Degradation mechanisms and advanced characterization and testing

Degradation Mechanisms in Solid Oxide Cells and Systems Workshop Proceedings 139

Solid Oxide Cell Degradation Operated in Fuel Cell and Electrolysis Modes: A Comparative Study on Ni Agglomeration and LSCF Destabilization M. Hubert1,2, J. Laurencin*1, P. Cloetens2, D. Ferreira Sanchez3, S. Pylypko1, M. Morales4, A. Morata4, B. Morel1, E. Siebert5 and F. Lefebvre-Joud1 1 Univ. Grenoble Alpes - CEA/LITEN, 17 rue des Martyrs, 38054, Grenoble, France 2 European Synchrotron Radiation Facility (ESRF), 38000, Grenoble, France 3 Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland 4 Catalonia Institute for Energy Research (IREC), Barcelona, Spain 5 Univ. Grenoble Alpes - CNRS, LEPMI, 1130 rue de la Piscine, 38402 Saint Martin d'Hères, France *Corresponding author e-mail address:jerome.laurencin@cea.fr

The electrochemical degradations of Solid Oxide Cells (SOCs) are generally attributed to several underlying phenomena such as electrode microstructural evolution, material chemical decomposition or electroactive sites poisoning by contaminants. Among them, it is generally considered that Ni agglomeration in the Ni-YSZ cermet and Lanthanum Strontium Cobalt Ferrite (LSCF) material destabilization are two prevalent mechanisms involved in the cell performance deterioration. Therefore, these two mechanisms have been specifically investigated by a coupled approach of long-term testing (1000≤t(h)≤9000) and post-test characterizations. The experimental results have then been analyzed in the frame of an in-house multi-scale model. The extent of Ni agglomeration has been characterized by three-dimensional electrode reconstructions obtained by X-ray nano-holotomography [1] (Fig. 1a). The electrode morphological properties, which have been measured on the 3D volumes, have revealed a substantial Ni coarsening over time. The compilation of all experimental data have allowed fitting the parameters of a physically-based law for Ni coarsening that was implemented in the modelling framework. The simulations have revealed that Ni agglomeration explains around 20-25% of the electrochemical degradation at 850°C after 1000 hrs of operation. However, the electrode microstructural evolution is found to be not affected by the cell polarizations. Therefore, the mechanism cannot explain the higher degradation rates recorded in electrolysis mode compared to the fuel cell ones. To explain such result, a set post-test characterizations have been performed to investigate the phase reactivity in the region of the CGO barrier layer [2]. They have revealed that Sr diffusion and formation of SrZrO3 secondary phase occur mainly during electrolysis operation (whereas the process is very limited in fuel cell mode) (Fig. 1b). The in-house multi-scale model has been used to interpret the role of the cell

operating mode on the LSCF destabilization mechanism. Based on the simulation results, a possible mechanism for the LSCF demixing and SrZrO3 formation has been proposed.

(a)

(b)Dimension (µm)

Dim

en

sio

n (

µm

)

(0b)(10b)(20b)(30b)(40b)(50b)(60b)(70b)

(b)

La0.6Sr0.4Co0.2Fe0.8O3

Ce0.9Gd0.1O1.95

LS

CF

-CG

O e

lectro

de

CG

O b

arrie

rla

ye

r

SrZrO3reconstructed slice

Fig. 1. (a) 3D rendering volume of a Ni-YSZ reconstruction obtained by X-ray nano-

holotomography, and (b) Reconstructed slice by synchrotron X-ray µdiffraction tomography for the

SrZrO3, La0.6Sr0.4Co0.2Fe0.8O3 and Ce0.9Gd0.1O1.95 crystalline structures

Acknowledgements The research leading to these results has

received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) Fuel Cells and Hydrogen Joint Undertaking (FCH-JU-2013-1) under grant agreement n° 621173 (SOPHIA project) and grant agreement n°621207 (ENDURANCE project).

Degradation mechanisms and advanced characterization and testing

February 17, 2017 ~ Barcelona, Spain 140

References [1] M. Hubert, Thesis, University of Grenoble, 2017.

[2] J. Laurencin, M. Hubert, D. Ferreira Sanchez, S. Pylypko, M. Morales, A. Morata, B. Morel, D. Montinaro, F. Lefebvre-Joud, E. Siebert, Electrochimica Acta (2017).

Solid Oxide Cell Degradation Operated in Fuel Cell and Electrolysis Modes: A Comparative Study on Ni Agglomeration and LSCF DestabilizationM. Hubert, J. Laurencin, P. Cloetens, D. Ferreira Sanchez, S. Pylypko, M. Morales, A. Morata, B. Morel, E. Siebert and F. Lefebvre-Joud

WORKSHOP PROCEEDINGS

DEGRADATION MECHANISMS IN SOLID OXIDE CELLS AND SYSTEMS

FEBRUARY 17, 2017BARCELONA, SPAIN

Solid Oxide Cell Degradation Operated in Fuel Cell and Electrolysis Modes: A Comparative Study on Ni Agglomeration

and LSCF Destabilization

M. Hubert(1,2), J. Laurencin(1), P. Cloetens(2), S. Pylypko(1), D. Ferreira Sanchez(3), M. Morales(4), A. Morata(4), D. Montinaro(5), B. Morel(1), E. Siebert(6), F. Lefebvre-joud(1)

Endurance, Sophia, Eco, Diamond, and SoctesQA projects: Jointed Workshop on SOCs Degradation

The 17th of February 2017, Barcelona

(1) CEA-Liten/LPH (Grenoble, France) (2) ESRF: European Synchrotron Radiation Facilites (Grenoble)(3) PSI: Paul Scherrer Institut (Villingen, Switzerland)(4) IREC: Catalonia Institute for Energy Research (Barcelona, Spain)(5) SOLIDpower(Mezzolombardo, Italy)(6) LEPMI-CNRS (Grenoble, France)

Context: Degradation of Solid Oxide Cells

� Main drawback: the degradation rates at stack levels are still too high for economic viability…� Main drawback: the degradation rates at stack levels are still too high for economic viability…

� Two main contributions:

(i) Ni coarsening in the Ni-YSZ cermet for the H2 electrode

(ii) LSCF destabilization and Sr reactivity with the electrolyte for the O2 electrode

� However…

� The basic underlying mechanisms not fully understood.

� The impact on performance loss not precisely quantified.

� The effect of SOFC versus SOEC modes not studied yet

� Two main contributions:

(i) Ni coarsening in the Ni-YSZ cermet for the H2 electrode

(ii) LSCF destabilization and Sr reactivity with the electrolyte for the O2 electrode

� However…

� The basic underlying mechanisms not fully understood.

� The impact on performance loss not precisely quantified.

� The effect of SOFC versus SOEC modes not studied yet

☺ High-temperature fuel cell and electrolyser are efficient energy-conversion systems that present several advantages (high performances, flexibility, etc.)☺ High-temperature fuel cell and electrolyser are efficient energy-conversion systems that present several advantages (high performances, flexibility, etc.)

1

Electrode

microstructural

evolutions

Active sites

poisoning by

contaminants

Material

chemical

decompositions

Inter-diffusion

and reactivity

between cell

components

Material and interface

instabilities activated at

high temperature under

polarization

Methodology:A coupled experimental and modeling approach

Electrochemical testing at the electrode and cell levels

Electrochemical testing at the electrode and cell levels

Microstructural characterization and

post-test analyses

Microstructural characterization and

post-test analyses

Multi-scale modeling from electrode to SRU

Multi-scale modeling from electrode to SRU

Validated models

Interconnect)x(P

2O

xrx dxx +

zr

l

0z =

Ionic conducting phase (electrolyte)

Mixed Electronic Ionic Conductor (electrode)

e -

(MIEC)oxO

−2O eelectrolyt

(g)2O(MIEC)•h

“Microstructural and material evolutions � Reactive mechanisms � loss in cell performances”

depending on the operating parameters (SOFC vs SOEC polarizations, temperatures, etc.)

“Microstructural and material evolutions � Reactive mechanisms � loss in cell performances”

depending on the operating parameters (SOFC vs SOEC polarizations, temperatures, etc.)

Functional layer

Ni-YSZsubstrate

2

Studied cell:Materials and initial performances

3

� A typical H2 electrode supported cell constituted with classical materials: � A typical H2 electrode supported cell constituted with classical materials:

� Initial cell performances measured on button cell

(active area=9 cm²)

� Initial cell performances measured on button cell

(active area=9 cm²)

� The cell exhibits good performances and is well representative of the classical SOEC/SOFC

technology.

� The cell exhibits good performances and is well representative of the classical SOEC/SOFC

technology.

- H2 electrode: Ni-YSZ- Electrolyte: + YSZ - Barrier layer: + CGO

- O2 electrode: + LSCF-CGO + LSCF - Collecting layer + LSC

- H2 electrode: Ni-YSZ- Electrolyte: + YSZ - Barrier layer: + CGO

- O2 electrode: + LSCF-CGO + LSCF - Collecting layer + LSC

T=800°C, H2/H2O=100/0

FU=60% at 0.5 A.cm-2

Long term testing:Degradation rates in fuel cell and electrolysis modes

Mode Temp. ( oC) H2/H2O SC/FU (%) j (A.cm -2) Ageing time (h) Degradationrates (mV.kh -1)

SOEC 850 50/50 60 -0.5 1500 9,7SOEC 850 50/50 60 -0.5 2000 8,8SOEC 850 50/50 60 -0.75 2400 15,8SOFC 850 50/50 60 +0.5 1000 3,5SOFC 750 100/0 60 +0.5 9000 13,5 (t<2000

hrs)4.5 (t>2000 hrs)

SOFC 850 100/0 80 +0.66 1200 57,2

� Overview of the test operating conditions and degradation rates� Overview of the test operating conditions and degradation rates

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

0 500 1000 1500 2000

Elapsed time (h)

Ce

ll vo

ltag

e (

V)

Electrolysis at i=-0.5 A.cm-2

Fuel Cell at i=+0.5 A.cm-2

OCV, H2/H2O=50/50, 850°C

Degradation rates between SOEC vs

SOFC can be compared

� The degradation rates are rather low and are in

the state-of-the-art of typical values reported in

literature.

� The degradation rates are rather low and are in

the state-of-the-art of typical values reported in

literature.

� The degradation rate are higher under

electrolysis current than under fuel cell current.

� The degradation rate are higher under

electrolysis current than under fuel cell current.

4

Ni agglomeration:3D electrode reconstructions by X-ray holotomography

5

� X-ray holotomography on the new beam line ID16A at ESRF (Grenoble):� X-ray holotomography on the new beam line ID16A at ESRF (Grenoble):

Fundamental of the measure: optical contrast according to an holographic

scheme.

� 3D reconstructions with a high field of view of 50 µm and high spatial resolution (<50 nm).

� Computation of the electrode properties (Rp, TPBs, Sp, etc.)

� 3D reconstructions with a high field of view of 50 µm and high spatial resolution (<50 nm).

� Computation of the electrode properties (Rp, TPBs, Sp, etc.)

� 3D volumes acquired at the cell inlet and outlet for each experiments:� 3D volumes acquired at the cell inlet and outlet for each experiments:

� No Ni depletion detected at the electrode/electrolyte interface. � No Ni depletion detected at the electrode/electrolyte interface.

� Impact of both (i) steam partial pressure and (ii) operating mode in

SOFC and SOEC modes very limited.

� Impact of both (i) steam partial pressure and (ii) operating mode in

SOFC and SOEC modes very limited.

� A clear increase of Rp_Ni over time.� A clear increase of Rp_Ni over time.

� Significant Ni agglomeration at 850oC� Significant Ni agglomeration at 850oC

A decrease of Sp_Ni/gas whereas Sp_Ni/YSZ not changedA decrease of Sp_Ni/gas whereas Sp_Ni/YSZ not changed

� The YSZ backbone limits the Ni agglomeration� The YSZ backbone limits the Ni agglomeration

6

Ni agglomeration:Parameters of a constitutive law for Ni agglomeration

� Rp_Ni=f(t,T) fitted by a standard power-law

coarsening model:

� Rp_Ni=f(t,T) fitted by a standard power-law

coarsening model:

(R�_��)� − (R�_�

� )� = k��� × t

withk��� ∝ e�∆�

���

(R�_��)� − (R�_�

� )� = k��� × t

withk��� ∝ e�∆�

���

� Best fit obtained for n=7-9:

� Inhibiting effect of YSZ.

� Consistent with a sintering mechanism controlled

by Ni surface diffusion.

� Best fit obtained for n=7-9:

� Inhibiting effect of YSZ.

� Consistent with a sintering mechanism controlled

by Ni surface diffusion.

� ξTPB=f(t,T) also fitted by the power-law model

with an exponent equal to 8 at different

temperatures.

� ξTPB=f(t,T) also fitted by the power-law model

with an exponent equal to 8 at different

temperatures.

� Substantial decrease of ξTPB that could affect the cell

performances.

� Substantial decrease of ξTPB that could affect the cell

performances.

n=8

n=8

7

Ni agglomeration:Modelling of cell performances over time

Cell model

Electrode micro-scale models� In-house multi

scale model.

� In-house multi

scale model.

0.4

0.6

0.8

1

1.2

1.4

1.6

-1 -0.5 0 0.5 1

Current density (A cm-2)

Ce

ll vo

ltag

e (

V)

Electrolysis Fuel Cell

T=850°C, H2/H2O=50/50Experimental curveSimulated data

T=800°C, H2/H2O=100/0Experimental curveSimulated data

(a)

(b)

� Accurate simulation of the polarization curves in

different conditions (T, polarization, gas feeding, etc.).

� Accurate simulation of the polarization curves in

different conditions (T, polarization, gas feeding, etc.).

� The Ni coarsening

represents a significant

part of the degradation

rates (∼30% in SOFC).

� It does not explain the

difference in SOFC and

SOEC modes.

� The Ni coarsening

represents a significant

part of the degradation

rates (∼30% in SOFC).

� It does not explain the

difference in SOFC and

SOEC modes.

� ξTPB=f(t,T) power-law implemented in the model:� ξTPB=f(t,T) power-law implemented in the model:

SOFC

T=850oC, H2/H2O=50/50, i=+0,5 A.cm-2SOEC

T=850oC, H2/H2O=50/50, i=+0,5 A.cm-2

8

LSCF destabilization and Sr reactivity with YSZ:Detection of a secondary Sr-rich phase in the barrier layer

� Synchrotron X-ray µfluorescence:� Synchrotron X-ray µfluorescence:

� Accumulation of Sr in the barrier layer for the SOEC

aged sample (T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2).

� Accumulation of Sr in the barrier layer for the SOEC

aged sample (T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2).

� SEM-EDX observations in the barrier layer:� SEM-EDX observations in the barrier layer:

� Sr-rich phase in SOEC mode in the voids of the

barrier layer at the electrolyte interface.

� A line of nano-pores at grains boundaries in the YSZ.

� Sr-rich phase in SOEC mode in the voids of the

barrier layer at the electrolyte interface.

� A line of nano-pores at grains boundaries in the YSZ.

� Quantification by image analysis of the Sr-rich

phase:

� Quantification by image analysis of the Sr-rich

phase:

� Formation of the secondary Sr-rich phase is

strongly promoted by an operation under

electrolysis current.

� Formation of the secondary Sr-rich phase is

strongly promoted by an operation under

electrolysis current.

� the amount of Sr-rich phase is much larger in electrolysis

mode than the quantity measured on the SOFC aged

samples.

� the amount of Sr-rich phase is much larger in electrolysis

mode than the quantity measured on the SOFC aged

samples.

nanopores

Sr-rich phase

9

LSCF destabilization and Sr reactivity with YSZ:Crystallographic nature of the Sr-rich phase in the barrier layer

� 2D synchrotron X-ray µdiffraction

tomography:

� 2D synchrotron X-ray µdiffraction

tomography:

� Samples prepared with

Xe PFIB

� Samples prepared with

Xe PFIB

2 tilted reconstructed slices across CGO layer for LSCF, CGO, and SrZrO3 crystalline phases

(deduced from each reconstructed diffracted pattern)

� The secondary Sr-rich phase detected by

SEM-EDX and µfluorescence is identified as

SrZrO3.

� The secondary Sr-rich phase detected by

SEM-EDX and µfluorescence is identified as

SrZrO3.

Pristine cell

SOEC: T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2

SOFC: T=850oC, H2/H2O=50/50, i=+0,5 A.cm-2

10

LSCF destabilization and Sr reactivity with YSZ:Detection of a secondary Co-rich phase in the barrier layer

� Synchrotron X-ray µfluorescence:� Synchrotron X-ray µfluorescence:

� Accumulation of Co in the barrier layer for the SOEC

aged sample (T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2).

� Accumulation of Co in the barrier layer for the SOEC

aged sample (T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2).

� SEM-EDX observations in the barrier layer:� SEM-EDX observations in the barrier layer:

� Formation Co-rich particles in SOEC mode in the

porosities in contact with SrZrO3 secondary phase.

� Formation Co-rich particles in SOEC mode in the

porosities in contact with SrZrO3 secondary phase.

� TEM-EDX characterization on

these particles:

� TEM-EDX characterization on

these particles:

� Co-rich particles identified as cobalt-ferrite type compound (perovskite orthorhombic Gd1-xSrxCo1-yFeyO3-d (0.84 ≤ x ≤ 0.88; 0.39 ≤ y ≤ 0.42).

� Co-rich particles identified as cobalt-ferrite type compound (perovskite orthorhombic Gd1-xSrxCo1-yFeyO3-d (0.84 ≤ x ≤ 0.88; 0.39 ≤ y ≤ 0.42).

� Precipitation of Co in the region of barrier

layer is concomitant with the formation of

SrZrO3 under SOEC conditions.

� Precipitation of Co in the region of barrier

layer is concomitant with the formation of

SrZrO3 under SOEC conditions.

11

LSCF destabilization and Sr reactivity with YSZ:Mechanism of LSCF destabilization under anodic polarization

� Computation with the model of the local operating parameters in the O2 electrode in the conditions of cell ageing

(T=850oC, H2/H2O=50/50, i=+/-0,5 A.cm-2)� Computation with the model of the local operating parameters in the O2 electrode in the conditions of cell ageing

(T=850oC, H2/H2O=50/50, i=+/-0,5 A.cm-2)

� Mechanism of LSCF destabilization and formation of SrZrO3 triggered under anodic polarization:� Mechanism of LSCF destabilization and formation of SrZrO3 triggered under anodic polarization:

••++→+ oLasurfxoLa VVSrOOSr ///

)(/

//////)(5.2

/ 2522252 CooLasurfxCo

xoLa VVVSrCoOCoOSr +++→++ ••

)(2)(2)( )( gasgassurf OHSrOHSrO →+

)(2)(2)()( )(234

)(2 YSZVOHSrZrOYSZOYSZZrOHSr ogasxoZgas r

••• ++→++

� Sr release driven by the depletion in oxygen vacancies in SOEC mode.� Sr release driven by the depletion in oxygen vacancies in SOEC mode.

� SrO evaporated under Sr(OH)2 hydroxide volatile molecules.� SrO evaporated under Sr(OH)2 hydroxide volatile molecules.

� Reactivity in the pores of the barrier layer.� Reactivity in the pores of the barrier layer.

� Because the high O2 inlet

flow rate, no significant

difference in Po2 for both

operating modes.

� Because the high O2 inlet

flow rate, no significant

difference in Po2 for both

operating modes.

� At the opposite of the SOFC

mode, strong oxygen

accumulation (i.e. vacancies

depletion) within LSCF in SOEC

mode.

� At the opposite of the SOFC

mode, strong oxygen

accumulation (i.e. vacancies

depletion) within LSCF in SOEC

mode.

� The SrO (and SrCox) passivation film

hinders the surface reactions.

� Sr loss induces a decrease in oxygen

chemical diffusivity for LSCF.

� Line of nanopores in YSZ decreases

its ionic conductivity

� The SrO (and SrCox) passivation film

hinders the surface reactions.

� Sr loss induces a decrease in oxygen

chemical diffusivity for LSCF.

� Line of nanopores in YSZ decreases

its ionic conductivity

� The LSCF destabilization could

explain the higher degradation

rates in SOEC compared to

SOFC.

� The LSCF destabilization could

explain the higher degradation

rates in SOEC compared to

SOFC.

Concluding remarks

� Long term testing in SOEC and SOFC modes (1000 h ≤ t ≤9000 h):

� Low degradation rates in SOFC mode (3,5 mV.kh-1 at T=850oC, H2/H2O=50/50, i=+0,5 A.cm-2)� Degradation rate three times higher in SOEC mode in comparable conditions (9 mV.kh-1 at

T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2)

� Long term testing in SOEC and SOFC modes (1000 h ≤ t ≤9000 h):

� Low degradation rates in SOFC mode (3,5 mV.kh-1 at T=850oC, H2/H2O=50/50, i=+0,5 A.cm-2)� Degradation rate three times higher in SOEC mode in comparable conditions (9 mV.kh-1 at

T=850oC, H2/H2O=50/50, i=-0,5 A.cm-2)

� Ni agglomeration studied with 3D reconstructions and modeling:

� The operating mode in SOFC vs SOEC has a negligible impact on Ni agglomeration.� Significant Ni coarsening at 850°C.� The Ni agglomeration represents around 30% of the loss in performances at 850°C

� Ni agglomeration studied with 3D reconstructions and modeling:

� The operating mode in SOFC vs SOEC has a negligible impact on Ni agglomeration.� Significant Ni coarsening at 850°C.� The Ni agglomeration represents around 30% of the loss in performances at 850°C

� LSCF destabilization and Sr reactivity with electrolyte:

� Formation of SrZrO3 in the porosities of the barrier layer strongly promoted in SOEC mode.� Mechanism proposed for LSCF destabilization under anodic polarization (SOEC mode).� The Sr loss and surface passivation is assumed to explain the higher degradation rates in

SOEC mode

� LSCF destabilization and Sr reactivity with electrolyte:

� Formation of SrZrO3 in the porosities of the barrier layer strongly promoted in SOEC mode.� Mechanism proposed for LSCF destabilization under anodic polarization (SOEC mode).� The Sr loss and surface passivation is assumed to explain the higher degradation rates in

SOEC mode

12

Thank you for your attention

Acknowledgements

European project: Endurance (lead by Prof. P. Piccardo)

European Project: Sophia (lead by Dr. R. Makkus)