Enhancing Hydride Thermodynamics Through Nanostructuring

The phase stability diagrams for the decomposition of magnesium alanate [Mg(AlH4)2] nanoparticles were constructed as a function of particle size and temperature to better understand nanostructuring of the complex hydrides for hydrogen storage. Relatively smaller nanoparticles of Mg(AlH4)2, MgH2, Al, and Mg ranging from 1 to 2 nm were directly calculated by using density functional theory (DFT) calculations, and cluster expansion and Monte Carlo simulation methods were developed to predict the phase stabilities of 2-100 nm nanoparticles. Our prediction demonstrates that bulk Mg(AlH4)2 can release hydrogen but its uptake reaction is unfavorable, and bulk Mg(AlH4)2 is metastable with respect to bulk MgH2. However, in the cases of nanoparticle systems, hydrogen release and its recharging may be possible by controlling the particle size and temperature, which may facilitate experimental studies to determine the thermodynamically favored reaction pathways for the dehydrogenation and hydrogenation processes of Mg(AlH4)2 nanoparticles. We also provide the equilibrium diagrams for Mg(AlH4)2 nanoparticle decomposition depending on a hydrogen partial pressure. Introduction Metal hydrides have recently attracted interest as hydrogen storage materials for transportation applications. In particular, magnesium hydride (MgH2) has gained a great deal of attention due to its high gravimetric and volumetric storage capacities, i.e., 7.6 wt. % H2 and 111 kg H2/m , respectively. Although MgH2 requires a high temperature to release H2 due to its high heat of formation, this may be overcome by adding Si to form Mg2Si; however, the hydrogenation reaction (i.e., the reverse reaction of releasing H2) has not been shown to readily occur, even at pressures up to 100 bar of H2 and at a temperature of 150 0C. This is likely in part due to kinetically unfavorable H2 dissociation on the Mg2Si surface. In addition to the Mg-based metal hydrides, complex hydrides including alanates ([AlH4] ) have recently gained attention as alternative hydrogen storage materials. Examples include NaAlH4, LiAlH4, KAlH4, Mg(AlH4)2, Na3AlH6, Li3AlH6, and Na2LiAlH6. Many of these materials have been known to release H2 upon contact with water with highly irreversible hydrolysis reactions, a process known as “one-pass” hydrogen storage. 6,7 For example, Mg(AlH4)2 can exothermically dehydrogenate at 1630C, but its direct rehydrogenation is not thermodynamically favorable. 8 As another example, LiAlH4 shows similar kinetic and thermodynamic barriers when it is rehydrogenated. The first dehydrogenation of LiAlH4 is an exothermic process with an approximate ∆H of –10 kJ/(mol of H2), 9 but its reverse process proceeds endothermically with ∆H of 9 kJ/(mol of H2) by which direct hydrogenation does not occur. To overcome the kinetic limitation to reversibility of hydrogenation process, alanates doped with titanium have been suggested not only to achieve kinetically enhanced dehydrogenation, but also to make the process reversible. For example, the reversible dehydrogenation of LiNa2AlH6 to LiH, 2NaH, and Al could be achieved by doping Ti with the complex hydride. Another well-known method for overcoming the kinetic barriers in the hydrogenation of complex hydrides is nanostructuring and nanocatalysis.A DFT-based theoretical study, for example, showed that as the MgH2 cluster size decreases below 19 Mg atoms, MgH2 becomes more destabilized, resulting in a significantly lower hydrogen desorption energy. An experimental study also found that upon size restriction of nanoparticles of NaAlH4, LiAlH4, and LiBH4, a drastic enhancement of the hydrogen desorption properties can be achieved. In nanoparticles of complex hydrides, predicting what phases would be more stable as a function of nanoparticle size may facilitate the rational design of nanostructured complex hydrides for hydrogen storage. In the current work, as a starting point toward understanding reversible dehydrogenation and rehydrogenation process of the complex hydrides, we construct a phase stability diagram of Mg(AlH4)2 nanoparticles as a function of particle size and temperature. For this, the following three steps were conducted: i) first-principles calculations for total energies of a series of configurations of complex hydride nanoparticles, ii) construction of a cluster expansion parameterized by the total energies calculated in step i), which enables rapid calculation of the total energy as a function of particle size and shape, and iii) thermodynamics predictions equipped with the cluster expansion-calculated properties of nanoparticles to calculate equilibrium phase boundaries. Methodology Nanoparticle decomposition depends on the composition and the size of the reaction products. To evaluate the energy of the reaction products as a function of nanoparticle size, we first directly determined the electronic structures and energies of a series of configurations of fully relaxed nanoparticles [Mg(AlH4)2, MgH2, Al, and Mg] with diameters up to ~2.3 nm by using density functional theory (DFT) calculations. DFT calculations were performed by using the Vienna Ab initio Simulation Package (VASP) with the projector-augmented wave (PAW) method. The Perdew and Wang (PW91) 22 generalized gradient approximation (GGA) exchange-correlation functional was used. A kinetic energy cutoff of 250 eV was used with a plane-wave basis set. The integration of the Brillouin zone was conducted at the Γ-point using first-order Methfessel-Paxton smearing with a width of 0.2 eV. Based on the DFT-calculated energies and structures of the nanoparticles, we employed the cluster expansion-based approach employed by Mueller and Ceder to predict the free energies of larger nanoparticles with diameters up to 100 nm. In this approach, the structure of the particle is represented on a lattice of sites, where each site can be either occupied or vacant. Site variables are assigned to each site, where the variable takes a value of +1 for an occupied site and -1 for a vacant site. The total energy of a nanoparticle configuration is then expanded as linear combination “cluster functions” as shown in Eq. (1). E s ( ) = Vcluster si