Lattice stability and high-pressure melting mechanism of dense hydrogen up to 1.5 TPa

Lattice stability and metastability, as well as melting, are important features of the physics and chemistry of dense hydrogen. Using ab initio molecular dynamics (AIMD), the classical superheating limit and melting line of metallic hydrogen are investigated up to 1.5 TPa. The computations show that the classical superheating degree is about 100 K, and the classical melting curve becomes flat at a level of 350 K when beyond 500 GPa. This information allows us to estimate the well depth and the potential barriers that must be overcome when the crystal melts. Inclusion of nuclear quantum effects (NQE) using path integral molecular dynamics (PIMD) predicts that both superheating limit and melting temperature are lowered to below room temperature, but the latter never reaches absolute zero. Detailed analysis indicates that the melting is thermally activated, rather than driven by pure zero-point motion (ZPM). This argument was further supported by extensive PIMD simulations, demonstrating the stability of Fddd structure against liquefaction at low temperatures.

Figure 1

Schematic of an energy surface and the strategy of using classical kinetic energy (or equivalently the equilibrium temperature) to explore the surface's main characteristics. Note that classical $T_m$ provides a practical estimate of the averaged barrier height, and $T_{sl}$ gives an assessment of the well depth of the initially solid phase.

Figure 2

Typical $Z$ curves in the P-T plane for ${\mathrm{H}}_{\mathrm{M}}$ at around 1.5 TPa calculated with AIMD simulations using the NVE ensemble and different $k$-point meshes.

Figure 3

Convergence of the classical melting curve of ${\mathrm{H}}_{\mathrm{M}}$ calculated with the $Z$-curve method using AIMD simulations in the NVE ensemble. Notice the large error that resulted from insufficient KPM sampling. The data of Chen are from [19].

Figure 4

Comparison of the melting curves of ${\mathrm{H}}_{\mathrm{M}}$ calculated with AIMD and AI-PIMD simulations. The latter uses eight beads to capture the NQE on the superheating limit. Corrections with respect to the classical superheating degree and $k$-point meshes are also plotted.

Figure 5

Variation of the projected PCF in ${\mathrm{H}}_{\mathrm{M}}$ with temperature at a pressure of $\sim 1.5\mathrm{TPa}$, after a long $(\sim 6\mathrm{ps})$ two-phase equilibrating. The arrow indicates the beginning of the homogeneous liquid feature. Lines are relatively shifted for presentation.

Figure 6

Comparison of the enthalpy of Fddd and the liquid phase of ${\mathrm{H}}_{\mathrm{M}}$, calculated using AI-PIMD at 50 K with 32 and 64 beads, respectively, and the extrapolated results to the infinite beads. Dotted and dash-dotted lines extrapolate the enthalpy to nearby pressures.