Direct molecular dynamics simulation of liquid-solid phase equilibria for two-component plasmas.

We determine the liquid-solid phase diagram for carbon-oxygen and oxygen-selenium plasma mixtures using two-phase molecular dynamics simulations. We identify liquid, solid, and interface regions using a bond angle metric. To study finite-size effects, we perform 27,648- and 55,296-ion simulations. To help monitor nonequilibrium effects, we calculate diffusion constants D(i). For the carbon-oxygen system we find that D(O) for oxygen ions in the solid is much smaller than D(C) for carbon ions and that both diffusion constants are 80 or more times smaller than diffusion constants in the liquid phase. There is excellent agreement between our carbon-oxygen phase diagram and that predicted by Medin and Cumming. This suggests that errors from finite-size and nonequilibrium effects are small and that the carbon-oxygen phase diagram is now accurately known. The oxygen-selenium system is a simple two-component model for more complex rapid proton capture nucleosynthesis ash compositions for an accreting neutron star. Diffusion of oxygen, in a predominantly selenium crystal, is remarkably fast, comparable to diffusion in the liquid phase. We find a somewhat lower melting temperature for the oxygen-selenium system than that predicted by Medin and Cumming. This is probably because of electron screening effects.

Direct molecular dynamics simulation of liquid-solid phase equilibria for a three-component plasma.

The neutron-rich isotope ²²Ne may be a significant impurity in carbon and oxygen white dwarfs and could impact how the stars freeze. We perform molecular dynamics simulations to determine the influence of ²²Ne in carbon-oxygen-neon systems on liquid-solid phase equilibria. Both liquid and solid phases are present simultaneously in our simulation volumes. We identify liquid, solid, and interface regions in our simulations using a bond angle metric. In general we find good agreement for the composition of liquid and solid phases between our MD simulations and the semianalytic model of Medin and Cumming. The trace presence of a third component, neon, does not appear to strongly impact the chemical separation found previously for two-component carbon and oxygen systems. This suggests that small amounts of ²²Ne may not qualitatively change how the material in white dwarf stars freezes. However, we do find systematically lower melting temperatures (higher Γ) in our MD simulations compared to the semianalytic model. This difference seems to grow with impurity parameter Q_{imp} and suggests a problem with simple corrections to the linear mixing rule for the free energy of multicomponent solid mixtures that is used in the semianalytic model.

Diffusion in Coulomb crystals.

Diffusion in Coulomb crystals can be important for the structure of neutron star crusts. We determine diffusion constants D from molecular dynamics simulations. We find that D for Coulomb crystals with relatively soft-core 1/r interactions may be larger than D for Lennard-Jones or other solids with harder-core interactions. Diffusion, for simulations of nearly perfect body-centered-cubic lattices, involves the exchange of ions in ringlike configurations. Here ions "hop" in unison without the formation of long lived vacancies. Diffusion, for imperfect crystals, involves the motion of defects. Finally, we find that diffusion, for an amorphous system rapidly quenched from Coulomb parameter Γ=175 to Coulomb parameters up to Γ=1750, is fast enough that the system starts to crystalize during long simulation runs. These results strongly suggest that Coulomb solids in cold white dwarf stars, and the crust of neutron stars, will be crystalline and not amorphous.

Diffusion of neon in white dwarf stars.

Sedimentation of the neutron rich isotope 22Ne may be an important source of gravitational energy during the cooling of white dwarf stars. This depends on the diffusion constant for 22Ne in strongly coupled plasma mixtures. We calculate self-diffusion constants D(i) from molecular dynamics simulations of carbon, oxygen, and neon mixtures. We find that D(i) in a mixture does not differ greatly from earlier one component plasma results. For strong coupling (coulomb parameter Γ> few), D(i) has a modest dependence on the charge Z(i) of the ion species, D(i)∝Z(i)(-2/3). However, D(i) depends more strongly on Z(i) for weak coupling (smaller Γ). We conclude that the self-diffusion constant D(Ne) for 22Ne in carbon, oxygen, and neon plasma mixtures is accurately known so that uncertainties in D(Ne) should be unimportant for simulations of white dwarf cooling.