Boccaccini2014.pdf

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 x x x ( 2 0 1 4 ) 1 e1 2

Available online at www.sciencedirect.com

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

Creep behaviour of porous metal supports for solid oxide fuel cells D.N. Boccaccini*, H.L. Frandsen, B.R. Sudireddy, P. Blennow, A˚.H. Persson, K. Kwok, P. Vang Hendriksen Department of Energy Conversion and Storage, Technical University of Denmark, Frederiksborgvej 399, DK-4000 Roskilde, Denmark

article info

abstract

Article history:

The creep behaviour of porous ironechromium alloy used as solid oxide fuel cell support

Received 9 May 2014

was investigated, and the creep parameters are compared with those of dense strips of

Received in revised form

similar composition under different testing conditions. The creep parameters were

23 July 2014

determined using a thermo-mechanical analyser with applied stresses in the range from 1

Accepted 24 July 2014

to 15 MPa and temperatures between 650 and 800
C. The GibsoneAshby and Mueller

Available online xxx

models developed for uniaxial creep of open-cell foams were used to analyse the results. The influence of scale formation on creep behaviour was assessed by comparing the creep

Keywords:

data for the samples tested in reducing and oxidising atmospheres. The influence of pre-

Metal supports

oxidation on creep behaviour was also investigated. In-situ oxidation during creep ex-

Creep

periments increases the strain rate while pre-oxidation of samples reduces it. Debonding of

High-temperature ferritic stainless

scales at high stress regime plays a significant role affecting the creep behaviour of the

steel

metal supports, in particular the stress exponent. The variation of the elastic modulus as

Solid oxide fuel cells

function of temperature and oxidation conditions was also determined by a high temperature impulse excitation technique. Additionally nano-indentation testing was performed in the metal oxide interface to elucidate the mechanical properties of the oxide scales and qualitative information about the oxide scale-metal interfacial bonding. Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Fuel cells and hydrogen (FCH) technologies are among the key innovations that Europe will have to rely on in order to reach its ambition of a low carbon economy. Hydrogen is an energy carrier like electricity but with the unequalled advantage of being storable in various forms and transportable in various

modes. In tandem with a fuel cell, hydrogen as a fuel provides the opportunity for a safe, carbon-free energy provision pathway, allowing flexible and decentralised power generation in multiple applications, with zero-emission at point of use. As such fuel cell and hydrogen technology is key in the European energy policy (e.g. enabling the storage and uptake of renewable intermittent energies) and in the European sustainable transport policy (e.g. providing clean transport).

* Corresponding author. Tel.: þ45 4677 4730; fax: þ45 4677 5858. E-mail addresses: [email protected], [email protected] (D.N. Boccaccini). http://dx.doi.org/10.1016/j.ijhydene.2014.07.138 0360-3199/Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

2

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 x x x ( 2 0 1 4 ) 1 e1 2

Because of its relevance for decarbonising critical European economic sectors, hydrogen and fuel cell technology can play an important role in the necessary technological transition and subsequently contribute to Europe's energy security, sustainability and competitiveness objectives. Nowadays one of the most promising fuel cell technologies is based on solid oxide electrolyte. A solid oxide fuel cell (SOFC) is an electrochemical device to convert chemical energy stored in fuels into electrical power with high efficiency. Solid cell components and high operating temperature, typically in the range from 650
C to 850
C, are the most important characteristics of an SOFC, as compared with other fuel cells [1]. In order to enhance the SOFC's long-term stability, reduce SOFC cost and potentially improve component reliability, recently development of metal supported SOFC (MS-SOFC) has been undertaken [1e9]. These MS-SOFCs are regarded as possible next generation fuel cells. The requirements for suitable metallic porous supports components of planar SOFCs are stringent, due to the influence of oxidising atmospheres at high temperatures in combination with stresses induced during thermal cycling caused by load cycles in stationary applications [10,11]. High temperature corrosion of porous alloys can be a severe process and effectively limit the operation regime of the MSSOFCs [12e16]. Moreover, when the MS-SOFCs are contacted with interconnectors, the thin electrolyte may bend as it rests on a soft metallic support. This of course also depends on the cathode layer. Therefore, the metal support materials must possess high creep resistance to avoid electrolyte bending. Furthermore, the metallic porous support must have a thermal expansion coefficient (CTE) similar to that of the electroactive ceramic components to avoid significant thermal stresses in the cell. The effect of CTE mismatch on the thermal stress distribution in a SOFC stack can be even greater than that of temperature gradients alone [17]. Although thermal stresses may not result in failure of the metallic porous supports immediately, they may generate creep deformation and cracks in the ceramic electrodes/electrolytes leading to the degradation of stacks integrity and electrochemical performance under long-term operation [4,18]. Therefore, porous metal supports in planar SOFC should provide adequate high-temperature creep resistance. So far mainly high temperature corrosion resistance was of concern, but this must be followed by creep resistance for a successful operation as fuel cell support. Consequently the study of thermal and mechanical properties of candidate support materials in terms of creep is of high importance for the performance and design of a reliable and durable SOFC stacks. Primary and secondary creep behaviour of MS-SOFCs metal supports must be experimentally characterised and modelled to be able to predict the response of the materials under service life conditions. Whether the primary or the secondary creep is more relevant in the study of metal supports is strictly related to the time scale of operation of the SOFCs. In this work, we focus on the characterisation and modelling of secondary creep. For metallic materials most creep experiments are conducted in uni-axial tensile mode [19]. However, uni-axial compression experiments are also used for brittle materials

to avoid stress crack propagation. For most materials creep properties are independent of loading direction [19]. In the creep experiments performed in reduction environment, hydrogen could induce corrosion embrittlement in the pure metal and porous support and thereafter exerts certain influences on corresponding material's properties. However, it has been clearly stated that the susceptibility to H2 embrittlement is greatly reduced with high Cr content [20]. The high Cr content employed in the formulation of the metal supports here investigated makes them not susceptible to the embrittlement. Furthermore, we did not observe any grain boundary cracking in the microstructure and this confirms that there is no embrittlement effect after testing or exposure to 9%H2. The aim of this study is to perform a thermo-mechanical analysis to acquire the creep parameters for the porous ferritic chromium stainless steel metal support (MS) (similar to the support described in Ref. [4]). The creep experiments were carried out by means of a thermo mechanical analyser (TMA) for stresses in the range from 1 to 15 MPa at temperatures between 650 and 800
C. Furthermore, the creep parameters of Crofer® 22 APU were also acquired and validated against values found in literature [21]. Subsequently the porous MS's behaviour is compared with that of Crofer® 22 APU, taken as zero porosity reference material. The presence of the oxide scales formed during preoxidation of samples before testing and also scale formation during creep testing can affect differently the strain rate of the porous metals. Therefore, isothermal and iso-stress experiments were carried out in both air and reducing atmosphere for in as-sintered samples to assess the influence of scale formation during creep. Experiments were also carried out in reducing atmosphere on pre-corroded samples for the assessment of scale thickness and its correlation to creep resistance. The results of the isothermal and iso-stress experiments allow the determination of the stress exponents and the activation energies of the creep process in the investigated materials. The variation of the elastic properties (i.e. Young's modulus) with temperature of the MS and Crofer® 22 APU was further investigated by high temperature impulse excitation technique (HT-IET). Additionally, nano-indentation testing was performed in the metal-oxide interface to obtain the mechanical properties (Young's modulus and hardness) of the scales and qualitative information about the scale-metal interfacial bonding.

Theory Creep of materials is the time-dependent deformation under an applied load. Generally it occurs at high temperature (thermal creep), but it can also takes place at room temperature in certain materials (e.g. lead, polymer or glass). In metals creep deformation generally occurs when they are subjected to a load at a temperature exceeding 60e70% of the melting temperature [22]. During typical creep experiments, strain (relative change in length) is measured as a function of elapsed time at constant temperature and stress level. Creep test data is presented as a

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

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 x x x ( 2 0 1 4 ) 1 e1 2

plot between time and strain known as “creep curve”. The slope of the creep curve corresponds to the creep rate d3/dt. A typical creep curve presents three characteristic regions [19,23,24]. I) Primary or transient creep: a period of decreasing creep rate where the resistance to creep increases until stage II II) Steady state creep: a period of roughly constant creep rate where the rate of strain-hardening is balanced by the rate of recovery III) Tertiary creep: the fracture stage region, where there is an accelerating creep rate due to the accumulated damage, and the reduction in the cross sectional area by necking or internal void formation. There are two main mechanisms, by which steady state creep takes place: diffusion creep and dislocation creep. Diffusion creep, driven by a gradient of free energy (chemical potential) created by the applied stress occurs by transport of material via diffusion of atoms within a grain [19,21,22,24,25]. Dislocation creep involves motion of dislocations. This mechanism of creep tends to dominate at high stresses and relatively low temperatures [22]. Since all mechanisms of steady-state creep are in some way dependent on diffusion and thermally activated processes, the dependence of creep on the absolute temperature can be described by an Arrhenius equation. Furthermore, the minimum secondary creep rate has a high dependence on stress, which is typically described by the Norton Power law [26]: 3 _ss

¼ AII  sn

(1)

where AII ¼ AexpðQ=RTÞ represents a temperature dependant material constant [24], s is the uniaxial applied stress, Q the activation energy, R the universal gas constant, A is the Dorn constant, n is the stress exponent. The stress provides a driving force for dislocation movement and diffusion of atoms. Thus, as the stress is increased, the rate of deformation also increases. Prediction of the power “n” value from first principles is not easy, as its value depends on which mechanism of creep is operating. For example, for diffusion creep “n” value is approximately 1, while for dislocation creep it is usually varying from metal to metal in the range 3e8 (the most common value is “n” ¼ 5) [21]. The activation energy, Q, is determined experimentally from the slope of a plot of the natural logarithm of creep rate against the reciprocal of temperature. The dependence of strain rate with porosity can be modelled with the GibsoneAshby [27] model developed for uniaxial creep of an open-cell reticulated foam. 3_

¼

 n 0:6 1:7ð2n þ 1Þ rð3nþ1Þ=2 AII sn nþ2 n

3

at al [28]. This model yields a simple expression for the steadystate deformation of cellular metals that obeys a power law creep: 3_

1þn=2 ðn1=2Þ

¼ FE

r

AII sn

(3)

where FE ¼ EF =Es , EF is the Young's modulus of the porous support and Es is the Young's modulus of the dense metal. Different models are reported in literature [29] for the determination of the elastic modulus of a foam with a good overview in Ref. [30]. For high relative density open-cell foams (r ~ 0.8), all the models give similar results, as can be observed in Fig. 10 of ref. [30]. We used equation (4) reported in Ref. [30] for high relative density open-cell foams:  EF ¼ Es

0:971  Vp 0:971

2:15


for 1  Vp > 0:2

(4)

where Vp is the pore volume fraction or foam porosity (the MS is 18%).

Experimental Materials and sample preparation The mechanical experiments were performed on the commercially available dense ferritic steel Crofer® 22 APU (Crofer) and in-house fabricated porous ferritic steel alloys (MS). The fabrication of the MSs was reported elsewhere [4]. The chemical composition of Crofer is composed of ~22 wt.% of chromium with ~0.5 wt.% of Mn and with some other minor elements (La,Ti) balanced by iron [21,31]. The composition of MS is similar to that of Crofer. Samples were laser cut with dimensions of (15 ± 1) mm  (0.3 ± 0.01) mm  (0.5 ± 0.01) mm and (60 ± 1) mm  (0.3 ± 0.01) mm  (10 ± 1) mm from both dense Crofer sheets and sintered porous MS for creep and HTIET testing, respectively. The MS samples and Crofer samples were laser cut perpendicular to the tape casting and cold rolling, respectively. Some of these samples prepared for creep testing were exposed to 850
C for 24 h in Ar/H2O/H2 atmosphere (pH2O/pH2 ¼ 9) (hereafter termed MS-24h). Samples before and after mechanical testing were metallographically prepared for microstructural investigations and the analysis was carried out using scanning electron microscope (SEM, Zeiss Supra 35). Furthermore some of these samples were etched in a solution of 5 ml HCl, 1 ml HNO3 and 3 ml Glycerol to reveal the microstructure of the metals through selective chemical attack. Then the samples were observed under an optical microscope (Aristomet, Leica) to assess the microstructure. The porosity was determined by mercury intrusion (Autopore IV 9500V1.05 from Micromeritics Instrument Corporation, Norcross, GA).

(2)

Testing methods where 3_ and r are the foam strain rate and relative density, respectively. At higher relative density, classical composite micromechanical theory is typically used to predict the deformation of porous solids, e.g. the variational model adapted by Mueller

High temperature impulse excitation technique (HT-IET) Rectangular samples of Crofer and MS were suspended in the nodes of their first bending vibration mode with platinumerhodium threads, and positioned in the HT-IET-furnace

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

4

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 x x x ( 2 0 1 4 ) 1 e1 2

(HT1600, IMCE, Genk, Belgium). The samples were subsequently subjected to a thermal cycle from room temperature to 700
C with a heating and cooling rate of 5
C/min and a dwell time of 2 h. The samples were excited periodically by the impact of a small ceramic projectile [32]. The resonant flexural frequencies of vibration are deduced from the digital signal by Fast Fourier Transform (FFT). The Young's and shear moduli are calculated using the ASTM E 1876-99 standard [33]. When the fundamental resonant frequency as well as the mass and the dimensions of the sample are accurately known, the Young's modulus can be calculated as [33] : E ¼ 0:9465

mff2 l3 bd3

T1

(5)

where m is sample mass, ff is the fundamental resonant frequency, b, l and d the width, length and thickness of the bar respectively and T1 a correction factor depending on the thickness-to-length ratio and Poisson ratio of the material. For the samples considered here the thickness-to-length ratio is small (0.005e0.008) and the value of T1 is very close to unity, and was therefore set to 1. The experiments were performed in air.

Creep experiments The creep rate was measured by uni-axial tension in a thermo-mechanical analyser (TMA) (NETZSCH, TMA 402 F1 Hyperion). The isothermal and iso-stress experiments, required for the determination of the creep parameters were conducted in reducing and oxidising atmosphere. The isothermal experiments were performed at 700
C for 72 h applying different loads to cover the stress range from 3 to 17 MPa. The temperature sweep, from room temperature to 700
C, dwell time and cooling back to room temperature, was performed in a flow of 50 ml min1 in air (pO2 ¼ 0.2 atm) or in 9% hydrogen in nitrogen (1024 < pO2 < 1012 atm, from hereafter written as 9%H2) at heating/cooling rates of 5
C min1. Iso-stress experiments were conducted at 650e700e750e800
C at 15 MPa for 72 h to determine the activation energy of the considered materials. This iso-stress experiments were performed in a flow of 50 ml min1 in dry 9% H2/N2 (1024 < pO2 < 1012 atm) at heating/cooling rates of 5
C min1. Additionally, creep experiments were performed on MS24h in identical reducing conditions as in the above experiments to assess the influence of the existing oxide scale on creep behaviour. Each specimen was held at the specified temperature for 15 min prior to applying the load. The creep strain-time curves at various temperatures were then obtained from these experiments.

Nano-indentation testing Nano-indentation is a useful testing method to determine mechanical response of small volumes and to probe the indentation response of bulk materials at an extremely low load regime, where indentation induced cracking can be largely avoided [34]. Nanoindentation experiments were performed on smoothcut surfaces of samples with a TI-950 TriboIndenter (Hysitron,

Inc., Minneapolis). The indenter shape and instrument frame compliance was calibrated by indenting a fused silica sample having an elastic modulus of 72 GPa. The tip is a Berkovich Indenter with a nominal radius of 50 nm. The machine compliance is 0.46 nm/mN. The unloading data are used to determine the mechanical properties based on the indentation theory [35], since the initial unloading portion of the loaddepth curve represents purely elastic recovery. These indentation experiments were conducted in a quasistatic mode with a displacement controlled loading of the indenter to the maximum displacement of 50 nm in 5 s, a 2 s hold segment and a 5 s unloading. The in-situ SPM imaging mode was used to rasterscan the sample zone prior to the indentations and to identify the spots most suited for the indentation experiment (smoothness and level). The in-situ SPM image was needed to identify the two distinct sample zones: the metal base and the oxide layer. The in-situ SPM images was used to study the interfaces between metal and oxide scales with high resolution and help to identify micro-cracks and debonding between metal and oxide. The indenter used for the imaging procedure was a Cube Corner indenter geometry. Moreover a Modulus mapping technique [36,37] was employed for high resolution mechanical mapping of the metal to oxide scale layer interface.

Results and discussion Influence of sample preparation Due to the relatively small sample dimensions, the laser cutting process might induce significant changes in the material. In order to observe these changes the microstructure of the laser cut region of the samples was compared with the microstructure of the non-laser cut region of the sample. Fig. 1 shows the microstructures of the both laser cut and non-laser cut samples. The microstructures are almost identical without any noticeable changes in pore size and pore distribution. This indicates that the laser cutting did not introduce any microscopic changes inside the material. It could be expected that for the MS and Crofer the creep might depend on the loading direction. As the porous samples is tape casted, there is a certain amount of orientation of the pore-former component in the green tape. This could give a tendency towards oriented microstructure in the finished, sintered, porous structure and can of course affect the mechanical properties. However, no significant orientation was discerned in the multiple micrographs analysed in this study and was therefore not considered. In the case of Crofer, the creep parameters may change as function of the cold rolling direction. To be able to completely disregard the effect of possible orientation during tape casting and cold rolling on creep behaviour more extensive position analysis is necessary. The measured total porosity of the MS was 18 ± 0.1%.

Elastic properties Fig. 2 shows the measured Young's modulus as a function of temperature of the dense Crofer and porous MS. The elastic

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

5

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 x x x ( 2 0 1 4 ) 1 e1 2

Fig. 1 e SEM micrographs showing the microstructure of the porous metal support investigated for a) laser cut and b) nonlaser cut specimens.

Creep properties of non-aged materials

This means that the strain hardening effect was rapidly recovered to make the creep strain increase at a constant rate soon after the load was applied. _ ” was deterThe minimum steady-state strain rate “3 ss mined from the recorded strain-time curve by performing the analysis of the first derivative with a commercial software for data processing (Origin®). Fig. 4 shows the strain-time curves of Fig. 3 together with their first derivate. The Double logarithmic plots of minimum strain rate vs applied stress (Norton

E-Modulus MS heating E-Modulus MS dwell E-Modulus MS cooling E-Modulus Crofer heating E-Modulus Crofer dwell E-Modulus Crofer cooling

240

Crofer

220

Young's modulus (E) (GPa)

modulus at room temperature before the thermal treatment was measured by IET to 60.0 ± 0.3 GPa. The Young's modulus “E” decreases more or less linearly with the increase in the temperature down to 32.5 ± 0.6 GPa at 700
C (this corresponds to a decrease of 45% from the room temperature value). In the case of Crofer, the room temperature elastic modulus is 229.0 ± 0.1 GPa which decreases to 160.7 ± 0.2 GPa at 700
C, a decrease of 30% with respect to the room temperature elastic modulus value. This decrease in Young's modulus with temperature is lower than that of MS, probably due to the effect of porosity [38]. After the thermal treatment, the Young's modulus of Crofer and MS are 231.5 ± 2.4 and 62.1 ± 0.1 GPa, respectively. Thus, the increments of room temperature Young's modulus values after the thermal treatment for Crofer and MS were 2.5 ± 0.3 GPa (~1.2%) and 2.0 ± 0.2 GPa (~3%) respectively. This increase in the value of the elastic modulus is due to the reduction of porosity caused by the formation of oxide scales in the samples. The Young's modulus of this scale was determined by nano-indentation testing to 247.9 GPa (s.d. 6.9 GPa) (see below), i.e. 20% higher than the Young's modulus of Crofer.

200 180 160 140 120 100 80

Metal support

60 40

Typical strain-time plots for the metal supports subjected to different stresses (3.5 MPa, 8.2 MPa, 11.5 MPa and 17.3 MPa respectively) in the first 24 h at 700
C are shown in Fig. 3. Only the first 1300 min are plotted to show the primary creep in detail. The creep curves at all the investigated stresses have a very short period (300e600 min) of primary creep.

0

100

200

300

400

500

600

700

800

Temperature (°C)

Fig. 2 e Young's modulus as a function of temperature during heat-up up to 700
C and cool-down of porous metal support (MS) and Crofer tested in air.

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

6

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 x x x ( 2 0 1 4 ) 1 e1 2

Fig. 3 e Creep curves of the metal support investigated for different applied loads at 700
C in 9% H2/N2 for a) 3.5 MPa, b) 8.2 MPa, c) 11.5 MPa and d) 17.3 MPa.

plots) are reported in Fig. 5. Table 1 reports all the experimental creep parameters deduced from these curves using Equation (1) together with the values of Crofer taken from literature [21]. Changes in the linearity in Norton plots are related to changes in the creep mechanism in the materials [21,26]. Fig. 5 shows a linear trend indicating that there is no significant change in the creep mechanism of the metal support, for the stress regime investigated (3e12 MPa). For applied stresses higher than 12e15 MPa a change in the creep mechanism could be expected in ferritic stainless steels, as reported by Kuhn [21] for Crofer. A slight change in linearity of Crofer samples and also for the metal support can also be seen at the higher stresses (12e15 MPa) of the Norton plots shown in Fig. 5 (see the curves for MS-9%H2 and Crofer), probably due to the above mentioned change in the creep mechanism. However, the high correlations coefficients obtained for all the Norton plots (reported in Table 1) suggest that no changes in the creep mechanism occur in the stress regime investigated (3e12 MPa). The stress exponent “n” and the AII constant for Crofer (Table 1) are in agreement with those found in literature [21].

The values of the creep parameters of the metal support in 9% H2 (MS-9%H2) are very close to those of Crofer, when tested in 9%H2 (Crofer 9%H2). This is expected, since both materials have comparable chemical compositions. However, the porous MS shows higher strain rate values than those obtained for Crofer (taken in this paper as nil porosity reference for modelling proposes), which results in the higher AII values of the Norton's power law (see Table 1). The increase of strain rate with porosity can be explained considering that for increasing porosities, the yield strength decreases and as consequence the strain rate increases. Fig. 6 shows the Arrhenius plot of the data obtained from the iso-stress experiments performed at 15 MPa for MS and Crofer materials. The obtained values of activation energy are reported in Table 1. Fig. 5 also shows the fitting curve of the model developed by GibsoneAshby [27] for uniaxial creep of an open-cell reticulated foam (eq. (2)) for the porous MS experimental data tested in 9%H2(MS-9%H2). The stress exponent “n” and AII constant for Crofer and the value of relative density of the porous MS were used as input in the model. Observing Fig. 5, it is evident that there is a lack of agreement of this model with

Fig. 4 e Creep curves of Fig. 3 and corresponding creep rates curves of the metal support investigated for different applied loads at 700
C in 9% H2/N2 for a,e) 3.5 MPa, b,f) 8.2 MPa, c,g) 11.5 MPa and d,h) 17.3 MPa. Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

7

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 x x x ( 2 0 1 4 ) 1 e1 2

Fig. 5 e Norton plots (secondary creep rate vs. stress) of porous ferritic steel supports and Crofer. Fitting models are also reported.

the experimental data (MS-9%H2). This could be expected as the model is developed for foams and is based on open structures with low relative densities (r ~ 0.2 and below). The assumption of the model is that the microstructure can be represented by thin beams, which is quite different from the structure seen in Fig. 1. Fig. 5 also shows fittings with the model developed by Mueller et al. [32] for intermediate and high relative density open-cell foams. In the Mueller's model labelled “Mueller IET” in Fig. 5, the Young's modulus of the porous MS and Crofer determined by IET and the stress exponent “n” of Crofer were used as model inputs to predict the creep behaviour of the porous MS. Mueller's model requires the Young's modulus as an input (see eq. (3)). The dynamic Young's modulus determined by the IET may differ from the static Young's modulus values determined by uniaxial tensile testing, as reported in literature [39]. For this reason, in the Mueller model result labelled “Mueller RG” in Fig. 5, the Young's modulus of the foam was calculated by eq. (4), which gives EF ¼ 128 GPa. Then, the ratio FE ¼ EF =Es is 0.65 and the fitting with the Mueller model give a better correlation with the experimental data, as Fig. 5 shows. In case a linear correlation between porosity and elastic modulus is assumed, as other authors reported for cast steels [40], the Young's modulus of the MS is 164 GPa, FE is 0.82, and the Mueller model returns a correlation coefficient of R2 ¼ 0.97 when fitting the experimental data (MS-9%H2). However, the actual Young's

Fig. 6 e Arrhenius plots obtained from the iso-stress experiments performed at 15 MPa for MS and Crofer.

modulus of the MS as determined by IET is quite different from that predicted by eq. (4) and also to that required for the attainment of an acceptable fitting in Mueller's model. This fact indicates the inadequacy of these models to study the MS investigated in this work. The activation energies of the materials were determined from the iso-stress experiments performed at 650e700e750e800
C at 15 MPa using Equation (1). The analysis of the Arrhenius plots obtained (not reported here) led to the attainment of the corresponding activation energy values, reported in Table 1. Fig. 7 a) and b) show post-mortem images of the microstructure of Crofer after the iso-stress creep experiments (9%H2eN2, 15 MPa) at 650
C and 800
C on etched samples, respectively. At higher magnification comparing Figures c,d, it is possible to observe some grain growth in the Crofer samples after the iso-stress creep experiments at 800
C. This slight increase in grain size has no effect on the creep behaviour of the material, as can be deduced from the linearity of the Arrhenius plot of Fig. 6, for the conditions of the iso-stress creep experiment performed in this work (800
C for 72 h at 15 MPa). However, for higher exposure times to creep stresses, it could be expected that the more sever grain growth likely occurring might affect the creep behaviour of Crofer.

Creep properties in air: influence of oxide scale growth The applied creep models assume that the foam stain-rate dependence on stress (through the stress exponent “n”) and

Table 1 e Creep parameters of porous metal support and Crofer 22 APU in different atmospheres. Sample name

Materials

MS-9%H2 MS-air MS-24h-9%H2

Metal support Metal support Metal support (corroded 24 h)

Crofer 9%H2 e

Crofer 22 APU H2 Crofer 22 APU [21]

Pre-corrosion treatment

Creep testing atmosphere

Stress range (MPa)

AII (MPan s1)

n

Q (KJ mol1)

R2

e e 24he850
C Ar/H2O/H2 (pH2O/pH2 ¼ 9) e e

9% H2/N2 Ambient 9% H2/N2

0e15 0e15 0e15

5.36Ee12 8.54Ee10 8.383Ee14

3.15 2.14 4.34

330 e e

0.99 0.97 0.99

9% H2/N2 e

0e12 <10

3.02Ee12 1.39Ee13

3.27 2.3

358 e

0.99

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

8

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 x x x ( 2 0 1 4 ) 1 e1 2

Fig. 7 e aed): Post-mortem optical micrographs of iso-stress crept samples after etching of Crofer material performed at 15 MPa for a,c) 650
C and b,d) 800
C. The comparison of the images at higher magnification(c,d, x50) put in evidence the occurrence of grain growth at 800
C.

temperature (through the activation energy Q, included in AII) are equal to those of the monolithic dense material [41]. However, as mentioned previously, these models do not take into account the effects of the growing oxidation scale. Recently models by Maharjan et al. [42] and Ruan et al. [43] were proposed for dense metals to describe the influence of residual stress evolutions in an oxide/metal system during isothermal oxidation. However, since these analytical models still require adaptation to porous systems, none of them were applied here. Further suitable analytical models are being developed by the authors to predict the creep behaviour of metal foams in the range of 18e20% porosity, including the effect of oxide growth.

The creep experiments performed in air (MS- air) show higher creep rates (highest AII) than under reducing conditions (MS-9%H2), while the stress exponent is observed to decrease (cf. Table 1). The simultaneous effects of oxide scale formation and creep can increase the creep rate [44]. This is in agreement with results from Asensio-Jimenez et al. [31], where increase in the strain rate as a consequence of scale growth in Crofer has been reported. In particular, they ascribed the higher strain rate obtained for the creep test performed in air to the development of tensile stresses in the dense metal by the oxide scale. After a given exposure time, the creep resistance of the metal substrate decreases and as a consequence the creep rate becomes sufficiently high to cause stress relaxation in the oxide layer [31]. Huczkowski et al. [45] and Zurek et al.

Fig. 8 e a) and b): Optical micrograph showing the post-mortem microstructure of MS etched samples iso-thermally tested in air at 3 MPa. At higher magnification (x50), it is possible to observe the formation of a thin scale layer inside the pores due to the in-situ oxidation occurring during the creep experiment, as indicated by the black arrows. Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

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 x x x ( 2 0 1 4 ) 1 e1 2

9

Fig. 9 e a) and b): post-mortem microstructure of pre-oxidised MS etched samples after the iso-thermal stress tests performed at 15 MPa. It is evident in the images the presence of a thick layer of scale from the pre-oxidation treatment, as indicated by the black arrows, which is responsible of a remarkable decrease in the strain rate. [46] proposed that this relief in stress due to oxide growth by creep of the metal substrate is responsible for the increase in the oxidation kinetics. Then, an increasing compressive growth stress (sox), which is built-up in the oxide scale, will generate an increasing tensile stress (sM) in the metallic substrate [31]: sM ¼ 2sox

x d

(6)

where x: is the thickness of the oxide layer formed on both sides of a flat substrate of thickness d. However, in the case of the porous metal supports investigated in this work, there must be a competition between the tensile stresses developed by the scale layer on the outer surface of the sample and the compressive stresses developed by the scales grown inside the pores. This could be the reason for the decrease of the stress exponent “n” in comparison with the specimens tested in 9%H2 (MS-9%H2). Fig. 8 a) and b)show the post-mortem microstructure of MS etched samples iso-thermally tested in air at 3 MPa. It can be observed in these figures the presence of a thin layer of scale inside the pores due to in-situ oxidation. This in-situ oxidation is responsible, as mentioned above, of changes in the creep behaviour of these materials, contributing with an increase in the strain rate. It is worthwhile to remark that some oxidation also occurs during the creep experiments performed in the reducing conditions set in this investigation (9%H2). In fact, the weight gain of the MSs after the creep experiments in 9%H2 is ~0.3% ± 0.1%. The reasons can be ascribed to the lack of complete inertness of the TMA chamber and also to the gas composition used during creep experiments (9%H2), which cannot completely avoid some possible nitrification of the metal. In the case of creep experiments performed in air, the weight gain is 1.13% ± 0.3%. We believe that this difference in weight gain between the experiments performed in 9% H2/N2 and the experiments performed in air is significant enough to permit the comparison of strain rate values and to assess the influence of oxidation growth in creep behaviour. It is also possible to see the influence of creep stresses in oxidation kinetics by comparing the weight gain of samples for specimens after creep test and specimens oxidised in the same TMA chamber without being subject to the stresses of creep

test. In particular, the weight gain of a MS sample tested at 15 MPa is ~50% higher than that of the sample oxidised in the same chamber without stresses. This difference can be ascribed to the fact that the creep stresses can break the scales formed during the experiments, then new metal areas are free to undergo further oxidation. It can be also due to the deformation of the crystalline lattice of the metal from the creep stresses, which could induce higher ion mobility leading to an increase of the oxidation kinetics of the metal. Further experimental studies must be carried out for a complete characterisation and modelled of the influence of the stress state on oxide growth of MSs.

Creep properties of the aged materials In pre-oxidised metal support samples, the weight gain after the pre-oxidation treatment was 3.17 ± 0.17 (D wt.%). Fig. 9 a) and b) show the post-mortem microstructure of the pre-oxidised MS etched samples after the iso-thermal stress tests performed at 15 MPa. It is evident in the images the presence of a thick layer of scale, which is responsible of a remarkable decrease in the strain rate. In this case, the creep experiments

Fig. 10 e Micrograph from post mortem SEM analysis performed on the pre-oxidised MSs. The black arrows indicate the delamination between the oxide scale and the metal.

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

10

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 x x x ( 2 0 1 4 ) 1 e1 2

were carried out in reducing conditions (9%H2). Fig. 5 shows that the strain rate of the pre-oxidised metal support sample (MS-24h-9%H2) is the lowest for all the investigated conditions (lowest AII), whereas the stress exponent “n” is higher than in the case of standard metal support in reducing conditions (MS-9%H2). In this case, the oxide growth due to the creep test is negligible (oxide was already formed during pre-oxidation at higher temperature), and the above mentioned mechanism of development of tensile stresses due to oxide growth is not present. The lower strain rate for the pre-oxidised metal support sample can be attributed to the increase in creep resistance due to the presence of the scale layer [44]. However, at a first glance, the increase in “n” could be considered anomalous, since it should be expected that due to the preoxidation state of the samples, the material behaviour might tend to that of a metal-ceramic composite (“n” of a ceramic is ~1 while “n” of a metal is ~>3). An explanation of this behaviour could be hypothesised from the post mortem SEM analysis (Fig. 10) of the samples after the creep experiments and the results obtained from the nano-indentation testing on the oxide scale-metal interface (Fig. 11). In the low stress regime (~3 MPa) the stresses developed during the creep testing are not high enough to debond the corrosion scale from the metal, and as consequence, the tensile stress is transferred through both the metal and the oxide layer. However, at high stress regime (>10 MPa), the stresses are high enough to debond the scale layer (see Fig. 10) and as consequence the tensile stresses are transferred through the metal only. This effect results in strain rates tending to that of the pure porous metal at the high stress regime. This combination of effects could increase the stress exponent “n” in comparison with MS in reducing conditions (MS-9%H2).

Nano-indentation The mechanical properties (Young's modulus and hardness) of the metal and scales present in the porous MS microstructure were also investigated by nano-indentation testing in the preoxidised sample before creep testing. Fig. 11 shows an in-situ SPM image of the scale-metal interface. On the oxide layers the in-situ SPM imaging is capable of guiding the indentations with very high precision being possible to place indentations in an oxidised area that is not regularly shaped. Therefore, the insitu SPM imaging is important to allow imaging of the tested region and precisely positioning before indentation. 15 indents were performed on the metal and on the oxide layer of the precorroded sample (24 h). The Young's modulus of the metal of the support was found to be 200 ± 5.2 GPa and the hardness 4.7GPa ± 0.1, while the Young's modulus and the hardness of the oxide scale layer were 247.9 ± 6.9 GPa and 31.5 ± 0.6 GPa. In addition to the mismatch in thermal expansion coefficient, the elastic modulus mismatch can be also considered a factor responsible for the debonding of the scale layer [47]. For this reason, modulus mapping was performed to determine the mechanical properties in the micro scale at the scale-metal interface with the highest resolution. Furthermore, from the modulus mapping reported in Fig. 12, it is also possible to identify an interface layer of a thickness of around 300 nm

Fig. 11 e In-situ SPM images of the scale-porous metal interface obtained from the nano-indentation testing a) before and b) after nano-indentation (pre-oxidised sample before creep testing).

between the oxide scale (Mn,Cr)3O4 and the metal that has a high elastic modulus. This layer could be ascribed to the Cr2O3 oxide scale layer, which is a typical corrosion product for this type of alloys [10,13]. It shows a severe mismatch with the elastic properties of the metal and the external (Mn,Cr)3O4 oxide scale layer. This gap in the elastic properties of adjacent layers can lead to localised high stress thermally induce zones, which can induce weakening of the metal-scale bonding interface for increasing ageing.

Conclusion The creep parameters of a porous ferritic stainless steel SOFC metal support are reported. The stress exponent of the porous metal support is close to the value obtained for the dense ferritic stainless steel in reducing conditions, indicating a similar creep mechanism in the metal. Instead the higher value of the pre-exponential constant of the porous samples

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

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 x x x ( 2 0 1 4 ) 1 e1 2

11

effect of stresses may permit higher ion mobility and as consequence increase the oxidation kinetics. However, the reasons of this behaviour are complex to explain and will be the subject of focused studies. In-situ oxidation resulted in an increase of the strain rate due to the development of tensile stresses in the metal. Modulus mapping by nano-indentation testing identifies a layer of higher stiffness between metal and oxide scale that could lead to the debond of the oxide layer due to the development of thermal stresses. The mechanical properties of the metal, oxide scale and their interfacial bonding strength affect the creep strain significantly. The good agreement of measured creep parameters of Crofer with the published literature validates the measurement techniques used in this study.

Acknowledgement The research leading to these results has received funding from EU's Fuel Cells and Hydrogen Joint Technology Initiative under grant agreement n
278257 (METSAPP) and from Energinet.dk under the project ForskEL 2012-1-10806. D.N.Boccaccini would like to thank Dr F. Teocoli for helping with TMA analysis and Dr Ude Hangen and Dr Jaroslav Lukes for their help to perform the nano-indentation testing.

references

Fig. 12 e Nano-DMA modulus mapping obtained by nanoindentation testing at the scale-metal interface for a) contact force (mN), b) amplitude (nm) and c) storage stiffness (N/m) showing a layer of high elastic modulus.

indicates a higher strain rate. The scale formation during creep testing or the presence of thick oxide scale layer (preoxidation treatment) before testing affects the creep behaviour differently. Pre-oxidised samples gained more weight due to debonding of the oxide scales facilitating in-situ oxide growth during the creep testing at high stresses that lead to an increase of the strain rate. Furthermore, crystalline lattice strain due to the

[1] Haile SM. Fuel cell materials and components. Acta Mater 2003;51:5981e6000. ˚ H, Stegk T, [2] Klemensø T, Nielsen J, Blennow P, Persson A Christensen BH, et al. High performance metal-supported solid oxide fuel cells with Gd-doped ceria barrier layers. J Power Sources 2011;196:9459e66. [3] Chiu Y-T, Lin C-K. High-temperature stress-rupture properties of a ferritic steel for solid oxide fuel cell interconnect. ECS Trans 2011;35:2581e90. The Electrochemical Society. [4] Blennow P, Hjelm J, Klemensø T, Ramousse S, Kromp A, Leonide A, et al. Manufacturing and characterization of metal-supported solid oxide fuel cells. J Power Sources 2011;196:7117e25. ˚ H, Pec anac G, [5] Glasscock JA, Mikkelsen L, Persson A Malzbender J, Blennow P, et al. Porous Fe21Cr7Al1Mo0.5Y metal supports for oxygen transport membranes: thermomechanical properties, sintering and corrosion behaviour. Solid State Ionics 2013;242:33e44. [6] Metcalfe C, Harris J, Kuhn J, Marr M, Kesler O. Progress in metal-supported axial-injection plasma sprayed solid oxide fuel cells using nanostructured NiO-Y0.15Zr0.85O1.925 dry powder anode feedstock. J Therm Spray Technol 2013;22:599e608. [7] Tucker MC. Progress in metal-supported solid oxide fuel cells: a review. J Power Sources 2010;195:4570e82. [8] Linderoth S. Solid oxide cell R&D at Risø National Laboratorydand its transfer to technology. J Electroceramics 2009;22:61e6. [9] McKenna BJ, Christiansen N, Schauperl R, Prenninger P, Nielsen J, Blennow P, et al. Advances in metal supported cells in the METSOFC EU consortium. Fuel Cells 2013;13:592e7.

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138

12

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 x x x ( 2 0 1 4 ) 1 e1 2

[10] Fontana S, Chevalier S, Caboche G. Metallic interconnects for solid oxide fuel cell: effect of water vapour on oxidation resistance of differently coated alloys. J Power Sources 2009;193:136e45. [11] Nicholls JR, Bennet MJ, Simms NJ, Hattendorf H, Smith AJ, Naumenko D, et al. Lifetime extension of FeCrAIRE-alloys in air. Potential role of enhanced Al reservoir and surface pretreatments (SMILER). In: Professor M Schu¨tze (KWIDechema, Frankfurt G, Quaddakkers WJ, editors. Nov. approaches to improv. high temp. corros. resist. Cambridge, England: Woodhead Publishing Limited; 2008. p. 129. [12] Sarasketa-Zabala E, Otaegi L, Rodriguez-Martinez LM, Alvarez MA, Burgos N, Castro F, et al. High temperature stability of porous metal substrates under highly humidified hydrogen conditions for metal supported solid oxide fuel Cells. Solid State Ion 2012;222e223:16e22. [13] Molin S, Gazda M, Jasinski P. High temperature oxidation of porous alloys for solid oxide fuel cell applications. Solid State Ion 2010;181:1214e20.  rez A. [14] Sotomayor ME, Ospina LM, Levenfeld B, Va Characterization of 430L porous supports obtained by powder extrusion moulding for their application in solid oxide fuel cells. Mater Charact 2013;86:108e15. [15] Jasinski P, Molin S, Kusz B, Gazda M. Evaluation of porous 430L stainless steel for SOFC operation at intermediate temperatures. J Power Sources 2008;181:31e7. [16] Molin S, Chen M, Hendriksen PV. Oxidation study of coated crofer 22 APU steel in dry oxygen. J Power Sources 2014;251:488e95. [17] Fergus JW. Metallic interconnects for solid oxide fuel cells. Mater Sci Eng A 2005;397:271e83. [18] Nakajo A, Kuebler J, Faes A, Vogt UF, Schindler HJ, Chiang LK, et al. Compilation of mechanical properties for the structural analysis of solid oxide fuel cell stacks. Constitutive materials of anode-supported cells. Ceram Int 2012;38:3907e27. [19] Kassner ME. Fundamentals of creep in metals and alloys. 2nd ed. Elsevier; 2009. [20] Kawashima A, Hashimoto K, Masumoto T. Hydrogen embrittlement of amorphous Fe-Cr-Mo alloys; 2013. [21] Kuhn B, Jimenez CA, Niewolak L, Hu¨ttel T, Beck T, Hattendorf H, et al. Effect of laves phase strengthening on the mechanical properties of high Cr ferritic steels for solid oxide fuel cell interconnect application. Mater Sci Eng A 2011;528:5888e99. [22] Frost HJ, Ashby MF. Deformation-mechanism maps. Oxford: Pergamon Press; 1982. [23] Scott JA, Dunand DC. Processing and mechanical properties of porous Fee26Cre1Mo for solid oxide fuel cell interconnects. Acta Mater 2010;58:6125e33. [24] Ashby F, Evans AG, Fleck NA, Gibson LJ, JWH and HNGW. Design for creep with metal foams. Met. foam. des. guid. Butterworth-Heinemann Elsevier Ltd; 2000. p. 103e12. [25] Nabarro FRN. Report of a conference on the strength of solids. Phys Soc Lond 1948:75e90. [26] Norton FH. The creep of steel at high temperatures. New York: McGraw-Hill; 1929. [27] Andrews EW, Gibson LJ, Ashby MF. The creep of cellular solids. Acta Mater 1999;47:2853e63. [28] Mueller R, Soubielle S, Goodall R, Diologent F, Mortensen A. On the steady-state creep of microcellular metals. Scr Mater 2007;57:33e6. [29] Roberts AP, Garboczi EJ. Elastic properties of model random three-dimensional open-cell solids. J Mech Phys Solids 2002;50:33e55.

[30] Despois J-F, Mueller R, Mortensen A. Uniaxial deformation of microcellular metals. Acta Mater 2006;54:4129e42. [31] Asensio-Jimenez C, Niewolak L, Hattendorf H, Kuhn B, Huczkowski P, Singheiser L, et al. Effect of specimen thickness on the oxidation rate of high chromium ferritic steels: the significance of intrinsic alloy creep strength. Oxid Met 2012;79:15e28. [32] Roebben G, Bollen B, Brebels A, Van Humbeeck J, Van der Biest O. Impulse excitation apparatus to measure resonant frequencies, elastic moduli, and internal friction at room and high temperature. Rev Sci Instrum 1997;68:4511. [33] ASTM E 1876e1899. Standard test method for dynamic Young’s modulus, shear modulus and poisson's ratio by impulse excitation of vibration; 2000. ASTM Committee E-28 on Mechanical Testing, subcommittee E2803 on Elastic Properties. n.d. [34] Abdoli H, Alizadeh P, Boccaccini D, Agersted K. Fracture toughness of glass sealants for solid oxide fuel cell application. Mater Lett 2014;115:75e8. [35] Oliver WC, Pharr GM. Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology. J Mater Res 2004;19:3e20. [36] Asif SA, Wahl KJ, Colton RJ. Nanoindentation and contact stiffness measurement using force modulation with a capacitive load-displacement transducer. Rev Sci Instrum 1999;70:2408. [37] Asif SAS, Wahl KJ, Colton RJ, Warren OL. Quantitative imaging of nanoscale mechanical properties using hybrid nanoindentation and force modulation. J Appl Phys 2001;90:1192e200. [38] Vasechko V, Singheiser L. Thermo-mechanical investigations of reoxidation-stable material concepts for solid oxide fuel cells. RWTH Aachen; 2014 [PhD thesis], http://darwin.bth. rwth-aachen.de/opus3/volltexte/2014/5064/. [39] Friedl O, Motz C, Peterlik H, Puchegger S, Reger N, Pippan R. Experimental investigation of mechanical properties of metallic hollow sphere structures. Metall Mater Trans B 2007;39:135e46. [40] Hardin RA, Beckermann C. Effect of porosity on the stiffness of cast steel. Metall Mater Trans A 2007;38:2992e3006. [41] Boonyongmaneerat Y, Dunand DC. Effects of strut geometry and pore fraction on creep properties of cellular materials. Acta Mater 2009;57:1373e84. [42] Maharjan S, Zhang XC, Xuan FZ, Wang ZD, Tu ST. Residual stresses within oxide layers due to lateral growth strain and creep strain: analytical modeling. J Appl Phys 2011;110:063511. [43] Ruan JL, Pei Y, Fang D. Residual stress analysis in the oxide scale/metal substrate system due to oxidation growth strain and creep deformation. Acta Mech 2012;223:2597e607. [44] Krzyzanowski M, Beynon JH, Farrugia DCJ. Oxide scale behavior in high temperature metal processing. Wiley-VCH Verlag GmbH & Co. KGaA; 2010. [45] Huczkowski P, Christiansen N, Shemet V, Piron-Abellan J, Singheiser L, Quadakkers WJ. Oxidation induced lifetime limits of chromia forming ferritic interconnector steels. J Fuel Cell Sci Technol 2004;1:30. € nsel M, Penkalla HJ, [46] Zurek J, Young DJ, Essuman E, Ha Niewolak L, et al. Growth and adherence of chromia based surface scales on Ni-base alloys in high- and low-pO2 gases. Mater Sci Eng A 2008;477:259e70. [47] Hutchinson JW, Suo Z. Mixed mode cracking in layered materials. Adv Appl Mech 1991;29:63e191.

Please cite this article in press as: Boccaccini DN, et al., Creep behaviour of porous metal supports for solid oxide fuel cells, International Journal of Hydrogen Energy (2014), http://dx.doi.org/10.1016/j.ijhydene.2014.07.138