Hydrogen Storage

Our calculations of total energies and (especially) kinetic properties require accurate crystal structures. However, the positions of hydrogen atoms are not measurable in standard X-ray diffraction experiments and, in some cases, neither are the positions of the metal atoms. Therefore, we necessarily employ crystal structure prediction methods in an effort to search for ground state structures.

When attempting to predict crystal structures, we often use a method called the prototype electrostatic ground state (PEGS) search. The code that we use, published by Eric Majzoub and Vidvuds Ozolins,¹ takes advantage of the ionic nature of complex metal hydrides by fixing the local arrangement of complex anion groups (for example, the AlH₄ group in NaAlH₄). Furthermore, it uses a very simple energy model consisting of electrostatic interactions and hard sphere repulsions (when atoms overlap) to quickly search for low-energy structures. We have had great success in using the PEGS method to predict completely unknown crystal structures, which we then refine using more accurate DFT calculations - a couple examples are show to the left.

Recently, we have used a genetic algorithm to search for ground state structures using energies directly from DFT.² This method has proven to be especially useful in difficult systems; for example, MgNH, where the ground state structure was exactly reproduced. We are still working to improve this method and to combine it with experimental measurements to efficiently and effectively find ground state crystal structures.

1. E.H. Majzoub and V. Ozolins, Phys. Rev. B. 77, 104115 (2008).

2. B. Meredig and C. Wolverton, Nat. Mat. 12, 123 (2012).

In studies of the kinetics of hydrogen storage reactions, we attempt to understand the atomic level processes that govern the rates at which hydrogen is released from (or absorbed into) complex metal hydrides. These studies have become increasingly important with the realization that many of the more promising storage materials have unacceptably slow reaction rates; in other words, your hydrogen-powered car will accelerate very slowly and will take a very long time to refill. However, an increased understanding of the reaction kinetics will significantly aid in the design of better catalysts and help to guide research away from materials that may be kinetically limited.

In our group, we have focused on two areas related to reaction rates: first, the interaction of H₂ with metal surfaces where it dissociates into atomic hydrogen; second, the formation and diffusion of bulk point defects that facilitate the mass transport. As an example of the latter, we have studied point defects in B₂₀H₁₆, a molecular crystal that is predicted to release hydrogen at a nearly ideal temperature and pressure by decomposing into pure boron and H₂.¹ In the figure to the left, we show this diffusion network for interstitial H₂ where polyhedra represent clusters of B₂₀H₁₆, red circles are stable interstitial sites for H₂, and red lines shown the allowed paths for jumps of H₂ between stable sites. Using these kinds of networks, we are able to calculate the diffusivity of point defects in complex metal hydrides and predict which will have suitable mass transport rates.

We are also actively working to understand the dissociation and diffusion of hydrogen on metal surfaces, which is a key step in the rehydrogenation of metal hydrides. These processes are especially responsive to catalysis, but this is an area that requires further study before we can begin to design new catalysts.

1. W.Q. Sun, C. Wolverton, A. Akbarzadeh, and V. Ozolins, Phys. Rev. B 83, 064112 (2013).