Evaluation of LSF based SOFC cathodes using cone-shaped electrodes and EIS

Solid State Ionics 344 (2020) 115096

Contents lists available at ScienceDirect

Solid State Ionics journal homepage: www.elsevier.com/locate/ssi

Evaluation of LSF based SOFC cathodes using cone-shaped electrodes and EIS

T

K. Kammer Hansen Department of Energy Conversion and Storage, Technical University of Denmark, DK-4000 Roskilde, Denmark

ABSTRACT

Seven compositions of (La1-xSrx)0.99FeO3-δ (x = 0, 0.05, 0.15, 0.25, 0.35, 0.50, 0.70) perovskites were synthesized using the glycine-nitrate method. The (La1xSrx)0.99FeO3-δ compounds were characterized with powder X-ray diffraction, dilatometry, total conductivity and electrochemical impedance spectroscopy on coneshaped electrodes using a Ce1.9Gd0.1O1.95 electrolyte. Dilatometry shows that the thermal expansion coefficient of the (La1-xSrx)0.99FeO3-δ compounds increases with increasing strontium content. Total conductivity finds is maximum for LSF35. The activity of the (La1-xSrx)0.99FeO3-δ based perovskites towards the electrochemical reduction of oxygen was strongly dependent of the strontium content, the activity being highest for the composition (La0.85Sr0.15)0.99FeO3-δ. The results indicates that Fe(III) is the catalytic active specie towards the electrochemical reduction of oxygen in a solid oxide fuel cell on (La1-xSrx)0.99FeO3-δ compounds. The results also show that oxide ion vacancies in the perovskite structure are important for the electrochemical reduction of oxygen. However, a negative effect of defect ordering (Sr2+ and oxide ion vacancies) is strongly indicated for the strontium rich compounds. Segregation of SrO to the surface might also play a detrimental effect on the activity towards the oxygen reduction reaction.

1. Introduction Solid oxide fuel cells (SOFC's) are convenient devices for environmental friendly production of heat and electricity [1]. A major obstacle that must be overcome before commercialization of the SOFC is a lowering of the costs. One way to achieve this is by decreasing the operation temperature. The international efforts have been very successful in this respect over the latest 25 years during which the operation temperature have been decreased from ca. 1000 °C to ca. 650 °C [2]. On the cathode side it is generally accepted that Co or FeeCo based perovskites have the highest performance [3]. However, Co or FeeCo based perovskites reacts with the commonly used electrolyte, YSZ (Yttria Stabilized Zirconia) forming insulating layers between the cathode and the electrolyte [4,5], lowering the overall performance of the fuel cell drastically. One way to overcome this is to add a barrier layer between the electrode and the electrolyte [6]. Another problem with Co or FeeCo based cathodes is that they have a high thermal expansion coefficient (TEC) above the TEC of the commonly used electrolytes [7]. To some extent this also counts for LSF. However, LSF do not contain carcinogenic Co and LSF with a low amount of Sr do have a favorable TEC (see this paper). La1-xSrxFeO3-δ (LSFx, where x is the Sr-cation percentage) perovskites are mixed electronic and ionic conductors [8–20]. The electronic conductivity is p-type, under oxidizing conditions, and both the electronic and ionic conductivity finds its maximum for the LSF50 perovskite [20].

A single systematic study of the activity of LSF based SOFC cathodes have been undertaken in the literature [21]. It was shown that strontium substituted ferrites has a much higher activity towards the reduction of oxygen than LaFeO3, but apart from this no systematic trend was found. Several attempts to use ferrites as SOFC cathodes have been reported in the literature [22–26]. It has been shown that LSF20 has a performance superior to manganate based cathodes, but inferior to cobalt based cathodes. The performance of LSF40 is at the same order of magnitude as manganate based cathodes ((La0.8Sr0.2)0.92MnO3-d and LSM40) [25,26]. To be able to characterize different electrodes in detail, the use of model electrodes can be highly useful. Model electrode studies in the literature are mainly based on thin film electrodes; see i.e. [27,28]. In this study LSFx perovskite compounds are studied as cathodes for the electrochemical reduction of oxygen in a SOFC using cone-shaped electrodes and EIS. The use of point (or cone-shaped) electrodes is a method to evaluate electrode materials. The cone-shaped electrodes technique has several advantages: the cone and electrolyte are sintered separately, and thereby formation of a reaction layer between the electrode and electrolyte during processing is avoided. The interface is ideally quite simple and the contact area can be calculated using Newman's formula [29]:

r=

1 4Rs

(1)

where Rs is the intercept with the real axis in the impedance plot at high

E-mail address: [email protected]. https://doi.org/10.1016/j.ssi.2019.115096 Received 29 August 2019; Received in revised form 2 October 2019; Accepted 10 October 2019 0167-2738/ © 2019 Elsevier B.V. All rights reserved.

Solid State Ionics 344 (2020) 115096

K.K. Hansen

frequency, and σ* is the specific conductivity of the electrolyte. r is the radius of the contact between the cone-shaped electrode and the electrolyte, from which the contact area can be calculated. Some measurements on ceramic cone-shaped electrodes have already been undertaken [30–34]. It seems like this technique is very useful when comparing different electrode materials, even though the unavoidable roughness of both the cone tip and the electrolyte surface make the electrode-electrolyte interface less ideal. In spite of this measurement on ceramic cone-shaped electrodes may be fairly reproducible [35]. 2. Experimental 2.1. Synthesis Powders of the LSFx perovskites were made by the glycine-nitrate process [36]. The A/B ratio in the ferrites was 0.99. One batch of each powder was made. Before starting the actual synthesis, aqueous solutions of metal-nitrates were made. The concentrations of metal ions, where determined with gravimetry. In short aqueous solutions of the metal-nitrates were mixed in the appropriate ratio in a beaker. Glycine was then added and the solutions were heated on a hot plate until they ignited. The resulting powders were then transferred to alumina crucibles and calcined at 1100 °C/12 h in air in a box furnace. The following metal-nitrates were used: La(NO3)3•6H2O (Alfa Aesar, 99.9%), Sr(NO3)2 (Alfa Aesar, 99%) and Fe(NO3)3•9H2O (Alfa Aesar, 98%). The phase purity of the LSF perovskites was checked by the use of powder Xray diffraction using a Stoe theta-theta diffractometer equipped with Cukα radiation. The diffractograms were recorded in the interval 2θ: 20 to 80o.

Fig. 1. Sketch of the experimental set-up. The set-up is a pseudo-three electrode set-up with a large counter/reference electrode (Pt) and a cone with a small contact area.

electrode. The platinum was added as a paste (Engelhard) and sintered in-situ at 800 °C. A Solartron 1260 in standalone mode was used for the electrochemical characterization of the cone-shaped electrodes. The frequency interval spanned was 1 MHz to 0.05 Hz (or 0.01 Hz at 600 °C). 5 points/decade and ten cycles were measured at each frequency. An amplitude of 24 mV was used throughout. The treatment of the data was done using the PC-DOS program ‘equivcrt’ by B.A. Boukamp [38]. In general the spectra were fitted with three (RQ)’s in series with a series resistance. Q is a constant phase element with the admittance:

2.2. Fabrication of samples

Y = Y0 (j /

Cylinders for fabrication of the bars and the cone-shaped electrodes were made as follows. The powders were pressed in appropriate dies and sintered at 1250 °C/12 h in air in a box furnace. The resulting bars were machined into bars, with the approximately dimensions 4*4*18 mm, and the cylinders were machined into cones. The cones had a base diameter of 7.5 mm and sides with an angle 45o. A pellet of CGO10 (Ce1.9Gd0.1O1.95) was used as electrolyte. The CGO10 pellet was made by mixing CGO10 powder (Rhodia) with stearic acid and glycerine in a ball mill overnight with EtOH. After drying the powder was pressed in an appropriate die and sintered at 1500 °C/2 h in a box furnace in air. The pellet was polished before use.

0)

n

(2)

where Y0 is a constant, ω is the cyclic frequency, and n is an exponent. Y0 and n are found from the fitting. 3. Results The results from the powder XRD experiments are summarized in Table 1, where the lattice parameters of the LSF compounds can be found. The lattice parameters are seen to depend on the strontium content. All the strontium containing LSF based perovskites belongs to the orthorhombic crystal system, whereas the strontium free perovskite is cubic. The results of the dilatometry measurements can be found in Table 2. The measurements are performed from 25 to 1000 °C. The thermal expansion coefficient (TEC) is seen to increase with increasing strontium content. The plot of the log of the total conductivities as a function of 1/T is shown in Fig. 2. The conductivity shows its maximum for LSF35. An example of a spectrum from the EIS measurements can be found in Fig. 3 together with the result from the fitting. The spectrum consists of three arcs. This is valid for all the compounds at all the measured temperatures. The ASR values as a function of Sr content at 700 °C is shown in Fig. 4. It is seen that the total ASR finds a minimum for x-values in the region of 0.15–0.25. The highest total ASR value is found for the strontium rich compound LSF70 with a total ASR of 43.6 Ωcm2 at 700 °C. The total ASR's measured on the seven ferrites at the three temperatures can be found in Table 3. The total ASR is the polarization resistance without the electrolyte resistance. The total ASR is lowest for the compound LSF15 at all temperatures. Tables 4 and 5 gives the contributions of the three individual arcs at 700 °C and 800 °C. They also depend on the amount of strontium in the LSF based perovskites, most noteworthy the low frequency arc.

2.3. Characterization of samples The dilatometry measurements were performed by the use of a Netzsch 402 dilatometer in the temperature range 30 to 1000 °C with heating and cooling ramps of 2 °C/min. The bars were kept at 1000 °C for 2 h before cooling down. The conductivity measurements were performed on the same bars, as follows: The bars where placed in an in-house build sample holder. Before being placed in the sample holder Pt-wires where attached to the end of the bars with Pt-paste. The two voltage probes were jointed with the bars with a spring load. The measurements were performed as follows: The bars where heated in increments of 50 °C with a heating range of 2 °C/min. At each temperature the samples were kept for 2 h. The resistance where measured with a Keithley micro-ohmmeter at each 5th min. The electrochemical characterizations of the iron-based perovskites were done in a set-up described elsewhere [37], see also Fig. 1. The measurements were done in air at temperatures of 800, 700 and 600 °C in the given order. The cone-shaped electrodes were kept at temperature for 24 h before recording of the electrochemical impedance spectrum. The measurements were repeated for the LSF05 and LSF50 cones on another electrolyte pellet. Platinum was used as a counter/reference 2

Solid State Ionics 344 (2020) 115096

K.K. Hansen

Table 1 Unit cell parameters for the synthesized compounds obtained from powder XRD data. All the compounds belong to the orthorhombic crystal system except the strontium free ferrite which belongs to the cubic crystal system. The numbers in brackets are the uncertainties. The unit cell parameters are given with the number of numbers found from the fitting of the data.

A/nm B/nm C/nm

LSF00

LSF05

LSF15

LSF25

LSF35

LSF50

LSF70

0.39286(9) – –

0.7640(3) 0.5543(3) 0.5341(3)

0.7655(3) 0.5578(3) 0.5339(9)

0.7636(3) 0.53348(21) 0.5565(3)

0.7632(3) 0.5563(3) 0.53419(22)

0.7631(3) 0.55616(3) 0.5344(3)

0.7632(3) 0.55606(23) 0.53409(20)

C = R(1

−Ζimag / Ω

The so called near-equivalent capacity of the medium and low frequency arcs can be calculated using the following equation [39]: (3)

n)/ n Y 1/ n, 0

The results are given in Table 6. It is seen that the compound LSF15 has the highest capacity.

ASR / Ωcm2

x 2FeFe

(5) (6)

• FeFe + FeFe

6000

8000

10000

40

The dissolution of SrFeO3 results in the formation of FeFe . The defect equilibrium is established through the following two reactions: • OOx + 2FeFe

4000

50

•

x VO•• + 2FeFe + 1/2O2

2000

Fig. 2. Total conductivities for LSF perovskites. The conductivity shows a maximum for LSF35.

(4)

• SrLa + FeFe + 3OOx

6.31 Hz

Zreal / Ω

Before starting the discussion it can be of use to outline the defect chemistry of the LSF perovskites. Dissolution of SrFeO3 in LaFeO3 can be described as (following the Kröger-Vink notation): LaFeO3

631 Hz 631 kHz

0

4. Discussion

SrFeO3

2000 1600 1200 800 400 0

Reaction [4] is the incorporation of oxygen into the perovskite matrix resulting in the annihilation of oxygen ion vacancies and the formation of tetravalent iron. The formation of oxygen vacancies is thermal activated, meaning that reaction 4 is displaced to the left with increasing temperature. This is confirmed in [40]. Reaction [5] is a charge disproportionation reaction. As mentioned in the introduction, it has been shown that both the electronic and ionic conductivity increases with increasing strontium content, until x equal to 0.5, where after they decreases [20]. The dilatometry measurements show that the TEC increases with increasing amount of strontium. The explanation for this is probably that the bonding energy for SrO is approximately 2/3 of the bonding energy of La2O3. This is due to size of Sr and La and charge. The TEC of the strontium rich ferrites is much too high to be used together with the commonly used electrolytes, CGO10 or YSZ (Yttria stabilized Zirconia). However, for the low strontium substituted ferrites, the TEC is well matched with CGO or YSZ. The plot of the electrical conductivity is shown in Fig. 2. The conductivity increases until LSF35, where after it decreases. The conductivity follows the small polaron hoping mechanism, until a characteristic temperature, where after the conductivity is metallic. The dependence of composition can be explained as follows. When strontium is substituted for lanthanum Fe(IV) is formed. This leads to increased conductivity. However, substitution of lanthanum with strontium also leads to oxygen vacancies and this results in decoupling of iron‑oxygen‑iron super exchange, lowering the conductivity. The super exchange was introduced by Goodenough [41]. This effect stems from the overlap between iron 3d-orbitals and oxygen 2p-orbitals. When there are oxygen vacancies this conduction pathway will be

30

20

10

0 0.0

0.2

0.4

0.6

0.8

x in La1-xSrxFeO3-δ Fig. 3. EIS spectrum for the perovskite LSF05 recorded in air at 700 °C. The measured data is plotted as circles and the solid line is the fitted data. The spectrum consists of three arcs.

discontinued, lowering the electronic conductivity. This is reason for the conductivity showing a maximum, when varying the strontium content. In this case it turns out to be a maximum in conductivity for LSF35. The interpretation of the EIS data can be done accordingly to the mechanism suggested by Siebert et al. [42] and Jerdal [43]. It should be noted that this is a simplified approach. A more correct approach is to use a transmission line circuit with systematic termination, for surface and interface. However, the simplified approach used in this study is useful for discussion of the properties of the ferrite based cathode materials. In this mechanism the high frequency arc is normally thought to be due to an exchange reaction (exchange of oxide ions) at the electrodeelectrolyte interface. The magnitude of this arc decreases when x is increased to 0.35 where after it increases. The near-equivalent capacity of this arc is found to be around 0.5

Table 2 Thermal expansion coefficients of the ferrites. The thermal expansion coefficients are seen to increase with increasing strontium doping.

−6

10

−1

K

LSF00

LSF05

LSF15

LSF25

LSF35

LSF50

LSF70

11.2 ± 0.3

12.2 ± 0.2

12.8 ± 0.1

15.2 ± 0.2

16.1 ± 0.3

19.6 ± 0.2

24.1 ± 0.2

3

Solid State Ionics 344 (2020) 115096

K.K. Hansen

Table 5 ASR and n values of the three individual arcs for the seven perovskite compounds measured in air at 800 °C. ASR values are given in Ωcm2. 800 °C

Table 3 Total ASR of the seven perovskite compounds measured in air at three different temperatures. The ASR values are given in Ωcm2. LSF00

LSF05

600 °C 700 °C 800 °C

117 16 2.4

18 (29) 3.5 (2) 0.8 (0.7)

LSF15

LSF25

LSF35

7.1 1.2 0.2

14 1.4 0.5

52 15 2.0

LSF50

LSF70

162 (91) 29 (11) 4.4 (1.7)

223 44 11

Table 4 ASR and n values of the three individual arcs for the seven perovskite compounds measured in air at 700 °C. ASR values are given in Ωcm2. 700 °C Compound LSF00 LSF05 LSF15 LSF25 LSF35 LSF50 LSF70

High frequency ASR

n 2

0.07 Ωcm 1.5 Ωcm2 0.5 Ωcm2 0.3 Ωcm2 0.4 Ωcm2 1.8 Ωcm2 2.11 Ωcm2

0.9 0.9 0.9 0.9 0.9 0.9 0.9

Medium frequency ASR

n 2

11 Ωcm 0.9 Ωcm2 0.4 Ωcm2 0.8 Ωcm2 1.3 Ωcm2 7.3 Ωcm2 5.4 Ωcm2

0.65 0.65 0.65 0.65 0.65 0.65 0.65

Low frequency ASR

N 2

5.9 Ωcm 1.0 Ωcm2 0.2 Ωcm2 0.2 Ωcm2 14 Ωcm2 17 Ωcm2 33 Ωcm2

Medium

Low

Compound

ASR

n

ASR

n

ASR

N

LSF00 LSF05 LSF15 LSF25 LSF35 LSF50 LSF70

0.03 Ωcm2 0.4 Ωcm2 0.1 Ωcm2 0.1 Ωcm2 0.1 Ωcm2 0.5 Ωcm2 0.6 Ωcm2

0.9 0.9 0.9 0.9 0.9 0.9 0.9

0.9 Ωcm2 0.2 Ωcm2 0.1 Ωcm2 0.1 Ωcm2 0.2 Ωcm2 0.2 Ωcm2 1.4 Ωcm2

0.65 0.65 0.65 0.65 0.65 0.65 0.65

1.2 Ωcm2 0.2 Ωcm2 0.05 Ωcm2 0.3 Ωcm2 1.8 Ωcm2 2.2 Ωcm2 8.6 Ωcm2

0.75 0.75 0.75 0.75 0.75 0.75 0.75

reaction at the surface of the cathode [42,43]. This arc reaches its minimum for the compounds LSF15 and LSF25 meaning that the redox reaction occurs most easily on these compounds. This implies that the amount of oxide ion vacancies (or the mobility of oxide ion vacancies) plays an important role in the redox reaction as the amount of oxide ion vacancies increases with increasing x. It should be noted that the oxygen surface exchange constant is coupled to the oxide ion conductivity, see i.e. [45]. A higher oxide ion conductivity leads to a higher oxygen surface exchange constant. But another parameter must also play an important role as the magnitude of this arc increases for x larger than 0.25. This parameter could be the amount of Fe(III), or ordering of oxide ion vacancies, lowering the mobility of the oxide ion vacancies. The ordering of oxide ion vacancies can occur for the compounds with high strontium content leading to a lowering of the mobility of the oxide ion vacancies. This has been shown for the compound SrFeO3-δ compared with the compound La0.5Sr0.5FeO3-δ [46]. This parameter is therefore most likely due to the amount of Fe(III), as the ionic conductivity, and thereby the oxygen surface exchange constant, increases when going from LSF35 to LSF50 [20], but the contribution from the low frequency arc increases. An additional effect can be segregation of SrO to the surface of the perovskite at high SrO contents. This will lead to a lower content of the thought catalytic active specie Fe(III) on the surface of the perovskite, thereby lowering the activity. The near-equivalent capacity of the medium and low frequency arcs can be calculated as for the high frequency arc. The results can be found in Table 6. It is seen that the near-equivalent capacitance attains a maximum for the compound LSF15, the most active perovskite of the perovskites investigated in this study. However, the capacity is too high for being a surface process, pointing towards the process for the low frequency being a bulk process. It is not possible from these data to give a full mechanism, but it could be that the medium and low frequency arcs are swapped in this case. That is the medium frequency arc is a redox reaction and the medium frequency arc is a diffusion process. It can be of use to compare the electrochemical properties of the ferrites studied here with the electrochemical properties of manganites studied in the literature [47,48]. For the manganites several findings have been found in the literature. It has been shown both that the activity towards the reduction of oxygen increases with increasing strontium content (at least until the strontium content is 0.5) [48] or that the activity finds its maximum for the intermediate compounds [48]. For the ferrites the same behavior as observed in [48] is confirmed in this study. For the cobaltites there is no such study in the literature to the best of my knowledge. However, unpublished data suggest that the activity

Fig. 4. The total ASR of the seven investigated LSF perovskites measured in air at 700 °C with EIS. The total ASR is seen to be lowest for the perovskite LSF15.

ASR/Ωcm2

High

0.75 0.75 0.75 0.75 0.75 0.75 0.75

μFcm−2 for all the compounds strongly suggesting that this arc is due to a double layer effect. The value is also in good agreement with the value found in [44] for a double layer capacitance. The medium frequency arc is attributed to diffusion of oxide anions in the bulk of the electrode [42,43]. This arc will therefore depend on the ionic conductivity of the electrode material. It is seen that the magnitude of this arc finds its minimum for x equal to 0.15. It is noteworthy that the contribution to the total ASR from this arc is highest for the two compounds LSF00 and LSF70. This is in good agreement with the fact that LSF00 only contains a very low amount of oxide ion vacancies, and that LSF70 contains ordered oxide ion vacancies, with a low oxide ionic conductivity to follow. It should also be noted that this arc might depend on the micro-structure of the electrode-electrolyte interface. This complicates matters as this arc will then be dependent on both micro-structure and ionic conductivity. It should be mentioned that this arc not is found on thin film electrodes. This is due to the larger surface area on planar thin film electrodes, compared to the cone-shaped electrodes, where most of the electrode area is covered due to the tight contact with the electrolyte. The low frequency arc is suggested to be due to a slow redox

Table 6 Capacities for the medium and low frequency arcs.

4

Compound

LSF00

LSF05

LSF15

LSF25

LSF35

LSF50

LSF70

C2/mFcm−2 C3/mFcm−2

0.2 15

0.04 62

19 1340

0.4 96

0.2 0.5

5 11

4 16

Solid State Ionics 344 (2020) 115096

K.K. Hansen

towards the reduction of oxygen increases with increasing strontium content [49]. The reason for this difference in behavior is probably that the electrons are localized on the manganites and the ferrites and delocalized on the cobaltites. The activity of the cobaltites is thus only determined by ionic conductivity. Based on these findings LSF15 is the best choice as a cathode materials, due to a relative low TEC and a high electro-catalytic activity.

Chem. 109 (1994) 134. [14] T. Ishigaki, S. Yamauchi, J. Mizusaki, K. Fueki, H. Naito, T. Adachi, J. Solid State Chem. 55 (1984) 50. [15] T. Ishigaki, S. Yamauchi, K. Kishio, J. Mizusaki, K. Fueki, J. Solid State Chem. 73 (1988) 179. [16] M.C. Kim, S.J. Park, H. Haneda, J. Tanaka, T. Mitsuhasi, S. Shirasaki, J. Mater. Sci. Lett. 9 (1990) 102. [17] M.C. Kim, S.J. Park, H. Haneda, J. Tanaka, S. Shirasaki, Solid State Ionics 40/41 (239) (1990). [18] J.E. ten Elshof, H.J.M. Bouwmeester, H. Verweij, Solid State Ionics 81 (1995) 97. [19] J.E. ten Elshof, H.J.M. Bouwmeester, H. Verweij, Solid State Ionics 89 (1996) 81. [20] M.V. Patrakeev, J.A. Bahteeva, E.B. Mitberg, I.A. Leonidov, V.L. Kozhevnikov, K.R. Poeppelmeier, J. Solid State Chem. 172 (2003) 219. [21] T. Takeda, R. Kanno, M. Noda, T. Tomida, O. Yamamoto, J. Electrochem. Soc. 134 (1987) 2656. [22] J.M. Ralph, A.C. Schoeler, M. Krumpelt, J. Mat. Science 36 (2001) 1161. [23] J.M. Ralph, C. Rossignol, R. Kumar, J. Electrochem. Soc. 150 (2003) A1518. [24] A. Mai, V.A.C. Haanappel, S. Uhlenbruck, F. Tietz, D. Stöver, Solid State Ionics 176 (2005) 1341. [25] F.S. Baumann, J. Fleig, G. Cristiani, B. Stuhlhofer, H.-U. Habenmeier, J. Maier, J. Electrochem. Soc. 154 (2007) B931. [26] K. Kammer, E.M. Skou, Solid State Ionics 175 (2005) 915. [27] D.N. Mueller, M.L. Machala, H. Bluhm, W.C. Chueh, Nat. Commun. 6 (2015) 1. [28] A. Schmid, G.M. Rupp, J. Fleig, Phys. Chem. Chem. Phys. 20 (2018) 12016. [29] J. Newman, J. Electrochem. Soc. 113 (1966) 501. [30] S.-I. Hashimoto, K. Kammer, P.H. Larsen, F.W. Poulsen, M. Mogensen, Solid State Ionics 176 (2005) 1013. [31] K. Kammer, Solid State Ionics 177 (2006) 1047. [32] K. Kammer, L. Mikkelsen, J.B. Bilde-Sørensen, J. Solid State Electrochem. 10 (2006) 934. [33] K.K. Hansen, M. Søgaard, M. Mogensen, Electrochem. Solid-State Lett. 10 (2007) B119. [34] S.-I. Hashimoto, K. Kammer, F.W. Poulsen, M. Mogensen, J. Alloys and Compounds 428 (2007) 256. [35] C.G. Graves, M. Mogensen, ECS Trans. 28 (11) (2010) 173. [36] L.A. Chick, L.R. Pederson, G.D. Maupin, J.L. Bates, L.E. Thomas, G.J. Exarhos, Matt. Lett. 10 (1990) 6. [37] M. Juhl, M. Mogensen, T. Jacobsen, B. Zachau-Christiansen, N. Thorup, E. Skou, Proc. Electrochem. Soc. ‘SOFC IV’ 524 (1995). [38] B.A. Boukamp, Solid State Ionics 20 (1986) 31. [39] T. Jacobsen, B. Zachau-Christiansen, L. Bay, S. Skaarup, F.W. Poulsen, et al. (Ed.), Proceedings of the 17th International Symposium on Materials Science; High Temperature Electrochemistry: Ceramics and Metals, 1996, p. 29 Roskilde, Denmark. [40] M. Kuhn, S. Hashimoto, K. Sato, K. Yashiro, J. Mizusaki, Solid State Ionics 195 (2011) 7. [41] J. Goodenough, Phys. Review 100 (1955) 564. [42] E. Siebert, A. Hammouche, M. Kleitz, Electrochim. Acta 40 (1995) 1741. [43] L.O. Jerdal, The Oxygen Electrode Process in Solid Oxide Fuel Cells, Ph.D. thesis NTNU, Norway, 1998. [44] N.L. Robertson, J.N. Michaels, J. Electrochem. Soc. 138 (1991) 1494. [45] J.A. Kilner, T.A. Ramanarayanan, W.L. Worrell, H.L. Tuller (Eds.), Proc. 2nd Symp. Ionic and Mixed Conducting Ceramics, PV-94-12, The Electrochem. Soc, 1994, p. 174. [46] J. Yoo, A. Verma, S. Wang, A.J. Jacobsen, J. Electrochem. Soc. 152 (2005) A497. [47] A. Hammouche, E. Siebert, A. Hammou, Mat. Res. Bull 24 (1989) 367. [48] K. Kammer Hansen, J. Electrochem. Sci. Eng. 8 (3) (2018) 255. [49] . K. Kammer Hansen, unpublished work, in preparation.

5. Conclusion The thermal expansion coefficient of the LSF compounds increases with increasing strontium content and is highest for LSF70. The total conductivity increases until LSF35. The activity of LSF based perovskites as cathodes for the electrochemical reduction of oxygen depends on the amount (or the mobility) of oxide ion vacancies and the amount of Fe(III). This means that the intermediate compound La0.85Sr0.15FeO3-δ is the most active towards the electrochemical reduction of oxygen in a SOFC of the LSF compounds investigated in this study. The effect of ordering of oxide ion vacancies for the strontium rich compounds cannot be excluded. Declaration of competing interest No conflict of interest. Acknowledgements Colleagues at the Department of Energy Conversion and Storage are thanked for fruitful discussions. Financial support from Energinet.dk through PSO-R&D-project no. 2006-1-6493 is gratefully acknowledged. References [1] N.Q. Minh, T. Takahashi, Science and Technology of Ceramic Fuel Cells, Elsevier Science B.V, 1995. [2] N.Q. Minh, J. Am. Ceram. Soc. 76 (1993) 563. [3] J.M. Ralph, C. Rossignol, R. Kumar, J. Electrochem. Soc. 150 (2003) A1518. [4] H.Y. Tu, T. Takeda, N. Imanishi, O. Yamamoto, Solid State Ionics 117 (1999) 277. [5] J.F. Gao, X.Q. Liu, D.K. Peng, G.Y. Meng, Catal. Today 82 (2003) 207. [6] C.C. Chen, M.M. Nasrallah, H.U. Anderson, J. Electrochem. Soc. 140 (1993) 3555. [7] G.Ch. Kostogloidis, Ch. Ftikos, Solid State Ionics 126 (1999) 143. [8] J. Mizusaki, T. Sasamo, W.R. Cannon, H.K. Bowen, J. Amer. Ceram. Soc. 65 (1982) 363. [9] J. Mizusaki, T. Sasamo, W.R. Cannon, H.K. Bowen, J. Amer. Ceram. Soc. 66 (1983) 247. [10] Y. Teraoka, H.-M. Zhang, S. Furukawa, N. Yamazoe, Chem. Lett. (1985) 1743. [11] J. Mizusaki, M. Ychihiro, S. Yamauchi, K. Fueki, J. Solid State Chem. 58 (1985) 257. [12] J. Mizusaki, M. Ychihiro, S. Yamauchi, K. Fueki, J. Solid State Chem. 67 (1987) 1. [13] S.E. Dann, D.B. Currie, M.T. Weller, M.F. Thomas, A.D. Al-Rawwas, J. Solid State

5