Rethinking the role of the giant planet instability in terrestrial planet formation models

Abstract

Advances in computing power and numerical methodologies over the past several decades sparked a prolific output of dynamical investigations of the late stages of terrestrial planet formation. Among other peculiar inner solar system qualities, the ability of simulations to reproduce the small mass of Mars within the planets’ geochemically inferred accretion timescale of ≲10 Myr after the appearance of calcium aluminum-rich inclusions (CAIs) is arguably considered the gold standard for judging evolutionary hypotheses. At present, a number of independent models are capable of consistently generating Mars-like planets and simultaneously satisfying various important observational and geochemical constraints. However, all models must still account for the effects of the epoch of giant planet migration and orbital instability; an event which dynamical and cosmochemical constraints indicate occurred within the first 100 Myr after nebular gas dispersal. If the instability occurred in the first few Myr of this window, the disturbance might have affected the bulk of Mars’ growth. In this manuscript, we turn our attention to a scenario where the instability took place after t≃ 50 Myr. Specifically, we simulate the instability’s effects on three nearly-assembled terrestrial systems that were generated via previous embryo accretion models and contain three large proto-planets (i.e. Earth, Venus and Theia) with 0.5 <m<1.0M⊕ and orbits interior to a collection of ∼Mars-mass embryos (a>1.3 au and m<0.2M⊕) and debris. While the instability consistently triggers a Moon-forming impact and efficiently removes excessive material from the Mars-region in our models, we find that our final systems are too dynamically excited and devoid of Mars and Mercury analogs. Thus, we conclude that, while possible, our scenario is far more improbable than one where the instability either occurred earlier, or at a time where Earth and Venus’ orbits were far less dynamically excited than considered here.