First principles modellingfor predicting and

understanding novel complex hydrides

Ole M. LøvvikCentre for Materials Science and Nanotechnology, University of Oslo, Norway

Institute for Energy Technology, Kjeller, Norway

FuncHy 2006:

5.3

3.7

2.9

5.6

3.0

2.0

7.0

3.0

2.7

1.9

1.4

3.0

0.0 2.0 4.0 6.0 8.0

LiAlH4

NaAlH4

KAlH4

Li3AlH6

Na3AlH6

K3AlH6

Mg(AlH4)2

Ca(AlH4)2

LiAlH4

NaAlH4

KAlH4

Li3AlH6

Na3AlH6

K3AlH6

Mg(AlH4)2

Ca(AlH4)2

Accessible hydrogen wt%:Alanates for hydrogen storage

Crystal structurepredictions

Spacegroup

# Type # atoms Model system

Ti3NiS6

K3Fe(CN)6

P21/n 14 Monoclinic 20 Na3AlF6 (α)Li3AlF6

Na3AlF6 (β)

Ti3NiS6

Fm-3m 225 Cubic 40 K3MoF6

P-1 2 Triclinic 20P21/n 14 Monoclinic 20

Pna21 33 Orthorombic 40

Immm 71 Tetragonal 20

R-3 148 Rhombohedral 20

E.g. Na3AlH6: Seven representative systemsFull relaxation; allows to leave space group/symmetry

Calculation detailsBand-structure DFTVASPGeneralized gradient appr. (GGA) PW91Projector augmented wave (PAW) methodSpin polarization allowedCut-off energy 780 eVOverall convergence: ~1 meV (0.01 kJ/mol)/unit cell

Predicted crystal structuresCompound Predicted

structureExpt. structure

LiAlH4 P21/c P21/cNaAlH4 I41 /a I41 /aKAlH4 Pnma PnmaLi3AlH6 R–3 R–3Na3AlH6 P21/n P21/nK3AlH6 P21/nMg(AlH4)2 P–3m1 P–3m1Ca(AlH4)2 Pbca

O. M. Løvvik, S. M. Opalka, H. W. Brinks, B. C. Hauback, Phys. Rev. B 69 (2004) 134117)O. M. Løvvik, O. Swang, Europhys. Lett. 67 (2004) 607.O. M. Løvvik, Phys. Rev. B. 71(2005) 144111.O. M. Løvvik, O. Swang, J. Alloys Comp. 404-406 (2005) 757-761.O. M. Løvvik, P. N. Molin, Phys. Rev. B. 72 (2005) 073201.O. M. Løvvik, O. Swang, S. M. Opalka, J. Mater. Res. 20 (2005) 3199 (Invited paper).

Thermodynamic stabilityFormation enthalpy:Hform(MnAlH(n+3)) =

E(MnAlH(n+3)) – n E(M) – E(Al) – (n+3)/2 E(H2)

Compound LiAlH4 NaAlH4 KAlH4 Li3AlH6 Na3AlH6 K3AlH6

Formation enthalpy (kJ/mol H2) -55.5 -54.9 -70.0 -102.8 -69.9 -78.5

Ca(AlH4)2

Ca

AlH

Non-layered structure

AlH4 tetrahedra,Al-H distance 162 pm

Ca 8-coordinated,Ca-H distance 220 pm

H wt. density: 5.9%H vol density: 70 kg/m3

Space group: Pbca

• O. M. Løvvik, Phys. Rev. B. 71 (2005) 144111.

-5 0 50

0.1

0.2

E (eV)

DO

S

H

-5 0 50

50

100

E (eV)

Total DOS

0

1

2

3Ca

0

0.5

1

Al

Ca(AlH4)2 density of states (DOS)

Thermodynamic stabilityFormation enthalpy:

Reaction enthalpy:

kJ/mol H2 Formationenthalpy

Reactionenthalpy

Mg(AlH4)2 -21.1 -6.2

Ca(AlH4)2 -59.4 -20.7

This is whyMg(AlH4)2is not reversible ...

Why are not Li and Mg alanates reversible?

Na and K alanates exhibit reversible hydrogenation, Li and Mg alanates not.Indicators of reversibility from calculations?Electronic structure: no clear signs.Crystal structure: Li3AlH6 stands out.Thermodynamic stability: Li3AlH6 stands out.The reason: Li small enough for Li3AlH6 to attain the R–3structure, making Li3AlH6 too stable.Similarly: Mg alanate too unstable.No easy generalization to other materials.

O. M. Løvvik, O. Swang, S. M. Opalka, J. Mater. Res. 20 (2005) 3199

O. M. Løvvik, O. Swang, J. Alloys Comp. 404-406(2005) 757-761

Prediction of new mixedalanate phases

LiAlH4

KAlH4NaAlH4

Li0.75Na0.25AlH4

Li0.5Na0.5AlH4

Li0.25Na0.75AlH4

K0.25Li0.75AlH4

K0.5Li0.5AlH4

K0.75Li0.25AlH4

K0.75Na0.25AlH4K 0.5Na 0.5

AlH 4

K 0.25Na 0.75A

lH 4

”Tetrahydrides”• O. M. Løvvik, O. Swang, Europhys. Lett. 67 (2004) 607.

Mixed tetrahydrides:

Prediction of new mixedalanate phases

Li3AlH6

K3AlH6Na3AlH6

Li2NaAlH6

LiNa2AlH6

KLi2AlH6

K2LiAlH6

Li2.5Na0.5AlH6

Li1.5Na1.5AlH6

Li0.5Na2.5AlH6

K2.5Li0.5AlH6

K1.5Li1.5AlH6

K0.5Li2.5AlH6

K2.5Na0.5AlH6K 2NaA

lH 6

K 1.5Na 1.5

AlH 6

KNa 2AlH 6

K 0.5Na 2.5

AlH 6

”Hexahydrides”

Mixed hexahydrides:

K2NaAlH6

K2LiAlH6

LiNa2AlH6

Phonon calculationsVibration frequencies from displacementcalculationsPhonon spectrum and partition functionIntegrated phonon density → Thermodynamics

P21/n: stable phase selectedfor thermodynamic predictions

Fm-3m: imaginary frequenciesinstability to transformation

Predicted Phonon Dispersion RelationsNa2LiAlH6 P21/n and Fm-3m Structures

Freq

uenc

y (T

Hz)

40

20

0

W WL Γ X KC BΓ X ZY A

40

20

0

10

30

Brillouin Zone Directions Brillouin Zone Directions

Thermodynamicpredictions

Na

Li

NaAl

activity(H)=3.9x10-3

300 K 100 Bar

AlNaAlH4

Li3AlH6

LiHNaAlH4

Li3AlH6

LiHNaAlH4

Na2LiAlH6

LiHNaHNa2LiAlH6

NaHNa3AlH6

Na2LiAlH6

NaAlH4

Na3AlH6

Na2LiAlH6

Optimum H reversibility{2Na : 1Li : 2Al : 9H}:

2NaAlH4+LiH300 K /100 Bar

Na2LiAlH6+Al300 K / 1 Bar

3/2 H2

2.6 wt%

2NaH+LiH+2Al400 K / 1 Bar

3/2 H2

2.6 wt%

Potential for 5.2 wt% Reversible H

5.3

3.7

2.9

5.6

3.0

2.0

7.0

3.0

4.7

2.6

2.7

1.9

1.4

3.0

2.6

0.0 2.0 4.0 6.0 8.0

LiAlH4

NaAlH4

KAlH4

Li3AlH6

Na3AlH6

K3AlH6

Mg(AlH4)2

Ca(AlH4)2

LiNa2AlH6

LiH+2NaAlH4

LiAlH4

NaAlH4

KAlH4

Li3AlH6

Na3AlH6

K3AlH6

Mg(AlH4)2

Ca(AlH4)2

LiNa2AlH6

LiH+2NaAlH4

Accessible hydrogen wt%:Alanates for hydrogen storage

Ti in NaAlH4 –experimental status

Nature of Ti additive not important (TiCl3, TiF3, Ti powder, Ti(OBu)4, etc.)Three different majority phases found:

Al3TiDispersed, amorphous Al1-uTiu (u = 0.07 – 0.15)TiH2

Samples with different Ti majority phases have comparablekineticsIs there an active minority phase? Doping or catalysis?Evidence for point defects (hydrogen vacancies), alanes(AlH3), and heterogeneous catalysis.

0-D (point) defects

Disregard majority phasesIf in the NaAlH4 lattice:

Ti substituting Al or Na, or at interstitial sites.Additional vacancies maybe produced.Surface or bulk?

Full relaxation; latticeconstants change?

O. M. Løvvik, S. M. Opalka, Phys. Rev. B. 71 (2005) 054103O. M. Løvvik, S. M. Opalka, Appl. Phys. Lett., 88 (2006) 161917

ResultsReference state dependent:

Gas phase: stable subst.Standard state: unstableAll relevant phases: highlyunstable substitution

Subsituted phases: significantchanges in cell parameters Unstable at least down to 3 mol% Ti.Least unstable at the surfaceO. M. Løvvik, S. M. Opalka, Phys.

Rev. B. 71 (2005) 054103O. M. Løvvik, S. M. Opalka, Appl. Phys. Lett., 88 (2006) 161917

The least unstable structureTi: not in the bulk.If in the lattice: near thesurface, replacing Al.Metastable at best.

Large local distortions.Increased Ti-H coordination from 4 to 8.Weak Ti-H bonds.

1-2D defectsin NaAlH4

Unknown if suchdefects exist.What are their energycosts?What are their effects?

NaAl

001

O. M. Løvvik, In prep.

1D: dislocations.2D: stacking faults, twins, anti phaseboundaries, grainboundaries.Unknown if suchdefects exist.What are their energycosts?What are their effects?

1-2D defects in NaAlH4

Relaxed structureVery similar to perfectlatticeMain difference: Na-Hcoordination reduced from 8 to 6Possible to observe?

Fault energy

00,10,20,30,40,50,60,7

0 2 4 6 8 10

Fault distance

Faul

t ene

rgy

(eV

)

Energy/unit cell

Energy/fault

Fault energy: 0.32 eV/fault = 0.21 J/m2

Ti substitutionnear faults

Investigated Al→Ti, Na →Ti, and interstitial TiInternal reference: Al →Ti most stableTi-H coordination: 8Results for Z = 6; 16.7 % Ti.Interpolated values for defect-free models

8.0

0.9

7.5Internalreference,energies in kJ/mol atom

3.1

0.9

1.2

Standard statereference,energies in kJ/mol atom

Defectspromote Ti substitution

Ti substitutionpromotesdefects

What about non-equilibrium structures?

Suppose that mono-dispersedTi is available.Investigated clusters withconstant number of atoms.⇒ reference is NaAlH4 + Ti.

A. Marashdeh, R. A. Olsen, O. M. Løvvik, G.-J. Kroes, Chem. Phys. Lett. 426 (2006) 180–186

The role of Ti?

End point: Na ↔ Ti.Na out of the structure.Start of decomposition?

The role of Ti?The ”zipper model”:

Ti acts as a slider.Catalysis, but not traditional.May be combined with H2splitting in the absorptionprocess.Ti substitution promoted by defects.

SummaryProbably no better alanate than NaAlH4.The LiH + NaAlH4 mixture interesting.Minority phase of Ti probably important in NaAlH4(de-)hydrogenation:

Ti substitution in bulk ruled out.Ti substitution promoted by defects.Non-equilibrium: Ti may enter lattice.Proposed model: zipper model with Ti-slider.

AcknowledgementsCoworkers:

Susanne Opalka, UTRCPeter Molin, UiOBjørn Hauback et al., IFERoar Olsen et al., Leiden UniversityOle Swang et al., Sintef

Funding: Nanomat, Norwegian Research Council