摘要

We have developed a procedure for crystal structure generation and prediction for ionic compounds consisting of a collection of cations and rigid complex anions. Our approach is based on global optimization of an energy functional consisting of the electrostatic and soft-sphere repulsive energies using Metropolis Monte Carlo (MMC) simulated annealing in conjunction with smoothing of the potential energy landscape via the distance scaling method. The resulting structures, or prototype electrostatic ground states (PEGS), are subsequently relaxed using first-principles density-functional theory (DFT) calculations to obtain accurate structural parameters and thermodynamic properties. This method is shown to produce the ground state structures of NaAlH4 and Mg(AlH4)(2), as well as the mixed cation alanate K2LiAlH6. For LiAlH4, the PEGS search produces a structure with a static DFT total energy equal to that of the experimentally observed structure; the latter is stabilized by vibrational contributions to the free energy. For mixed-valence hexa-alanates, XY AlH6, where X= (Li, Na, K), and Y = (Mg, Ca), the PEGS method predicts six unsuspected structure types, which are not found in the existing structure databases. The PEGS search yields energies that are, on the average, better than the best database structures with the same number of atoms per unit cell, demonstrating the predictive power and usefulness of the PEGS structures. In addition to the recently synthesized LiMgAlH6 compound, we predict that LiCaAlH6, NaCaAlH6, and KCaAlH6 are also thermodynamically stable with respect to phase separation into other alanates and metal hydrides. In contrast, NaMgAlH6 and KMgAlH6 are slightly unstable (by less than 3 kJ/mol) relative to the phase separation into NaAlH4, KAlH4, and MgH2. We suggest that solid-state ion-exchange reactions between X3AlH6 (X= Li, Na, K) and YCl2 (Y = Mg, Ca) could be used to synthesize the predicted mixed-valence hexa-alanates.

  • 出版日期2008-3