We report a molecular dynamics study of the heterogeneous crystallization of high-pressure glassy water using (plastic) ice VII as a substrate. We focus on the thermodynamic conditions P ∈ [6–8] GPa and T ∈ [100–500] K, at which (plastic) ice VII and glassy water are supposed to coexist in several (exo)planets and icy moons. We find that (plastic) ice VII undergoes a martensitic phase transition to a (plastic) fcc crystal. Depending on the molecular rotational lifetime τ, we identify three rotational regimes: for τ > 20 ps, crystallization does not occur; for τ ∼ 15 ps, we observe a very sluggish crystallization and the formation of a considerable amount of icosahedral environments trapped in a highly defective crystal or in the residual glassy matrix; and for τ < 10 ps, crystallization takes place smoothly, resulting in an almost defect-free plastic fcc solid. The presence of icosahedral environments at intermediate τ is of particular interest as it shows that such a geometry, otherwise ephemeral at lower pressures, is, indeed, present in water. We justify the presence of icosahedral structures based on geometrical arguments. Our results represent the first study of heterogeneous crystallization occurring at thermodynamic conditions of relevance for planetary science and unveil the role of molecular rotations in achieving it. Our findings (i) show that the stability of plastic ice VII, widely reported in the literature, should be reconsidered in favor of plastic fcc, (ii) provide a rationale for the role of molecular rotations in achieving heterogeneous crystallization, and (iii) represent the first evidence of long-living icosahedral structures in water. Therefore, our work pushes forward our understanding of the properties of water.
Similar conclusions apply to the cases of P = 6 GPa and T = 200 K; P = 7 GPa and T = 100 K and T = 200 K; and P = 8 GPa and T = 100 K and T = 200 K. At these conditions, molecular rotations are mostly inactive.
Similar conclusions apply to the higher temperature of T = 500 K, and therefore, we limit ourselves to presenting the results for T = 400 K.