DFT modeling of novel materials for hydrogen storage

Transcription

1 DFT modeling of novel materials for hydrogen storage Tejs Vegge 1, J Voss 1,2, Q Shi 1, HS Jacobsen 1, JS Hummelshøj 1,2, AS Pedersen 1, JK Nørskov 2 1 Materials Research Department, Risø National Laboratory, DTU 2 CAMD, Department of Physics, NanoDTU, Technical University of Denmark (DTU)

2 Outline Complex storage materials new challenges Density Functional Theory (DFT) and computational tools Examples of research using DFT calculations hydrogen desorption from NaAlH 4 nanoparticles hydrogen diffusion and the role of Ti-based catalysts characterization of new borohydride super structures screening new borohydride alloy systems metal ammines for indirect hydrogen storage Perspectives on the role of DFT calculations

3 Complex storage materials new challenges Conventional metal hydrides: good properties; too low capacity Novel materials with significantly higher capacities complex metal hydrides: NaAlH 4, LiBH 4 and Mg(BH 4 ) 2 ammonia based metal salts: LiNH 2 and Mg(NH 3 ) 6 Cl 2 Challenges: thermodynamics, kinetics and reversibility Production and characterization is challenging difficult synthesis, reactive, explosive and toxic materials characterization of light elements, insulators, reactive compounds Fundamental storage problems are well suited for DFT calculations catalysis, hydrogen desorption and diffusion structural information and thermodynamic stability

4 Calculational schemes for hydrogen storage materials A constant trade-off between accuracy and system size/time Specific processes require different levels of accuracy mass and heat transfer, fluid dynamics (analytical methods) grains, grain boundaries (semi-empirical potentials) breaking chemical bonds, formation/surface enthalpies (DFT) van der Waals forces, semi-conductor band gabs (full-ci) The electronic interactions are essential for complex hydrides (DFT) Replace the many body electronic wavefunction with elec. density Computationally demanding (~10 2 atoms and pico second MD) Going from ground state energies (0K) and the time-scales of atomic vibrations to experimental conditions.

5 Simulating rare events Rate-limiting processes (e.g. hydrogen dynamics) occur on time scales that are very different from the computations Most relevant events are rare Adapted from Henkelman (2006)

6 From ground state energies to finite temperature behavior The free energy is essential for the stability of most storage materials G(T) = E 0k elec+ E vib (T) T*S vib (T) + E trans (T) + E rot (T) + P*V The vibrational energy and entropy can be obtained from the phonon DOS The phonon frequencies are determined from the dynamic matrix, which is calculated from the Hellmann-Feynman forces E vib (T)= Ke and Tanaka, PRB 94, (2005) S vib (T)=

7 Examples of recent DFT-based research Hydrogen desorption from NaAlH 4 - particle shape and Ti-based catalysts

8 Particle properties and desorption (NaAlH 4 /LiAlH 4 ) Hydrogen desorption vs. particle size Determining the particle shape Barriers for H 2 desorption Catalytic bonding sites (e.g. Ti) Andreasen, Vegge and Pedersen, J. Solid State Chem 178, 3664 (2005) Vegge, Phys. Chem. Chem. Phys. 8, 4853 (2006) Lohstroh and Fichtner, Phys Rev. B 75, (2007)

9 Ti-based dopants for complex hydrides Possible catalytic roles of Ti-dopants improved ab-/desorption of hydrogen improved hydrogen diffusion improved mass transport Wang et al. J. Alloys Compd. 391, 245 (2005) Løvvik and Opalka, Appl. Phys. Lett. 8, 4853 (2006) Marashdeh, Olsen, Løvvik and Kroes, Chem. Phys. Lett. 426, 180 (2006) Vegge, Phys. Chem. Chem. Phys. 8, 4853 (2006) B.v. Colbe et al., Angew. Chem, Int. Ed. 45, 3663 (2006)

10 Hydrogen desorption barriers Open surfaces and Ti-dopants lower desorption barriers TiCl 3 facilitates formation of Na-vacancy and lowers the barrier for H 2 desorption significantly Strong effect in the near surface region. Catalytic bulk effects are required to explain the full effect of Ti (001) Ti@Na 2v (001) (110) (001) Ti@Na 2v (001) Vegge, Phys. Chem. Chem. Phys. 8, 4853 (2006)

11 Hydrogen diffusion processes in NaAlH 4 /Na 3 AlH 6 (QENS) Quasielastic neutron scattering as a probe for hydrogen dynamics Four samples (30-400K): NaAlH 4 and Na 3 AlH 6 (un-/ticl 3 -doped) No hydrogen diffusion observed below 250K Limited hydrogen dynamics prior to the onset of desorption Voss, Shi, Jacobsen, Zamponi, Lefmann and Vegge, J Phys Chem B 111, 3886 (2007) Shi, Voss, Jacobsen, Lefmann, Zamponi and Vegge, J Alloys Compd , 469 (2007)

12 Hydrogen dynamics in NaAlH 4 (DFT) Vacancy mediated hydrogen dynamics Barrier for hydrogen diffusion in NaAlH 4 : E act = 0.31 ev Barrier for localized hydrogen dynamics: E act = 0.44 ev Hydrogen is trapped by Ti in doped NaAlH 4 Shi, Voss, Jacobsen, Lefmann, Zamponi and Vegge, J Alloys Compd , 469 (2007)

13 Hydrogen diffusion in Na 3 AlH 6 (DFT) Vacancy mediated hydrogen dynamics Long range hydrogen diffusion: E act =0.75 ev, l =3.0 Å Local hydrogen dynamics (on the AlH 5 -complex): E act =0.41 ev, l=2.5å Titanium can lower vacancy formation energy marginally Voss, Shi, Jacobsen, Zamponi, Lefmann and Vegge, J. Phys. Chem. B 111, 3886 (2007).

14 Elastic Incoherent Structure Factor (EISF) NaAlH 4 : q-independence long range diffusion Na 3 AlH 6 : q-dependence localized dynamics Shi, Voss, Jacobsen, Lefmann, Zamponi and Vegge, J Alloys Compd , 469 (2007)

15 Mobile species in NaAlH 4 and Na 3 AlH 6 Na-vacancy diffusion in Na 3 AlH 6 (0.12 ev) Resent muon work suggests AlH 4- defects AlH 3 formation and diffusion in NaAlH 4 Palumbo et al. J. Phys Chem B 109, 1168 (2005) Lohstroh and Fichtner, Phys Rev. B 75, (2007) Kadono et al. Phys Rev (in press)

16 Borohydrides: crystal structures, analysis and prediction

17 Structural analysis and predictions from DFT Structural stability analysis from the phonon dispersion (free energy) Many phases are separated by small energy differences Calculating phase transition temperatures for known materials (LiBH 4 ) Low barrier rotations and dynamical disorder should be considered (harmonic approx.) Tool in prediction of new structures Łodziana and Vegge, Phys. Rev. Lett. 93, (2004) Łodziana and Vegge, Phys. Rev. Lett. 97, (2006)

18 Mg(BH 4 ) 2 : symmetry constrained optimization and annealing Experiments agree on very large unit cells of Mg(BH 4 ) 2 What is the significance of these super structures (P6 1 and Fddd*)? DFT calculations require symmetry constrained annealing and optimization: mapping the atomic degrees of freedom Wyckoff positions P6 1 Her et al., Acta Cryst. B63, (2007) Cerny et al., Angew. Chem. Int. Ed. 46, 5765 (2007)

19 Symmetry constrained annealing/structure relaxation For spacegroup #70 Mg(BH 4 ) 2, symmetry constraints reduce number of ionic degrees of freedom from 2109 to 62 Symmetry constrained relaxation of magnesium borohydride 704- atoms-unit cell

20 Symmetry constrained annealing/structure relaxation Spacegroup #70 Mg(BH 4 ) 2 supposed to have disorder with half a unit cell correlation length along the a-axis Gradient projection allows inclusion / relaxation of shift as additional degree of freedom:

21 Electronegativity heat of formation of metal borohydrides Search for simple descriptors (electronegativity ) LiBH 4 is too stable destabilization by alloying? 3d-transition metals and Li in a 1:1 ratio Lithium Scandium Borohydride Nakamori et al., Phys Rev B 74, (2006)

22 Metal ammines (hydrogen pill) Storing hydrogen indirectly as NH 3 in a salt (MgCl 2 )

23 Thermal decomposition of metal ammines - Mg(NH 3 ) 6 Cl 2 4/6 1/6 1/6 4+2/8 1/8 1/8 Christensen et al., J. Matter. Chem. 15, 4106 (2005) Sørensen at al.(2007)

24 Decomposition pathways for metal ammines Determining the decomposition pathway (different for Mg and Ca) Reproduce the experimental binding energies Lowering of the residual desorption temperature by alloying Vegge et al. Woodhead Publishing (2007)

25 Perspectives on DFT modeling in hydrogen storage DFT calculations can provide a better understanding of the interactions between the catalysts and storage medium Integrated experimental and DFT work on specific processes, e.g. hydrogen dynamics and ab-/desorption Assist in the determination of novel complex structures Use DFT calculations to design new hydrogen storage materials with optimized thermodynamic properties Model systems can give reliable finite temperature properties Fast screening studies selecting materials for synthesis