Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Computational search of anion redox Li-ion battery composition space

Dan Davies
February 28, 2021

Computational search of anion redox Li-ion battery composition space

MRS Fall Meeting 2019, EN02.

Dan Davies

February 28, 2021
Tweet

More Decks by Dan Davies

Other Decks in Research

Transcript

  1. Computational search of anion redox
    Li-ion battery composition space
    Dan Davies
    MRS Fall Meeting EN02
    2nd December 2019
    @danwdavies

    View full-size slide

  2. Redox processes in metal oxides
    M O
    M redox: Mn à Mn+1
    Most battery cathodes
    O redox: O2- à O1-
    2p6 à 2p5
    Localised “oxygen hole”

    View full-size slide

  3. Oxygen redox
    Proc. Roy. Soc. A, 231, 404 (1955)

    View full-size slide

  4. Oxygen redox
    Proc. Roy. Soc. A, 231, 404 (1955)
    S = ½
    hole on O
    R. Schnadt, J. Schneider, Phys. Kondens. Materie, 11, 19 (1970)

    View full-size slide

  5. Anion redox processes
    M redox: O redox:
    Mn à Mn+1 O2- à O1-
    “Anion redox”

    View full-size slide

  6. High-throughput materials screening
    LOW COST
    HIGH COST
    Band gap
    Electron energies
    Likely crystal
    structure
    Sustainability …
    Thermodynamic
    stability
    Dynamic stability
    Electronic structure
    New energy materials
    Chemical heuristics
    Data-driven models
    Automated
    calculations
    (DFT, hybrid DFT
    and beyond)

    View full-size slide

  7. Anion redox cathodes
    1. Electrostatic environment of O
    From Madelung potentials
    2. Orbital overlap
    From electronic density of states
    analysis

    View full-size slide

  8. LiMO2
    and Li2
    MO3
    systems
    Li
    O
    E.g:
    LiCoO2
    LiNiO2
    E.g:
    Li2
    RuO3
    Li2
    MnO3

    View full-size slide

  9. Electrostatic environment of O
    Madelung site potential:
    In practice calculated
    using Ewald method:

    View full-size slide

  10. O Madelung potentials
    Oxygen is very sensitive to its crystal environment: High
    variation in potential within the same crystal structure type
    1st row
    2nd row 3rd row

    View full-size slide

  11. Madelung potentials: local O
    environment
    In a disordered supercell: Madelung site potential decreases when
    changing the local O coordination environment from TM à Li

    View full-size slide

  12. Madelung potentials: local O
    environment
    In a disordered supercell: Madelung site potential decreases when
    changing the local O coordination environment from TM à Li

    View full-size slide

  13. Hole localization: the ionic picture
    Atomic contributions Structural contributions
    d0 Strongly favour O
    holes
    Increasing d-electrons
    disfavours M holes (in
    line with M ionisation
    energy)
    J. B. Torrance, R. M. Metzger, Role of the Madelung energy in hole conductivity of
    copper oxides, Phys. Rev. Lett. 1989

    View full-size slide

  14. Orbital overlap from DFT calculations
    E
    M
    O
    U
    Δ
    Generate possible
    spin orderings
    Full relaxation
    DOS calculation
    for lowest energy
    structure
    1M. K. Horton et al., High-throughput prediction of the ground-state collinear magnetic
    order of inorganic materials using Density Functional Theory, npj Comp. Mater. 2019
    , , …
    Magnetic ordering1 Hybrid DFT
    • HSE06 (25% screened HF
    exact exchange)
    • 64 Å3 K-point density
    • 600 eV energy cut off

    View full-size slide

  15. Orbital overlap
    −4 −2 0 2 4
    (nergy (eV)
    ArE. unitV
    TotDO D2S
    0g (V)
    2 (S)
    −4 −2 0 2 4
    (nergy (eV)
    ArE. unitV
    TotDO D2S
    Cu (d)
    2 (S)
    MgO VBM is ~100% O
    p-states
    Cu2
    O VBM is ~90% Cu
    d-states
    At the two extremes for binary oxides, the VBM can be
    dominated by O p-states or M d-states

    View full-size slide

  16. At the two extremes for binary oxides, the VBM can be
    dominated by O p-states or M d-states
    −4 −2 0 2 4
    (nergy (eV)
    ArE. unitV
    TotDO D2S
    0g (V)
    2 (S)
    −4 −2 0 2 4
    (nergy (eV)
    ArE. unitV
    TotDO D2S
    Cu (d)
    2 (S)
    LiCoO2
    VBM is ~40% O p-states, ~60% Co
    d-states
    −4 −2 0 2 4
    (nergy (eV)
    ArE. unitV
    TotDO D2S
    Co (d)
    2 (S)
    Orbital overlap

    View full-size slide

  17. p-DOS analysis: LiMO2
    vs Li2
    MO3
    LiWO2
    Li2
    WO3
    −2 −1 0 1 2
    (nergy (eV)
    ArE. uniWV
    ToWDO D2S
    W (d)
    2 (S)
    −2 −1 0 1 2
    (nergy (eV)
    ArE. unitV
    LiRuO2
    Li2
    RuO3
    −2 −1 0 1 2
    (nergy (eV)
    ArE. unitV
    −2 −1 0 1 2
    (nergy (eV)
    ArE. unitV

    View full-size slide

  18. p-DOS analysis: LiMO2
    vs Li2
    MO3
    20%-90% range
    Increase in
    p-character across
    each TM row
    (Large JT effect for
    Fe and
    reorganisation to
    square planar for Pt)
    Fraction of O p-states in upper 1 eV of VBM
    1st row 2nd row 3rd row

    View full-size slide

  19. Summary and ongoing work
    • Structural and electronic factors influence
    anion redox processes in metal oxides
    • Qualitative trends can be extracted from
    electrostatic considerations
    • The fraction of O p-character can vary from
    20% – 90%
    • The materials design challenge is to tune the
    fraction deterministically

    View full-size slide

  20. Acknowledgements
    • Aron Walsh
    • David Scanlon (UCL)
    • Keith Butler (Sci-ML, STFC)
    • Benjamin Morgan (Bath)
    • Alex Squires (Bath)

    View full-size slide

  21. Extra: Hole localization: the ionic picture

    View full-size slide