Draft version October 9, 2019 Typeset using LATEX twocolumn style in AASTeX62 The Possible Astrometric Signature of a Planetary-mass Companion to the Nearby Young Star TW Piscis Austrini (Fomalhaut B): Constraints from Astrometry, Radial Velocities, and Direct Imaging Robert J. De Rosa,1 Thomas M. Esposito,2 Lea A. Hirsch,1 Eric L. Nielsen,1 Mark S. Marley,3 Paul Kalas,2, 4 Jason J. Wang,5, ∗ and Bruce Macintosh1 1Kavli Institute for Particle Astrophysics and Cosmology, Stanford University, Stanford, CA 94305, USA 2Department of Astronomy, University of California, Berkeley, CA 94720, USA 3NASA Ames Research Center, MS 245-3, Mountain View, CA 94035, USA 4SETI Institute, Carl Sagan Center, 189 Bernardo Ave., Mountain View CA 94043, USA 5Department of Astronomy, California Institute of Technology, Pasadena, CA 91125, USA ABSTRACT We present constraints on the presence of substellar companions to the nearby (d 7:6 pc), young (440 40 Myr) K4Ve star TW Piscis Austrini, the wide ( 0.3 pc) companion to the∼ A4V star Fo- malhaut.± We combined absolute astrometry from Hipparcos∼and Gaia with literature radial velocity measurements and dedicated high-contrast imaging observations obtained with Keck/NIRC2 to achieve sensitivity to brown dwarf and planetary-mass companions (& 2 MJup) over many decades of orbital 3 period (. 10 yr). The significant astrometric acceleration measured between the Hipparcos and Gaia catalogues, reported previously in the literature, cannot be explained by the orbital motion of TW PsA around the barycenter of the Fomalhaut triple system. Instead, we find that it is consistent with the reflex motion induced by an orbiting substellar companion. The combination of astrometry, radial +0:7 velocities, and a deep L0 imaging dataset leads to a constraint on the companion mass of 1:2 0:6 MJup. However, the period of the companion is poorly constrained, with a highly multi-modal period− poste- rior distribution due to aliasing with the 24.25 year baseline between Hipparcos and Gaia. If confirmed through continued astrometric or spectroscopic monitoring, or via direct detection, the companion to TW PsA would represent a choice target for detailed atmospheric characterization with high-contrast instruments on the upcoming James Webb Space Telescope and Wide Field Infrared Survey Telescope. Keywords: astrometry { stars: individual (TW PsA) { techniques: high angular resolution { techniques: radial velocity arXiv:1910.02965v1 [astro-ph.EP] 7 Oct 2019 ∗ 51 Pegasi b Fellow 2 1. INTRODUCTION joint analysis of absolute astrometry from Hipparcos and The field of astrometric detection of exoplanets will Gaia, a deep high-contrast imaging dataset, and a long- soon be transformed with the discovery of thousands term radial velocity record. We discuss the measured of planetary-mass companions via the astrometric reflex astrometric acceleration, previously noted by (Kervella motion induced on their host star measured by Gaia et al. 2019), in Section2, where we demonstrate it can- (Perryman et al. 2014). While these discoveries will not be explained by the orbit of TW PsA around Fo- span a wide range of orbital periods, a small subset of malhaut, and present limits on the period and mass of them will be at separations wide enough, and around an orbiting companion consistent with the measured ac- stars young enough, that they will be amenable to direct celeration. The long-term radial velocity record is pre- detection and characterization with ground- and space- sented in Section3, and the deep coronagraphic imag- based instrumentation. Previous ground-based surveys ing dataset in Section4. Finally, we present joint con- searching for the astrometric signal induced by an or- straints on the properties of a companion consistent with biting planet have typically targeted fainter stars due the astrometric signal in Section5. to the requirement to have similar-magnitude reference stars (e.g., Sahlmann et al. 2014), whilst space-based 2. ASTROMETRY observations have focused on characterizing known sys- Absolute astrometry for TW PsA was included in tems (e.g., Benedict et al. 2017). These observations both the Hipparcos (HIP 113283; ESA 1997; van are typically expensive, and dedicated astrometric mis- Leeuwen 2007a) and the recent Gaia (Gaia DR2 sions such as Gaia are required to have sufficient sensi- 6604147121141267712; Gaia Collaboration et al. 2018) tivity around a large enough number of stars for a blind catalogues. In the Hipparcos catalogue, the star appears search to be worthwhile. While we await the release typical compared to other stars within a factor of two of the final Gaia astrometric catalog, the intermediate in Hp-band flux. The catalogue goodness of fit metric +2:19 data releases (e.g. Gaia Collaboration et al. 2018) can is 1:33 compared to 1:30 1:39 for the similar brightness be combined with previous measurements from Hippar- stars. This metric is equivalent− to χ2 = 99:5 for TW cos to search for planetary-mass companions to nearby 2 +71 PsA, (χ = 131 47 for the comparison sample), corre- stars (e.g. Brandt 2018; Kervella et al. 2019). −2 sponding to a χν = 1:2 given the 87 one-dimensional TW Piscis Austrini (TW PsA) is a K4Ve star (Keenan astrometric measurements of the star and the five model & McNeil 1989) at a distance of 7.6 pc (Gaia Collabora- parameters. The quality of fit was good enough that no tion et al. 2018). It shares a similar three-dimensional attempt was made to fit either additional acceleration velocity to the naked-eye star Fomalhaut (Luyten 1938) terms or a stochastic solution (van Leeuwen 2007b). and the faint M-dwarf LP 876-10 (Mamajek et al. 2013), We performed a similar check of the quality of the as- forming a wide triple system that extends several de- trometry in the Gaia Data Release 2 catalogue. We grees across the sky. The Fomalhaut system is young, selected stars within a factor of two in G-band flux, with an estimated age of 440 40 Myr (Mamajek 2012). ± that had good quality parallax and photometric mea- This system has been well-studied given its proximity. surements as in Lindegren et al.(2018). The good- The primary hosts an extended debris disk (Holland ness of fit metric for TW PsA was unusually low at et al. 1998), within which a candidate exoplanet with +35:1 4.7 compared with 19:1 9:7 for stars in the compari- a circumplanetary dust ring has been detected via di- son sample, indicating a− relatively good quality of fit of rect imaging at optical wavelengths (Kalas et al. 2008). the five parameter model to the astrometric measure- LP 876-10 also hosts a debris disk (Kennedy et al. 2014), 2 2 +3560 ments (χ = 136 for TW PsA and χ = 960 530 for unusual given the apparent rarity of debris disks around the comparison sample). The reduced χ2 is somewhat− M-dwarfs. Thermal-infrared measurements of TW PsA 2 2 +15:7 high, χν = 2:0 (χν = 4:3 1:9 for the comparison sam- between 24 and 100 µm do not any show evidence of ple), suggesting that the uncertainties− on the individual emission in excess of the predicted photospheric flux astrometric measurements are slightly underestimated. (Carpenter et al. 2008; Kennedy et al. 2014). Shan- No significant astrometric excess noise was reported in non et al.(2014) propose that all three stars originate the catalogue. from the same birth cluster, with Fomalhaut and LP 876-10 forming as a binary and TW PsA captured from 2.1. Astrometric acceleration the cluster into a weakly bound orbit. While Fomal- haut shows evidence of a planetary system (Kalas et al. The acceleration of TW PsA between the Hipparcos 2008), searches for giant planets around TW PsA via and Gaia missions was estimated by computing three proper motion differentials; (1) µ µ the differ- direct imaging (Heinze et al. 2010; Durkan et al. 2016; G − H Nielsen et al. 2019) and radial velocity (Butler et al. ence between the two catalogue proper motions, (2) µ µ the difference between the proper motion com- 2017) have thus far yielded nothing. H − HG In this paper we present constraints on the presence puted from the absolute position of the star in the two of wide-orbit giant planets around TW PsA with a catalogues and the Hipparcos proper motion, and (3) µ µ the same but with respect to the Gaia proper G − HG 3 motion. The proximity of TW PsA (7.6 pc) causes PsA (0.725 M ; Demory et al. 2009) and LP 876-10 a non-negligible change in its apparent proper motion (0.18 M ). across the sky simply due to perspective effects. With- The orbital motion of TW PsA around Fomalhaut out correcting for these, the estimated position of the was noted by Kervella et al.(2019) as being a poten- star at the Gaia epoch using Hipparcos astrometry will tial source of the astrometric acceleration measured be- be systematically offset by 0.2 mas, biasing the derived tween the Hipparcos and Gaia astrometry. The mag- 1 µHG proper motion by 10 µas yr− . The magnitude of nitude of the tangential velocity differential (∆vtan = ∼ 1 this effect is comparable to the size of the uncertainty 18:7 6:4 m s− , Kervella et al. 2019) seems inconsis- ± on µHG given the precision of the absolute astrometric tent with the change in orbital velocity expected for measurement of the star within both catalogues. an orbital period of close to 8 Myr (Mamajek 2012). We account for this effect using the prescription laid The escape velocity from Fomalhaut is estimated to be 1 out in Butkevich & Lindegren(2014), propagating the 210 m s− (Mamajek 2012), so it would be surprising for Hipparcos astrometry from the Hipparcos to the Gaia the velocity of TW PsA to change by close to 10% of epoch (and vice versa) before computing the difference.