Planetesimal Accretion in the Solar System and Beyond: An Application to Systems of Tightly-Packed Inner Planets

Abstract

Planetesimals are the smallest gravitationally bound objects to play a role in the planet formation process. In a bottom-up fashion, these bodies are thought to collide and grow to form protoplanets, which are roughly Mars-sized objects, and then eventually coalesce into larger worlds. Throughout this process, some planetesimals are left behind and can persist for billions of years after the formation of a planetary system. This is believed to be the principal source of asteroid and Kuiper belt objects, among many other small body populations in the present-day solar system. In addition to providing clues about the process of planet formation, planetesimals can also transport planets themselves, through weak but numerous gravitational interactions. Planetesimals around other stars sometimes collide and generate dust, producing a distinct observational marker that can be used to infer their presence. In this thesis, I use state of the art N-body simulations to understand the dynamics that govern planetesimal interactions and growth. For the first time, I follow this growth process using bodies with masses comparable to those predicted by planetesimal formation models. Upon doing so, I show that certain dynamical mechanisms involving mean-motion resonances only operate with sufficiently high resolution, and the effects of these mechanisms place a number of constraints on the planet formation process, including tracing the initial sizes of planetesimal formation and using collisionally-generated dust to infer the orbital properties of unseen planets in nearby disks. I also use these resolution capabilities to directly follow the growth of a system of terrestrial planets, starting from planetesimals. In doing so, I assess the viability of an in-situ formation model for systems of tightly-packed inner planets (STIPs), which appear to be a common outcome of planet formation. I also use these simulation results to train a neural network to generate a larger set of post-planetesimal accretion phase initial conditions, which I leverage to construct a statistical sample of simulated planetary systems for comparison with observations.