Na and Al are immiscible, even in the melt.  The compound NaAlH₄ exists as an ionic molecular solid, with AlH₄^- anions bound with polar covalent Al-H bonds.  Likewise, Na₃AlH₆, is composed of AlH₆^3- anions, charge balanced by three Na⁺ cations. In-situ single crystal Raman scattering studies have shown that the AlH₄^- anions are stable up to the melt in NaAlH₄ (E.H. Majzoub, V. Ozolins, K.F. McCarty, Phys. Rev. B, 71, 024118, 2005), limiting the explanations of enhanced sorption kinetics through transition-metal "doping" procedures commonly used for this compound.   Our group studies the structure, lattice dynamics, and thermodynamic properties of these materials to develop higher hydrogen capacities and better hydrogen sorption kinetics.

We have developed a suite of Monte Carlo global optimization techniques, using basin-hopping and potential energy smoothing using both Metropolis and non-conventional algorithms, which can predict ground state crystal structures and structures close to the ground state in many of the complex anionic hydrides, as well as prototype structures for electrostatically dominated nano-clusters.  This approach allows us to search the space of materials difficult to access experimentally, and to search for new potential hyrdogen storage materials. The method we use is called PEGS, for "prototype electrostatic ground states." [Prototype electrostatic ground state approach to predicting crystal structures of ionic compounds: Application to hydrogen storage materials, E.H. Majzoub, V. Ozolins, Phys. Rev. B, 77, 104115, (2008)]

The PEGS method is quite robust. In addition to predicting ground state structues of ionic crystals, it is also able to address the more complicated issue of crystal polymorphs. A polymorph is simply a variation on a crystal structure, and many crystals phase tranform into different structues as a function of temperature, or pressure, for example. Several of the polymorphs of calcium borohydride have been predicted using the PEGS method as shown below.

While recent interest has focused on complex metal hydrides such as NaAlH4 and Ca(BH4)2, these compounds are not as easily tunable (as are the interstitial metallic hydrides) through alloying with other metal atoms due to the strongly ionic character of the cohesive energy. However, the complex hydrides are superior on a wt.% hydrogen basis, and are the preferred materials for vehicular transport. In order to address thermodynamic tunability, we investigate these materials at the nano-scale, where the ratio of surface to bulk atoms impacts the energetics. Recent theoretical work by Wagemans et al. [J. Am. Chem. Soc., 127, 16675, 2005] and others indicate that small clusters of MgH2, for example, can significantly lower the desorption enthalpy with respect to bulk. Small metal or hydride clusters may be incorporated into nanoporous frameworks such as block polymer templates, for example, to prevent agglomeration and perhaps even improve tunability through particle/surface interactions.

The figure below shows the total free energy, including entropy, of small clusters of MgH2, calculated using first-principles density functional theory, with structure prototypes generated using the stochastic methods described above. The free energy of small clusters is expectedly larger than that of the bulk and are therefore "destabilized" with respect to the bulk.

Nanocluster hydrides may be housed in a porous framework with controlled size nanopores. We are currently investigating incorporation of small clusters of complex hydrides into these frameworks in collaboration with Sandia National Laboratories.