You reap what you sow: On the impact of nuclei morphology on seeded molecular dynamics simulations.

This article investigates the validity of using seeded molecular dynamics (MD) with artificially constructed spherical face-centered cubic (fcc) seeds to study crystal nucleation from Lennard-Jones (LJ) melts. By comparing these constructed seeds to "unbiased seeds" extracted from brute force nucleation trajectories, the study finds significant differences in their properties and committor probability distributions, indicating that the commonly used reaction coordinate (RC)—the size of the largest crystalline cluster measured by the number of crystal-like atoms (Nq6)—is a poor descriptor of nucleation progress in LJ systems. Incorporating crystallinity metrics, such as the Beckham–Peters RC that combines cluster size and crystallinity, improves agreement but does not fully reconcile the differences, suggesting additional factors like cluster shape or polymorph may be relevant. The results imply that cluster equilibration is slow relative to growth, challenging classical nucleation theory assumptions and questioning the reliability of seeding methods based on spherical fcc seeds and simple cluster size RCs for accurately estimating nucleation rates and free energy landscapes in LJ melts.