Abstract

Most past and ongoing direct imaging surveys of exoplanets are limited by their reliance on thermal emission, resulting in a biased understanding of planet demographics at wide separations. Hot-start

planets have higher luminosities at young ages and are therefore preferentially favored to be found in high-contrast imaging observations. In contrast, warm- and cold-start planets---those that experience a loss of gravitational potential energy through accretion luminosity---are more difficult to detect, and sub-Jovian low-mass planets at wide separation remain largely unexplored. Strong Halpha emission from young accreting planets can be used to overcome these limitations. This approach provides an alternative assessment of planet occurrence rates regardless of their formation history and enables the search for sub-Saturn-mass planets for the first time. In this talk, I will present results from a deep Halpha imaging survey of 227 members of the ~2 Myr star-forming region IC 348 obtained using HST WFC3-UVIS. I will introduce a deep learning-based image classification model that has helped us identify new accreting protoplanet candidates, some of which could be sub-Jovian-mass planets. This study provides the largest demographic constraints onlong-period accreting protoplanets, offering the first clues about a potentially hidden population of low-mass or cold-start planets at wide separations.