LSM–YSZ interactions and anode delamination in solid oxide electrolysis cells

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

LSMeYSZ interactions and anode delamination in solid oxide electrolysis cells Michael Keane, Manoj K. Mahapatra, Atul Verma 1, Prabhakar Singh* Center for Clean Energy Engineering, Department of Chemical, Materials, and Biomolecular Engineering, University of Connecticut, 44 Weaver Rd., Unit 5233, Storrs, CT 06269-5233, USA

article info

abstract

Article history:

Symmetric cells of the configuration air/LSM//YSZ//LSM/air have been fabricated and

Received 22 June 2012

electrically tested under impressed voltage conditions to understand the anode delami-

Received in revised form

nation behavior commonly observed during the operation of solid oxide electrolysis cells

16 August 2012

(SOEC). Electrical performance degradation has been measured with time at various

Accepted 24 August 2012

applied voltages ranging from 0 to 0.8 V with respect to OCV, and cell component micro-

Available online 21 September 2012

structural and chemical changes have been examined. Post-test observations indicate the development of a weak anodeeelectrolyte interface leading to the delamination of the

Keywords:

anode from the electrolyte surface. Microstructural analysis of the anodeeelectrolyte

Solid oxide electrolysis cell

interface revealed extensive morphological and chemical changes including the formation

Strontium-doped lanthanum

of lanthanum zirconate, an uneven porous interface, and localized grain boundary porosity

manganite

in the electrolyte. An anode delamination mechanism based on morphological change and

Anodeeelectrolyte interface

compound formation at the anodeeelectrolyte interface is proposed.

Delamination

Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

Lanthanum zirconate

1.

Introduction

Hydrogen is essential for the upgrading and reforming of petroleum products. Other potential large-scale hydrogen applications include the production of synthetic hydrocarbon fuels via the FischereTropsch process, and direct use as a transportation fuel in emerging hydrogen fuel cell vehicles. Currently 96% of the commercial hydrogen is produced by conventional steam reforming and partial oxidation of abundant natural gas and liquid hydrocarbons [1,2]. Since both of the above processes produce significant carbon emissions, development of large-scale hydrogen production technologies with reduced emissions of greenhouse gases is of great commercial interest. Hydrogen produced from water electrolysis, as opposed to the reformation of hydrocarbons, offers

reserved.

the potential to be carbon-free when paired with a renewable energy source such as wind. The high temperature electrolysis of steam using solid oxide electrolysis cells (SOEC) has also been reported to show thermodynamic efficiency in excess of 50% [3]. An SOEC is comprised of a dense oxygen ion-conducting electrolyte layer and two porous electrodes. This mode of operation, whereby electrical energy is used to convert water into hydrogen and oxygen, is essentially the reverse of the electrode processes operating in a solid oxide fuel cell (SOFC). The electrolyte is typically a dense yttria-stabilized zirconia (YSZ). The fuel electrode (SOEC cathode) is usually a porous nickeleYSZ cermet composite whereas the air electrode (SOEC anode) is commonly a porous lanthanum strontium manganite (LSM) or a LSMeYSZ composite [4]. For large scale

* Corresponding author. Tel.: þ1 860 486 8379; fax: þ1 860 486 8378. E-mail address: [email protected] (P. Singh). 1 Current address: SiEnergy Systems, 85 Bolton Street, Cambridge, MA 02140, USA. 0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijhydene.2012.08.104

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

hydrogen production, the cells are connected in series using electronically conducting interconnects [5]. The open circuit voltage (OCV) of each cell in a stack varies from 0.9 to 1 V, depending on the exposure temperature and oxygen potential across the electrolyte (determined by the anode and the cathode gas compositions). The voltage applied to each cell is typically thermal neutral voltage, approximately 1.3 V. At this voltage, the thermal energy consumed by the electrochemical reactions is equal to the thermal energy produced due to ohmic heating in the cell, resulting in a highly efficient process. Higher voltages are avoided as the additional ohmic heating could result in localized overheating and high thermal gradients in the SOEC stack [6]. One of the most important developmental barriers in SOEC technology is that of long-term degradation, as high as 20% reduction in hydrogen production rate per 1000 h [7]. The degradation has been classified into two broad groups: (a) electrode poisoning due to gas phase contamination, and (b) gradual delamination of the electrode at the anodee electrolyte interface. The performance degradation associated with chromium poisoning of the air electrode [5,8] and silica poisoning of the fuel electrode [9e11] are well documented for high temperature electrochemical systems including SOEC and SOFC. It has been suggested that the transport of chromium from steel interconnects to the electrodeeelectrolyte interface could lead to the air electrode delamination [12]. The observations remain inconclusive as tests conducted in the absence of chromium-containing materials have also resulted in the electrode delamination [13]. Anode delamination from the electrolyte results in reduced electrochemically active area and increased ohmic loss; several processes for the delamination of the electrode have been postulated, including high oxygen pressure development, morphological changes in the LSM anode, and electrolyte grain boundary separation [5,13e16]. The anode delamination has been considered as the largest contributor to cell performance degradation [5]. It has been proposed that the delamination of the anode is due to the build up of high oxygen pressure at the anodee electrolyte interface as well as within the electrolyte due to differences in ionic and electronic conductivity between the electrode and the electrolyte. For a primarily electronic conducting anode such as LSM, high oxygen pressure can build up within the closed pores on the anodeeelectrolyte interface [17]. However, this mechanism for mechanical separation has been refuted by Momma et al., who observed that a platinum anode did not degrade significantly or delaminate in a long term experiment [13]. Authors showed that high oxygen pressures developed at the platinumeYSZ interface were not sufficient to cause delamination. Knibbe et al. [18] have proposed that the observed electrical performance degradation in the electrolysis cells consisting of an LSMeYSZ anode results from the formation of porosity in the YSZ electrolyte grain boundaries near the anodee electrolyte interface and not by the delamination of the anode. The porosity formation in the electrolyte is postulated to be due to high oxygen partial pressure build up because of an increase in electromotive potential in the YSZ electrolyte near the interface. Kaiser et al., on the other hand, concluded that the electrode delamination could be caused by the presence of doped aluminum in YSZ and formation of an electronically

16777

conducting MnAl2O4 layer near the anodeeelectrolyte interface [14]. Authors also suggest that the delamination can be prevented by coating a layer of pure YSZ near the interface. This is inconsistent with the observations of three groups that have observed delamination using a pure YSZ electrolyte [13,15,19]. Chen et al. concluded that delamination of the LSM anode from the electrolyte results from LSM disintegration near the anodeeelectrolyte interface [20]. The authors proposed that the electrode disintegration results from the development of tensile strains due to lattice shrinkage, caused by the formation of cation vacancies in the over-stoichiometric (oxygen excess) stability field of LSM at high oxygen partial pressures. However, this study is limited to a single, moderate current density. Authors do not report any resistive phase formation (e.g. lanthanum zirconate) between the LSM and YSZ despite the well-known reactivity between these materials [21]. In order to further the understanding of SOEC anode delamination, chemical and morphological changes at the anodeeelectrolyte interface have been investigated in this study using electrochemical tests with a symmetric cell configuration. Unlike the conventional full cell configuration, the selected test configuration simplifies the cell assembly and eliminates sealing requirements as both electrodes of the cell are exposed only to air. The electrochemical reaction associated with the oxygen reduction at the cathode remains identical to the cathodic reaction in an SOFC, while the oxygen ion oxidation at the anode is identical to the anodic reaction in an SOEC. The cell arrangement therefore provides a simple configuration where air electrode behavior in both SOFC and SOEC conditions can be compared. The use of symmetric cells also eliminates the possibility of degradation at the SOEC cathode due to gas phase nickel transport, coarsening and contamination resulting from silica. All ceramic cell construction excludes cathode poisoning due to gaseous species such as Cr6þ emanating from commonly used stainless steel interconnect materials. The symmetric cells were electrochemically tested at various operating voltages to measure the overall electrochemical degradation and examine the chemical, structural and morphological changes at the anodeeelectrolyte interfaces. A mechanism for the electrode delamination has been developed based on experimental results and observations.

2.

Experimental

2.1.

Symmetric cell fabrication

25 mm diameter symmetric button cells (air/LSM//YSZ//LSM/ air) consisting of 190 mm thick (ZrO2)0.92(Y2O3)0.08 (YSZ) electrolyte (Fuel Cell Materials) and 25 mm thick (La0.8Sr0.2)0.98MnO3x (LSM) electrode (Fuel Cell Materials) were fabricated for electrical testing and electrode delamination study. Concentric circular electrodes (anode ¼ 10 mm and cathode ¼ 20 mm) were applied on both sides of the electrolyte disc by screen printing technique using 105 mesh screen. The smaller diameter of the anode relative to the cathode ensured that the entire anode would be electrochemically active

16778

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

Thermocouple

Furnace

0.8

2.5

0.7 cm2 Resistance/

Potentiostat and impedance analyzer

Electrolyte Cathode and current collection

Data acquisition software

Oxygen

0.5

1.5

0.4

1

0.3 0.2

Non-ohmic Resistance

0.1

Ohmic resistance

0.5

Current density/A·cm-2

2

0.6

Anode and current collection

Current

Fig. 1 e Schematic of symmetric cell test apparatus.

0

0

0

20

40

60

80

100

Time/hr

during testing. Electrodes were subsequently sintered in air for 2 h at 1200  C. The electrochemical active area of the cell anode was calculated to be 0.8 cm2. Silver screen current collector (Alfa Aesar, 50 mesh) with silver wires (Alfa Aesar, 0.25 mm) were attached to each electrode using silvere palladium contact paste (Electro-Science Laboratories Inc., 15% Pd). Sintering of the current collector was performed in air for 1 h at 850  C. A schematic of the experimental test set up is shown in Fig. 1. A 25 mm diameter alumina tube was used to support the symmetric cell assembly, and the leads from a multi-channel potentiostat (VMP2, Bio-Logic) were attached to the assembled button cell. A type K (chromelealumel) thermocouple was placed within 5 mm of the cell to monitor the operating temperature.

Fig. 3 e Plots of average ohmic and non-ohmic resistances and current density during 100 h tests with 0.8 V applied.

rate maintained at 300 sccm. Although typical SOECs are operated at 0.3e0.4 V above OCV [6], cells in the current study were tested in a wide voltage range (0e0.8 V above OCV) to assess the degradation under simulated nominal and accelerated cell operating conditions.

a 1.2

Electrochemical testing

The cells were heated to 840  C at 3  C/min in flowing air. Using the potentiostat, a pre-determined constant voltage was applied for 100 h and the cell current was recorded every 60 s through the duration of the test. Impedance measurement was performed at four hour intervals using a 10 mV alternating current in the frequency range from 100 mHz to 200 kHz. Experiments were repeated several times at each imposed voltage condition to ensure reproducibility. All tests were performed under flowing air (2% water vapor) with flow

Ohmic Resistance (Ω)

1

2.2.

0.8 0.6

0.4 0.2 0 0

20

40

60

80

100

60

80

100

Time (hr)

b

1

Non-ohmic Resistance (Ω)

0.9

1 4 hours

0.8

16 hours 40 hours

-Z'' /

0.6

80 hours

0.4

0.8 0.7

50 sccm air

0.6 0.5 0.4 0.3 300 sccm air

0.2 0.1 0

0.2

0

20

40 Time (hr)

0 0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Z' /

Fig. 2 e Nyquist plots of impedance spectra obtained from a cell tested with 0.8 V from 4 to 80 h. For clarity, only four spectra are shown.

Fig. 4 e Plots of a) ohmic, and b) non-ohmic resistance changes with time for four cells tested at 0.8 V. The two tests shown with - and A were operated with 50 sccm flowing air, while the tests shown with C and: were operated with 300 sccm flowing air.

16779

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

Fig. 5 e a) LSM anode (right) delaminated from YSZ electrolyte (left) after testing for 100 h at 0.8 V, b) untested symmetric cell (anode side) consisting of an electrolyte disk, screen-printed electrodes, and silver mesh and wires attached with silver paste.

2.3.

Characterization

As-fabricated and tested cells were analyzed for morphological and chemical changes. Bulk electrode and electrodee electrolyte interfaces were examined for morphological changes and reaction products formation. Interfaces were further examined after dissolving the electrode in concentrated hydrochloric acid at room temperature for two hours in order to remove residual LSM from the electrolyte surface. An FEI Quanta 250 FEG scanning electron microscope (SEM) and energy-dispersive X-ray spectroscopy (EDS, attached to SEM) was used for the morphological and elemental distribution study. A Bruker AXS D-8 Advance X-ray diffractometer (XRD) was used for the identification of compounds present in both as-fabricated and tested cells.

3.

Results

Experimental results are presented in terms of (1) cell electrical performance degradation and changes in ohmic and non-ohmic impedances with applied voltage, (2) lanthanum zirconate (La2Zr2O7) formation at the anodeeelectrolyte interface and (3) morphological changes at the electrodee electrolyte interfaces.

3.1. Electrochemical measurements and post-test observations Fig. 2 shows a typical impedance spectra exhibited by the symmetrical cell assembly at 0.8 V. The skewed semicircle in the impedance plane represents the contributions from electrolyte (ohmic) resistance, oxygen surface exchange and

migration, as well as gas-phase oxygen transport. The highfrequency intercept represents ohmic resistance, while the semicircle diameter represents the sum of non-ohmic contributions [22,23]. Fig. 3 shows ohmic and non-ohmic resistances, as well as current density as a function of time for a cell tested at 0.8 V. The trend in resistance changes is similar to those observed at 0.3 V and 0.5 V (figures omitted for brevity). The 0.8 V tested cells show an initial improvement in both ohmic and non-ohmic performance with subsequent increase in both resistances after 20 h; the majority of the cell degradation after 4 h is due to ohmic resistance degradation. The non-ohmic resistances of cells tested at 0.5 V and 0.3 V begin to increase after approximately 50 h and 80 h, respectively. Current density for 0.8 V tested cells typically reaches a maximum of approximately 2.0 A/cm2 after an initial improvement. Fig. 4a represents the reproducibility of the ohmic resistance trend for four cells tested at 0.8 V. Using air flow rates below 300 sccm brought about variance in nonohmic resistance due to mass transport limitations, while increasing the air flow above 300 sccm had no effect on nonohmic resistance, indicating minimal contribution from mass transport limitations (Fig. 4b). Therefore, an air flow rate of 300 sccm was used for all experimental conditions. After completion of the electrochemical tests, cells were visually examined for electrode integrity and attachment with the electrolyte. A complete delamination of the anode from the electrolyte of each cell tested at 0.8 V was observed during the cell disassembly (Fig. 5a). The anode left behind a darkcolored imprint on the electrolyte after delamination. Anodes from the cells tested at 0.5 V and 0.3 V also showed anode delamination, however, the degree of delamination (area fraction of the anode delaminated) varied from 80% to 90%. The cathode layer from the untested cells, on the other

Table 1 e Rates of ohmic resistance (RU) and non-ohmic resistance (Rnon-ohmic) degradation between 20 and 90 h, and degree of post-test anode delamination as a function of applied voltage. Applied voltage (V) 0 0.3 0.5 0.8

RU degradation rate (U cm2/1000 h)

Rnon-ohmic degradation rate (U cm2/1000 h)

Degree of anode delamination

0.16 0.73 2.31 5.63

13.94 1.51 0.89 3.83

0% 80% 85% 100%

16780

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

YSZ Kα YSZ Kβ LZ

Intensity / a.u.

0.8 V 0.5 V 0.3 V OCV Untested

YSZ Substrate 25

30

35

40

45

50

55

60

2θ / °

Fig. 6 e X-ray diffraction patterns of electrolyte surfaces after hydrochloric acid treatment. Patterns of 0.8 V, 0.5 V, and 0.3 V cells are taken on the anode-side electrolyte surfaces. (:): (ZrO2)0.92(Y2O3)0.08 (Cu Ka, JCPDS 01-0704431); (-): (ZrO2)0.92(Y2O3)0.08 (Cu Kb, JCPDS 01-070-4431); (C): La2Zr2O7 (Cu Ka, JCPDS 00-050-0837).

hand, remained well adherent to the electrolyte in all tests (Fig. 5b). Table 1 summarizes the extent of anode delamination observed after the electrochemical measurements for the cells tested at four different voltages. Also shown are the average degradation rates of ohmic and non-ohmic resistances from 20 to 90 h after the beginning of each test. The 0.8 V tested cells exhibit an initial improvement in the performance for the first 20 h. Degradation rates of ohmic and non-ohmic resistances and degree of anode delamination increase with voltage bias level.

3.2.

Compound formation

Electrodeeelectrolyte interaction and interfacial compound formation were examined using the X-ray diffraction technique. Fig. 6 shows X-ray diffraction patterns from electrolyte

Intensity / a.u.

Anode side

Cathode side 30

35

40

45

50

55

surfaces of electrochemically tested and untested cells after dissolving of the electrodes in hydrochloric acid. Diffraction patterns are shown from as-received, as-fabricated, and OCV cell as well as electrically tested cells at 0.3, 0.5, and 0.8 V for 100 h. As the observed peaks are dominated by the yttriastabilized zirconia electrolyte substrate, the y-scale (peak intensity) was magnified to reveal the lower-intensity peaks of reaction compounds. The lanthanum zirconate (La2Zr2O7) diffraction pattern (JCPDS 00-050-0837) is identified for the electrically tested cells. The intensity of the zirconate peaks increases with the applied voltage. XRD patterns of the cathode and anode side surfaces of the electrolyte of electrically tested cells (Fig. 7) show that lanthanum zirconate is only present on the anode side of the cells. Also visible in Figs. 6 and 7 are small residual peaks of YSZ from copper Kb radiation.

3.3.

YSZ Kα YSZ Kβ LZ

25

Fig. 8 e Scanning electron micrograph of as-received electrolyte surface. Scale bar is 4 mm.

60

2θ / °

Fig. 7 e X-ray diffraction patterns of active electrolyte surfaces after testing at 0.8 V and dissolve electrodes with hydrochloric acid. (:): (ZrO2)0.92(Y2O3)0.08 (Cu Ka, JCPDS 01070-4431); (-): (ZrO2)0.92(Y2O3)0.08 (Cu Kb, JCPDS 01-0704431); (C): La2Zr2O7 (Cu Ka, JCPDS 00-050-0837).

Morphological observations and chemical analysis

The surface morphology of the as-received YSZ electrolyte is shown in Fig. 8. The surface shows typical dense granular structure with well-defined grain boundaries and localized isolated pores formed during sintering. Fractured electrolyte cross section showed the YSZ to be uniformly dense throughout its thickness. Fig. 9 shows the LSM electrode surface morphology before and after electrochemical measurements and delamination. The as-sintered electrodes (Fig. 9a) are 30e40% porous with a particle size in the range of 1e2 mm. The thermal grooving on the LSM grains is evident on the surface. Electrode surface morphology (in contact with the electrolyte) does not show noticeable changes due to testing (Fig. 9b). This observation is distinct from the reported LSM disintegration [20]. The electrodeeelectrolyte interface morphologies (obtained after the dissolution of electrode) of untested and electrochemically tested cells are shown in Fig. 10. Fig. 10a

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

16781

increases. The elevated ridges are wider and the contacting electrolyte surfaces show more pronounced roughening as well as higher coverage due to the formation of surface reaction product. Many of the YSZ grain boundaries have opened up into long connected pores. Fig. 10d shows the anode YSZ surface from cell exposed to 0.8 V for 100 h. The reaction product coverage is further exaggerated compared to the 0.3 V and 0.5 V cells. Grain boundary pores are also present. Energy dispersive X-ray spectroscopy of the anode-side electrolyte after testing at 0.8 V showed some lanthanum, strontium, and manganese left by the anode after delamination, in addition to the expected zirconium and yttrium from the electrolyte (Fig. 11a). After treating with hydrochloric acid, the strontium and manganese were removed, but lanthanum remained, which is consistent with the XRD detection of lanthanum zirconate phase (Fig. 11b). In summary, an increase in applied voltage on symmetric cells increases: 1. 2. 3. 4.

Degradation rate Fraction of anode area delaminated from electrolyte Lanthanum zirconate at anode-side electrolyte surface Morphological changes on anode-side electrolyte surface including elevated ridge broadening and grain boundary porosity formation

4.

Fig. 9 e Scanning electron micrographs of LSM electrode surfaces. Scale bar is 4 mm a) electrode surface as sintered, b) anode surface after delamination in a cell tested at 0.8 V.

shows the surface morphology as developed on the YSZ after screen printing and sintering of the electrodes. The electrolyte maintains its dense granular structure with evidence of the formation of a peripheral impression of the electrode in the form of elevated ridges. The observed surface morphology was not altered by subjecting cells to 850  C for 100 h with no applied voltage (OCV). Furthermore, the morphology on the cathode side was not altered in cells tested under applied voltage. Fig. 10b shows the electrochemically active anodee electrolyte interface of cells exposed to 0.3 V for 100 h. The active YSZ surface shows broadening of the elevated ridges and formation of small particles that cover part of the active YSZ surface. These changes are present throughout the entire anodeeelectrolyte interfacial area. Fig. 10c shows the anodee electrolyte interface from cell exposed to 0.5 V for 100 h. The electrolyte surface morphology appears similar to that observed on cell tested at 0.3 V, however, the extent of compound formation and grain boundary delineation

Discussion

Electrochemical testing of cells under impressed voltage conditions showed an increase in the both ohmic and nonohmic resistance with the applied voltage. Lanthanum zirconate formed at the anodeeelectrolyte interface, and the anode-side electrolyte surface experienced morphological changes including grain boundary porosity formation. A mechanism inclusive of changes in the electrical performance, interface morphology changes and compound formation has been developed and presented.

4.1.

Electrochemical observations

The electrochemical measurements show a correlation between applied bias voltage, anode delamination and increase in the both ohmic and non-ohmic resistances (Table 1). The initial decrease in both ohmic and non-ohmic resistances (Fig. 3) is explained by well-documented electrode conditioning process under cathodic current [24,25] and in LSMeYSZ composite electrodes under anodic current [26]. Jiang et al. have also proposed a re-equilibration process of surface strontium being responsible for the observed decrease in electrode resistance under cathodic currents [24]. Backhaus-Ricoult et al. related the improvement to activation of the manganite cathode due to an extension of the active area for oxygen incorporation from the triple phase boundary line to the free electrolyte surface due to mixed-conducting characteristics induced by manganese doping of the electrolyte [25]. The activation periods are 20, 50, and 80 h for cells tested at 0.8, 0.5, and 0.3 V, respectively, as these are the times at which the non-ohmic resistances reach their minimum values. The

16782

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

Fig. 10 e Scanning electron micrographs of anode-side electrolyte surface morphologies after dissolving LSM in hydrochloric acid. Each cell was tested at a different voltage for 100 h. Scale bar is 4 mm. Black arrows indicate lanthanum zirconate particles and white arrows indicate YSZ grain boundary porosity. a) Open circuit voltage (0 V), b) 0.3 V, c) 0.5 V, d) 0.8 V.

end of the activation period can be interpreted as the time when the rate of degradation equals the rate of activation. Lower voltage extends the activation period because the activation processes are slower at low currents. The long activation period in the 0.3 V tested cell is responsible for a net decrease in non-ohmic resistance as shown in Table 1. The subsequent increase in the ohmic and non-ohmic resistance is consistent with loss of contact and delamination at the anodeeelectrolyte interface as demonstrated theoretically [27] and experimentally [18]. The non-ohmic resistance increase is explained through the loss of triple phase boundary length resulting in a decrease in the area available for oxygen reduction and evolution. The observed increase in the ohmic and non-ohmic resistances is attributed to the morphological and chemical changes at the anodee electrolyte interface. Degradation at the cathode side is considered negligible as no visible morphological or chemical changes were observed at the cathodeeelectrolyte interface after electrochemical testing.

4.2. Compound formation and morphological observations The microscopic observations and X-ray diffraction indicate that major electrodeeelectrolyte interfacial changes are limited to the anodeeelectrolyte interface of cells operated

under electrical bias conditions. The interface morphology and reaction products formed under the experimental conditions remain similar; however, the severity changes with the applied bias. Since lanthanum zirconate was identified only on the anodeeelectrolyte interface of cells tested under applied voltage, the formation is attributed to the evolution of oxygen at that interface. Lanthanum zirconate formation at the anodee electrolyte interface is attributed to the following oxidative reactions (strontium is excluded for simplicity) [21,28,29]: LaMnO3þd þ zZrO2 þ

 0  d  d 3z þ O2 4La1z MnO3þd0 2 4

z þ La2 Zr2 O7 2 LaMnO3 þ ZrO2 þ 0:25O2 40:5La2 Zr2 O7 þ MnO2

(1)

(2)

Reaction (1) does not seem likely as it would require the LSM A-site deficiency to be further increased. The Gibbs free energy of reaction (2) decreases with higher oxygen pressures due to oxygen being a reactant, as follows:   0:5 aLZ aMO DG ¼ DG0 þ RTln aLM aZO a0:25 O

(3)

In Eq. (3), DG0 is standard Gibbs free energy, R is the gas constant, T is temperature, and the terms in parentheses are

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

16783

Fig. 11 e Energy-dispersive X-ray spectroscopy of anode-side active electrolyte surface after testing at 0.8 V: a) before hydrochloric acid treatment, b) after treatment.

the activities of the species of reaction (2) in terms of the reaction quotient. Using the thermodynamic data of Chen [28], the Gibbs free energy of the reaction becomes negative at oxygen partial pressure above approximately 3.7 atm, indicating a favorable forward reaction. Calculations for the oxygen pressure build up at the anodeeelectrolyte interface yield a theoretical maximum of 1013 atm at 0.8 V and 104 atm at 0.3 V (based on equations from [17]), and hence the formation of lanthanum zirconate should be thermodynamically favorable as observed experimentally. As the oxygen partial pressure at the cathodeeelectrolyte interface will be no greater than atmospheric oxygen pressure, lanthanum zirconate formation by reaction (2) is not expected on the cathode side, which is also in agreement with the experimental observations. The formation of pores along the YSZ grain boundaries (Fig. 10c and d) is associated with development of high internal oxygen pressure (as calculated) near the anodee electrolyte interface. The high pressure results from the

oxidation of oxygen ions to oxygen gas at defects or closed pores near the YSZ surface, which leads to additional pore formation and damage to the electrolyte [17,30]. The as-sintered sample revealed the formation of an impression of the LSM in the YSZ in the form of ridges as explained by the nucleation of epitaxial (Mn)eYSZ at the LSMeYSZ triple phase boundary (TPB) during sintering [21,31,32]. It has been proposed that the diffusion of manganese into the YSZ results in an excess of lanthanum in the LSM near the electrodeeelectrolyte interface and subsequent surface diffusion of La3þ and Zr4þ cations toward the TPB causes the formation of ridges and a lanthanum zirconate phase. Although the evidence for cation diffusion is seen in the form of thermal grooving in both LSM (Fig. 9a) and YSZ (Fig. 10a) [33], lanthanum zirconate did not form in detectable quantities in the as-sintered condition in the current study. During voltage application, the ridges broaden due to additional surface diffusion and lanthanum zirconate formation (Fig. 10bed).

16784

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

a

As - sintered LSM on YSZ • YSZ ridge formation at triplephase boundary due to surface cation migration

LSM

YSZ

b

Initiation of voltage application • Oxygen ions diffuse through YSZ bulk and grain boundaries

O2-

c

O2-

5.

O2-

Continued voltage application • La2 Zr2O 7 formation at the anode – electrolyte interface • Porosity formation in the electrolyte grain boundaries

La2 Zr2 O 7

O2-

d

lead to fracture and the observed delamination. It is known that lanthanum zirconate has a significantly lower thermal expansion coefficient (7.0  106  C1) than either YSZ (9.6  106  C1) or LSM (11.6  106  C1) [29]. These findings are in contrast to previous observations [17,20] which attribute delamination solely to mechanical breakdown of the electrolyte or electrode material. The high oxygen pressure at the interface will exacerbate the delamination of the anode and electrolyte.

End of electrical testing • Increased surface coverage of La 2 Zr2 O 7 and weakening of the interface • Increased porosity formation at the YSZ grain boundaries

Conclusions

The anode delamination in the solid oxide electrolysis cells has been examined by imposing a voltage bias on symmetric cells of the configuration air/LSM//YSZ//LSM/air. Lanthanum zirconate formation and morphological changes at the anodee electrolyte interface increase in severity with applied bias, while the cathodeeelectrolyte interface remains relatively unchanged. A mechanism for anode delamination comprising of interfacial compound formation, YSZ grain boundary porosity development, and other morphological changes has been proposed.

Fig. 12 e Schematic of chemical and morphological changes at the SOEC anodeeelectrolyte interface: a) assintered LSMeYSZ interface, b) voltage application initiated, c) voltage application in progress, d) voltage application complete.

Acknowledgments

4.3.

references

Anode delamination

Anode delamination is proposed to result from the development of an uneven anodeeelectrolyte interface, as well as the formation of lanthanum zirconate under electrical testing as schematically shown in Fig. 12. The as-sintered interface (Fig. 12a) shows the initial ridge formation of YSZ at the triple phase boundary due to cation diffusion and surface migration as previously discussed [21]. Under electrochemical testing, oxygen ions transport through the YSZ electrolyte via bulk and grain boundary diffusion from the cathode to the anode side. Oxygen ions at the YSZ grain boundaries and the anodeside LSMeYSZ interface (Fig. 12b) are oxidized to oxygen gas to provide electrons necessary for current flow. The YSZ grain boundaries develop pores near the interface due to high oxygen pressure. Lanthanum zirconate and manganese dioxide similarly form according to reaction (2), and accumulate at the LSMeYSZ interface (Fig. 12c). The poor conductivity of lanthanum zirconate [21] is also responsible for degradation and for part of the ohmic resistance increase. The processes continue with electrical testing, resulting in the coverage of entire anode-side LSMeYSZ with reaction products severely weakening the interface (Fig. 12d). Stresses developed at the interface due the mismatch in the thermal expansion coefficients of lanthanum zirconate, yttria stabilized zirconia and lanthanum strontium manganite may also

The authors thank ConocoPhillips and Idaho National Laboratory for financial support. Technical discussions with Dr. Boxun Hu and Na Li are greatly acknowledged. Mark Drobney and Joseph Csiki are acknowledged for assisting with cell test set up design and assembly.

[1] Ramachandran R, Menon RK. An overview of industrial uses of hydrogen. Int J Hydrogen Energy 1999;23(7):593e8. [2] Elder R, Allen RWK. Nuclear heat for hydrogen production: coupling a very high/high temperature reactor to a nuclear power plant. Prog Nucl Energy 2009;51:500e25. [3] O’Brien JE, McKellar MG, Harvego EA, Stoots CM. Hightemperature electrolysis for large-scale hydrogen and syngas production from nuclear energy e summary of system simulation and economics analyses. Int J Hydrogen Energy 2010;35:4808e19. [4] Yang C, Coffin A, Chen F. High temperature solid oxide electrolysis cell employing porous structured (La0.75Sr0.25)0.95MnO3 with enhanced oxygen electrode performance. Int J Hydrogen Energy 2010;35:3221e6. [5] Mawdsley JR, Carter JD, Kropf AJ, Yildiz B, Maroni VA. Posttest evaluation of oxygen electrodes from solid oxide electrolysis stacks. Int J Hydrogen Energy 2009;34:4198e207. [6] O’Brien JE, Stoots CM, Herring JS, Hartvigsen JJ. Performance of planar high-temperature electrolysis stacks for hydrogen production from nuclear energy. Nucl Technol 2007;158: 118e31. [7] Sharma VI, Yildiz B. Degradation mechanism in La0.8Sr0.2CoO3 as contact layer on the solid oxide electrolysis cell anode. J Electrochem Soc 2010;157(3):B441e8. [8] Ge L, Verma A, Goettler R, Lovett D, Singh Raman RK, Singh P. Oxide scale morphology and Cr evaporation characteristics

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 6 7 7 6 e1 6 7 8 5

[9]

[10] [11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

of alloys for balance of plant applications in solid oxide fuel cells. Metall Mater Trans A, submitted for publication. Ebbesen SD, Graves C, Hauch A, Jensen SH, Mogensen M. Poisoning of solid oxide electrolysis cells by Impurities. J Electrochem Soc 2010;157(10):B1419e29. Singh P, Vora SD. Vapor phase silica transport during SOFC operation at 1000  C. Ceram Eng Sci Proc 2005;26(4):99e110. Zhang K, Mahapatra MK, Verma A, Ge L, Bhowmick S, Singh P. Long term stability of Ni-YSZ SOFC anode. Lucerne, Switzerland: European Fuel Cell Forum; 2010. Hino R, Haga K, Aita H, Sekita K. R&D on hydrogen production by high-temperature electrolysis of steam. Nucl Eng Des 2004;233:363e75. Momma A, Kato T, Kaga Y, Nagata S. Polarization behavior of high temperature solid oxide electrolysis cells (SOEC). J Ceram Soc Jpn 1997;105(5):369e73. Kaiser A, Monreal E, Koch A, Stolten D. Reactions at the interface La0.5Ca0.5MnO3-YSZ/Al2O3 under anodic current. Ionics 1996;2:184e9. Guan J, Minh N, Ramamurthi B, Ruud J, Hong J, Riley P, et al. High performance flexible reversible solid oxide fuel cell. Final technical report, performed under DOE cooperative agreement DE-FC36e04GO14351. Torrance, CA.: GE Global Research Center; 2004e2006. Lim H, Virkar AV. A study of solid oxide fuel cell stack failure by inducing abnormal behavior in a single cell test. J Power Sources 2008;185:790e800. Virkar AV, Nachlas J, Joshi AV, Diamond J. Internal precipitation of molecular oxygen and electromechanical failure of zirconia solid electrolytes. J Am Ceram Soc 1990;73(11):3382e90. Knibbe R, Traulsen ML, Hauch A, Ebbesen SD, Mogensen M. Solid oxide electrolysis cells: degradation at high current densities. J Electrochem Soc 2010;157(8):B1209e17. Sohal MS, O’Brien JE, Stoots CM, Herring JS, Hartvigsen JJ, Larsen D, et al. Critical causes of degradation in integrated laboratory scale cells during high-temperature electrolysis. INL/EXT-09e16004 prepared for the US. Idaho Falls, ID: Department of Energy, Office of Nuclear Energy, Under DOE Idaho Operations Office, Contract DE-AC07-05ID14517, Idaho National Laboratory; May 2009. Chen K, Jiang SP. Failure mechanism of (La, Sr)MnO3 oxygen electrodes of solid oxide electrolysis cells. Int J Hydrogen Energy 2011;36:10541e9.

16785

[21] Mitterdorfer A, Gauckler LJ. La2Zr2O7 formation and oxygen reduction kinetics of the La0.85Sr0.15MnyO3, O2(g)jYSZ system. Solid State Ionics 1998;111:185e218. [22] Holme TP, Pornprasertsuk R, Prinz FB. Interpretation of low temperature solid oxide fuel cell electrochemical impedance spectra. J Electrochem Soc 2010;157(1):B64e70. [23] Adler SB. Mechanism and kinetics of oxygen reduction on porous La1xSrxCoO3d electrodes. Solid State Ionics 1998;111: 125e34. [24] Jiang SP, Love JG, Zhang JP, Huong M, Ramprakash Y, Hughes AE, et al. The electrochemical performance of LSM/ zirconia-yttria interface as a function of a-site nonstoichiometry and cathodic current treatment. Solid State Ionics 1999;121:1e10. [25] Backhaus-Ricoult M, Adib K, St. Clair T, Luerssen B, Gregoratti L, Barinov A. In-situ study of operating SOFC LSM/ YSZ cathodes under polarization by photoelectron microscopy. Solid State Ionics 2008;179:891e5. [26] Liang M, Yu B, Wen M, Chen J, Xu J, Zhai Y. Preparation of LSM-YSZ composite powder for anode of solid oxide electrolysis cell and its activation mechanism. J Power Sources 2009;190:341e5. [27] Gazzarri JI, Kesler O. Non-destructive delamination detection in solid oxide fuel cells. J Power Sources 2007;167: 430e41. [28] Chen M. Understanding the thermodynamics at the LaMnO3-YSZ interface in SOFC. Ph.D. dissertation. ETH Zurich; 2005. [29] Labrincha JA, Frade JR, Marques FMB. La2Zr2O7 formed at ceramic electrode/YSZ contacts. J Mater Sci 1993;28:3809e15. [30] Virkar AV. Mechanism of oxygen electrode delamination in solid oxide electrolyzer cells. Int J Hydrogen Energy 2010;35: 9527e43. [31] De Ridder M, van Welzenis RG, Denier van der Gon AW, Brongersma HH, Wulff S, Chu W-F, et al. Subsurface segregation of yttria in yttria-stabilized zirconia. J Appl Phys 2002;92(6):3056e64. [32] Clausen C, Bagger C, Bilde-Sørensen JB, Horsewell A. Microstructural and microchemical characterization of the interface between La0.85Sr0.15MnO3 and Y2O3-stabilized ZrO2. Solid State Ionics 1994;70/71:59e64. [33] Mullins WW. Theory of thermal grooving. J Appl Phys 1957; 28(3):333e9.