First-principles calculations of the structural, dynamical, and electronic properties of liquid MgO

Abstract

The structural, dynamical, and electronic properties of liquid MgO have been investigated over a wide range of pressure ͑0 to ϳ 240 GPa͒ and temperature ͑3000-10 000 K͒ using first-principles molecular dynamics ͑FPMD͒ within the framework of density-functional theory and the pseudopotential approximation. Our results show that the liquid structure is highly sensitive to compression: the Mg-O coordination number increases from 5 at zero pressure to 7 at high pressure. The Grüneisen parameter and heat capacity are found to increase upon twofold compression by 40% and 20%, respectively. The dynamical behavior of the liquid phase is characterized by the diffusion coefficient, which is found to decrease with increasing pressure and to increase with increasing temperature in a way that can be accurately characterized by an Arrhenius relationship with activation energy and volume of 0.85 eV and 1.3 Å 3 , respectively. The calculated electronic density of states show that the electronic structure of the liquid phase differs substantially from that of the crystalline phase: the liquid has no band gap and a density of states at the Fermi level increases with increasing volume and temperature.

Spin crossover in liquid (Mg,Fe)O at extreme conditions

“…We find a higher result for ⟨Z FeO ⟩ by as much as 1.0 when Fe is in the high-spin state as compared to the low-spin state, which reflects the larger spatial extent of the 3d-like orbitals in the former electronic configuration and hence a larger ionic radius. The Mg-O coordination is very close to previous ab initio results on pure MgO [15] and appears independent of the spin state of Fe in the liquid. We find ⟨Z MgO ⟩ ≈ 5.2 at 11.4Å 3 /atom, very close to the experimental value of 4.95 at the same atomic volume [47].…”

“…3. The diffusivity of (Mg,Fe)O appears thus overall fairly similar to that of MgO [15], the small but systematic difference being possibly attributable to the difference in exchange-correlation functional (PBEsol + U vs LDA) and the different chemical composition. The diffusivities of high-spin Fe, low-spin Fe, Mg, and O reveal a weak dependence of D on element and spin state [31], the weak dependence on spin state being at odds with earlier results on Fe diffusivity in crystalline (Mg,Fe)O, where the diffusivity was found to vary by a factor of three between high-spin and low-spin Fe [42].…”

“…Experimental work utilizing shockwave experiments [6], and the laser-heated diamond anvil cell [7] has produced valuable results, but many properties of these liquids remain weakly constrained. Quantum-mechanical, finite-temperature density-functional theory (DFT) molecular dynamics (MD) [8,9] has proved to be a powerful and accurate theoretical method for predicting the properties of liquid silicates [10][11][12][13], oxides [14,15], and metals [16] at extreme conditions. In this approach, no assumptions are made on the shape of the electronic charge density, and the effects of temperature are included in both the electronic and ionic degrees of freedom.…”

Electronic redistribution around oxygen atoms in silicate melts by ab initio molecular dynamics simulation

“…With production runs of only 5ps, it is not possible to check with much accuracy if the imposed density (see below) for the melt under investigation corresponds to P~0 in the simulation. On the other hand it is known that the use of the generalized gradient approximation (GGA) for the exchange correlation energy tends to underestimate by ~2-3% the density of silicate minerals when the use of the local density approximation (LDA), a priori less accurate, leads (fortuitously) to a better estimation [8,10]. In fact GGA remedies some of the weaknesses of LDA, and especially its tendency to overbinding, and the slight underestimation of the density comes from a lack of dispersion energy between atoms [23,24].…”

“…With regard to mantle minerals and their melts, the number of studies using ab initio molecular dynamics simulations (AIMD) is increasing late years and now cover a rather large range of composition (e.g. SiO2 [5,6], MgO [7,8], MgSiO3 [9,10], Mg2SiO4 [11][12], Na2SiO4 [13] and CaO(0.12)Al2O3(0.21)SiO2(0.67) [14]). …”