Abstract
We examine the effect of spatial resolution on initial mass ejection in grid-based hydrodynamic simulations of binary neutron star mergers. The subset of the dynamical ejecta with velocities greater than ∼0.6c can generate an ultraviolet precursor to the kilonova on approximately hour timescales and contribute to a years long nonthermal afterglow. Previous work has found differing amounts of this fast ejecta, by one to two orders of magnitude, when using particle-based or grid-based hydrodynamic methods. Here, we carry out a numerical experiment that models the merger as an axisymmetric collision in a corotating frame, accounting for Newtonian self-gravity, inertial forces, and gravitational wave losses. The lower computational cost allows us to reach spatial resolutions as high as 4 m, or ∼3 × 10−4 of the stellar radius. We find that fast ejecta production converges to within 10% for a cell size of 20 m. This suggests that fast ejecta quantities found in existing grid-based merger simulations are unlikely to increase to the level needed to match particle-based results upon further resolution increases. The resulting neutron-powered precursors are in principle detectable out to distances ≲200 Mpc with upcoming facilities.We also find that head-on collisions at the freefall speed, relevant for eccentric mergers, yield fast and slow ejecta quantities of order 10−2
M
⊙, with a kilonova signature distinct from that of quasi-circular mergers.