Lipowsky et al. reply

String-like cooperative motion in homogeneous melting

Despite the fundamental nature and practical importance of melting, there is still no generally accepted theory of this ubiquitous phenomenon. Even the earliest simulations of melting of hard discs by Alder and Wainwright indicated the active role of collective atomic motion in melting and here we utilize molecular dynamics simulation to determine whether these correlated motions are similar to those found in recent studies of glass-forming (GF) liquids and other condensed, strongly interacting, particle systems. We indeed find string-like collective atomic motion in our simulations of "superheated" Ni crystals, but other observations indicate significant differences from GF liquids. For example, we observe neither stretched exponential structural relaxation, nor any decoupling phenomenon, while we do find a boson peak, findings that have strong implications for understanding the physical origin of these universal properties of GF liquids. Our simulations also provide a novel view of "homogeneous" melting in which a small concentration of interstitial defects exerts a powerful effect on the crystal stability through their initiation and propagation of collective atomic motion. These relatively rare point defects are found to propagate down the strings like solitons, driving the collective motion. Crystal integrity remains preserved when the permutational atomic motions take the form of ring-like atomic exchanges, but a topological transition occurs at higher temperatures where the rings open to form linear chains similar in geometrical form and length distribution to the strings of GF liquids. The local symmetry breaking effect of the open strings apparently destabilizes the local lattice structure and precipitates crystal melting. The crystal defects are thus not static entities under dynamic conditions, such as elevated temperatures or material loading, but rather are active agents exhibiting a rich nonlinear dynamics that is not addressed in conventional "static" defect melting models.