2018AJ....156..292T

Understanding the distribution and occurrence rate of small planets was a fundamental goal of the Kepler transiting exoplanet mission, and could be improved with K2 and Transiting Exoplanet Survey Satellite (TESS). Deriving accurate exoplanetary radii requires accurate measurements of the host star radii and the planetary transit depths, including accounting for any "third light" in the system due to nearby bound companions or background stars. High-resolution imaging of Kepler and K2 planet candidate hosts to detect very close (within ∼0.''5) background or bound stellar companions has been crucial for both confirming the planetary nature of candidates, and the determination of accurate planetary radii and mean densities. Here we present an investigation of the effect of close companions, both detected and undetected, on the observed (raw count) exoplanet radius distribution. We demonstrate that the recently detected "gap" in the observed radius distribution (also seen in the completeness-corrected distribution) is fairly robust to undetected stellar companions, given that all of the systems in the sample have undergone some kind of vetting with high-resolution imaging. However, while the gap in the observed sample is not erased or shifted, it is partially filled in after accounting for possible undetected stellar companions. These findings have implications for the most likely core composition, and thus formation location, of super-Earth and sub-Neptune planets. Furthermore, we show that without high-resolution imaging of planet candidate host stars, the shape of the observed exoplanet radius distribution will be incorrectly inferred, for both Kepler- and TESS-detected systems.