PowerPoint-Pr

1
Experimental Investigation and Simulation of
Oxygen Transport in SOFC Materials

1. Motivation and systems ZrO2, LaGaO3, LaMnO3
2. Experimental Tracer diffusion in electrolyte
3. Experimental Tracer diffusion in LSM/YSZ pair
4. Modelling Static lattice gt migration mechanism
5. Modelling Molecular dynamics gt diffusion

Martin Kilo, Christos Argirusis, Günter
Borchardt, Rob A. JacksonTU Clausthal, Institut
für Metallurgie, Robert-Koch-Str. 42D-38678
Clausthal-Zellerfeld, Germany Keele University,
School of Physics and Chemistry, Keele, Staffs
ST5 5GB / UK
2
Motivation Oxygen mixed and ionic conductors

• Most common examples doped ZrO2 or doped
  perovskites, e.g. (La0.8Sr0.2)(Ga0.8Mg0.2)O3-d
  , LSGM, LaxSr1-xMnO3-d, LSM
• Doping with aliovalent cations leads to fast
  oxygen diffusion, but usually to slow cation
  diffusion

n 1 YZr1'
n 2 CaZr2'
sT A(x) exp(-Ea(x) / RT)
x' ? 0.08 0.12
3
Open questions

• Experimental
• What is the practical connection between the
  experimental oxygen diffusion coefficient and
  conductivity?
• Oxygen diffusion under applied electrical field
• Influence of thermal ageing on oxygen diffusion
• Simulation of oxygen diffusion
• Static lattice Mechanism of transport
• Molecular dynamics Transport coefficients
• Finite element modelling Simulation of real
  systems

4
Gaseous tracer 16O / 18O
gas tracer
Nat. 0.2 18O Tracer gt90 18O Surface
limited x(18O,x0) lt 90
c(x,t)-c0(cs-c0)(erf(x/2(DOt)0.5)-exp(hxh2
DOt)erf(x/2(DOt)0.5h(DOt)0.5))
5
Isotopic lateral and depth distribution SIMS
Analysis
VG SIMS-lab
Quadrupole detection 7 kV Ar Detection of
positively/negatively charged ions Charge
compensation with flood gun
Cameca 3f/5f
Magnetic sector field gt10 kV O/- Detection of
positively charged ions Charge compensation by
conducting layer
6
ZrO2-systems CSZ, ScSZ, YSZ working conditions
CaO-ZrO2 (Duran, J.Mat.Sci. 22 1987 4348)
Sc2O3-ZrO2 (Ruh, J.Am.Ceram.Soc. 60 1977 399)
Y2O3-ZrO2 (Suzuki, SSI 81 1995 211)
Working regions
All ZrO2-systems have a cubic part of the phase
diagram with fast oxygen transport and slow
cation transport
Cation diffusion
Oxygen diffusion
(red line T 1000 C)
Mostly single crystals
7
Oxygen transport Self diffusion
Oxygen Diffusion- Maximum in D(x) like sT, MD -
?H not strongly dependent on Y2O3 content -
Haven ratio no simple T-function - Fuel cell
Field, ageing
8
Oxygen diffusion Activation enthalpies and ageing
Methods ml mechanical loss DC dielectrical
conductivity DO self diffusion dl
dielectric loss
Ageing Preannealing decreases Oxygen diffusion
coefficient for x(Y2O3) ? 8mol
9
Static oxygen diffusion Summary

• Different experimental methods reveal different
  information
• Self diffusion
• Activation enthalpy of oxygen diffusion lowest
• Oxygen diffusion is dependant on thermal history
• Oxygen diffusion under working condition of SOFC
  ?
• Conductivity
• Conductivity nonlinear gt association. What
  are the contributions of association and
  migration?
• Mechanical loss
• What is the difference between local and
  diffusive jumps?

10
Oxygen incorporation into SOFC electrolyte
Three possible mechanisms - 3 phase boundary
(3PB) - electrode surface - through electrode
2PB
Surface diffusion
LSM
300nm
2PB
3PB
ZrO2
20 µm
20 µm
11
Model system for SOFC electrode/electrolyte
PLD of LSM on YSZ at 800 C
LSM stripes 20 µm wide LSM layer 300 nm thick
12
Experimental setup
Pt contact
LSM structured cathode
YSZ
Pt ink referenceelectrode
Pt ink counter electrode
13
Oxygen exchange in cathode / YSZ
FEM calculation of oxygen distribution after
diffusion from a line source
FEMlab
LSM
YSZ
Assumptions Line source at the 2PB DYSZ gtgt
DLSM k1 at 2PB (LSM/YSZ) ? k2 at 2PB (18O/LSM)
? k3 at 2PB (18O/YSZ) 0
14
Experimental results LSM/YSZ-10
15
Depth profile analysis
Oxygen content under dense LSM, LSM stripe, free
YSZ
LSM surface, dense, unstructured
Crater 200x200 µm
on the YSZ (LSM free area)
on the LSM stripe
16
18O content Variation of overpotential
17
Oxygen diffusion under field Summary

• The bulk path seems to be very sensitive
  regarding the applied cathodic overpotential.
  Even at low cathodic overpotentials, the bulk
  path is blocking.
• The 3PB is more active at low cathodic
  overpotentials. The higher the cathodic
  overpotential, the more inactive becomes the
  3PB.
• The solid/solid interface-resistance is clearly
  visible with SIMS and depends on the applied
  overpotential.

18
ZrO2 Modelling oxygen migration

• Migration energies, hopping energies, migration
  pathways
• Association energies

19
Migration pathway from static lattice calculations
Code GULP (J. Gale, London)
- Single jump between two vacancies in undoped
ZrO2 ?E(O2-) lt 0.2 eV - Equilibrium position
of O2- ion (0.333,0.25,0.25)
20
Association energy from supercell calculations
Supercells of 444 unit cells, varying Y/Zr
content Association energy difference between
supercell lattice energy and perfect lattice
energies
21
Summary static lattice calculations

• Results
• Low migration energy, high association energy
• Oxygen vacancies affects local oxygen
  surroundings
• Limitation of static lattice calculations
• Calculation of one single jump
• Assumption of a perfect or at least well-defined
  surrounding
• Temperature effects difficult to describe
• Molecular dynamics
• Information as function of temperature and time
• More realistic description of highly disordered
  systems
• Trajectory allows conclusions on jump mechanisms
• But Slow diffusion difficult

22
Oxygen diffusion Molecular dynamics on YSZ
Cubic unit cell
Jumps between one or two Y ions are less likely
than between two Zr ions ? Restricted diffusion
path for high dopant level
23
MD Oxygen diffusion coefficient in YSZ

• Maximum in D similar to experimental point,
  but higher values of D
• Like experimental observed, ?H independent of
  x(Y2O3) .
• At high x(Y2O3), D independent of x(Y2O3)

24
MD Oxygen diffusion coefficient in LSGM-8282
Activation enthalpy close to the experimental
values Diffusion goes along (110) direction
25
MD Oxygen diffusion coefficient in ULSM
Two activation enthalpies due to local hopping
Sketch of migration pathway along (100) T
1200K, 1250 ps green oxygen pink, grey La,
Sr red Mn
26
Summary of computer simulation results

• Static lattice calculations
• Migration energies too low
• Supercell method good estimation of association
  energies
• ? What are the limitations ?

• Molecular Dynamics calculations
• Diffusion coefficients similar to the experiment
• Activation enthalpies of O almost identical
• ? Percolation network ?

27
Conclusions I Experimental results

• Static oxygen diffusion experiments
• Activation enthalpy of oxygen diffusion lowest
• Oxygen diffusion is dependant on thermal history

• Oxygen diffusion under SOFC conditions
• Even at low cathodic overpotentials, the bulk
  path is blocking
• The 3PB is less active at high cathodic
  overpotential

? How are the diffusivities affected ?
28
Conclusions II Modelling results

• Static lattice calculations O/ZrO2
• Estimate of association energies using
  supercells
• Molecular dynamics on YSZ
• Diffusion coefficients and activation energies
  are close to the experimental values
• Existence of percolation pathways?
• Molecular dynamics on LSGM, ULSM
• Oxygen migration only along (110)
• Localised jumps according to the cation
  surrounding of A- and B-sublattices

29
Future perspectives

• Dynamic oxygen diffusion
• YSZ/ULSM Variation of time, polarisation, p(O2)
• Variation of the cathode material
• Oxygen exchange coefficient at the solid/solid
  interface?
• Anode/Electrolyte Hydrogen
• Computer simulations
• Atomistic modelling of oxygen transport across
  interfaces solid/solid and gas/solid
• Modelling of oxygen transport under electrical
  field
• Other materials LSCF, Apatites
• Advanced methods QM, finite elements

30
Acknowledgements

• M. Weller, MPI Stuttgart Experimental
  results
• Prof. P. Schmidt, TU Darmstadt Use of
  computer centre
• Deutsche Forschungsgemeinschaft (DFG)
  Financial support