Tao
Wei
a,
Qin
Zhang
a,
Yun-Hui
Huang
*a and
John B.
Goodenough
b
aState Key Laboratory of Material Processing and Die & Mould Technology, School of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China. E-mail: huangyh@mail.hust.edu.cn
bTexas Materials Institute, ETC 9.102, The University of Texas at Austin, 1 University Station, C2200, Austin, TX 78712, USA
First published on 2nd November 2011
The fabrication of a solid oxide fuel cell (SOFC) with symmetrical anode/cathode electrodes is a practical strategy to reduce strain, promote co-sintering of the electrodes, and hence to improve stability and reliability. Non-stoichiometric Sr2Co1+xMo1−xO6 (x = 0.1, 0.15, 0.2) (SCMO) double perovskites are designed to be used as symmetrical anode/cathode electrodes for symmetrical SOFCs with La0.8Sr0.2Ga0.83Mg0.17O3 (LSGM) as electrolyte. As an individual anode or cathode, SCMO shows considerable electrochemical performance. A typical cell with Sr2Co1.15Mo0.85O6 as the symmetrical anode/cathode electrodes, the maximum power density reaches as high as 460 mW cm−2 at 800 °C. The thermal expansion coefficient (TEC) of SCMO ranges from 6 × 10−6 to 15 × 10−6 K−1 from room temperature to 900 °C, matching well that of the conventional LSGM electrolyte.
Cobalt-based single-perovskite cathode materials for SOFCs, such as SrCo0.8Fe0.2O3 (SCF) and Ba0.5Sr0.5Co0.8Fe0.2O3, exhibit excellent electronic and O2−-ion conductivities in air, and are also good catalysts for oxygen oxidation.9,10 They are especially attractive for practical operation at intermediate temperature even at low temperature. However, spin-state transition makes thermal expansion coefficients (TECs) of the Co-based single perovskites too high to match well with those of the electrolytes La0.8Sr0.2Ga0.83Mg0.17O2.815 (LSGM) and Ce0.8Sm0.2O2 (SDC). For example, the TEC of SCF is 30 × 10−6 K−1,11,12 whereas the TECs of LSGM and SDC are 10–12.4 × 10−6 K−1 in the temperature range of 20–900 °C.13–15 The mismatch in TEC prevents the cobalt-based single-perovskite materials from application in symmetrical SOFCs.16
In previous works, we have found that the double perovskites Sr2MMoO6 (M = Mg, Mn, Fe, Co, Ni) show an excellent electrochemical performance when used as anode materials for SOFCs.17–19 Among them, the Co-based Sr2CoMoO6 anode exhibits remarkably high power density in both H2 and CH4 fuels. According to our measurements, the TECs are 6 × 10−6 to 15 × 10−6 K−1 from room temperature (RT) to 900 °C for the Co-based double perovskites, matching well with those of SDC or LSGM. The oxygen content of Sr2CoMoO6±δ can be varied from oxygen-rich to oxygen-deficient, depending on the atmosphere. In order to yield oxygen vacancies in both the anodic and the cathodic atmospheres, we have designed Sr2Co1+xMo1−xO6−δ (SCMO, x = 0.1, 0.15, 0.2) in which high-valent Mo ions are partially substituted by low-valent Co ions in order to retain oxygen vacancies without compromise of electronic conductivity in both the reducing atmosphere at the anode and the oxidizing atmosphere at the cathode. By using the Sr2Co1+xMo1−xO6−δ as symmetrical anode/cathode electrodes, symmetrical SOFCs have been fabricated. The structure and the electrochemical performance have been systematically investigated.
The phase of the samples was checked by X-ray powder diffraction (XRD) with a Philips X'Pert PRO diffractometer using filtered Cu-Kα radiation. The diffraction profiles were analyzed with a Rietveld refinement program, RIETAN 2000, for identification of crystal structure and lattice parameters. The electronic conductivity was measured under H2 and in stagnant air from 350 to 850 °C with a standard four-point probe method on an RTS-8 digital instrument. Simultaneous TG and DTA were obtained by a Stanton STA 781 thermal analyser at a heating rate of 5 °C min−1 from RT to 900 °C in 5% H2/Ar (100 ml min−1) and air atmosphere. Thermal expansion data were collected with a dilatometer (NETZSCH STA449c/3/G) under a 5% H2/Ar flow (100 ml min−1) at a rate of 5 °C min−1 between RT and 900 °C. The oxidation state of the Co ions was examined by X-ray photoelectron spectroscopy (XPS, MULT1LAB2000, VG). Morphologies were observed with field scanning electron microscopy (SEM, Sirion 200, Holland).
For single fuel cell fabrication, SCMO, SDC, SCF and “NiO+SDC” (weight ratio 65:
35) powders were made into terpineol-based slurries for the electrodes. The electrolyte disk LSGM was about 20 mm in diameter and 0.3 mm thick. In order to clearly evaluate the electrochemical performance of SCMO as anode and cathode, we fabricated a series of 9 different single testing cells with SCMO as separate anode, separate cathode or as anode and cathode at the same time (symmetrical electrodes). SDC served as a buffer layer where SCMO was used as anode and LSGM as electrolyte to prevent inter-diffusion between SCMO and LSGM. SDC slurry was screen-printed onto one side of the LSGM disk followed by firing at 1300 °C in air for 1 h; then the SCMO slurry was screen-printed onto the SDC buffer layer and baked at 1100 °C for 2 h. The SCF cathode slurry was screen-printed onto another side of the LSGM and sintered at 1000 °C for 2 h. With SCMO as cathode, “NiO+SDC” was used as the anode; the anode was sintered at 1250 °C for 4 h. With SCMO used as both anode and cathode for the symmetrical fuel cells, SCMO slurry was screen-printed on two sides of LSGM disk and then sintered at 1100 °C for 2 h. Table 1 lists all the 9 fuel cells that are marked from Cell 1 to Cell 9.
Single cell | Sample 1 | Sample 2 | Sample 3 |
---|---|---|---|
1/SDC/LSGM/SCF (Cell 1) | 2/SDC/LSGM/SCF (Cell 2) | 3/SDC/LSGM/SCF (Cell 3) | |
NiO/SDC/LSGM/1 (Cell 4) | NiO/SDC/LSGM/2 (Cell 5) | NiO/SDC/LSGM/3 (Cell 6) | |
1/LSGM/1 (Cell 7) | 2/LSGM/2 (Cell 8) | 3/LSGM/3 (Cell 9) | |
Average valance | 2.12 | 2.57 | 2.77 |
Co3+ content (%) | 11.6 | 56.7 | 77.1 |
σ in air/S cm−1 | 1.13 | 3.75 | 5.12 |
σ in H2/S cm−1 | 7.83 | 8.52 | 4.16 |
E a in air/eV | 0.21 | 0.08 | 0.16 |
E a in H2/eV | 0.15 | 0.12 | 0.32 |
For electrochemical testing, Ag paste was used as a current collector instead of the Pt paste used in our previous work, to avoid any influence of catalytic activity from Pt. The cells were tested in a vertical tubular furnace. Pure H2 (30 ml min−1) was used as fuel and air served as the oxygen source. Electrochemical characterizations were performed under ambient pressure. Electrochemical impedance spectra (EIS), power output and overpotential for the cells were measured with an Arbin FCTS and Princeton PAR2273 electrochemical working station. Power density and current density were calculated for the effective area of the cell (0.6 × 0.4 = 0.24 cm2).
![]() | ||
Fig. 1 XRD patterns for Sr2Co1+xMo1−xO6−δ (x = 0.1, 0.15, 0.2) samples obtained at room temperature: (a) x = 0.1, (b) x = 0.15, and (c) x = 0.2. |
x | 0.1 | 0.15 | 0.2 |
---|---|---|---|
Space group | I4/m | I4/m | I4/m |
a/Å | 5.5782(3) | 5.5753(8) | 5.5702(9) |
c/Å | 7.9475(4) | 7.9328(3) | 7.9213(9) |
V/Å3 | 247.300(9) | 247.097(4) | 245.785(7) |
d/g cm−3 | 5.2573(2) | 5.2759(1) | 5.2893(3) |
Co–O1 (Å)(×2) | 2.0387(2) | 2.0381(4) | 2.0372(1) |
Co–O2 (Å)(×4) | 2.0324(1) | 2.0036(2) | 1.9931(1) |
Mo–O1 (Å)(×2) | 1.9120(1) | 1.9187(4) | 1.9234(5) |
Mo–O2 (Å)(×4) | 1.9350(5) | 1.9412(8) | 1.9456(7) |
Sr–O1 (Å)(×4) | 2.7896(1) | 2.7867(6) | 2.7858(7) |
Sr–O2 (Å)(×4) | 2.8001(6) | 2.8010(3) | 2.7932(1) |
Sr–O2 (Å)(×4) | 2.8022(4) | 2.8033(6) | 2.8051(1) |
O1 (x y 0) | x = y = 0.2423(6) | x = y = 0.2445(3) | x = y = 0.2469(9) |
O2 (0 0 z) | z = 0.2434(7) | z = 0.2514(3) | z = 0.2571(7) |
g M | 0.744(9) | 0.698(9) | 0.672792 |
ξ | 0.4898 | 0.3978 | 0.3456 |
R wp (%) | 5.90 | 5.54 | 4.40 |
R p (%) | 4.13 | 3.97 | 3.27 |
Further structural information can be seen from Table 1. The bond length 〈Mo–O〉 is shorter than that of 〈Co–O〉, which is due to the smaller ionic radius of Mo6+ (0.59 Å) than that of the Co2+ or Co3+ ions.20 The 〈Sr–O〉 bond length is longer than 〈Mo–O〉 and 〈Co–O〉 for the large cation radius of Sr2+ (1.44 Å, twelve-fold).20 For the double-perovskite, Co and Mo ions at the B-sites usually show some antisite disorder, which affects the electrochemical performance.18 The value of the order parameter can be calculated as ξ = 2(gM − 0.5) from the refined occupancy of Co cations at the correct site (gM). The ordered parameter ξ for SCMO samples with x = 0.1–0.2 drops from 0.490 to 0.346 as x increases. For Sr2Co1+xMo1−xO6−δ, increasing x results in the increase of Co3+ content. The radius and valence of Co3+ are closer to those of Mo6+ as compared with Co2+, which in return leads to more disorder in the cationic array of Co and Mo ions. Additionally, Co ions are excessive and Mo ions are deficient in the non-stoichiometric SCMO structure, so some Co ions necessarily occupy the Mo sites.
XPS is a characterization technique that enables one to study the changes in chemistry as well as electrostatic potentials of atoms in materials.21 The measured spectra of the three samples are shown in Fig. 2. For Sample 1, the peaks of binding energy at 780.1 and 795.6 eV are assigned to Co2+ 2p3/2 and 2p1/2, respectively, which indicates that most of the cobalt in Sample 1 is bivalent. For Sample 2 and Sample 3, the peaks shift to lower binding energy 779.8 and 794.6 eV, indicative of an increase in Co3+ content.22 The Co valence in SCMO obtained from XPS is in good accord with that of the iodometric titration.
![]() | ||
Fig. 2 XPS spectra for Sr2Co1+xMo1−xO6−δ with x = 0.1, 0.15 and 0.2 at room temperature. |
Fig. 3 shows the electronic conductivity for the SCMO samples over the temperature range from 350 to 850 °C. The ordered double-perovskite Sr2CoMoO6 has alternating CoO6/2 and MoO6/2 corner-shared octahedra. The Co-rich non-stoichiometry provides sufficient Co3+ or Co2+ and leads to low cationic ordering in structure. The activation energies (Ea) for the samples in the interval 350 < T < 850 °C (except for Sample 3 in air) are listed in Table 1.
![]() | ||
Fig. 3 Temperature dependence of electrical conductivity for Sr2Co1+xMo1−xO6−δ (x = 0.1, 0.15, 0.2): (a) in air, (b) in H2. |
In a reducing atmosphere, all the cobalt of Sr2CoMoO6−δ are Co2+ and the molybdenum is reduced to give Mo5+–O–O–Mo6+ conductive interaction. In air, all the Mo of Sr2CoMoO6+δ are Mo6+, and the cobalt is oxidized to give poorer Co2+–O–O–Co3+ conductive interaction; the conductivity is only 0.22 S cm−1 at 850 °C in air.18 In contrast, the conductivity in air of Sample 3, Fig. 3a, is 6.5 S cm−1 at 400 °C and exhibits a metallic temperature dependence in the range 400 < T < 700 °C. The Co-rich Sample 3 necessarily has Co on the Mo sites as well as a higher degree of antisite disorder, which creates percolating pathways of Co–O–Co interaction responsible for most of the electronic conduction below 700 °C. At higher temperatures, there appears to be an increasing contribution to the electronic conductivity from the non-percolating regions. In a reducing atmosphere, where all the cobalt are Co2+, we conjecture that the reduced Mo concentration and increased oxygen-vacancy concentration in Sample 3 create centers that trap electron of the Mo6+/Mo5+ couple in Anderson-localized state below a mobility edge in any narrow Mo-4d band to give a semiconductive temperature dependence of the conductivity with a reduced charge-carrier concentration.
Since the perovskite lattice does not allow oxygen interstitials, oxygen-rich samples contain cation vacancies as occurs in the La1−xMn1−xO3 system.23Oxygen-rich samples are not oxide-ion conductors. The TGA data attained in air (Fig. 4a) show the onset of a progressive weight loss from 210 °C for all three Co-rich samples. The amount of oxygen released on heating decreases from Sample 1 to Sample 3, which is consistent with the stronger octahedral-site preference of Co3+. The TGA data of Fig. 4b show that in 5% H2/Ar atmosphere, all three Co-rich SCMO samples undergo three steps of weight loss. Bound water would be lost first; reduction of Co3+ to Co2+ would occur before reduction of Mo6+. We note that the removal of oxygen from between two Mo would trap a pair of electrons in a Mo5+–Mo5+ bond between two five-fold-coordination Mo atoms.
![]() | ||
Fig. 4 TG and DTA curves for Sr2Co1+xMo1−xO6−δ (x = 0.1, 0.15, 0.2) recorded from room temperature to 900 °C: (a) in air, (b) in 5% H2/Ar (100 ml min−1). |
Fig. 5 shows the linear thermal expansion (ΔL/L0) and thermal expansion coefficient (TEC) for SCMO samples from RT to 900 °C. In Fig. 5a, it is observed that the ΔL/L0 is approximately linearly dependent on temperature for these three samples. Generally, TECs for the Co-based perovskites are quite high, the values often reach to 20–30 × 10−6 K−1.24 The TEC values for the SCMO samples vary within the range 6 to 15 × 10−6 K−1 from RT to 900 °C, which is lower than the TEC values reported for conventional perovskites rich in Co3+ ions that undergo a transition from the low-spin to a higher-spin state with increasing temperature. A small increase in the ΔL/L0 value from Sample 1 to Sample 3 is consistent with a higher concentration of Co3+ ions. Low TEC values for SCMO match well with the standard electrolyte materials. For example, the TEC is 11 × 10−6 K−1 for Zr0.84Y0.16O1.92 and 10–13 × 10−6 K−1 for LSGM.25,26 Therefore, the SCMO samples provide a good thermal expansion compatibility with the electrolytes. Moreover, TEC changes only within a narrow range over temperature for all the SCMO samples, which is favorable to a stable cell geometry during operation of the SOFC.
![]() | ||
Fig. 5 (a) Thermal expansion (ΔL/L0) and (b) the corresponding thermal expansion coefficients (TECs) as functions of temperature for Sr2Co1+xMo1−xO6−δ (x = 0.1, 0.15, 0.2) measured in air from room temperature to 900 °C. |
Although the electrolyte-supported single cells hinder us from obtaining the optimum fuel-cell performance owing to the high ohmic drop across the electrolyte, it is useful to compare the performances of the anode or cathode materials.27,28Fig. 6 shows the cell voltage and power density as functions of current density for the cells with SCMO respectively as anode, cathode, and symmetrical electrode at 800 °C. Fig. 6a shows the performance of the SCMO/SDC/LSGM/SCF cells, corresponding to Cell 1, 2 and 3 as listed in Table 1. Cell 1 exhibits a maximum power density (Pmax) of 660 mW cm−2, the highest one among all the 9 cells. Cell 2 shows almost equal Pmax as Cell 1, but the Pmax for Cell 3 is only 540 mW cm−2. The high power density for Cells 1 and 2 is ascribed not only to their higher electronic conductivity, but also to a better oxide-ion conductivity due to the formation of more oxygen vacancies in the double-perovskite structure (see Fig. 4).
![]() | ||
Fig. 6 Cell voltage and power density as functions of current density for the 9 series single fuel cells: (a) SCMO as anode, (b) SCMO as cathode, (c) SCMO as symmetrical anode/cathode electrodes. All the cells operate at 800 °C. |
Fig. 6b shows the performance of the cells with configuration NiO+SDC/SDC/LSGM/SCMO, corresponding to Cells 4, 5 and 6 in Table 1. The cells with SCMO as cathode material also show considerable power density. Interestingly, the power density of Cell 6 with Sample 3 as cathode is higher than those of Cell 4 and 5 with Sample 1 and 2 as cathodes. The order of Pmax is totally different from that with SCMO as anode. The value of Pmax at 800 °C is about 540 mW cm−2 for Cell 6 (Sample 3 as cathode) and 520 mW cm−2 for Cell 5 (Sample 2 as cathode). For Cell 4 (Sample 1 as cathode), the performance is somewhat poorer. Among the three samples, Sample 3 has the highest Co3+ content and the lowest Co–Mo ordering. Its order parameter ξ is only 0.35, which means that the structure is close to a single perovskite. For the Co-based cathode materials, the main contribution to the catalytic activity for oxygen decomposition comes from Co3+ ions, and the Co3+–O–Co2+ couples provide small-polarization conductivity.
We focus on symmetrical fuel cells in which the SCMO layers serve as the anode and cathode. Fig. 6c exhibits the performance of symmetrical cells with the SCMO/LSGM/SCMO configuration, corresponding to Cells 7, 8 and 9. Cell 8 with Sample 2 as both cathode and anode shows a power density as high as 460 mW cm−2 at 800 °C. To the best of our knowledge, this is one of the highest power densities for the SOFCs with symmetrical electrodes. So far, La0.75Sr0.25Cr0.5Mn0.5O3, La0.8Sr0.2Sc0.8Mn0.2O3, and (La,Sr)(TiFe)O3 have been reported as symmetrical electrodes for SOFCs.4,5,29,30 For example, the SOFC with La0.75Sr0.25Cr0.5Mn0.5O3 as symmetrical electrodes was reported to show the highest power output of 500 mW cm−2 at 950 °C.5 For symmetrical cells, the electrode material should exhibit considerable conductivity in air and in the reductive fuel atmosphere. Sample 2 meets the above requirement. From Fig. 6, we can see that Sample 1 shows the best performance as an anode material (at 800 °C, Pmax = 660 mW cm−2 with LSGM as electrolyte), which is consistent with our previous report.19,20 If Sample 1 is used as a cathode, the power densities are lower than where Sample 1 is used as an anode. For example, Pmax at 800 °C is 412 mW cm−2 with Sample 1 only as cathode and 390 mW cm−2 with Sample 1 as both cathode and anode. In the symmetrical cell, Sample 1 can provide sufficient oxygen transport at the anode side, but it cannot give enough catalytic activity for oxygen reduction at the cathode side. Thus the electrochemical process is limited by the cathode side, leading to a relatively low power output. Sample 2, on the other hand, exhibits remarkable power density as both anode and cathode; it shows the highest power output among the three samples when it is used as a symmetrical anode/cathode material.
Fig. 7 shows the morphologies of the SCMO electrodes, SDC buffer layer, and LSGM electrolyte of Cell 7. Fig. 7a shows the cross-section of Cell Sample 1/LSGM/Sample 1. Fig. 7b shows the morphology of anode layer (Sample 1 sintered at 1100 °C for 2 h) on the disk of LSGM electrolyte. The SCMO particles are porous. The porous microstructure will not only provide sufficient three-phase boundaries (TPBs) for adsorption and diffusion of fuels, but also ensure efficient charge transfer from electrode to electrolyte or from electrolyte to electrode. In Fig. 7c, the upper portion shows the porous SDC buffer layer, and the bottom shows the dense LSGM electrolyte. We can also find from Fig. 7d that the SCMO layer has a compact connection with LSGM.
![]() | ||
Fig. 7 SEM images: (a) cross-section of Cell 7 with configuration of Sample 1/LSGM/Sample 1 after running in H2, (b) Sample 1 sintered at 1100 °C on the disk of LSGM for 2 h, (c) cross-section between LSGM and SDC buffer layer, (d) cross-section between Sample 1 layer and LSGM electrolyte. |
Fig. 8 shows typical plots of the overpotential vs. the current density for the cells with SCMO as symmetrical cathode/anode measured at 800 °C in air and pure H2, respectively. The electrode overpotential reflects the catalytic activity of oxygen reduction at the cathode and hydrogen oxidation at the anode, which can be determined by the exchange current density. In Fig. 8a, the overpotential for the SCMO anode is lower than that for SCMO cathode, which demonstrates that the reaction rate of oxygen reduction at the SCMO cathode surface is the rate limiting process in the SOFC. The overpotential decreases from Sample 1 to Sample 3 where SCMO is the cathode owing to the increasing Co3+ content and conductivity. If SCMO is the anode, the overpotential (in H2) increases with increasing x; Sample 3 has the lowest electronic conductivity in H2 and shows the highest overpotential. In general, a high overpotential leads to a poor electrochemical performance of the cell. For instance, Sample 1 shows a high overpotential as cathode (in air), giving rise to a notable degradation in power density. In addition, we can see from Fig. 8 that the overpotentials in air are much higher than those in H2 for all the SCMO samples, which indicates that the power in the symmetrical cell is predominantly limited by the cathode.
![]() | ||
Fig. 8 Overpotential as a function of current density for the Sr2Co1+xMo1−xO6−δ (x = 0.1, 0.15, 0.2) single cells operating at 800 °C: (a) in air, (b) in H2. |
We further evaluated the behavior of symmetrical electrodes in H2 and in air by using the SCMO samples as the working and the counter electrodes of a symmetrical half-cell where the same gas is fed to both electrodes. Shown in Fig. 9 is the impedance at 800 °C for the SCMO electrodes measured on the symmetrical half-cell under open circuit conditions in air and H2. Fig. 9a shows the impedance of the half-cell electrode is low in air, which demonstrates that SCMO is a suitable candidate cathode material. As reflected in the TGA curves (Fig. 4), the SCMO samples possess oxygen vacancies in air to provide an oxide-ion conductivity. Oxygen reduction can not only take place at TPBs, but also be spread over all the cathode surfaces. Sufficient oxygen reduction will certainly give rise to a low polarization resistance. Fig. 9b show the polarization resistances tested in H2. We can see that the polarization resistances in H2 are much larger than those in air. Co-rich SCMO anodes also show considerable power output. In this regard, we can deduce that in the anode atmosphere the dissociative adsorption of hydrogen on the SCMO anode surface is not rate-limiting. In H2, an SCMO anode does not cause additional polarization resistance. The electrochemical process needs a continuing source of oxide ions to react with H2. However, since the half-cell is exposed to H2, there is no oxygen source to continually produce oxide ions for the anode reaction, which may cause excess polarization resistance. All the polarization resistances are consistent with conductivity and power-density data for the SCMO samples. High electronic and oxide-ion conductivity facilitates low polarization resistance and high power density.
![]() | ||
Fig. 9 Impedance spectra measured at 800 °C for the symmetrical half cells using Sr2Co1+xMo1−xO6−δ (x = 0.1, 0.15, 0.2) as symmetrical anode/cathode electrodes: (a) in air, (b) in H2. |
This journal is © The Royal Society of Chemistry 2012 |