Oxygen’Exchange’in’Thin’Layers’of’SOFC’Cathode’ · 2014-07-28 · J. Electrochemical Society, 154, B439-B445 (2007). Chao, Gerdes, Kitchin, Salvador (CMU) Electrochemical

Carnegie MellonSalvador Research Group, h2p://neon.materials.cmu.edu/salvador/

Oxygen Exchange in Thin Layers of SOFC Cathode

1

Paul Salvador

Department of Materials Science and EngineeringCarnegie Mellon UniversityPi:sburgh, PA 15206

Lu Yan Philip Tsang K. R. Balasubramaniam Hui DuShanling Wang Robin Chao Lam Helmick

Sarthak Havelia Oleg Maksimov Joanna Meador

EISA vs SBA-15 HT

Hard Template EISA

Sintered particles vs.

Open pores

1um 1um

200 nm 200 nm

Funded by DOE -‐ SECA, Thanks to L. Wilson, W. Surdoval, B. White, P. Burke, Shailesh Vora

Wednesday, July 27, 11

http://neon.materials.cmu.edu/salvador/



Outline

2

• Utility of thin layers- Functioning SOFCs- Basic science- Fuel cell collaboration

• Microstructure and oxygen exchange- Orientation of relaxed layers- Dislocations and strain- Extended boundaries in polycrystalline layers- Free surfaces

• Conclusions




Carnegie MellonSalvador Research Group, h2p://neon.materials.cmu.edu/salvador/ 3

Infiltra>on: Surface Ac>ve nanopar>cles for SOFCs

What are the opQmal materials to use for infiltraQon?

What are the surface proper.es of cathode materials?

What are the mechanisms of enhanced performance / degradaQon?

What are the proper.es of nanosized par.cles?

What are the effects of the support on the proper.es of cathodes?

REVIE

W

www.advmat.de

cathode impedance decreased from greater than 3V ! cm2 to lessthan 0.5V ! cm2 upon application of a cathodic potential. Theresults with LCF–YSZ electrodes also rule out the possibility thatSr migration to the electrolyte interface might be responsible fordeactivation,[104] since no Sr was present in this electrode.

If the deactivation model we have presented is correct, thereare important implications. It is usually assumed that deactiva-tion is associated with solid-state reactions. Approaches used toavoid deactivation therefore generally involve inserting a barrierlayer to prevent contact between the perovskite and the electrolytelayers. Deactivation by solid-state reaction with the YSZelectrolyte undoubtedly occurs for some materials (e.g., LSCoand YSZ) and the use of barrier layers does help stabilize theperformance. However, the strategy needs to be quite different forstabilizing electrode performance when deactivation is due tosintering of the perovskite. One could attempt to prevent theformation of a dense layer (barrier layers may help achieve this ifthe interfacial energies between the perovskite and the barriermaterial prevent film formation) or to increase ionic conductivitythrough the perovskite layer. In either case, the approach tocathode development would be quite different from thosedesigned to avoid solid-state reactions.

3.3. LSCF–YSZ Electrodes

Sr-doped LaFe0.8Co0.2O3 (LSCF) can be considered a subset ofLSF. LSCF has better electronic conductivity than LSF and asimilar ionic conductivity.[104–106] It exhibits very good initialperformance as an SOFC cathode when a doped-ceria layer isused to separate it from the YSZ.[107,108] It is also used in thecurrent collector of some fuel-cell designs.[109] Very recently, workhas appeared in which an LSCF cathode was formed byinfiltration into porous YSZ.[110] Chen et al. have reported thatan infiltrated LSCF–YSZ electrode had an impedance as low as0.047V ! cm2 at 800 8C, comparable to the performance ofinfiltrated LSCo–YSZ and LSF–YSZ electrodes.[27,45] Unfortu-nately, the stability of infiltrated LSCF–YSZ electrodes was notreported.

Given the instability of LSF–YSZ and LSCo–YSZ electrodes, itseems likely that there will be stability issues with LSCF as well.The major issue is whether or not the Co in the LSCF will remainin the perovskite phase, so that La2Zr2O7 formation will beprevented. A study of La0.8Sr0.2Mn0.8Co0.2O3 (LSCM) infiltrationinto porous YSZ showed reasonable initial performance;however, the LSCM–YSZ electrodes were found to exhibitstability similar to that of LSCo–YSZ electrodes. Performancedeclined dramatically, even at 700 8C.[97] Therefore, we suggest itis likely that the LSCF electrodes will also undergo solid-statereactions with YSZ. Clearly, the stability of the infiltrated LSCFelectrodes needs to be checked given the intriguing performancethat was achieved.

3.4. LSM–YSZ Electrodes

Some of the key properties of LSM have already been discussedearlier in this review and have been reviewed extensivelyelsewhere.[33] Under typical cathode conditions, LSM has an

electronic conductivity greater than 200 S cm"1 but has negligibleionic conductivity. It undergoes a solid-state reaction with YSZ toform La2Zr2O7 above 1 250 8C,[66] but LSM–YSZ mixtures arestable at lower temperatures. Because LSM–YSZ composites arethe standard material for cells with YSZ electrolytes, there is avery large body of work characterizing the performance of theseelectrodes.

Much of the effort in improving LSM–YSZ electrodes hascentered on developing an optimal composite structure. Forexample, Virkar and coworkers have developed mathematicalmodels to determine what that ideal structure should be.[55] Asdiscussed earlier and shown in Figure 3, it is common practice forfuel-cell developers to engineer LSM–YSZ electrodes so that theregion near the electrolyte, the functional layer, has a differentmicrostructure from that of the region farther from theelectrolyte, the current-collection layer. Micrographs of thefunctional and conduction layers show that the functional layertends to have smaller pores and less overall porosity. This reflectsthe need to enhance the concentration of TPB sites. Highporosities are not required for high gas-phase transport rates inthin layers. Infiltration procedures are ideal for preparing thefunctional layer, but are not really needed for the current-collection layer.

The performance of electrodes formed by infiltration of LSM isvery good and comparable to that of conventional LSM–YSZcomposites. For example, Armstrong and Virkar preparedinfiltrated LSM electrodes using nitrate-salt solutions andachieved power densities as high as 1.2W cm"2 for a celloperating in hydrogen at 800 8C.[78]

As with conventional LSM–YSZ composites, structure appearsto be critical. Sholklapper et al. reported a power density of0.3W cm"2 at a temperature of only 650 8C for a cell prepared byinfiltrating a porous YSZ scaffold with LSM nanoparticles.[81] Thestructure of these electrodes is shown in the SEMmicrographs inFigure 10 and demonstrates that the YSZ scaffold is coated withsmall LSM particles. The picture in Figure 10 is similar to that of

Figure 10. Cross-sectional SEM image showing the microstructure of aLSM–YSZ cathode that was fabricated by infiltrating a porous YSZ scaffoldwith LSM nanoparticles. Reproduced with permission from reference [81].Copyright 2006 The Electrochemical Society.

Adv. Mater. 2009, 21, 943–956 ! 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 951

REVIE

W

www.advmat.de

implies a change in the surface area of the LSF phase. If electrodeperformance is limited by the rate of surface oxidation and thesurface area decreases, the performance would be expected todecline proportionally. A smaller active surface area could alsoexplain the decrease in impedance with increasing currentdensity. In the Butler–Volmer picture of electrode reactions, ratesare expected to accelerate as one moves farther from equilibrium.However, a plot of the cathode impedance did not show theexpected logarithmic dependence between current and over-potential that would be expected for Butler–Volmer kinetics.[85]

Furthermore, while we cannot totally rule out this model, it seemsunlikely that the slope of the V–i relationship would remainconstant in going from anodic to cathodic polarization if thesurface kinetics were limiting.

An alternative picture for understanding deactivation of theLSF–YSZ electrodes is shown schematically in Figure 8. Based onthe SEM results, the LSF deposits calcined at 850 8C exist as smallparticles on the YSZ scaffold, separated by gaps that allow gas-

phase oxygen to diffuse to the YSZ interface. This is depicted inFigure 8a, which is similar to the diagram in Figure 2b. Whenthese particles sinter, either over time at operating conditions orfollowing high-temperature calcination, the LSF forms a dense,polycrystalline layer over the YSZ scaffold, as depicted inFigure 8b. Because the ionic conductivity of LSF is much lowerthan that of YSZ (8! 10"4 S cm"1 for LSF and 1.89! 10"2

S cm"1 for YSZ at 700 8C[20,101]), the transport of oxygen ionsthrough the LSF could be limiting. The coupling of oxygen-iontransport up the YSZ ‘‘fingers’’ and through the LSF film wouldthen be responsible for the current-dependent impedances.

Data for cathode performance with infiltrated Ca- and Ba-doped LaFeO3 (LCF and LBF)[101] provide additional support forthe picture in Figure 8. LCF and LBF have nearly the sameelectronic conductivity as that of LSF, but their ionic conductiv-ities are significantly lower. LCF, in particular, has an ionicconductivity that is 50 times lower than that of LSF at 700 8C.Following calcination at 850 8C, the initial performance of LCF–YSZ and LBF–YSZ electrodes was indistinguishable from thatobserved with LSF–YSZ electrodes. Assuming the morphology ofthe electrodes is similar to that shown in Figure 8a, it isreasonable that the ionic conductivity would not be critical in thiscase.

Following calcination at 1 100 8C, SEM showed that LSF, LCF,and LBF each tended to form a dense film over the YSZ, similar tothat pictured schematically in Figure 8b. Cathode performancefor each of the three composite cathodes also decreaseddramatically. Figure 9 shows impedance data at open circuitand at 100mA cm"2 for three cells made with identical anodesand electrolytes, but with infiltrated cathodes based on LSF, LBF,or LCF.[101] The impedance at open circuit was much larger forthe electrodes based on LCFand LBF, as would be expected for themodel in Figure 8, given the significantly lower ionicconductivities of the LCF and LBF. Similar to the findings withLSF, the impedance of LCF–YSZ and LBF–YSZ cathodesexhibited a strong current dependence. With LCF–YSZ, the

Figure 7. SEM images of a) the porous YSZ scaffold, b) LSF film on theYSZ scaffold after calcination at 850 8C, c) LSF film after testing for 1000 has 700 8C, and d) LSF film after calcination at 1100 8C. Reproduced withpermission from reference [85]. Copyright 2007 The ElectrochemicalSociety.

Figure 8. Schematic diagram of infiltrated LSF–YSZ cathode a) aftercalcining in air at 850 8C and b) after calcining at higher temperatures(>1000 8C) or long term aging.

0.0

1.0

2.0

3.0

4.0

-1.0 0.0 1.0 2.0 3.0 4.0Z Re / cm-2

-Z Im

/ c

m-2

Figure 9. Impedance spectra obtained from fuel cells with infiltratedelectrodes as a function of the active component, LSF (&), LBF (#),and LCF (~), used in the cathode. The composite cathodes in each ofthese cells were calcined at 1100 8C. The filled symbols show datameasured at open circuit while the open symbols were obtained at acurrent density of 100mA cm"2. Reproduced with permission fromreference [101]. Copyright 2008 The Electrochemical Society.

950 ! 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2009, 21, 943–956

Scholklapper et al. (LBL), Electrochem. Solid-State Lett., 9, A376 (2006)

W. Wang et al. (Penn)J. Electrochemical Society, 154, B439-B445 (2007).

Chao, Gerdes, Kitchin, Salvador (CMU)Electrochemical Society Trans, 35, 2387-2399, (2011).

LSM on YSZLBL

LSCF on YSZPenn

LSM on LSCFCMU





Probe the nature of atomic scale surface chemistry or interface crystallographyrather than the device scale micro-structural perturbations in SOFC conditions:

T = 500 - 900 °C, PO2 ≈ 10-5 - 1 atm, Overpotential ≈ 0 - 0.4 V

Conceptual Thin Film Sample: Proxy to Crystals

Need High Quality Samples with Controlled Microstructural Features:

Single Crystals or Thin Films

Mixed Conductor

Surface

Substrate

Mass Uptake / Release

Transient Conductivity

Surface Electronic StatesSurface Structure

Surface Stoichiometry

Theory / Modeling /Simulation

Reaction Pathways

Surface Chemical Species

• thickness < D / k

• surface sensitive : k

• characterization of film quality / microstructural states

• compare properties to bulk / films

FILMS ARE GOOD PROXIES FOR BULK CERAMICS





SOFC Fuel Cell Collaborators

5

Wonyoung Lee, Khabiboulakh Katsiev, Helia Jalili, Bilge YildizDepartment of Nuclear Science and Engineering, Massachusetts Institute of Technology

Lu Yan, Robin Chao, Herb Miller, Shen Dillon, K. R. Balasubramanian, Shanling Wang, and Hui Du

Gregory Rohrer, John Kitchin Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, PA

Tim T. Fister, Dillon D. Fong, Peter M. Baldo, Matthew J. Highland, Kee-Chul Chang, Terry Cruse, and Brian Ingram,

Paul H. Fuoss, Jeffrey A. Eastman, H. You, and Michael KrumpeltMaterials Science Division, Argonne National Laboratory, Argonne, Illinois

Supported by the DOE-SECA program and DOE-NETL.

Paul Ohodnicki, Kirk Gerdes, Randy GemmanDepartment of Energy, National Energy Technology Laboratory, Morgantown, West Virginia

Michael Weir, Stefan Krause, Clemens HeskeDepartment of Chemistry, University of Nevada, Las Vegas





Orienta1on Dependence

6

Sasaki et al. Solid State Ionics. 161: 145-154 (2003)!

Kchem (111) > Kchem (110) > Kchem (100) !

YSZ Single CrystalsLSM Films





Films have surface ac1vated response to ECR

7

Linear relationship of Relaxation and Thickness indicates the response is

Surface Limited.

Oxidation and Reduction are Similar

Crystallographic Anisotropy Exists by ≈ 75 %





Crystallographic Anisotropy in Apparent EA

8

Different Orienta@on have different ac@va@on energies and kchem’s

Ac@va@on Energies are on the order of Literature Values

Ac@va@on Energies are Orienta@on Dependant





Relaxed Films:Misfit Dislocations from Surface

Strained Films:Inherited Dislocations

Controlling Disloca1ons in Films

Low Dislocation (LD) Substrate

High Dislocation (HD) Substrate Low Mismatch (LM) Substrate

High Mismatch (HM) Substrate

How do Dislocations Impact Surface Properties?

Thickness





Rocking Curve Widths and Disloca1on Content

10

Inte

nsity

/ (a

rb u

nits

)

w / °

On NGO

On STO

600 nm

600 nm 50 nm

50 nm FWHM = 0.12 °

FWHM = 0.032 °

FWHM = 0.016 °

FWHM = 0.024 °

22.9 23.0 23.1 23.2 23.3 23.4 23.5 23.6 23.7 23.8 23.9 24.0





Anomalous Relaxa1on Behaviors

11

significant substrate effect: STRAIN ( <100 nm).

significant substrate effect: DISLOCATION ( >100 nm).

•  surface exchange : on STO > on NGO

�

g(t) = 1− exp(−K chem tL

)

K chem = Lthickness

τ relaxation time





Ac1va1on energy vs thickness and substrate

12

50 nm 600 nm on NGO

on NGO

on STO on STO

Ea = 1.5 eV Ea = 0.75 eV

Ea = 0.75 eV

Ea = 1.5 eV ln K

chem

/ cm

s-1

1000/T / K-1 1000/T / K-1

Thick Films on Different Substrates

Thin Films on Different Substrates

• Two different ac@va@on Energies: two different processes

• Thick Films are more similar to one another and are generally more ac@ve

• at higher temperature, thinnest film on STO is most ac@ve

• for thin films, values are almost and order of magnitude different





Dependence on thickness and temperature

13

-8.0

-7.5

-7.0

-6.5

-6.0

-5.5

-5.0

-4.5

-4.0

0 200 400 600 800

log

Kch

em, r

ed /c

ms-1

Thickness/nm

-8.0

-7.5

-7.0

-6.5

-6.0

-5.5

-5.0

-4.5

-4.0

0 200 400 600 800 Thickness / nm

on STO on NGO

strained

strained

strained

strained relaxed

relaxed

relaxed

relaxed

at 600 °C

at 800 °C

on STO

on NGO

• Solid horizontal arrows represent thickness independent regimes

• Loca@ons of coherent strain a full relaxa@on are marked

• Behaviors are somewhat complex...





Explana1on of Epitaxial Film Observa1ons

14

!

-15.00

-14.00

-13.00

-12.00

-11.00

-10.00

-9.00

-8.00

0.90 1.00 1.10 1.20

ln K

chem

, red

/ cm

s-1

1000/T / K-1

Kr -50nm-STO Kr-100nm-STO Kr-300nm-STO Kr-600nm-STO

-18.00

-17.00

-16.00

-15.00

-14.00

-13.00

-12.00

0.90 1.00 1.10 1.20

ln K

chem

/ c

m s-1

1000/T / K-1

Kr-50nm-NGO Kr-100nm-NGO Kr-200nm-NGO Kr-600nm-NGO

Ea=1.5 eV

Ea=0.75 eV

Ea=1.5 eV

Ea=0.75 eV

on STO on NGO

50nm 100nm

600nm 300nm

50 nm 100 nm 200 nm 600nm





Textured Film Surfaces

15

Substrate

Single Crystal !

Perovskite !

Substrate!

(111)!

Epitaxial !Perovskite !Cathode!

Surface!

Single Crystal !

Perovskite !

Substrate!

(100)!


Surface!

Single Crystal !

Perovskite !

Substrate!

(110)!


Surface!

Single Crystal !

Electrolyte!

Substrate!

(100)!


Surface!

Single Crystal !

Electrolyte!

Substrate!

(111)!


Surface!

Films Allow for Surface / Microstructural Control

Substrate!

Polycrystal!Perovskite !Cathode!

Surface!

Single Crystal Specific surface

BULK SURFACES

Multi-variant Specific surface / boundaries

Poly-Xtal Ensemble

(100)! (110)! (111)! (100)! (110)! all!

How do we synthesize these materials?

INFILTRATES

Buffer!

Can the Strain Free Surface and Grain Boundaries be Separated?





ECR Proper1es fit to 2 Separate Processes

16

0

0.2

0.4

0.8

0.6

1.0

1.2 N

orm

aliz

ed c

ondu

ctiv

ity g

(t)

0 200 400 600

Time / s

600 C - oxidation ECR data

slow process data fitting

fast process data fitting

0

0.2

0.4

0.8

0.6

1.0

1.2

Time / s

800 C - oxidation

0 200 400 600

0

0.2

0.4

0.8

0.6

1.0

1.2

0 10 20 30

Time / s

800 C - oxidation

ECR data

Data fitting ECR data Data fitting

(a) (b)

�

g( t) =σ t −σ final

σ final −σ initial

=1− Aexp(−K1,chemtL

) − (1− A)exp(−K2,chemtL

)

< 800 °C, A≠1, two Kchem : a fast and a slow processes. ≥ 800 °C, A=1, one Kchem : faster response at 600 °C diminishes with temperature.





Temperature dependence of two mechanisms

17

-15

-14

-13

-12

-11

-10

-9

-8

0.8 0.9 1 1.1 1.2

ln K

ch

em

/ c

ms-1

1000/T (K-1)

lnKred-2

lnKoxi-2

lnKred-1

lnKoxi-1

Ea,1=1.57 eV

Ea,2=0.87 eV

ln k

chem

/ cm

s-1

lnkoxi,1

lnkoxi,2

lnkred,1

lnkred,2

I

(a) kgrain kgb kgb

(b)

1.  la O’ et al, J. Electrochem sec. (2009). 2.  De Souza et al, Mater. lett. (2000). 3.  Yan et al, Solid state ionics. (2011). 4.  Kan et al, Solid state ionics. (2010). 5.  Yasuda et al, J Solid State Chem. (1996).

Our data agree with literature:

Kchem is on the order of 10-5cm/s.1-5

Ea1: 1.48 eV for microelectrode1, 1.32 eV dense pellet2,

Ea2: 0.8 eV (100) and (111) on STO3, 0.07-0.8 eV powder4.

I

(a) kgrain kgb kgb

(b)





Rela1ve Contribu1on to Responses

18

Temp / K 883 923 986 1088 1191 A 0.6 0.7�� 0.8 1� 1�

koxi, 1 !10-6 / cms-1 0.668 1.09 4.20 21.8 32.4 koxi, 2 !10-6 / cms-1 12.0 14.0 29.2 N/A�� N/A��kred, 1 !10-6 / cms-1 0.996 1.67 6.42 41.1 55.8 kred, 2 !10-6 / cms-1 15.0 23.1 48.4 N/A�� N/A��

!

�

g(t) =σ t −σ final

σ final −σ initial

=1− Aexp(−k1,chem tL

) − (1− A)exp(−k2,chem tL

)

Kgrain Kgb Kgb

(b) Kgrain

Kgb Kgb

(c)

Low T High T

(a)





600 nm LSM on YSZ (111) with different grain size

RMS : 3.1 nm Sa : 4.2 um2

850°C annealed 24 hr

0.5 um

900°C annealed 24 hr

0.5 um

RMS : 3.3 nm Sa : 4.3 um2

Grain size increases

Post-annealing after 750°C deposition

Grain size: ≈ 100 nm Grain size: ≈ 150 nm

Temp / K 883 923 986 1088 1191 A small grain 0.6 0.7� 0.8 1 1 A big grain 0.7 0.75� 0.84 1 1

Kchem values for the fast process and the slow process are the same for both grain sizes.





Comparison of Textured and Epitaxial (110)

20

Ea,2=0.87 eV

Ea,1=1.57 eV Ea=1.2 eV

(110) LSM-YSZ fast process

(110) LSM-YSZ slow process

(110) LSM-STO

ln k

oxi /

cm

s-1

*N100

*S100





Summary

21

Two apparent processes occuring on the surface (Ea) for Kchem .

These were interpreted as belonging to: (1) the native surface response of individual grains/variants and (2) the variants boundaries / grain boundaries of the textured films.

The first (native surface) process : EA,1 ≈ 1.5 eV, the second (extended defect) process EA,2 ≈ 0.75 eV. The Kchem,2 values are almost 3 orders of magnitude higher than the Kchem,1 values at low temperatures (< 700°C)

Depends on the density of the defects.

At higher temperatures, the data can be fit with one Kchem Intermediate value of EA indicate that both processes contribute to overall exchangeThe native surface, higher activation energy process is competitive with

The native surface processes are Strain dependent Orientation Dependent Substrate dependent




Documents

Oxygen’Exchange’in’Thin’Layers’of’SOFC’Cathode’ · 2014-07-28 · J. Electrochemical Society, 154, B439-B445 (2007). Chao, Gerdes, Kitchin, Salvador (CMU) Electrochemical