We investigate the thermoelastic properties of close-packed phases of iron at pressures up to 400 GPa and temperature to 6000 K using a tight-binding total-energy method and the cell model of the vibrational partition function. The calculated properties are in good agreement with available static and shock-wave experimental measurements. The compressional behavior of a number of thermoelastic parameters is found to resemble that of a prototypical oxide ͑MgO͒ supporting some aspects of universal behavior at high compression: the product of thermal expansivity ␣ and bulk modulus is found to be nearly independent of compression at high pressure and the logarithmic volume derivative of ␣ is found to decrease with compression. In contrast to the behavior of MgO, ␣K T and ␦ T of iron are found to depend strongly on temperature due to contributions from the thermal excitation of electrons. Significant decrease of the elastic constants of hcp iron with temperature was found.

A tight-binding total-energy model parametrized to first-principles linearized augmented plane-wave computations is applied to study elasticity and elastic anisotropy in Fe, Xe, and Si at high pressures. We find that the model works well in reproducing the compression, electronic structure, phase relations, and elasticity in these diverse materials. In Xe, for example, the same parametrization works well over a fivefold compression range from a van der Waals solid to a dense metal. We find that the cubic close-packed structures are all more anisotropic than hexagonal close packed and that at high pressures the elastic anisotropy approaches that of any central force nearest-neighbor model. We find that long-range, nonorthogonal parametrizations are necessary for greatest accuracy.

We present a parametrization of the interaction potential for Al3+ in water from first principles calculations. We have performed a critical study of the Al3+–water interaction using sequences of correlation consistent basis sets that approach the complete basis set limit and include core-valence correlation effects. We suggest as minimum theoretical requirements treatment of the electron correlation at the MP2 level of theory using a triple zeta quality basis set that accounts for the effect of core-valence correlation. The latter amounts for an increase of ∼5 kcal/mol (3%) to the stabilization energy, a shortening of 0.015 Å in the Al–O distance, and an increase of 22 cm−1 in the harmonic frequency of the Al–O vibration. This is the first time that core-valence effects were investigated for this system. The stabilization energy of the Al3+(H2O) cluster is 201 kcal/mol and the corresponding Al–O bond length is 1.719 Å at the MP2 level of theory with the cc-pwCVQZ basis set. This minimum is metastable with respect to the Al2++H2O+ asymptote since even the second ionization potential (IP) of Al is larger than the first IP of water. The hexa-aqua cluster Al3+(H2O)6 is, however, stable upon dissociation to Al3+(H2O)5+H2O by 64.8 kcal/mol, demonstrating the capacity of “effective” solvation in stabilizing the charge on the cation. The optimal structures of the n=5 and 6 clusters (having C2v and Th symmetries, respectively) and their harmonic vibrational frequencies are the first ones reported at the MP2 level with basis sets of this size. Core-valence correlation effects for the n=6 cluster are found to be of similar magnitude with those observed for the n=1 cluster. The stabilization energy of the n=6 cluster with respect to its fragments is 723.7 kcal/mol and the corresponding Al–O distance is 1.911 Å. These results were used in order to parametrize a pairwise-additive interaction potential for aluminum–water interaction that was grafted onto the Toukan–Rahman interaction potential for water. The potential model reproduces the ab initio results for Al3+(H2O)6 within 2.0 kcal/mol for the stabilization energy and 0.003 Å for R(Al–O) distance. Using this potential we estimated the enthalpy of solvation of Al3+ to be −1106±6 kcal/mol, therefore favoring the lower value of the experimentally obtained data (−1115 and −1140 kcal/mol, respectively). In addition, we calculate the first peak of the Al–O radial distribution function at 1.885 Å, in excellent agreement with x-ray diffraction studies that suggest a peak at 1.882±0.004 Å. We compute the first peak of the Al–H radial distribution function at 2.473 Å and the average angle between the plane of a water molecule and the Al–O vector at −28.27°.

We present a clear and rigorous derivation of the Ewald-like method for calculation of the electrostatic energy of the systems infinitely periodic in two-dimensions and of finite size in the third dimension (slabs). We have generalized this method originally developed by Rhee et al. [Phys. Rev. B 40, 36(1989)] to account for charge-dipole and dipole-dipole interactions and therefore made it suitable for treatment of polarizable systems. This method has the advantage over exact methods of being significantly faster and therefore appropriate for large-scale molecular dynamics simulations. It however involves a Taylor expansion which has to be demonstrated to be of sufficient order. The method was extensively benchmarked against the exact methods by Leckner and Parry. We found it necessary to increase the order of the multipole expansion from 4 (as in original work by Rhee et al.) to 6. In this case the method is adequate for aspect ratios (thickness/shortest side length of the unit cell) ≤ 0.5. Molecular dynamics simulations using the transferable/polarizable model by Rustad et al. were applied to study the surface relaxation of the nonhydroxylated, hydroxylated, and solvated surfaces of α-Fe2O3 (hematite). We find that our nonhydroxylated structures and energies are in good agreement with previous LDA calculations on α-alumina by Manassidis et al. [Surf. Sci. Lett. 285, L517, 1993]. Using the results of molecular dynamics simulations of solvated interfaces, we define end-member hydroxylated-hydrated states for the surfaces which are used in energy minimization calculations. We find that hydration has a small effect on the surface structure, but that hydroxylation has a significant effect. Our calculations, both for gas-phase and solutionphase adsorption, predict a greater amount of hydroxylation for the α-Fe2O3 (012) surface than for the (001) surface. Our simulations also indicate the presence of four-fold coordinated iron ions on the (001) surface.