Ab initiostudies on high pressure phases of ice

Abstract

The pressure-induced transition of H2O into the ice X phase, characterized by symmetric hydrogen bonding, is studied using ab initio molecular dynamics combined with ultrasoft pseudopotentials. A good description of the hydrogen bond is obtained only after gradient corrections to the local-density approximation are included. The transition into ice X is predicted at 49 GPa, in good agreement with experiment, when proton quantum fluctuations are treated within mean-field theory. Molecular-dynamics simulations show that a mode-softening description of the transition is appropriate.PACS numbers: 64.70. Kb, ice has an unusually rich phase diagram [1]. The dominant hydrogen-bonding interactions give rise to a sequence of rather open structures which become more close packed with increasing pressure. Nevertheless, the phases ice I through ice IX, which occur at pressures up to a few tens of GPa, all still have intact water molecules as their basic building blocks. At a sufficiently high pressure, the molecular picture is expected to break down entirely. For the moderately high pressure ice VII and VIII phases, it is found that the H atoms become more symmetrically bonded with increasing pressure; in fact the covalently bonded O-H distance increases while the hydrogen-bonded O-O distance decreases with increasing pressure [2,3]. Thus, in the simplest picture, the transition would occur by having H atoms shift to the fully symmetric positions midway between the two neighboring O atoms, leaving the oxygen lattice intact. Such a speculative phase, in which the distinction between the covalent and hydrogen bonds has been eliminated, is called ice X.Despite considerable discussion in the literature [4][5][6][7][8][9], there is to date no compelling direct evidence for the existence of ice X. Experimentally, Raman spectra of ice VIII up to 50 GPa at 100 K demonstrate the appearance of a new band above 40 GPa [4] which could reflect such a transition; and Brillouin scattering studies up to 67 GPa at 300 K also give an indication of a transition at 44 GPa from an anomaly in the behavior of the longitudinal sound velocity [5]. Earlier work based on extrapolation of Raman spectra from lower pressures at room temperature also suggested that a symmetric hydrogen-bond structure might form in ice VII at pressures of about 75 ± 20 GPa [6]. However, little is experimentally known about the positions of the protons in the high pressure phase, because direct measurements like neutron or electron diffraction are difficult. Theoretically, studies based on empirical interatomic potentials have supported the notion of a transition to the ice X structure. An early theoretical calculation by Holzapfel [7] predicted a transition to symmetric hydrogen bonding in ice VII at pressures between 35 and 80 GPa. The interaction of the hydrogen atoms with the two nearest-neighbor oxygen atoms was approximated by equivalent Morse potentials. Later, a calculation by Stillinger and Schweizer [8] predicted a transition at roughly 60 GPa. This work included...

Ab initio total energy studies of minerals using density functional theory and the local density approximation

“…This implies that a further extension to the model, a gradient correction, is necessary. This is similar to what has been observed by Lee et al (1992) for ice, and our own preliminary gradient corrected calculations carried out at the time of writing show that the incorporation of a gradient correction yields lattice constants which are very close (less than 0.5%) to the experimental data.…”

Ab initio theory of planetary materials

“…O linkage, or by its pressure-induced symmetrization. First-principles simulations have reproduced the pressure-induced symmetrization at low temperature [200,201], although the precise value of the transition pressure could only be obtained by properly treating quantum effects on the protons [202]. The calculated infrared activity of ice across the symmetrization [203] compares favorably with experiments.…”

Monte Carlo Simulation of Interlayer Molecular Structure in Swelling Clay Minerals. 1. Methodology

“…The development ofhigh quality ab initio calculations of water moleculeor Na § clay mineral interaction energies remains a challenge for future research (Saner 1989, Lasaga 1992, Bleam 1993). Semi-empirical estimates, for clusters comprising a water molecule or Na + interacting with a clay mineral fragment making up about two dozen atoms, are available (Saner et al 1984, Delville 1991; but they do not yet give a good account of the electronelectron correlations and overlap that are expected to contribute to the short-range repulsions and van der Waals interactions important to H-bonded systems, such as interlayer water (Matsouka et al 1976, Lee et al 1992. Given this state of affairs, we shall take the view that, to capture efficiently and accurately the shortrange interactions between a water molecule or Na + and a 2:1 clay mineral, it is necessary to reduce the size of the clay mineral fragment to one that can be described by the more rigorous ab initio calculations--in practice, perhaps three or four atoms.…”