Maximum superheating and undercooling: Systematics, molecular dynamics simulations, and dynamic experiments

Abstract

The maximum superheating and undercooling achievable at various heating (or cooling) rates were investigated based on classical nucleation theory and undercooling experiments, molecular dynamics (MD) simulations, and dynamic experiments. The highest (or lowest) temperature Tc achievable in a superheated solid (or an undercooled liquid) depends on a dimensionless nucleation barrier parameter beta and the heating (or cooling) rate Q. beta depends on the material: beta[equivalent]16pigamma_sl³/(3kTmDeltaH_m²) where gammasl is the solid-liquid interfacial energy, DeltaHm the heat of fusion, Tm the melting temperature, and k Boltzmann's constant. The systematics of maximum superheating and undercooling were established phenomenologically as beta= (A0–b log10Q)thetac(1–thetac)2 where thetac = Tc/Tm, A0 = 59.4, b = 2.33, and Q is normalized by 1 K/s. For a number of elements and compounds, beta varies in the range 0.2–8.2, corresponding to maximum superheating thetac of 1.06–1.35 and 1.08–1.43 at Q~1 and 10^12 K/s, respectively. Such systematics predict that a liquid with certain beta cannot crystallize at cooling rates higher than a critical value and that the smallest thetac achievable is 1/3. MD simulations (Q~10^12 K/s) at ambient and high pressures were conducted on close-packed bulk metals with Sutton-Chen many-body potentials. The maximum superheating and undercooling resolved from single- and two-phase simulations are consistent with the thetac-beta-Q systematics for the maximum superheating and undercooling. The systematics are also in accord with previous MD melting simulations on other materials (e.g., silica, Ta and epsilon-Fe) described by different force fields such as Morse-stretch charge equilibrium and embedded-atom-method potentials. Thus, the thetac-beta-Q systematics are supported by simulations at the level of interatomic interactions. The heating rate is crucial to achieving significant superheating experimentally. We demonstrate that the amount of superheating achieved in dynamic experiments (Q~10^12 K/s), such as planar shock-wave loading and intense laser irradiation, agrees with the superheating systematics.