Scholarly article on topic 'Evaluation of Ca3(Co,M)2O6 (M=Co, Fe, Mn, Ni) as new cathode materials for solid-oxide fuel cells'

Evaluation of Ca3(Co,M)2O6 (M=Co, Fe, Mn, Ni) as new cathode materials for solid-oxide fuel cells Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Keywords
{"Solid-oxide fuel cells" / "Hexagonal perovskite cathode" / "Electrochemical performance"}

Abstract of research paper on Materials engineering, author of scientific article — Fushao Li, Long Jiang, Rui Zeng, Tao Wei, Yingxian Xu, et al.

Abstract Series compounds Ca3(Co0.9M0.1)2O6 (M=Co, Fe, Mn, Ni) with hexagonal crystal structure were prepared by sol–gel route as the cathode materials for solid oxide fuel cells (SOFCs). Effects of the varied atomic compositions on the structure, electrical conductivity, thermal expansion and electrochemical performance were systematically evaluated. Experimental results showed that the lattice parameters of Ca3(Co0.9Fe0.1)2O6 and Ca3(Co0.9Mn0.1)2O6 were both expanded to certain degree. Electron-doping and hole-doping effects were expected in Ca3(Co0.9Mn0.1)2O6 and Ca3(Co0.9Ni0.1)2O6 respectively according to the chemical states of constituent elements and thermal-activated behavior of electrical conductivity. Thermal expansion coefficients (TEC) of Ca3(Co0.9M0.1)2O6 were measured to be distributed around 16×10−6 K−1, and compositional elements of Fe, Mn, and Ni were especially beneficial for alleviation of the thermal expansion problem of cathode materials. By using Ca3(Co0.9M0.1)2O6 as the cathodes operated at 800°C, the interfacial area-specific resistance varied in the order of M=Co<M=Fe<M=Ni<M=Mn, and the over-potential increased in the order of M=Fe≈M=Co<M=Mn<M=Ni. Among all of these compounds, Ca3(Co0.9Fe0.1)2O6 showed the best electrochemical performance and the power density as high as ca. 500mWcm−2 at 800°C achieved in the single cell with La0.8Sr0.2Ga0.83Mg0.17O2.815 as electrolyte and Ni–Ce0.8Sm0.2O1.9 as anode. Ca3(Co0.9M0.1)2O6 (M=Co, Fe, Mn, Ni) can be used as the cost-effective cathode materials for SOFCs.

Academic research paper on topic "Evaluation of Ca3(Co,M)2O6 (M=Co, Fe, Mn, Ni) as new cathode materials for solid-oxide fuel cells"

Available online at www.sciencedirect.com

SCienCeDireCt Progress in Natural

CrossMark -viwiv®i#iisvi Science

Materials International

ELSEVIER Progress in Natural Science: Materials International 25 (2015) 370-378 -

www.elsevier.com/locate/pnsmi www.sciencedirect.com

Original Research

Evaluation of Ca3(Co,M)2O6 (M = Co, Fe, Mn, Ni) as new cathode materials

for solid-oxide fuel cells

Fushao Lia,b, Long Jiangb, Rui Zengb, Tao Weib, Yingxian Xua, Fan Wanga, Yunhui Huangb,n

aSchool of Chemistry and Chemical Engineering, Qujing Normal University, Qujing, Yunnan 655011, PR China bSchool of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan, Hubei 430074, PR China

Received 20 July 2015; accepted 31 August 2015 Available online 31 October 2015

Abstract

Series compounds Ca3(Co0.9M0.1)2O6 (M = Co, Fe, Mn, Ni) with hexagonal crystal structure were prepared by sol-gel route as the cathode materials for solid oxide fuel cells (SOFCs). Effects of the varied atomic compositions on the structure, electrical conductivity, thermal expansion and electrochemical performance were systematically evaluated. Experimental results showed that the lattice parameters of Ca3(Co0.9Fe0.1)2O6 and Ca3(Co0.9Mn0.1)2O6 were both expanded to certain degree. Electron-doping and hole-doping effects were expected in Ca3(Co0.9Mn0.1)2O6 and Ca3(Co0.9Ni0.1)2O6 respectively according to the chemical states of constituent elements and thermal-activated behavior of electrical conductivity. Thermal expansion coefficients (TEC) of Ca3(Co0.9M0.1)2O6 were measured to be distributed around 16 x 10"6K" 1 and compositional elements of Fe, Mn, and Ni were especially beneficial for alleviation of the thermal expansion problem of cathode materials. By using Ca3(Co0.9M0.1)2O6 as the cathodes operated at 800 °C, the interfacial area-specific resistance varied in the order of M = Co < M = Fe < M = Ni < M = Mn, and the over-potential increased in the order of M=Fe e M = Co < M = Mn < M = Ni. Among all of these compounds, Ca3(Co0.9Fe0.1)2O6 showed the best electrochemical performance and the power density as high as ca. 500 mW cm" 2 at 800 °C achieved in the single cell with La0.8Sr0.2Ga0.83Mg0.17O2.815 as electrolyte and Ni-Ce0.8Sm0.2O19 as anode. Ca3(Co0.9M0.1)2O6 (M = Co, Fe, Mn, Ni) can be used as the cost-effective cathode materials for SOFCs.

© 2015 The Authors. Production and hosting by Elsevier B.V. on behalf of Chinese Materials Research Society. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Solid-oxide fuel cells; Hexagonal perovskite cathode; Electrochemical performance

1. Introduction

To cater for the human being's ambition to build the decentralized, hydrogen-based energy economy, solid-oxide fuel cells (SOFCs) technology has a tremendous potential to provide the future societies with the efficient, environmentally friendly, fuel-extendable electric power [1,2]. Currently, lowering the operation temperature to the intermediate range of around 500-800 °C is one of the main goals in SOFCs research, namely IT-SOFCs [3]. However, a great challenge then confronted is to search for the active cathode materials with accepted cost at lowered temperatures [4-6]. In the last

nCorresponding author.

E-mail address: huangyh@mail.hust.edu.cn (Y. Huang). Peer review under responsibility of Chinese Materials Research Society.

decades, cobalt containing compounds with ABO3 perovskite crystal structures have been intensively studied as the most promising candidate cathode materials for IT-SOFCs due to their prominent electro-catalytic activity and impressive electrical conductivity, chiefly like Bai _ xSrxCoi _ yFeyO3 _ s, Lai _xSrxCoO3 _ s, Sr1 _xRExCoO3 _ s(RE=rare earth metal), REBaCo2O5+s, and their derivatives [7-12]. Nevertheless, a huge problem with these cobaltite-based perovskites is their rather large thermal expansion coefficient (TEC), which may rather disable the cathode function layer by peeling it off unpredictably from the oxygen-ions conductive electrolytes, such as Y2xZri _ 2xO2 _x(YSZ), Cei _xSmO _ s(SDC), Lai _ xSrxGa1 _yMgyO3 _ s(LSGM), etc. An immediate solution to this problem is to tune the sublattice compositions with doping elements. But in most cases, this effect is almost negligible

http://dx.doi.org/10.1016/j.pnsc.2015.09.002

1002-0071/© 2015 The Authors. Production and hosting by Elsevier B.V. on behalf of Chinese Materials Research Society. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

compared with the great gap in TEC between the pairs of these cell components. On the other hand, once the dopant is introduced, discrepancy in ion radius between the host and guest ions can only be counteracted either by tilting or by cooperative rotation, or by displacement of the M-O polyhedron subcells, and consequently the structural distortion is almost inevitable. And in some extreme cases the first-order phase transition or collapse of the original lattice may then ensue, which will sharply decrease the electronic and ionic transport properties. Furthermore to some heavily-doped compounds with perorveskite structure, surface elements segregation or enrichment from A-sites is yet another hard-handling problem, such as surface Sr-enrichment/segregation on La1 _xSrxCoO3_s surface [13,14]. So, it is natural to think that, if those key electro-catalytic properties of cobaltite perovskites can still be retained, finding other cobalt containing structures with likewise robust lattices would be another strategic way.

In this regard, Ca3Co2O6 (CCO) is a good candidate that was recently explored as a promising cathode material in our group [15]. As a member of large fascinating oxide family A3n+3mA'nB3m+nO9m+6n [16], this interesting compound (m — 0, n — 1, A — Ca, A ' — B — Co) crystallizes into hexagonal crystal structure (R-3c space group) with pseudo 1D features. The Ca atoms are evenly distributed around the [Co2O6]K pillars which consist of alternative face-shared CoO6 octahedrons and distorted CoO6 trigonal prisms (Fig. 1). Another interesting character of this structure is that the constituent Co-ions are packed into two groups of six-coordinated polyhedrons. So, there are actually three classes of M-O polyhedral cages with three types of central metal ions in this lattice. Diversity of M-O polyhedron subcells increases the structural generality to a wide spectrum of atomic compositions, and this may be an unexplained reason why a large number of metal-oxides adopt this stacking mode [16]. Besides, the formula of this compound is so fascinating and its constituent elements are so plain and so easily accessible in the earth mantle

(compared with rare earth metal and other alkali earth mental elements as in the afore-mentioned materials) that we are much intrigued by the idea of adopting it as the cathode material for SOFCs. On account of this, our interest is now more aroused by further asking what new appreciated properties will result if we use the Ca3Co2O6 lattice as the basic crystallographic framework to accommodate more flexible atomic compositions. According to our previous study, this inquiry soon becomes a constructive suggestion because the overall performance of pure Ca3Co2O6 are far from being optimized as a cathode material. Its TEC is still a little large and its electrical conductivity is relatively low (~ 5-9 S cm _1 over the temperature range of 300-800 °C), and so on.

Therefore, in order to achieve the coordinated contributions from the multi-atomic compositions, original Ca3Co2O6 lattice was used as the basic structural prototype to accommodate the more functional atoms so that the Ca3Co2O6-based cathode materials with more synthetically improved performances can be achieved. In view of structural compatibility, only those elements which are neighbors to the cobalt in the elements periodic table are going to be considered, namely Fe, Mn, Ni. Accordingly, Ca3(Co,M)2O6 (M — Co, Fe, Mn, Ni) series compounds were prepared by sol-gel route with citric acid as the chelating agents and then systematically evaluated as the cathode materials for the IT-SOFCs. The low level of the M concentrate in Ca3(Co,M)2O6 was fixed to the 0.1 mole fractions of total (Co,M) sites so that each composition's beneficial contributions can be identified one by one while the key electro-catalytic contribution from the cobalt' s unique redox couples is still retained.

2. Experiment

2.1. Materials preparation and cell fabrication

Polycrystalline Ca3(Co09M01)2O6 (M—Co, Fe, Mn, Ni) powders were prepared from citric acid complex precursors.

Fig. 1. Crystal structure of Ca3Co2O6. Left: hexagon cell; Right: projected view along c-axis.

As the raw materials, Ca(NO3)2 • 4H2O, C4H6O4Co • 4H2O, Ni(NO3)2 • 6H2O, C4H6O4Mn • 4H2O, Fe(NO3)3 • 9H2O of analytical grade in stoichiometric ratio were dissolved into a minimum amount of de-ionized water under continuous stirring. The citric acid and ethylene glycol were then added as the chelating regent and the disperser respectively; the molar ratio of total metal ions to citric acid was set as 1:2. This admixture of solution was then put on a hot plate to evaporate until a dried homogeneous gel was obtained, and thus made gel was first burned at 400 °C for 5 h and then annealed at 900 °C for 10 h both in air with interment grinding. The resultant powders were left for use and further heat treatment according to characterization items.

Oxide electrolytes La0.8Sr02Ga0.83Mg0.17O2.815 (LSGM) dense wafers and Ce0.8Sm0.2O1.9 (SDC) powders were prepared via solid-state reaction, as described in our previous works [17,18]. The anode composite was made by thoroughly mixing the NiO and SDC fine powders in weight percentage ratio of 65:35.

Single cells were fabricated with an electrolyte-supported technique. 230( 7 5) ^m thick LSGM disc was used as electrolyte, NiO-SDC composite as anode, and SDC as the buffer layer between the anode and the electrolyte to prevent inter-diffusion of cationic species. The SDC slurry was screen-printed onto LSGM disc and sintered at 1300 °C for 2h in stagnant air. The anode layer (~ 30 ^m) was attained by screen-printing the "NiO-SDC" slurry on the electrolyte and successively baking at 1250 °C for 4 h. The cathode slurry was screen-printed onto another side of the electrolyte and fired at 950 °C for 5 h to achieve porous texture with layer thickness of ~ 20 ^m. To minimize current collector's catalytic effect on the both of anode and cathode sides, Ag grids is made by drying a small amount of stuck Ag paste. The cell was mounted onto an alumina tube and meticulously sealed by Ag paste. H2 was fed to the anode side as a fuel at a flow rate of 60 mL min _1 (STP) while the cathode side was exposed to the ambient air.

Symmetric cells with conformation of cathodeiLSGMicathode were employed to test electrochemical impedance spectroscopy (EIS). The fabrication method of the symmetric cells is almost same as that of single cells except that the cathode slurry was simultaneously screen-printed onto the both sides of LSGM disc. 3-electrode cells were fabricated as per literature with some minor modifications [19]. Briefly, Pt paste was screen-printed onto the opposite side of electrolyte-supported cathode to act as the counter electrode (C-E), aligned correctly in any possibility with the cathode and sintered together with cathode. Ag paste was painted aside cathode as the reference electrode (R-E). The distance between the cathode and R-E was 2 mm; the area ratio of the cathode to C-E was set about 1:2.

2.2. Characterization and electrochemical test

Crystal structure and phase purity of the samples were monitored by X-ray powder diffraction (XRD, Philips X'Pert PRO diffractometer) in Bragg-Brentano reflection geometry with Cu Ka radiation at 40 kV and a receiving slit of 0.2-0.4 mm.

The diffraction patterns were collected at room temperature by step scanning in the range of 10° < 20 < 80° with the scan rate of 5° min _1 and cell parameters were obtained by running Whole-Profile-Fitting program with Jade. The binding state of compositional elements in Ca3(Co,M)2O6 (M—Co, Fe, Mn and Ni) was analyzed using a MULT1LAB2000 X-ray Photoelectron Spectrometer incorporating. The incident radiation was monochromatic Al Ka X-rays (1486.6 eV). Narrow high-resolution scans were run to obtain O1s and Co2p level spectra with 0.05 eV steps. Base pressure in the analysis chamber was 1.0 x 10 _9 Torr and during sample analysis was 1.0 x 10 _ 8 Torr. All binding energies were referenced to the C1s peak (285 eV) arising from adventitious carbon. Morphologies of sample powders were imaged by high resolution electronic images, and the micrographs were taken by a scanning electron microscope (SEM, Hitachi: S4500) working at an electron accelerating voltage of 20 kV.

The thermal expansion was measured on the rectangular-shaped bar samples (5 mm x 5 mm x 20 mm) from RT to 1000 °C at a heating rate of 5 °C min _1 by using a dilatometer (NETZSCHSTA449c/3/G). Temperature dependence of electrical conductivity was studied with a modified four-probe method on the manually sampled RTS-8 digital instrument in the stagnant air. To prepare specimens for the conductivity measurement, as-prepared sample powders were pressed into a pellet with diameter of 13 mm and thickness of 1 mm under a pressure of 100 MPa followed by sintering at 900 °C for 10 h.

Electrochemical experiments were performed on a 2273 electrochemical system driven by the software package PowerSuite. EIS across symmetric cells were collected around open potential (Eocp) using a voltage disturbance signal of 10 mV amplitude, and the frequency varied from 100 kHz to 10 mHz. The program of ZsimpWin was used to construct the equivalent circuit models and to simulate the EIS measurement based on the least-squares method. Cathodic polarization was performed on the 3-electrode cell to characterize the polarization behaviors of the composite cathodes and I-V curve was measured on the single cells to demonstrate the power density output.

3. Results and discussion

Phase impurity of as-prepared Ca3(Co09M01)2O6 (M—Co, Fe, Mn, Ni) powder by the sol-gel route were first checked by XRD. Phase-pure Ca3Co2O6 with hexagonal crystal structure can be easily attained by annealing at 850-950 °C. The representative XRD patterns of all Ca3(Co09M01)2O6 (M—Co, Fe, Mn, Ni) samples at RT are presented in Fig. 2. All samples' primary peaks of XRD patterns can all be well indexed to the hexagonal subcells. Evidently, Fe- and Mn-ions have good solubility in solid solvent of Ca3Co2O6. However, a minor amount of impurity phases appear in the sample with M — Ni which can be discerned as the CoNiO2 and CaO separately, and this may be explained by the dominating bivalent state of oxidized nickel and thus it is hard for the crystal field of Ca3Co2O6 to compensate the charges once such a high level of Co-ions is removed by the Ni2 + ions. Despite this fact, all of the compositions are still referred as

Ca3(Co09M01)2Og for simplicity in this article regardless of the known secondary phases and actual stoichiometric constituents. Back to Fig. 2, another interesting observation which can be made is that the samples of M = Fe and Mn seem to have higher crystalline purity as faintly suggested by the smaller humps imposed on the baselines at 20 e 20 °C compared with M = Co, and meanwhile their baselines are also smoother. This theoretical amorphous hump is almost totally flattened in the case of M=Mn. So, good ionic-radius tolerance and well-balanced bonding network can be totally reached in the case of M=Fe and Mn.

Besides, positional shift of the main diffraction peaks differs from one sample to another, indicative of different degrees of expanded hexagonal lattice. The lattice parameters calculated by the Jade program are listed in Table 1. As M is set as Fe and Mn, the lattice parameters a, b, and c are all increased compared with that of the parent Ca3Co2O6. This lattice expansion can be ascribed to three factors: one is the substitution of Fe and Mn ions with larger radius for Co-ions, and the second is the cationic valence change due to the substitution, and the third is the difference in the ionic-occupation site. As pointed out, Ni2 + and Fe3 + ions tend to reside at trigonal prism site, 6a (0, 0, 1/4) of Co [20,21], and Mn4 + ion like to occupy the octahedral site, 6b (0, 0, 0) [22]. As the cathode materials of SOFC, expansion of lattice helps to support higher degree of freedom for oxygen ionic movement and in turn to accelerate the kinetics of oxygen reduction reaction.

Chemical states of elements O and Co are among the most critical factors in determining the ionic and electronic transport properties of a cobaltite conductor, and the binding state of

20 40 60 80

Fig. 2. XRD patterns of Ca3(Co0.9M01)2O6 powders after fully annealed at 900 °C in air.

Table 1

Lattice parameters of Ca3(Co0.9M0.1)2O6 obtained from Jade programa.

each element is assessed by the XPS of O1s and Co2p core-level. The O1s spectrum of Ca3Co2O6 sample consisted of two components around 529.5 and 532.1 eV (Fig. 3a) respectively, and one should reasonably be assigned to the surface oxygen species (Osur/) while another to the lattice ones (OLattice) [23,24]. Superimposed upon this profile, evolution of O1s spectra with the varied atomic compositions shows the little change in shape, implying that ordered or disordered state of oxygen vacancy remained essentially undisturbed in spite of incorporation of guest M-anions. When the M in Ca3(Co0.9M0.1)2O6 is Mn, there is a slight shift in peak-position and this result is likely correlated with change in balanced M-O bonding strength due to the electron-doping effect of Mn.

Side by side, Fig. 3b shows the asymmetric Co2p core-level spectra in the Ca3(Co0.9M0.1)2O6 series compounds. In all samples, there is one dominating valence state as implied by monotonous fluctuation of the main peaks, reasonably corresponding to Co3 + (components Co2p1/2 and Co2p3/2). Conforming to the drawn conclusions from O1s, the difference in binding scenario of Co2p appears when M is Mn. Between the main asymmetric peaks of Co2p, there is an obviously newly-appeared satellite peak between the two main ones of Ca3(Co0 9Mn01)2O6, theoretically corresponding to Co2 + for the electron doping effect of Mn since the preferred state of the oxidized manganese is + 4, which can denote more electrons to reduce high valence of Co-ions to low valence state.

The electrical conductivity (s) of Ca3(Co09M01)2O6 (M = Co, Fe, Mn, Ni) as the function of heating temperature is subsequently shown in Fig. 4a. All compounds behave like a thermal-activated semiconductor in the temperature range between 300 °C and 800 °C and their apparent electrical conductivity differ only slightly from each other. As it is well-known, B-site sublattice cations are commonly responsible for the electron transfer in transition-metal oxides through strong overlapping B-O-B bonds in accordance with Zerner double-exchange mechanism, illustrated in Eq. (1), and the overlapping between oxygen 2p orbits with transition metal 3d orbit results in the electronic conduction involving B cations with different valence states.

b«+ _ O2- - B(n+1)+ ^ B(n+1)+ - O- - B("+1)+

^ B(n+1)+ - O2- - Bn+ (1)

Among, Mn can effectively improve the electrical conductivity of Ca3(Co0.9M0.1)2O6 over the whole temperature range while Fe can also do in the temperature range of cell operation.

1 □ CoNiO2 o CaO M = Mn

1, , 1 ,i CaCoO PDF#51-0311 II ,1 ,!i ......................

Sample Space group a (A) b (A) c (A) a (deg.) f (deg.) y (deg.)

M = Co R-3c 9.074 (3) 9.074 (3) 10.377 (2) 90 90 120

M=Fe R-3c 9.080 (5) 9.080 (5) 10.380 (1) 90 90 120

M=Mn R-3c 9.087 (3) 9.087 (3) 10.424 (7) 90 90 120

aData for M= Ni discarded for low reliability.

524 526 528 530 532 534 536 538 540 Binding energy (eV)

780 790 800

Binding energy (eV)

Fig. 3. XPS of (a) O1s and (b) Co2p in Ca3(Co0.9M01)2O6.

—m- M - Co

j M - Fe

k M - Mn

1 M - Ni

J*/" ik.'.'

w 1,»'

300 400 500 600 700 800

T (oC)

"■■-T. A

-H-M - Co

-(J- M - Fe -A- M - Mn -{»- M - Ni

1.2 1.4 1.(

1000/T (K-1)

Fig. 4. (a) Temperature dependent electrical conductivity (s), (b) Arrhenius plots of log (sT) vs 1000/T for Ca3(Co0.9Mo.i)2O6 in the air.

As for Ni, the higher electrical conductivity should be expected if all target amount of alivalent Ni2 + ions can be successfully introduced into the lattice. The fact that the actual electrical conductivity of Ca3(Co09Ni01)2O6 is not superior to Ca3Co2O6 at higher temperatures may be explained by the presence of insulating impurity phases like CaO.

On the other hand, knowledge of conduction mechanism can help to learn more about structural information, and so the Arrhenius plots of ln(sT) vs 1000/T is shown in Fig. 4b. For the M — Co and Fe sample, the corresponding curve is hard to be linearly fitted over the whole investigating temperatures, which means that the nature of conduction at lower temperatures differs from that in higher temperature range. As for M — Mn and Ni, a nearly straight line can be essentially fitted. Considering the 1D feature of crystallographic structure, introduction of manganese and nickel anions may enhance the inter-pillar charge-exchange on the whole investigating temperature range, which may occur only in high temperature range for M— Co, and Fe.

In the fabrication and operation of the SOFC, there are concerns regarding the thermal expansion behavior of a cathode material, and the large TEC mismatch between the cell' s adjacent components make it difficult to achieve a firm connection. So, the TECs of all samples were measured in air when the temperature was ranged from 25 °C to 1000 °C, as shown in Fig. 5. All specimens' expansions are almost linearly changed with heating temperatures between 200 °C and

950 °C after initial conditioning of the measure system. Based on the differential calculation additionally, the average TECs of all Ca3(Co0 9M01)2O6 samples are estimated to distribute around 16 (10 _ 6K _1), which are all much closer to the above-mentioned oxide electrolytes than most Co-based per-ovskite cathodes (typically above 20). Compositional elements of Fe, Mn and Ni are especially appreciated since they can further reduce the TEC of Ca3(Co0.9M0.1)2O6 to lower levels as opposed to that of Ca3Co2O6. What is more, the continuously progressing change of thermal expansion with temperature demonstrates the lattice stability on the one hand, and, on the other hand, implies that there is no ordering phase of oxygen vacancies formed during the heating in all members, which is viewed as a prerequisite for producing effective oxygen ionic mobility [25].

Kinetics of oxygen reduction reaction on the title cathodes was first studied by EIS with the LSGM-supported symmetric electrodes, and the corresponding Nyquist plots of the cells are shown in Fig. 6a. All curves contain essentially two heavily overlapped capacitance arcs arising from different relaxation of electrode process, and this can be further confirmed by the Bode plots in Fig. 6b since each curve contains primarily two time constants (arrows pointing C1 and C2 respectively in Fig. 6b), which means that ORR is governed by at least two different electrode processes. Based on this results, the cathode-electrolyte interface can be modeled correspondingly by the equivalent circuit of Rohm(R1Q1)(R2Q2) (floating upon Fig. 6a), in which, the overall

ohmic resistance, Rohm, includes the electrolyte resistance, the electrode ohmic resistance, and the lead resistance. On the all arcs, the high-frequency resistance is probably associated with chargetransfer processes (R1). The low-frequency arc is ascribed to diffusion processes (R2), including the adsorption-desorption of oxygen, oxygen diffusion at the gas-cathode interface, and the surface diffusion of intermediate oxygen species [19,26]. The difference between the real axes intercepts of the impedance plot is considered to be the cathode polarization resistance (Rp=R1 + R2), commonly normalized as area specific resistance (ASR). On the whole, ASR increases much unfavorably with the decreasing temperature as indicated by the ln(ASR)-1000/T lines (Fig. 6c), and so the title cathode materials are not suitable for running in the

0.014 0.012 0.010

^ 0.008

300 400 500 600 700 800 900 1000 1100 T (oC)

Fig. 5. Temperature dependent thermal expansion (AL/Lo) of Ca3(Co0.9M0.1)2O6 measured in air.

relatively lower temperature ranges. Compared to Ca3Co2O6 at 800 °C, the ASR is almost the same except for Ca3(Co0.9Mn0.1)2O6 and the ASRs increase in the order of Ca3(Co0.9Mn0.1)2O6 > CaB^cNn^ > Ca3(Co0.9Fe0.1)2 O6 > Ca3(Co09Co01)2O6. However, ASRs of Ca3(Co0 9Mn01)2O6 and Ca3(Co0.9Ni0.1)2O6 are less sensitive to the decreasing temperatures compared with those of Ca3(Co0.9Fe01)2O6 and Ca3(Co0.9Co0.1)2O6 as demonstrated by the magnitudes of active energy (Ea) values calculated from the Arrhenius plots of ASRs in Fig. 6c. The difference in this property is probably caused by the difference in electrical conductivity since conductivities of Ca3(Co0.9M)2O6 (M= Ni, Mn) are also less sensitive to temperatures compared with M=Co and Fe (Fig. 5). Good electrical conductivity equally facilitates the charge transfer between cathode and neutral adsorbed oxygen-species in the process of ORR. It is also worth noting that the magnitudes of the Ea fitted from thermal evolution of ASRs are extraordinarily low compared with those of other highly active cathodes [27].

Since EIS of a symmetric cell comprises both the anodic and the cathodic branches of an electrode in principle, it is necessary to study the overpotential behavior of the cathode on the 3-electrode platform. Fig. 7 collects the current-over-potential curves of Ca3(Co0.9M01)2O6 cathodes. Polarizing extent of all electrodes is rather sensitive to operating temperature as exemplified by Ca3Co2O6 in the Fig. 7a. Of our particular concern is the effect of varied atomic composition on overpotential within the cell (Fig. 7b). Ca3(Co0.9Ni0.1)2O6 cathode exhibit the largest polarized extent

0.25 0.20 ^ 0.15

I 0.10

0.05 0.00

-n-M = Co -(J-M-Fc w M = Mn -T-M = Ni

-0.05 ..................

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

10 100 1000 10000 100000 /(Hz)

0.92 0.96

1.00 1.04 1.08 1000/7* (k"1)

1.12 1.16

Fig. 6. Electrochemical impedance analysis of symmetrical cells: (a) Nyquist plots of Ca3(Co0.9M0.1)2O6 electrodes at 800 °C in the air and (b) the corresponding Bode plots; (c) Arrhenius plots of ASRs of the electrodes with corresponding values of active energy (Ea).

1 800 °C

< 750 °C

700 "C

—v— 600 °C

0.5 1.0 1.5

Current (A.cm-2)

0.4 0.6 0.8 Current (A.cm-2)

Fig. 7. (a) Effect of operation temperatures on the overpotential of Ca3Co2O6; (b) Comparison of overpotential between Ca3(Co0.9M01)2O6 serial cathodes at 800 °C ( ohmic drop across LSGM layer strictly controlled by fixing thickness to 240 ^m).

1.2 —u— 800°C

fete1 Ji>f>fi0m*\. —a— 750°C

1.0 —A— 700 °C

g —W— 650°C

200 g £

0.5 1.0

Current (A.cm2)

200 fe

Current (A.cm2)

200 £

Current (A-cm2)

200 g £

1.0 1.5

Current (A.cm)

Fig. 8. Current dependence of potentials and power density of single cells with Ca3(Co0.9M01)2O6 cathode at different temperature: (a) M=Co, (b) M=Fe, (c) M=Mn, (d) M=Ni.

than any other ones. The polarization extent varies in the sequence of Ca3(Co0.9Ni0.i)2O6 > Ca3(Co09Mn01)2O6 > Ca3(Co0.9Fe0.i)2O6 e Ca3Co2O6. Considering the lowered cobalt content, it is remarkable that Ca3(Co0.9Fe01)2O6 exhibit the same overpotential with Ca3Co2O6.

Fig. 8 displays the their voltage and power density as a function of the current density with the Ca3(Co0.9M01)2O6 samples as cathodes working in a pure H2. A 240-^m-thick LSGM disc was used as electrolyte and Ni-SDC as anode. Open circuit potential (Eocp) of all cells reach the value nearly as high as 1.15 V at 800 °C, almost reaching to the Nernst limit. It is clearly seen that the Ca3(Co0 9M01)2O6 (M = Co, Fe) cathodes are superior to those of Ca3(Co0.9M01)2O6 (M=Mn, Ni) in term of electro-catalytic activity, and highest power

density close to ca. 500 mW cm " 2 at 800 °C is achieved from the single cell with Ca3(Co09Fe01)2O6 as the cathode, Fig. 9b. The power density order is Ca3(Co09Fe01)2O6 > Ca3Co2O6 > Ca3(Co09Mn01)2O6 > Ca3(Co09Ni01)2O6, consistent with the order based on the cathodic polarization curve on the 3-electrode cells.

Fig. 9 shows the SEM micrographs of cathode/electrolyte interface and electrode surface. The cross-view SEM image between cathode and electrolyte is displayed in Fig. 9a. All the cathodes produce the similar porous backbone with thicknesses of about 20 ^m. It is clear that no obvious phase boundary was observed, indicative of a good connection between cathodes and electrolyte. However, different surface morphologies of cathode backbone were clearly seen, as displayed in Fig. 9b-e.

Fig. 9. (a) Cross-sectional SEM images of Ca3(Co0.9Fe01)2O6 (M = Co) cathode/electrolyte interface, and high-magnification SEM images revealing the microstructural details of cathode surface: (b) M = Co, (c) M=Fe, (d) M=Mn, (e) M=Ni.

In contrast with Ca3Co2O6, Ca3(Co09M01)2O6 (M — Fe, Mn) exhibit finer the crystalline grains, and this can improve the cathode performances in at least two ways: (1) alleviating the TEC to a certain extent as predicted from Young-Laplace formula; (2) increasing the internal surface for oxygen exchange by increasing more active sites, which in turn tend to lessen the polarization resistance based on the Adler or ALS model [28,29]:

2F2 V (1 - e)aC0Dkk

where t, e, Dk, k and a are tortuosity, fractional porosity, ionic oxygen diffusion constant, gaseous oxygen surface exchange coefficient and internal surface area/unit volume, respectively, and C0 is the surface concentration of oxygen.

Additionally, there are almost no any invisible secondary particles on the primary ones in the sample of M— Fe and Mn. Returning to the discussion based on baselines in XRD profiles, Fe and Mn atomic compositions in Ca3(Co0.9M0.1)2O6 seem to act as purifying agents for the crystallization. Comparatively, there are many secondary particles evolving out on surface of Ca3Co2O6 particles which can be viewed as amorphous species

from the XRD baseline. In this respect, Fe and Mn seem to act as the unique compositions to promote the structural homogeneity. As for sample with M— Ni, the secondary phase is quite evident either by SEM examination or by XRD detection.

4. Conclusions

Cationic substitutions are employed to enhance the overall performance of Ca3Co2O6 cathode of SOFC. The studies show that the cationic substitution in Ca3Co2O6 has great effects on the structure, electrical conductivity, thermal expansion and electrochemical performance. Lattice of Ca3Co2O6 can be expanded by Fe and Mn substitution. The electrical conductivity of Ca3Co2O6 can be improved by Mn substitution and the thermal expansions can be well suppressed by Fe-, Mn-and Ni-substitutions. As cathode material for SOFC, Ca3(Co09Fe01)2O6 exhibits the power density as high as ca. 500 mWcm_2 at 800 °C. The difference in the electro-catalytic property can be mainly attributed to the difference in ion-electron mixed transporting properties and cathode-electrolyte interface structure. Our experiment demonstrates that one-dimensional compounds of Ca3(Co09M01)2O6 are promising cost-effective cathode materials.

Acknowledgment

This work was supported by the Natural Science Foundation of China (Grants 513111014 and 21175050), and the PCSIRT (No. IRT14R18). In addition, the authors thank the Analytical and Testing Center of Huazhong University of Science and Technology for XRD and SEM measurements.

References

[1] E.D. Wachsman, K.T. Lee, Science 334 (6058) (2011) 935-939.

[2] Z. Shao, et al., Angew. Chem. Int. Ed. Engl. 50 (8) (2011) 1792-1797.

[3] Z. Shao, W. Zhou, Z. Zhu, Prog. Mater. Sci. 57 (4) (2012) 804-874.

[4] J. Richter, et al., Monatshefte Chem. - Chem. Mon. 140 (9) (2009) 985-999.

[5] A.J. Jacobson, Chem. Mater. 22 (3) (2010) 660-674.

[6] C. Sun, R. Hui, J. Roller, J. Solid State Electrochem. 14 (7) (2009) 1125-1144.

[7] M.G. Bellino, et al., J. Am. Chem. Soc. 129 (11) (2007) 3066-3067.

[8] Z. Shao, S.M. Haile, Nature 431 (7005) (2004) 170-173.

[9] Y. Li, et al., Chem. Mater. 23 (22) (2011) 5037-5044.

[10] L. Dieterle, et al., Adv. Energy Mater. 1 (2) (2011) 249-258.

[11] L. Jiang, et al., J. Power Sources 232 (0) (2013) 279-285.

[12] F. Dong, et al., Int. J. Hydrog. Energy 37 (5) (2012) 4377-4387.

[13] Z. Cai, et al., Chem. Mater. 24 (6) (2012) 1116-1127.

[14] E.J. Crumlin, et al., Energy Environ. Sci. 5 (3) (2012) 6081-6088.

[15] T. Wei, et al., Sci. Rep. 3 (2013).

[16] H.-C. zur Loye, et al., CrystEngComm 14 (1) (2012) 23.

[17] T. Wei, et al., Chem. Mater. 24 (8) (2012) 1401-1403.

[18] Y.-H. Huang, et al., Chem. Mater. 21 (11) (2009) 2319-2326.

[19] W. Zhou, et al., J. Mater. Chem. 21 (39) (2011) 15343-15351.

[20] A. Jain, S. Singh, S. Yusuf, Phys. Rev. B 74 (17) (2006).

[21] S. Kawasaki, M. Takano, T. Inami, J. Solid State Chem. 145 (1) (1999) 302-308.

[22] V.G. Zubkov, et al., J. Solid State Chem. 160 (2) (2001) 293-301.

[23] M.E. Lynch, et al., Energy Environ. Sci. 4 (6) (2011) 2249-2258.

[24] E. Mutoro, et al., Energy Environ. Sci. 4 (9) (2011) 3689-3696.

[25] W. Zhou, et al., Int. J. Hydrog. Energy 35 (3) (2010) 1356-1366.

[26] E.P. Murray, T. Tsai, S.A. Barnett, Solid State Ion. 110 (3-4) (1998) 235-243.

[27] W. Zhou, et al., Angew. Chem. Int. Ed. 52 (52) (2013) 14036-14040.

[28] S.B. Adler, J. Electrochem. Soc. 143 (11) (1996) 3554.

[29] B.C.H. Steele, K.M. Hori, S. Uchino, Solid State Ion. 135 (1-4) (2000) 445-450.