 days or so. Ultra-short period planets with these radii are rare, and it may be that since LTT 9779 b likely hasa large mass, it can hold onto a high fraction of its atmosphere. It could also have migrated to its current positionover a longer dynamical timescale, 10 years, not leaving enough time to blow-off a large fraction of its atmosphere byphotoevaporation.Assuming energy-limited atmospheric escape, and adopting the current mass, radius and orbital separation ofLTT 9779 b, we estimate mass loss rates of 2 − × g s − during the saturated phase of X-ray emission (forefficiencies of 525%; Owen & Jackson 2012; Ionov et al. 2018). Assuming the X-ray evolution given by Jackson et al.(2012) and the corresponding extreme-ultraviolet emission by Chadney et al. (2015) and King et al. (2018) we estimatea total mass loss of 29 M ⊕ . Considering instead the hydrodynamic calculations by Kubyshkina et al. (2018), this massloss increases to be greater than the total mass of the planet, and employing the detailed atmospheric escape evolutionmodel of Lopez (2017) suggests that the planet could have had an atmospheric mass fraction of up to ∼
60% of thetotal planet mass, or around half that of Saturn ( ∼ M ⊕ ). This means that LTT 9779 b could not have formed insitu with properties close to those we measure here, ruling out such a model. Conversely, adopting an initial planetmass and radius equal to that of Jupiter, we estimate a mass-loss of 5.5 × g over the current age of the system,which would only be ∼
3% of the total initial planet mass. Therefore, we can be sure that if the planet began as aJupiter-mass gas giant, photoevaporation cannot be the sole mechanism that removed most of its atmosphere.One possible mechanism for atmospheric loss is Roche Lobe Overflow (RLO; Valsecchi et al. 2015). Planets withmasses of ∼ M J orbiting solar-mass stars can fill their Roche Lobes for orbital periods approaching 12 hours. Forprogenitor hot Jupiters with large cores ( ∼ M ⊕ ), the initial migration inwards to the RLO orbit is driven by tidalinteraction with the host star. The migration can then reverse as mass is stripped from the planet at a rate of 10 - − g s − and continues on for a Gyr or so, assuming the escaping material settles in an accretion disk aroundthe star and transfers its angular momentum back to the planet. The planet can migrate outwards, reaching anorbital period of ∼ − Table 1 . Stellar properties of LTT 9779Alternative Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . LTT 9779TIC 183985250 TESSHIP 117883 HIPPARCOS2MASS J23544020-3737408 2MASSTYC 8015-1162-1 TYCHOCatalogue DataRA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (J2000) 23h54m40.60s TESSDEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (J2000) -37d37m42.18s TESSpm RA (mas yr − ) 247.615 ± DEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (mas yr − ) -69.801 ± π . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (mas) 12.403 ± ± ± ± ± ± ± s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (mag) 8.02 ± ± ± ± T eff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (K) 5445 ± SPECIES log g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) 4.43 ± SPECIES [Fe / H] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) +0.25 ± SPECIES v sin i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (km s − ) 1.06 ± SPECIES v mac . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (km s − ) 1.98 ± SPECIES T eff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (K) 5496 ± ZASPE log g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) 4.51 ± ZASPE [Fe / H] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) +0.24 ± ZASPE v sin i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (km s − ) 1.7 ± ZASPE T eff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (K) 5499 ± SPC log g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) 4.47 ± SPC [m / H] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) +0.31 ± SPC v sin i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (km s − ) 2.2 ± SPC T eff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (K) 5443 +14 − ARIADNE log g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) 4 . +0 . − . ARIADNE [Fe / H] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (dex) +0.27 ± ARIADNE M (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ( M (cid:12) ) 1 . +0 . − . SPECIES + MISTM (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ( M (cid:12) ) 1 . +0 . − . YY + GAIAM (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ( M (cid:12) ) 0 . +0 . − . ARIADNE R (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ( R (cid:12) ) 0.95 ± SPECIES + MISTR (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ( R (cid:12) ) 0.92 ± (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ( R (cid:12) ) 0.949 ± ARIADNE L (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (L (cid:12) ) 0 . ± .
04 YY + GAIAL (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (L (cid:12) ) 0 . ± . ARIADNE M V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (mag) 5.30 ± Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (Gyr) 2 . +2 . − . SPECIES + MISTAge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (Gyr) 1 . +1 . − . YY + GAIA ρ (cid:63) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (g cm − ) 1 . +0 . − . YY + GAIASpectral Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G7V This work < S
HARPS > . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.148 ± < logR (cid:48) HK,HARPS > . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . -5.10 ± P rot,v sin i (days) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 This work Table 2 . Transit, orbital, and physical parameters of LTT 9779 b. Forthe prior descriptions, which are the expected probability distributionsfor each of the fit parameters, N ( µ, σ ) represents a normal distributionwith mean µ and standard deviation σ , whereas U ( a, b ) and J ( a, b ) rep-resent a uniform prior and Jeffrey’s prior, both defined between points a and b , respectively (see Methods: Global Modelling for more informa-tion).Parameter Prior Value Light-curve parameters P (days) N (0 . , .
1) 0 . ± . T (days) N (2458354 . , .
1) 2458354 . ± . U (0 ,
1) 0 . +0 . − . r U (0 ,
1) 0 . +0 . − . ρ (cid:63) (kg/m ) N (1810 , +125 − q , TESS U (0 ,
1) 0 . +0 . − . q , TESS U (0 ,
1) 0 . +0 . − . q ,NGTS U (0 ,
1) 0 . +0 . − . q ,NGTS U (0 ,
1) 0 . +0 . − . RV parameters
K (m s − ) U ( − , . +0 . − . e ω (deg) 90 90 γ Coralie (m s − ) N (0 , − . +2 . − . γ HARPS (m s − ) N (0 , − . +0 . − . σ Coralie (m s − ) J (10 − , . +2 . − . σ HARPS (m s − ) J (10 − , . +0 . − . Derived parameters R p / R (cid:63) – 0 . +0 . − . a/ R (cid:63) – 3 . +0 . − . i – 76 . ± . p ( M ⊕ ) – 29.32 +0 . − . M ⊕ R p ( R ⊕ ) – 4.72 ± R ⊕ (cid:104) T eq (cid:105) a (K) – 1978 ± a (AU) – 0 . +0 . − . ρ p (g cm − ) – 1 . ± . a Equilibrium temperature using equation 4 ofM´endez & Rivera-Valent´ın (2017) with A = 0 . β = 0 .
5, and (cid:15) = 1.
Correspondence and requests for materials should be addressed to James S. Jenkins ([email protected]).1.1.
Acknowledgements
Funding for the TESS mission is provided by NASA’s Science Mission directorate. We acknowledge the use ofpublic TESS Alert data from pipelines at the TESS Science Office and at the TESS Science Processing Opera-tions Center. This research has made use of the Exoplanet Follow-up Observation Program website, which is op-erated by the California Institute of Technology, under contract with the National Aeronautics and Space Admin-istration under the Exoplanet Exploration Program. Resources supporting this work were provided by the NASAHigh-End Computing (HEC) Program through the NASA Advanced Supercomputing (NAS) Division at Ames Re-search Center for the production of the SPOC data products. JSJ and NT acknowledge support by FONDECYTgrants 1161218, 1201371, and partial support from CONICYT project Basal AFB-170002. MRD is supported byCONICYT-PFCHA/Doctorado Nacional-21140646/Chile and Proyecto Basal AFB-170002. JIV acknowledges sup-port of CONICYT-PFCHA/Doctorado Nacional-21191829. This work was made possible thanks to ESO Projects0102.C-0525 (PI: D´ıaz) and 0102.C-0451 (PI: Brahm). RB acknowledges support from FONDECYT Post-doctoralFellowship Project 3180246. This work is partly supported by JSPS KAKENHI Grant Numbers JP18H01265 andJP18H05439, and JST PRESTO Grant Number JPMJPR1775. The IRSF project is a collaboration between NagoyaUniversity and the South African Astronomical Observatory (SAAO) supported by the Grants-in-Aid for Scientific Re-search on Priority Areas (A) (Nos. 10147207 and 10147214) and Optical & Near-Infrared Astronomy Inter-UniversityCooperation Program, from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japanand the National Research Foundation (NRF) of South Africa. We thank Akihiko Fukui, Nobuhiko Kusakabe, Ku-miko Morihana, Tetsuya Nagata, Takahiro Nagayama, and the staff of SAAO for their kind support for IRSF SIRIUSobservations and analyses. CP acknowledges support from the Gruber Foundation Fellowship and Jeffrey L. BishopFellowship. This research includes data collected under the NGTS project at the ESO La Silla Paranal Observatory.NGTS is funded by a consortium of institutes consisting of the University of Warwick, the University of Leicester,Queen’s University Belfast, the University of Geneva, the Deutsches Zentrum f¨ur Luft- und Raumfahrt e.V. (DLR;under the ‘Großinvestition GI-NGTS), the University of Cambridge, together with the UK Science and TechnologyFacilities Council (STFC; project reference ST/M001962/1 and ST/S002642/1). PJW, DB, BTG, SG, TL, DP andRGW are supported by STFC consolidated grant ST/P000495/1. DJA gratefully acknowledges support from theSTFC via an Ernest Rutherford Fellowship (ST/R00384X/1). EG gratefully acknowledges support from the Davidand Claudia Harding Foundation in the form of a Winton Exoplanet Fellowship. MJH acknowledges funding from theNorthern Ireland Department for the Economy. MT is supported by JSPS KAKENHI (18H05442, 15H02063). AJ, RB,and PT acknowledge support from FONDECYT project 1171208, and by the Ministry for the Economy, Development,and Tourism’s Programa Iniciativa Cient´ıfica Milenio through grant IC 120009, awarded to the Millennium Institute ofAstrophysics (MAS). PE, AC, and HR acknowledge the support of the DFG priority program SPP 1992 Exploring theDiversity of Extrasolar Planets (RA 714/13-1). We acknowledge the effort of Andrei Tokovinin in helping to performthe observations and reduction of the SOAR data.1.2.
Author Contributions
JSJ led the
TESS precision radial-velocity follow-up program, selection of the targets, analysis, project coordination,and wrote the bulk of the paper. MD, NT, and RB performed the HARPS radial-velocity observations, PT observedthe star with Coralie, and MD analysed the activity data from these sources. NE performed the global modeling, withPCZ performing the TTV analysis, and RB, MGS, and AB performing the stellar characterisation using the spectraand evolutionary models. PAPR worked on the
EMPEROR code and assisted in fitting the HARPS radial-velocities. EDLcreated a structure model for the planet, and in addition to GWK and PJW, performed photoevaporation modeling.JNW performed analysis of the system parameters. DRC led the Keck NIRC2 observations and analysis. GR, RV,DWL, SS, and JMJ have been leading the
TESS project, observations, organisation of the mission, processing ofthe data, organisation of the working groups, selection of the targets, and dissemination of the data products. CEH,SM, and TK worked on the SPOC data pipeline. CJB was a member of the TOI discovery team. SNQ contributedto TOI vetting, TFOP organization, and TRES spectral analysis. JL and CP helped with the interpretation of thesystem formation and evolution. KAC contributed to TOI vetting, TFOP organization, and TFOP SG1 ground-basedtime-series photometry analysis. GI, FM, AE, KIC, MM, NN, TN, and JPL contributed TFOP-SG1 observations.JSA, DJA, DB, FB, CB, EMB, MRB, JC, SLC, AC, BFC, PE, AE, EF, BTG, SG, EG, MNG, MRG, MJH, JAGJ,TL, JM, MM, LDN, DP, DQ, HR, LR, AMSS, RHT, RTW, OT, SU, JIV, SRW, CAW, RGW, PJW, and GWK arepart of the NGTS consortium who provided follow-up observations to confirm the planet. EP and JJL helped withthe interpretation of the result. CB performed the observations at SOAR and reduced the data, CZ performed thedata analysis, and NL and AWM assisted in the survey proposal, analysis, and telescope time acquisition. All authorscontributed to the paper. 1.3.
Author Information
Reprints and permissions information is available at . Correspondence and requests formaterials should be addressed to James S. Jenkins, [email protected]
Competing Interests
The authors declare that they do not have any competing financial interests.
Figure 1. Transit lightcurves and phase folded RVs for LTT 9779
The left top panel shows the discovery
TESS lightcurve averaging over 33 transits, with the center and right panels showing the single transit follow-up photometry fromNGTS and LCOGT for LTT 9779, with a 0.8 day period, and including the associated transit model using the parameters shownin Table 2. The full data set is shown by the small points and the binned data is superimposed on these as the larger and darkerpoints with associated uncertainties. The lower panel shows the 31 HARPS radial-velocities in blue and 18 Coralie measurementsin orange (see Table 3), also folded to the period of the planet, and with their respective uncertainties. The mean uncertaintyof the Coralie velocities is a factor of 10 larger than those from HARPS. The best fit model is again overplotted on the data.The
TESS
NGTS and LCOGT photometry, along with the HARPS and Coralie velocities were fit simultaneously to ensure thebest constraints possible on the planet parameters, along with a more accurate description of the overall uncertainties.2.
EXTENDED DATA0 m a ss [ M ⊕ ] r ad i u s [ R ⊕ ] Figure 2. LTT 9779 b in the period-mass and period-radius planes.
The top plot shows all currently confirmed planetswith a fractional mass uncertainty below 30%, separated in colour by their detection method. The lower plot shows all currentlyconfirmed transit planets with a fractional radius uncertainty below 5%. LTT 9779 b is clearly isolated in the Neptune desertin period-mass-radius space, meaning it is heavily decoupled from the current populations of known exoplanets. - Earth - Venus Uranus Neptune LTT 9779bNGTS-4bRocky USPPlanets - K-78b, K2-229b, K-10b, Corot-7b, K2-141b - HD 3167b, K2-131b, WASP-47e, 55 Cnc e, K2-106b % F e % F e % F e R o c k % H O % H O
10 0 % H O . g c m . g c m . g c m Figure 3. LTT 9779 b in the mass-radius plane.
The plot includes all non-gas giant ultra-short period planets withwell constrained Doppler masses. LTT 9779 b is marked by the red square. Structure models from Zeng et al. (2016) areplotted as solid curves and labelled depending on the bulk composition of the planet. The models range from a 100% iron coreplanet, through to a 100% water world. The ultra-short period planets all agree with rocky-iron compositions, explained byphotoevaporation of their primordial atmospheres. LTT 9779 b is significantly larger, indicating that it has a residual hydrogenand helium atmosphere around the core. Dashed iso-density curves are shown in green for reference, highlighting the similardensities between Neptune and LTT 9779 b. For reference, Venus, Earth, Uranus, and Neptune are represented in the plot. Photoevaporation Roche Lobe OverflowDisk Accretion Large Core Mass
LTT 9779bNGTS-4bNeptune 55 Cancri eCorot-7bWASP-18b
Figure 4. Distribution of planetary densities as a function of host star metallicity for currently known transitingplanets with orbital periods less than 1.3 days.
The sample is split into those with masses less than 0.1 M J (green circles;ultra-short period planet proxies) and those with masses above (red circles; ultra-hot Jupiter proxies). The ultra-hot NeptuneLTT 9779 b and longer period NGTS-4 b are clearly labelled in the figure. The blue curve is a power law described by3 . × [( N Fe /N H ) / ( N Fe /N H ) (cid:12) ] − . + 0 .
8, which bounds the regions governed by the physical processes that determine the planetbulk properties. Those physical processes and the direction in which they move planets are shown in the plot. R e l a ti v e f l ux Time (BJD - 2457000) R e s i du a l s ( pp m ) Figure 5. Normalised
TESS pre-search data conditioning timeseries photometry for LTT 9779 with the optimalmodel (black curve) overplotted on the data (top). The model residuals are shown in the lower panel.3.
METHODS
TESS
Photometry Treatment
TESS observed the star LTT 9779 (HIP 117883, TIC 183985250, TOI 193) using Camera 2 (CCD 4), between UT2018 Aug 23 and Sep 20 (JD 2458354.11439 − TESS alert on the 4th of October2018, and assigned the code
TESS
Object of Interest (TOI) 193. As part of the alert process candidate vetting, thelight curve modeling did not show any hints of abnormality, such that no transit depth variations were apparent, noPSF centroiding offsets were found, and no secondary eclipses report, giving rise to a bootstrap false alarm probability(FAP) of 2 × − .In Figure 5 we show the TESS pre-search data conditioning light curve for LTT 9779, after removal of points thatwere flagged as being affected by excess noise. Given the quiescent nature of the star, the photometric light curveis fairly flat across the full time series, with the small transits (1584 ±
43 ppm) readily apparent to the eye. Thissimplified the modeling effort, giving rise to the small residual scatter shown in the figure.
Follow-up NGTS Photometry <
2. A total of 6502 images were obtained, each with an exposuretime of 10 s using the custom NGTS filter (520 −
890 nm). The observations were taken with the telescope slightlydefocussed to avoid saturation. The telescope guiding was performed using the DONUTS auto-guiding algorithm(McCormac et al. 2013), which resulted in an RMS of the target location on the CCD of only 0.040 pix, or 0.2 (cid:48)(cid:48) ,. Dueto this high precision of the auto-guiding, the use of flat fields during the reduction of the images was not required.Comparison stars were chosen manually and aperture photometry was performed on the images using a customaperture photometry pipeline. The wide field-of-view provided by NGTS enabled the selection of a good number ofsuitable comparison stars, despite LTT 9779 being a relatively bright star. When combined, the resulting photometryshowed the transit signal of TOI-193 with a depth and transit centre time consistent with the
TESS photometry. Thecombined NGTS light curve has a precision of 170 ppm over a half hour timescale, which is a comparable to the
TESS precision of 160 ppm over this timescale (for a single transit).
Dilution Probability
Given the reality of the transit as ’on-source’, the issue of dilution of the light curve by a foreground or backgroundstar is considered in a probabilistic sense. In this case, we aim to test the probability of having a blended star so closeto the star angularly on the sky, that the AO observations would not have detected it. The AO sensitivity deterioratesquickly below 0.5 (cid:48)(cid:48) or so, with low sensitivity to objects with angular separations of 0.1 (cid:48)(cid:48) or less on sky.3.1.
Background or Foreground Contaminant
With this in mind, we used the Besan¸con galactic model (Robin et al. 2003) to generate a representative star fieldaround the position of LTT 9779, with the aim of testing the likelihood of having a diluted star that significantly affectsthe transit parameters. The model has been used in a similar manner previously. For instance, in Fressin et al. (2013)they applied the model to test the probability that each of the Kepler transit planet candidates in their study was theresult of a blended eclipsing binary. We selected all stars within a one square degree box surrounding our target, downto the magnitude limit of V = 21 that the model provides. This gave rise to over 2200 stars to work with, for whichwe randomly assign positions in RA and Dec using a uniform random number generator, constrained to be within theselected box boundaries. We then ran the simulation 10 million times to generate a representative sample, recordingall the events where a star passed within a separation of 0.1 (cid:48)(cid:48) from LTT 9779, and finally normalising by the numberof samples probed. The test returned only 48 events, providing a probability to have such a close separation betweentwo stars in this field of only 4.8 × − (0.0005%).Although the probability we found is very small, it is actually an upper limit. Blending by stars as faint as 21stmagnitude for instance, does not affect the transit depth enough to push the radius of the planet above the Neptunedesert. The faintest population of stars in our test, which also represents the most abundant population, biases theprobability to larger values. For instance, if we take the mean of the final bin (20.5 magnitudes), we have a magnitudedifference from LTT 9779 of 10.2, which relates to an effect at the level of 83 ppm, only 5% of the observed transitdepth. If LTT 9779 b is truly a hot Jupiter, then in order to push it out of the Neptune desert, given its orbital periodand current radius, we require dilution from a star of ∼ × − (0.00001%),ruling out the possibility that dilution of the light curves is the reason the planet falls in such an isolated part of theparameter space. 3.2. Binary Star Contaminant
Although a non-bound stellar contaminant is unlikely to be diluting the transit of LTT 9779 b sufficiently to pushit out of the Neptune desert, a binary companion is likely to have a higher chance to be present and tightly separatedto LTT 9779. Therefore, we performed Monte Carlo simulations to test how likely having a stellar binary that isbright enough to dilute the transit lightcurve sufficiently would be. We simulated 10 binary systems, drawing thesystem parameters from the probability density functions (PDFs) calculated in Raghavan et al. (2010). Here the5orbital log-period PDF in days is a normal distribution with mean of 5.03 and standard deviation of 2.28. Most otherparameters like eccentricity, the orbital angles, and the mass ratio, were simulated using uniform distributions withintheir respective bounds. Only the system inclinations were drawn from a cosine PDF.When simulating the systems, we normalized each by the fraction of the orbital period that the secondary starwould spend within 0.1 (cid:48)(cid:48) of the primary. Therefore, systems that never approached within this angular separation wereassigned a fractional time ( T f ) of zero, those that always were found within this limit were assigned a value of unity,and the rest were assigned a value between 0 − P = ( (cid:80) n = N t n =1 P n T f,n ) /N t , where the probability P is the sum totalof fractional probabilities P n T f,n across all samples, normalised by the total number of samples N t . Finally, we thennormalised by the 46% fraction of such stars found to exist in binaries.With these simulations we arrived at a value of 13.5% for the probability that LTT 9779 has a binary companion thatcould be found within an angular separation of 0.1 (cid:48)(cid:48) at any one time. Although this is a relatively large probability,this is integrated across all binary mass fractions, and therefore does not take into account that only a small massrange is permitted by the spectral analysis. When we account for the cross correlation function analysis discussedbelow, the probability drops to essentially zero, since the larger secondary masses required to affect the transit depthsufficiently are all ruled out. Gaia Variability
Another way to probe for very closely separated stars on the sky is to study the measurements made by Gaia,in particular the excess noise parameter (cid:15) and the Tycho-Gaia astrometric solution (TGAS) discrepancy factor ∆ Q (Lindegren et al. 2012; Michalik et al. 2014; Lindegren et al. 2016). These can be used to look for excess variability inthe observations that are indicative of blended starlight from a foreground or background star, spatially close enoughthat they can not be resolved by the instrument.Both the (cid:15) and ∆ Q parameters are listed in Gaia DR1 as standard outputs, however Gaia DR2 only reports theexcess noise, which turns out to be unreliable for stars with G < ∼
13 (Lindegren et al. 2018). ∆ Q measures thedifference between the proper motion derived in TGAS and the proper motion derived in the Hipparcos Catalog (vanLeeuwen 2007). Also, Rey et al. (2017) utilised both ∆ Q and (cid:15) to show the lack of binarity for some stars. ∆ Q is expected to follow a χ distribution with two degrees of freedom for single stars. The Gaia DR1 (cid:15) and ∆ Q forLTT9779 are 0.394 (with a significance of 134.281) and 2.062, respectively. According to Lindegren et al. (2016), allsources obtain a significant excess source noise of ∼ > > Q threshold of 15.086 for a star to be well behaved (at a significance level of 1%), Lindegren et al.reduces this threshold to 10. This means that any star with ∆ Q <
10 is considered to be astrometrically well be-haved, again showing that this star is highly likely to be uninfluenced by contaminating light from a background object.
Follow-up Spectroscopy
NRES Spectroscopy
In order to aid in characterisation of the host star we used the LCO robotic network of telescopes (Brown et al. 2013)and the Robotic Echelle Spectrographs (NRES; Siverd et al. 2018). We obtained 3 spectra, each composed of 3 x 1200sec exposures, on UT 2018 Nov 5, 8, and 9. All three spectra were obtained with the LCO/NRES instrument mountedon a 1 m telescope at the LCO CTIO node. The data were reduced using the LCO pipeline resulting in spectra withSNR of 61 −
73. We have analyzed the spectra using SpecMatch while incorporating the Gaia DR2 parallax using themethod described by Fulton & Petigura (2018). The resulting host stars parameters contributed to those listed inTable 1. 3.4.
TRES Spectroscopy
We obtained two reconnaissance spectra on the nights of UT2018-11-04 and UT2018-11-05 using the TillinghastReflector Echelle Spectrograph (TRES; F˝ur´esz et al. 2008) located at the Fred Lawrence Whipple Observatory (FLWO)in Arizona, USA. TRES has a resolving power of ∼ − ∼
35 at 5200˚A. The spectra were then reduced and extracted as described6in Buchhave et al. (2010), whereby the standard processing for echelle spectra of bias subtraction, cosmic ray removal,order-tracing, flatfielding, optimal extraction (Horne 1986), blaze removal, scattered-light subtraction, and wavelengthcalibration was applied. The Spectral Parameter Classification tool (Buchhave et al. 2012) was used to measure thestellar quantities we show in Table 1. 3.5.
HARPS Spectroscopy
Upon examination of the light curve we decided to perform high cadence follow-up spectroscopic observations withthe High-Accuracy Radial velocity Planet Search spectrograph (HARPS; Pepe et al. 2000) installed at the ESO 3.6mtelescope in La Silla, in order to fully cover the phase space. We started observing LTT 9779 on Nov 6th 2018. From aninitial visual examination of the spectra and cross correlation function (CCF) of the online Data Reduction Software(DRS), it was consistent with no evidence of blending with other stellar sources nor as being a fast rotator or active,based on the width of the CCF and various activity indicators.We acquired 32 high-resolution (R ∼ ∼ value. This is likely due to increased stellar activity noise that affects thebluest orders the most, combined with relatively low signal-to-noise ratios, and therefore removing these orders allowshigher precision to be reached. It was with this data that we performed the EMPEROR (Pe˜na Rojas & Jenkins 2020)fitting, providing the independently confirmed and constrained evidence for LTT 9779 b (Figure 6). For instance, theDoppler orbital period was found to be 0.7920 ± . d , in excellent agreement with that provided by the TESStransit fitting, and allowing the period to be constrained in the joint fit to one part in 80’000 (0.001%). We also usedthis spectra to test if possible spectral line asymmetries and/or activity related features could be driving the signal.In particular, we searched for linear correlations between the spectral bisector inverse slope measurements and theradial-velocities (see Figure 7), along with performing period searches using Generalized Lomb Scargle periodograms(Zechmeister & K¨urster 2009) and Bayesian methods with the EMPEROR code. The Spearman correlation coefficientbetween the BIS and RVs is found to be 0.22 with a p-value of 0.22, meaning there exists no strong statistical evidenceto reject the null hypothesis that such a weak correlation has arisen by chance. From the periodogram analyses, no sta-tistically significant periodicities were detected with false alarm probabilities of less than 0.1%, our threshold for signaldetection. We also performed the same analyses on the full width at half maximum of the HARPS cross-correlationfunction, and chromospheric activity indicators like the S , H α , and HeI indices, again with no statistically significantresults encountered.Finally, we also reprocessed the HARPS spectra to generate CCFs with binary masks optimised for spectral typesbetween G2-M4, but across a wider ±
200 km s − range in velocity to check for weaker secondary CCFs that could bedue to additional, nearby companions. We took a typical HARPS LTT 9779 spectrum and injected mid-to-late M starspectra with decreasing SNRs, until we could not detect the M star CCFs no more, providing an upper limit on themass of any contaminating secondary. From analysis of the mean flux ratio between the M stars and LTT 9779, wefound that we should be able to detect stellar contaminants down to a mass of 0.19 M (cid:12) , using the mass-luminosityrelation of Benedict et al. (2016), however no companion CCFs were detected. Such a companion would have amagnitude difference of over 7.5, and since we previously calculated above that a maximum magnitude difference of5.5 would be required to push LTT9779b out of the Neptune Desert, the limits permitted by the CCF analysis showthat a diluted companion would not change the conclusions of our work. Median Absolute Deviation = median( | x i − median ( x ) | Coralie Spectroscopy
Additional phase coverage was performed using the Coralie spectrograph installed in the 1.2 m Swiss LeonhardEuler Telescope at the ESO La Silla Observatory in Chile. Coralie has a spectral resolution of ≈ ≈ − , which allowed the identification theKeplerian signal with an amplitude of 20 − . Table 3 . Radial-velocities of LTT 9779JD - 2450000 RV Uncertainty Instrument(m s − ) (m s − )8429.51804 -10.59 0.86 HARPS8430.54022 -16.91 0.74 HARPS8430.59553 -9.41 0.68 HARPS8430.67911 1.99 0.79 HARPS8430.76201 13.40 1.21 HARPS8431.51068 6.71 0.61 HARPS8431.64346 16.09 0.83 HARPS8431.69130 14.98 0.87 HARPS8431.73217 8.41 0.55 HARPS8432.50941 12.77 0.73 HARPS8432.65689 -7.23 0.94 HARPS8432.69804 -13.45 1.06 HARPS8432.72573 -18.32 4.02 HARPS8464.53817 -25.17 1.02 HARPS8464.64153 -16.81 1.11 HARPS8464.68616 -10.08 1.27 HARPS8465.53024 0.00 0.85 HARPS8465.59314 10.82 0.84 HARPS8465.64411 12.09 0.86 HARPS8465.68104 15.61 1.12 HARPS8466.52022 14.89 1.03 HARPS8466.58232 8.12 0.90 HARPS8466.63157 2.49 1.09 HARPS8466.66865 -2.85 1.10 HARPS8481.53213 14.93 0.94 HARPS8481.57805 12.72 0.84 HARPS8482.53643 -8.75 0.74 HARPS8482.57255 -11.89 0.82 HARPS8482.60140 -16.09 0.90 HARPS8483.52686 -24.82 0.80 HARPS8483.59338 -20.68 1.12 HARPS8483.61557 -18.95 0.93 HARPS8438.56440 -14.80 4.50 CORALIE Figure 6. Independently constrained system parameters from the
EMPEROR
MCMC runs of the 31 HARPS radial-velocities.
From top to bottom we show the posteriors of the velocity amplitude, the orbital period, and the eccentricity ofthe orbit. Overplotted on each histogram is a gaussian distribution with the same input parameters as those calculated fromthe posterior distributions. We also show the values obtained from the distributions. The histograms reveal that the signal iswell constrained with the current data in hand, and the period in particular is in excellent agreement with that from the
TESS lightcurve. 8438.62857 -7.40 4.60 CORALIE8438.72084 10.40 5.00 CORALIE8439.56828 35.30 5.60 CORALIE8439.64481 3.80 4.80 CORALIE8439.70910 -11.70 5.20 CORALIE8440.56824 4.90 4.70 CORALIE8440.64498 -13.20 4.70 CORALIE8440.70927 -27.70 5.00 CORALIE8441.57027 -16.30 4.20 CORALIE8441.66132 -17.50 4.60 CORALIE8441.74898 1.00 4.50 CORALIE8442.56932 -0.60 4.50 CORALIE8442.64202 11.60 4.90 CORALIE8442.70651 0.60 5.00 CORALIE8443.57400 20.10 5.00 CORALIE8443.64711 -0.70 4.70 CORALIE8443.71686 5.60 4.80 CORALIE
Follow-up High-angular Resolution Imaging
NIRC2 at Keck9 − − −
10 0 10 20 30
Radial Velocity (m s − ) − − B I S ( m s − ) CORALIETERRA
Figure 7. Spectral line bisector inverse slope measurements as a function of the radial-velocities.
The orangediamonds and blue circles relate to measurements made using HARPS and Coralie, respectively. The best fit linear trend isshown by the dashed line, and a key in the upper left indicates the origin of the data points.
As part of our standard process for validating transiting exoplanets, we observed LTT 9779 with infrared high-resolution adaptive optics (AO) imaging at Keck Observatory (Ciardi et al. 2015). The Keck Observatory observationswere made with the NIRC2 instrument on Keck-II behind the natural guide star AO system. The observations weremade on UT 2018 Nov 22 following the standard 3-point dither pattern that is used with NIRC2 to avoid the leftlower quadrant of the detector which is typically noisier than the other three quadrants. The dither pattern step sizewas 3 (cid:48)(cid:48) and was repeated twice, with each dither offset from the previous one by 0 . (cid:48)(cid:48) .The observations were made in the narrow-band Br − γ filter ( λ o = 2 . λ = 0 . µ m) with an integration timeof 2 seconds with one coadd per frame for a total of 18 seconds on target. The camera was in the narrow-angle modewith a full field of view of ∼ (cid:48)(cid:48) and a pixel scale of approximately 0 . (cid:48)(cid:48) per pixel. The Keck AO observationsshow no additional stellar companions were detected to within a resolution ∼ . (cid:48)(cid:48) FWHM (Figure 8 left).The sensitivities of the final combined AO image were determined by injecting simulated sources azimuthally aroundthe primary target every 45 ◦ at separations of integer multiples of the central source’s FWHM (Furlan et al. 2017).The brightness of each injected source was scaled until standard aperture photometry detected it with 5 σ significance.The resulting brightness of the injected sources relative to the target set the contrast limits at that injection location.The final 5 σ limit at each separation was determined from the average of all of the determined limits at that separationand the uncertainty on the limit was set by the rms dispersion of the azimuthal slices at a given radial distance. Thesensitivity curve is shown in the left panel of Figure 8, along with an inset image zoomed to primary target showingno other companion stars.HRCam at SOARIn addition to the Keck observations, we also searched for nearby sources to LTT 9779 with SOuthern AstrophysicalResearch (SOAR) speckle imaging on 21 December 2018 UT, using the high resolution camera (HRCam) imager.Observations were performed in the I -band, which is a similar visible bandpass to that of TESS . Observations consistedof 400 frames, consisting of a 200 ×
200 binned pixels region of interest, centered on the star. Each individual frame is6.3 (cid:48)(cid:48) on a side, with a pixel scale of 0.01575 (cid:48)(cid:48) and 2 × ∼
11 s, and using an AndoriXon-888 camera. More details of the observations and processing are available in Ziegler et al. (2020).The 5 σ contrast curve and speckle auto-correlation function image are shown in the right panel of Figure 8. Nonearby sources were detected within 3 (cid:48)(cid:48) of LTT 9779, down to a contrast limit of 6 − I -band. Wecan also rule out brighter background blends very close to the star, down to around 0.1 (cid:48)(cid:48) separation. Combining the0 Figure 8. Companion sensitivity for the Keck NIRC2 adaptive optics imaging and the SOAR Adaptive OpticsModule (SAM).
For NIRC2 (left), the black points represent the 5 σ limits and are separated in steps of 1 FWHM ( ∼ . (cid:48)(cid:48) );the purple represents the azimuthal dispersion (1 σ ) of the contrast determinations (see text). The inset image is of the primarytarget showing no additional companions within 3 (cid:48)(cid:48) of the target. For SAM (right) the black curve also represents the 5 σ limit, and the black data points mark the sampling. The inset also shows the speckle image of the star, constructed from theAuto-Correlation Function. results from Keck and SOAR, we can be rule out background blended eclipsing binaries contaminating the TESS large aperture used to build the LTT 9779 light curve.
Stellar Parameters
To calculate the stellar parameters for LTT 9779 we used four different methods, with three of them applied to thethree different sets of spectra we obtained from NRES, TRES, and HARPS, and a photometric method that usedour new tool
ARIADNE . For the NRES spectra, we used the combination of SpecMatch and Gaia DR2 to perform thespectral classification, following the procedures explained in Fulton & Petigura (2018). TRES spectral observationsused the Spectral Parameter Classification (
SPC ; Buchhave et al. 2012) tool to calculate the stellar parameters, whereaswe used the Spectroscopic Parameters and atmosphEric ChemIstriEs of Stars (
SPECIES ; Soto & Jenkins 2018) andthe Zonal Atmospheric Stellar Parameters Estimator (
ZASPE ; Brahm et al. 2017) algorithms to analyse the HARPSspectra. Details of these methods can be found in each of the listed publications, yet in brief,
SPC and
ZASPE calculatethe parameters by comparing the spectra to Kurucz synthetic model grids (Kurucz 1992), either by direct spectralfitting, or by cross correlation. In this way, regions of the spectra that are sensitive to changes in stellar parameterscan allow parameters to be estimated by searching for the best matching spectral model.On the other hand,
SPECIES uses an automatic approach to calculate equivalent widths for large numbers of atomicspectral lines of interest, Fe i for instance. The code then calculates the radiative transfer equation using MOOG (Sneden1973), applying ATLAS9 model atmospheres (Castelli & Kurucz 2004), and converges on the stellar parameters usingan iterative line rejection procedure. Convergence is reached once the constraints of having no statistical trend betweenabundances calculated from Fe i and Fe ii for example, reaches a pre-determined threshold value.Each of these three methods return consistent results for the majority of the bulk parameters, in particular thestellar effective temperature is in excellent agreement, with a mean value of 5480 ±
42 K, along with the surface gravity1(log g ) of the star, which is found to be 4.47 ± ± +0 . − . M (cid:12) .For the radius, we used the ARIADNE code (Vines & Jenkins 2020), which is a new python tool designed to automat-ically fit stellar spectral energy distributions in a Bayesian Model Averaging framework. We convolved Phoenix v2(Husser et al. 2013), BT-Settl, BT-Cond (Allard et al. 2012), BT-NextGen (Hauschildt et al. 1999; Allard et al. 2012),Castelli & Kurucz (2004), and Kurucz (1993) model grids with commonly available filter bandpasses: UBVRI; 2MASSJHK s ; SDSS ugriz; ALL-WISE W1 and W2; Gaia G, RP, and BP; Pan-STARRS griwyz; Stromgren uvby; GALEXNUV and FUV; Spitzer/IRAC 3.6 µ m and 4.5 µ m; TESS; Kepler; and NGTS creating six different model grids, whichwe then interpolated in Teff − log g − [Fe/H] space. ARIADNE also fits for the radius, distance, Av and excess noiseterms for each photometry point used, in order to account for possible underestimated uncertainties. We used the SPECIES results as priors for the T eff , log g , and [Fe/H], and the distance is constrained by the Gaia DR2 parallax,after correcting it by the offset found by Stassun & Torres (2018). The radius has a prior based on GAIAs radiusestimate and the Av has a flat prior limited to 0.029, as per the re-calibrated SFD galaxy dust map (Schlegel et al.1998; Schlafly & Finkbeiner 2011). We performed the fit using dynestys nested sampler (Speagle 2020), which returnsthe Bayesian evidence of each model, and then afterwards we averaged each model posterior samples weighted by theirrespective normalized evidence. This returned a final stellar radius of 0.949 ± R (cid:12) .As LTT 9779 b appears as an odd-ball when scrutinising its mass and radius, we want to be sure that the stellarradius is not biased in the sense that the star is really an evolved star, much larger than the stellar modelling predicts,and hence the planet is more likely a UHJ. Although we have arrived at the same values from three different analysesand instrumental data sets, we can add more confidence to the results by studying the stellar density throughout theMCMC modelling process, when assuming the planet’s orbit is circular. In this case, we place a log-uniform prioron the stellar density, constrained to be within 100 − − , and then study how it changes as a function ofR p / R (cid:63) .We find that the distribution is bimodal (Figure 9), with the most likely stellar density region given by the lower,more densely constrained part of the parameter space in the figure. The upper mode in the figure, pushing towardshigher stellar densities and lower values of R p / R (cid:63) , is arguing towards the star being an M dwarf, which is ruled out bythe high resolution spectroscopic data, and is inconsistent with our global-modelling effort (less probable part of theposterior space). This mode is also only consistent with a very narrow set of limb-darkening coefficients, all of whichare inconsistent at several sigma with theoretical models, whereas the lower, more probable mode, has a wide range ofpossible limb-darkening coefficients, which are all in agreement with theoretical models. Therefore, this test rules outa more evolved state for the star in either case, with the higher probability mode being in excellent agreement withthe results from the stellar modelling.Finally, for the confirmation of the transit and radial-velocity parameters it is prudent to analyse the activity of thestar, in order to assess the impact that any activity could have on the measurements. From the above analyses wefind the star to be a very slow rotator, with a HARPS v sin i limit of 1.06 ± − , lower than the projected solar v sin i value (1.6 ± − ) determined from HARPS spectral analysis (Pavlenko et al. 2012), indicating a slowlyrotating, and therefore inactive star. Given the calculated radius of the star, such a slow rotation gives rise to an upperlimit of the rotation period to be 45 d. If the planetary orbit is aligned with the stellar plane of rotation, such that wecan assume the inclination angle is the same, then this value is the absolute rotation period. Kepler Space Telescopedata analysis of old field stars of this spectral type, have rotation periods ranging from a few days for the youngeststars, with a peak around 20 d, and a sharp fall after this with a tail reaching up to almost 100 d (McQuillan et al.2014). A rotation period of 45 d would place LTT 9779 in the upper tail of the Kepler distribution, indicating the staris old, and agreeing with the combined age estimate of 2.0 +1 . − . Gyrs. This result would also suggest that the activityof the star should be weak. We calculate the activity using the Ca ii HK lines, following the analysis procedures andmethods presented in Jenkins et al. (2006, 2008, 2011, 2017). We find the star to be inactive, with a HARPS S -indexof 0.148 ± logR (cid:48) HK,HARP S of -5.10 ± ∼ Planet-to-star radius ratio ( R p / R * ) S t e ll a r den s i t y ( k g / m ) Figure 9. Stellar density as a function of R p / R (cid:63) when modelling the TESS light curve with a log-uniform prior on thestellar density and the planetary eccentricity constrained to be zero. properties can be found in Table 1.
Global Modelling
As stated in the main text, the global modeling of the data was performed using juliet (Espinoza et al. 2019).This code uses batman (Kreidberg 2015) to model the transit lightcurves and radvel (Fulton et al. 2018) to modelthe radial-velocities. We performed the posterior sampling using MultiNest (Feroz et al. 2009) via the PyMultiNestwrapper (Buchner et al. 2014).The fit was parameterized by the parameters r and r , both having uniform distributions between 0 and 1, whichare transformations of the planet-to-star radius ratio p and impact parameter b that allow an efficient exploration ofthe parameter space (Espinoza 2018). In addition, we fitted for the stellar density by assuming a prior given by thevalue obtained with our analysis of the stellar properties, assuming a normal prior for this parameter with a meanof 1810 kg/m and standard deviation of 130 kg/m . We parameterized the limb-darkening effect using a quadraticlaw defined by parameters u and u ; however, we use an uninformative parameterization scheme (Kipping 2013) inwhich we fit for q = ( u + u ) and q = u / (2 u + 2 u ) with q and q having uniform priors between 0 and 1. Forthe radial-velocity parameters, we used wide priors for both the systemic radial-velocity of each instrument and thepossible jitter terms, added in quadrature to the data.For the photometry, we considered unitary dilution factors for the TESS
NGTS and LCOGT photometry afterleaving them as free parameters and observing that it was not needed based on the posterior evidence of the fits.This is consistent with the a-priori knowledge that the only source detected by Gaia DR2 within the
TESS apertureis a couple of faint sources to the south-east of the target, the brighter of which has ∆ G = 5 . TESS passband, this would imply a dilution factor
D > . TESS photometry, no extra noise model nor jitter term was needed tobe added according to the bayesian evidence of fits incorporating those extra terms. For the NGTS observations, weconsidered the data of the target from the nine different telescopes as independent photometric datasets (i.e., havingindependent out-of-transit baseline fluxes in the joint fit), that share the same limb-darkening coefficients. We initiallyadded photometric jitter terms to all the NGTS observations, but found that fits without them for all instrumentswere preferred by looking at the bayesian evidences of both fits. For the LCOGT data, we used gaussian process intime to detrend a smooth trend observed in the data. A kernel which was a product of an exponential and a matern3/2 was used, and a jitter term was also fitted and added in quadrature to the reported uncertainties in the data— this was the model that showed the largest bayesian evidence. We note that fitting the lightcurves independentlyprovides statistically similar transit depths to the joint model, showing that all are in statistical agreement. Finally,an eccentric orbit is ruled out by our data with an odds ratio of 49:1 in favor of a circular orbit; the eccentric fit,performed by parameterizing the eccentricity and argument of periastron via S = √ e cos ω and S = √ e sin ω , givesan eccentricity given our data of e < .
058 with a 95% credibility.With all the photometry in hand, we could also compare individually each light curve transit model to test if theyare in statistical agreement, or any biases exist, such that the radius measurement is biased. We proceeded to again fiteach light curve independently with juliet , recording the transit model depths to test for statistical differences. Asexpected, we found the
TESS photometry produced the most precise value ( T d, TESS = 2299 +320 − ppm), with the LCOand NGTS fits arriving at values of T d,LCO = 1925 +620 − ppm and 1594 +980 − ppm, respectively. All three are in statisticalagreement. We also jointly modeled the LCO and NGTS lightcurves to provide a more constrained comparison withthe TESS photometry, and found a value of T d,LCO + NGT S = 1678 +540 − ppm, again in statistical agreement with the TESS value. Therefore, we can be confident that all three instruments provide a similar description for the planet’sphysical size.
Transit Timing Variations
The Transit Timing Variations (TTVs) of LTT 9779 b was measured using the
EXOFASTv2 (Eastman et al. 2013;Eastman 2017) code.
EXOFASTv2 uses the Differential Evolution Markov chain Monte Carlo method (DE-MC) toderive the values and their uncertainties of the stellar, orbital and physical parameters of the system. For the TTVanalysis of LTT 9779 b we fixed the stellar and orbital parameters to the values obtained from the global fit performedby
SPECIES and juliet , except for the transit time of each light curve and their baseline flux.In a Keplerian orbit, the transit time of an exoplanet follows a linear function of the transit epoch number (E): T c ( E ) = T c (0) + P E (1)Where P is the orbital period of the exoplanet and T c (0) is the optimal transit time in a arbitrary zero epoch andcorresponds to the time that is least covariant with the period and has the smallest uncertainty. Our best-fitted valuefrom EXOFASTv2 is: T c (0) = 2458354 . ± . EXOFASTv2 ’s fitting, resulting in 33 parameters to fit. The best fit results are shown in Figure 10,where the grey area corresponds to the 1 σ of the linear ephemeris shown in Equation (1).We found no evidence of a clear periodic variation in the transit time. The RMS variation from the linear ephemerisis σ = 181 . σ limit, if we remove them the RMS deviation is reduced to155.9 sec. On the other hand, the reduced chi-squared is χ red = 1 .
23, which is an indicator that the transit times fitaccordingly with the proposed linear ephemeris.In conclusion, the existence of transit timing variations in LTT 9779 b is not evident for the time-span of our transitdata. In addition, with the apparent lack of another short period signal in the RV data, this suggest that there is noother inner companion in the planetary system. Any other tertiary companion must be far from LTT 9779 b, suchthat the gravitation or tidal interactions are small, and the linear trend in the RVs might be pointing in that direction.
Metallicity Analysis
The correlation between the presence of giant planets and host star metallicity has been well established (Gonzalez1997; Fischer & Valenti 2005; Jenkins et al. 2017; Maldonado et al. 2018), along with the apparent lack of any correlationfor smaller planets (Jenkins et al. 2013; Buchhave et al. 2012, 2014). We studied the small sample of known USP4 O - C [ m i n ] Figure 10. Observed minus computed mid-transit times of LTT 9779 b.
The residuals (TTV) of the transit timesare shown considering the proposed linear ephemeris. The dashed line corresponds to zero variation and the grey area is thepropagation of 1 σ uncertainties, considering the optimal transit time from EXOFASTv2 and the period from juliet . The epoch0 is the first light curve obtained by
TESS and is also the corresponding epoch of the optimal transit time. The TTV valuesshown in this plot fit accordingly with the proposed linear ephemeris ( χ red = 1 . planets and Ultra Hot Jupiters (UHJs, the gas giant planets with orbit periods of less than 1 day), using values takenfrom the TEPCat database (Southworth 2011), whilst recalculating metallicities for those where we could find theirspectra, ( ∼ half the sample), using SPECIES (Soto & Jenkins). We found a similar general trend, whereby the USPplanets tend to orbit more metal-poor stars when compared with the UHJs, however the sample is small enough thatsingle outliers bias the statistics, therefore we extended slightly the orbital period selection out to 1.3 days, increasingthe sample by over 55%. With this updated sample, we find a Kolmogorov-Smirnov (KS) test probability of only 1%that the USP planets and UHJs are drawn from the same parent population.A couple of notable exceptions to the trend here are the planets 55 Cancri e and WASP-47 e, both small USP planetsthat orbit very metal-rich stars. However, there exists additional gas giant planets in these systems, meaning theystill follow the overall picture. If we exclude these two, the KS probability drops to 0.1% that the populations arestatistically similar. The diversity of USP planets is high, therefore many more detections are needed to statisticallyconstrain the populations in this respect. We also require more UHJs to build up a statistical sample, since the subsolarmetallicity of WASP-43 can also bias the tests. If we look at the density-metallicity parameter space (Figure 4), thereare indications of a general trend whereby the low-density planets are mostly UHJs orbiting metal-rich stars, and thehigher density USP planets orbit more metal-poor stars.3.7.
Data availability
Photometric data that support the findings of this study are publically available from the Mikulski Archive for SpaceTelescopes (MAST; http://archive.stsci.edu/) under the TESS Mission link. All radial-velocity data re available fromthe corresponding author upon reasonable request. Raw and processed spectra can be obtained from the EuropeanSouthern Observatorys data archive at http://archive.eso.org.3.8.
Code Availability
All codes necessary for the reproduction of this work are publically available through the GitHub repository, asfollows:5
EMPEROR: https://github.com/ReddTea/astroEMPEROR
Juliet: https://github.com/nespinoza/juliet
SPECIES: https://github.com/msotov/SPECIES
ARIADNE:
CERES: https://github.com/rabrahm/ceres
ZASPE https://github.com/rabrahm/zaspe REFERENCES
Allard F., Homeier D., Freytag B., 2012, PhilosophicalTransactions of the Royal Society of London Series A,370, 2765Anglada-Escud´e G., Butler R. P., 2012, ApJS, 200, 15Benedict G. F. et al., 2016, AJ, 152, 141Brahm R., Jord´an A., Espinoza N., 2017, PASP, 129,034002Brahm R., Jord´an A., Hartman J., Bakos G., 2017,MNRAS, 467, 971Brown T. M. et al., 2013, PASP, 125, 1031Buchhave L. A. et al., 2010, ApJ, 720, 1118Buchhave L. A. et al., 2014, Nature, 509, 593Buchhave L. A. et al., 2012, Nature, 486, 375Buchner J. et al., 2014, A&A, 564, A125Casasayas-Barris N. et al., 2019, A&A, 628, A9Castelli F., Kurucz R. L., 2004, xxx, 1Chadney J. M., Galand M., Unruh Y. C., Koskinen T. T.,Sanz-Forcada J., 2015, Icarus, 250, 357Ciardi D. R., Beichman C. A., Horch E. P., Howell S. B.,2015, ApJ, 805, 16Dotter A., 2016, ApJS, 222, 8Eastman J., , 2017, EXOFASTv2: Generalizedpublication-quality exoplanet modeling codeEastman J., Gaudi B. S., Agol E., 2013, Publications of theAstronomical Society of the Pacific, 125, 83Ehrenreich D. et al., 2015, Nature, 522, 459Espinoza N., 2018, Research Notes of the AmericanAstronomical Society, 2, 209Espinoza N., Kossakowski D., Brahm R., 2019, MNRAS,490, 2262Feroz F., Hobson M. P., Bridges M., 2009, MNRAS, 398,1601F˝ur´esz G., Szentgyorgyi A. H., Meibom S., 2008, in SantosN. C., Pasquini L., Correia A. C. M., Romaniello M., eds,Precision Spectroscopy in Astrophysics. pp 287–290Fischer D. A., Valenti J., 2005, ApJ, 622, 1102Fressin F. et al., 2013, ApJ, 766, 81Fulton B. J., Petigura E. A., 2018, AJ, 156, 264Fulton B. J., Petigura E. A., Blunt S., Sinukoff E., 2018,PASP, 130, 044504 Furlan E. et al., 2017, AJ, 153, 71Gaia Collaboration et al., 2018, A&A, 616, A1Gaia Collaboration et al., 2016, A&A, 595, A1Gonzalez G., 1997, MNRAS, 285, 403Hauschildt P. H., Allard F., Baron E., 1999, ApJ, 512, 377Horne K., 1986, PASP, 98, 609Husser T. O., Wende-von Berg S., Dreizler S., Homeier D.,Reiners A., Barman T., Hauschildt P. H., 2013, A&A,553, A6Ionov D. E., Pavlyuchenkov Y. N., Shematovich V. I., 2018,MNRAS, 476, 5639Jackson A. P., Davis T. A., Wheatley P. J., 2012, MNRAS,422, 2024Jenkins J. M. et al., 2016, in Software andCyberinfrastructure for Astronomy IV. p. 99133EJenkins J. S., Jones H. R. A., Pavlenko Y., Pinfield D. J.,Barnes J. R., Lyubchik Y., 2008, A&A, 485, 571Jenkins J. S. et al., 2006, MNRAS, 372, 163Jenkins J. S., Jones H. R. A., Tuomi M., D´ıaz M., CorderoJ. P., Aguayo A., et al. 2017, MNRAS, 466, 443Jenkins J. S. et al., 2013, ApJ, 766, 67Jenkins J. S. et al., 2011, A&A, 531, A8King G. W. et al., 2018, MNRAS, 478, 1193Kipping D. M., 2013, MNRAS, 435, 2152Kreidberg L., 2015, Publications of the AstronomicalSociety of the Pacific, 127, 1161Kubyshkina D. et al., 2018, A&A, 619, A151Kurucz R., 1993, ATLAS9 Stellar Atmosphere Programsand 2 km/s grid. Kurucz CD-ROM No. 13. Cambridge,13Kurucz R. L., 1992, in Barbuy B., Renzini A., eds, IAUSymposium Vol. 149, The Stellar Populations ofGalaxies. p. 225Li J., Tenenbaum P., Twicken J. D., Burke C. J., JenkinsJ. M., Quintana E. V., Rowe J. F., Seader S. E., 2019,PASP, 131, 024506Lindegren L. et al., 2018, A&A, 616, A2Lindegren L. et al., 2016, A&A, 595, A4Lindegren L., Lammers U., Hobbs D., O’Mullane W.,Bastian U., Hern´andez J., 2012, A&A, 538, A786