Can metal-rich worlds form by giant impacts?

Context. Astronomical observations revealed the existence of exoplanets whose densities are far higher than what is expected from cosmochemistry. This high-density planetary population may account for 9% of terrestrial planets, suggesting the existence of processes that form planets with compositions dramatically different from their starting materials. Aims. A commonly invoked theory is that these high-density exoplanets are the metallic cores of super- Earth-sized planets whose rocky mantle was stripped by giant impacts. Here we aim to test this hypothesis. Methods. To maximize the likelihood that metal-rich giant-impact remnants form, we model the late orbital instability of tightly packed super-Earths orbiting a host star at small stellocentric distances ("compact systems"). We combine orbital dynamics, impact physics, and machine learning to explore the stability and collisional evolution of 100 observed compact systems. In each unstable compact system, we assume that the super-Earths undergo giant impacts and explore 1000 possible collision scenarios. We repeat the simulations with different initial conditions, such as the initial masses and composition of the super-Earths. Results. We find that giant impacts are capable of stripping the mantles of super-Earths and form metal-rich worlds as massive and large as the observed high-density exoplanets. However, we also find that, in most of the explored scenarios, mantle-stripping giant impacts between super-Earths are unlikely to occur at rates sufficient to explain the size and currently estimated abundance of the observed high-density exoplanets. We explain this as the interplay of three factors: the size of the super-Earths being in most cases smaller than 2 Earth radii; the efficiency of mantle stripping decreasing with increasing planetary size; and the likelihood of compact system instability decreasing with increasing average sizes of the planets in the compact system. Conclusions. We conclude that most of the observed high-density exoplanets are unlikely to be metal-rich giant-impact remnants.