The Role of Stellar Mass in High­Contrast ImagingThe Role of Stellar Mass in High­Contrast Imaging Justin R. Crepp, John A. JohnsonJustin R. Crepp, John A. Johnson (a.k.a.,(a.k.a., aa simplesimple explanationexplanation ofof California Institute of TechnologyCalifornia Institute of Technology arXiv:1103:4910 (accepted to ApJ) allall exoplanetexoplanet imagesimages toto date)date) (1)(1) Why Why have have the the first first handful handful of of exoplanet exoplanet images images been been (5)(5) Table Table 1 1 and and our our simulation simulation results results are are consistent consistent with with the the around M,K and A stars, but no spectral types in between? around M,K and A stars, but no spectral types in between? (3)(3) Two Two additional additional observational observational results results support support presence of two different planet formation mechanisms: one operating presence of two different planet formation mechanisms: one operating This must be telling us something about planet formation.This must be telling us something about planet formation. interpretation interpretation of of Beta Beta Pic Pic b b and and the the HR HR 8799 8799 planets planets as as close close to to stars stars that that results results in in strong strong correlations correlations between between star star mass mass forming by “core­accretion”: forming by “core­accretion”: and and planet planet properties, properties, and and another another operating operating further further from from stars stars (or (or (i) Beta Pic b and HR 8799 bcde fall in the possibly possibly at at all all separations) separations) where where there there exists exists little little correspondence correspondence same sector of the mass / semi­major axis Beta Pic b plot as RV planets (Currie et al. 2011). between star mass and planet properties.between star mass and planet properties. High metallicity!

(ii) Spectra of HR 8799 b indicate a very HR 8799 bcde Some observational constraints on the high metal content (Barman et al. 2011). presence of ultra-wide companions: ~1% ~1% “Extrapolating the inferred companion Directly imaged planetary­mass (m<15M ) companions with stellar primaries (http://exoplanet.eu/). mass function to <13M , we find that Fig. 1: Directly imaged planetary­mass (m<15MJup) companions with stellar primaries (http://exoplanet.eu/). J Mode 1 Fig. 3 – Mass ratio vs. semi­major axis for planetary mass companions from Currie et al. 2011. Given sufficient subdeuterium-burning “planetary” There currently exists an apparent dichotomy between the detection of planets around A­stars in the solar companions, if able to form through gravo- ~80 AU Mode 1 time baselines, Doppler RV teams will eventually discover companions with the same orbital period as Beta Pic turbulent fragmentation, exist in wide neighborhood and ultra­wide separation planets around M, K stars in clusters. time baselines, Doppler RV teams will eventually discover companions with the same orbital period as Beta Pic orbits around ~1% of Sun-like stars” b and HR 8799 e (indeed, many RV trends already exist in current data sets!). Given their co­rotation and orbit ~5% configuration, it is difficult to justify invoking separate formation mechanisms for HR 8799 e and HR 8799 b. Metchev & Hillenbrand 2010 ~20% Further, HR 8799 e has a high metal content, and our simulations of an extended (a<100 AU) RV population indicate that it is natural to image the first exoplanets around nearby A­type stars. “... naïve estimate of the frequency (of Mode 2 Mode 2 (2)(2) The The Doppler Doppler technique technique has has shown shown that that strong strong ultra-wide planetary-mass companions) is approximately 4% ...” correlations exist between star mass and planet properties: correlations exist between star mass and planet properties: (4)(4) How How do do we we explain explain the the bottom bottom portion portion of of Table Table 1? 1? Ireland et al. 2011 A­starA­star M­starM­star (1) the occurrence rate of gas giant planets correlates with star mass (Johnson et al. 2010) Observations of a group of stars having the same distance and Observations of a group of stars having the same distance and (2) planets orbiting A­stars are more massive than those orbiting lower­mass stars (Bowler et al. 2010) Fig. 5 – The simplest explanation for all exoplanet images to date: two modes of formation. Companions that form close to their star (3) planets orbiting A­stars have wider orbits than those orbiting lower­mass stars (Bowler et al. 2010) age age represents represents a a special special case case in in high­contrast high­contrast imaging, imaging, exhibit very strong correlations between star mass and planet properties: planet occurrence rate, semimajor axis extent, and mass all rise with increasing star mass. Mode 1 (blue region) is likely “core­accretion” and is also responsible for creating the Doppler RV planet Can Can extrapolation extrapolation of of the the radial radial velocity velocity (RV) (RV) planet planet neutralizing two important trade­offs between low­mass stars neutralizing two important trade­offs between low­mass stars population. Another mechanism, Mode 2 (orange region) has a lower occurrence rate at wide separations. Our simulations indicate that companions with ultra­wide orbits may exhibit very little correspondence between star mass and planet properties. These results population to separations accessible to high­contrast population to separations accessible to high­contrast and high­mass stars. and high­mass stars. Simulations of a tight cluster show that Simulations of a tight cluster show that are consistent with the predictions of Boley et al. 2009. instruments explain the top portion of Table 1? instruments explain the top portion of Table 1? there is a situation where low­mass stars can dominate both there is a situation where low­mass stars can dominate both YES! YES! (6)(6) There is another important effect to consider: There is another important effect to consider: young stellar clusters young stellar clusters the the number number ANDAND efficiency efficiency of of detections. detections. This This occurs occurs when when are are subject subject to to an an observational observational bias bias related related to to small small number number statistics. statistics. all all correlations correlations between between star star mass mass and and planet planet properties properties are are Low­mass stars produce the first Low­mass stars produce the first ultra­wide separationultra­wide separation exoplanet images exoplanet images removed (the red curves below). removed (the red curves below). irrespective of correlations between star mass and planet properties. irrespective of correlations between star mass and planet properties.

Cluster N D (pc) Age (Myr) AB Dor 89 34+/-26 70

Argus 64 106+/-51 40

Beta pic 50 31+/-21 10

Tuc-hor 49 48+/-7 30

Columba 41 82+/-30 30

Eta Cha 24 108+/-9 6

Carina 23 85+/-35 30

TW hydra 22 48+/-13 8 Fig. 6 – The number of detections in a (tightly bound) young stellar cluster always favors Octans 15 141+/-34 20 low­mass stars. Since the stellar IMF falls off faster than the planet occurrence rate grows, MK stars are the first spectral­types to produce planet images. This bias is exacerbated by the fact that: (i) half of star systems are binaries and high­contrast teams avoid binaries; (ii) the youngest stars are at a distance of ~140 pc, forcing the overall planet occurrence rate to take on its value at wide separations; (iii) the occurrence rate of (7)(7) Conclusions: Conclusions: ultra­wide separation planets appears to be only several percent!

(i)(i) Extrapolation of the Doppler RV planet population to wide (tens of AU) Extrapolation of the Doppler RV planet population to wide (tens of AU) separations results in an excellent match between simulations and separations results in an excellent match between simulations and observations to date, both detections and non­detections.observations to date, both detections and non­detections. Fig. 2 – Planet detection numbers (top) and rates (bottom) as a function of host star mass for a 50 pc age­and­ (ii)(ii) HR 8799 bcde and Beta Pic b likely formed by the same process as those of the HR 8799 bcde and Beta Pic b likely formed by the same process as those of the brightness selected ground­based imaging survey using contrast levels comparable to present­day instruments. Doppler RV population. Over­plotted are non­detections from a number of previous programs. Extrapolation of the Doppler RV Doppler RV population. population to wide semimajor axes yields an excellent agreement between simulations and observations to (iii) A­stars are ideal high­contrast imaging targets for reasons in addition to their (iii) A­stars are ideal high­contrast imaging targets for reasons in addition to their date, in terms of both absolute and relative detection rates. Thus far, only 2 planetary systems have been Fig. 4 – Planet detection numbers (top) and rates (bottom) as a function of host star mass for simulations imaged from the ground in the parameter space considered above (bright, young, nearby stars): HR 8799 and intrinsic youth and ability to serve as a bright beacon for “extreme” AO systems.intrinsic youth and ability to serve as a bright beacon for “extreme” AO systems. of a young stellar association synthesized to resemble Tucana­Horologium. When correlations between Beta Pic. Both are A­stars. Therefore, the slope of the detection rate curve as a function of star mass must be star mass and planet properties are removed, the number of detections essentially follows the IMF. (iv) Observations of a young, tightly bound cluster represents a special case in (iv) Observations of a young, tightly bound cluster represents a special case in positive. This occurs when correlations exist between star mass and planet properties. high­contrast imaging where MK­stars dominate the number of detections.high­contrast imaging where MK­stars dominate the number of detections.