Bismuth and niobium co-doped barium cobalt oxide as a promising cathode material for intermediate temperature solid oxide fuel cells

Journal of Power Sources 295 (2015) 33e40

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Bismuth and niobium co-doped barium cobalt oxide as a promising cathode material for intermediate temperature solid oxide fuel cells Shaofei He a, Shiru Le b, *, Lili Guan a, Tao Liu a, Kening Sun b, ** a b

Department of Chemistry, School of Science, Harbin Institute of Technology, NO. 92 Xidazhi Street, Harbin, 150001, PR China Academy of Fundamental and Interdisciplinary Sciences, Harbin Institute of Technology, NO. 2 Yikuang Street, Harbin 150080, PR China

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


BaBi0.05Co0.95yNbyO3d (0.1  y  0.2) is investigated as cathode for IT-SOFCs.
BaBi0.05Co0.8Nb0.15O3d presents the biggest lattice parameter 4.719 Å.
With Nb doping, the electrical conductivities and TEC values gradually decrease.
Maximum power density of single cell with y ¼ 0.15 reaches 1.23 W cm2 at 750  C.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 14 January 2015 Received in revised form 10 June 2015 Accepted 25 June 2015 Available online xxx

Perovskite oxides BaBi0.05Co0.95yNbyO3d (BBCNy, 0  y  0.2) are synthesized and evaluated as potential cathode materials for intermediate temperature solid oxide fuel cells (IT-SOFCs). Highly charged Nb5þ successfully stabilizes the cubic perovskite structure to room temperature with Nb substituting content y  0.1. The phase structure, thermal expansion behavior, electrical conductivity and electrochemical performance of BBCNy with cubic phase are systematically studied. The samples exhibit excellent chemical compatibility with GDC and have sufficiently high electrical conductivities. However, the thermal expansion coefficients of BBCNy samples are nearly twice those of the most commonly used electrolyte materials YSZ and GDC, which is a major drawback for application in IT-SOFCs. The polarization resistances of BBCNy with y ¼ 0.10, 0.15 and 0.20 on GDC electrolyte are 0.086, 0.079 and 0.107 U cm2 at 700  C, respectively. Even though the YSZ electrolyte membrane and GDC barrier layer are approximately 50 mm and 10 mm in thickness, the highest maximum power density (1.23 W cm2) of the single cell Ni-YSZjYSZjGDCjBBCN0.15 is obtained at 750  C. Good long-term stability of the single cell with BBCN0.15 cathode is also demonstrated. These results demonstrate that BBCNy perovskite oxides with cubic structure are very promising cathode materials for IT-SOFCs. © 2015 Elsevier B.V. All rights reserved.

Keywords: Solid oxide fuel cell Cathode Cubic perovskite structure Intermediate temperature Electrochemical performance

1. Introduction

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (S. Le), [email protected] (K. Sun). http://dx.doi.org/10.1016/j.jpowsour.2015.06.134 0378-7753/© 2015 Elsevier B.V. All rights reserved.

Solid oxide fuel cells (SOFCs) are highly efficient and environmentally friendly devices to convert chemical energy in fuels directly to electrical power as compared to traditional thermal power generation plants [1e3]. Recently, considerable efforts are focused on lowering the working temperature to an intermediate

34

S. He et al. / Journal of Power Sources 295 (2015) 33e40

range of 600e800  C [4]. Unfortunately, with lowering the operating temperature, the cathode polarization increases more rapidly than other polarizations, which is severely detrimental for performances of the fuel cells [5]. Therefore, it is critical to exploit novel cathode materials with superior catalytic activity for IT-SOFCs. Due to the superior catalytic activity for oxygen reduction reaction (ORR) at reduced temperatures, oxygen-deficient conductors with mixed ionic and electronic conductivities have been comprehensively and thoroughly developed [6e8]. Among the large variety of mixed conducting oxides, BaCoO3d is a promising parent component for many perovskite oxides with diverse properties [9]. On the one hand, Ba2þ with the large ionic radius (1.60 Å) is beneficial to create large free volume [10] and sufficient oxygen vacancy [11,12] for oxygen ions transportation in the bulk; on the other hand, the low binding energy of BaeO can enhance the fast oxygen diffusion in the bulk and the quick oxygen exchange over the surface [13]. Therefore, the cathode materials based on BaCoO3d should have promising performances. Reported as an oxygen ion conductor, bismuth oxide has a high oxygen ion conductivity due to the small binding energy of BieO bond [14]. Niu et al. [15,16] investigated the effect of Bi replacing in A-site of Bi1xSrxFeO3d and found that the electrode performance was promoted effectively. Zhou et al. [17] studied the B-site ordered double perovskite Ba2Bi0.1Sc0.2Co1.7O6d and found that it showed promising performance as a cathode material for IT-SOFCs. However, Shao et al. [14] demonstrated that a low Bi doping concentration in BaBixCo0.2Fe0.8xO3d was favorable due to the large expansion of Bi at high temperatures resulted from the thermal reduction reaction of Bi5þ to Bi3þ. Therefore, in this work, only 5 mol% Bi will be doped in B-site of BaCoO3d. However, just doping 5 mol% Bi cannot stabilize the simple cubic perovskite structure to room temperature due to the large ionic radius mismatch between Ba and Bi/Co. To enhance the structure stability, one of the most reliable and effective strategies is to dope proper cations in B-site. Nagai et al. [18] found that Nb2O5 was the most effective dopant for the structure stability of SrCoO3d-based oxides. Dong et al. [5] further demonstrated the effectiveness of Nb doping in B-site for stabilizing the cubic perovskite structure of BaFeO3 and found that a small amount of Nb doping further enhanced the electrochemical activity of BaNb0.05Fe0.95O3d. In addition, Zhou et al. [19] reported that SrSc0.175Nb0.025Co0.8O3d achieved extremely outstanding performance due to the presence of favorable transport paths by doping a small amount of Nb. Therefore, the substitution of Co by Nb may not only stabilize the cubic perovskite structure to room temperature, but also promote the performances of cathode materials. Recently, Wang et al. [20] studied Nb-doped BaBi0.05Co0.95O3d as oxygen permeable membranes and excellent oxygen permeation performance was found for BaBi0.05Co0.8Nb0.15O3d, which was comparable to that of the state-of-the-art Ba0.5Sr0.5Co0.8Fe0.2O3d membrane. However, to the best of our knowledge, its performance as a cathode material for IT-SOFCs is still unreported so far. In this work, Nb-doped BaBi0.05Co0.95O3d perovskite oxides, i.e. BaBi0.05Co0.95yNbyO3d, are synthesized as cathode materials for IT-SOFCs. The crystal structure, oxygen vacancy concentration, thermal expansion behavior, electrical conductivity and electrochemical performance of BaBi0.05Co0.95yNbyO3d are systematically evaluated as a function of Nb doping content. 2. Experimental 2.1. Powder synthesis and cell fabrication BaBi0.05Co0.95yNbyO3d (BBCNy, y ¼ 0.0e0.2) powders were synthesized by a conventional solid state reaction method [20].

Analytical grade BaCO3 (Aladdin, China), Bi2O3 (Aladdin, China), Co3O4 (Aladdin, China) and Nb2O5 (Aladdin, China) were used as the raw materials. Stoichiometric amounts of the raw materials were weighted and well mixed by high-energy ball milling at 300 rpm for 12 h using ethanol as the liquid medium. After drying, the primary powder was pressed into a pellet and pre-calcined at 900  C for 10 h in air. Then, the pellet was ground thoroughly and subsequently pressed into a membrane and sintered at 1100  C for 10 h in air. The as-obtained sample was ball-milled for 6 h to decrease the average particle size. Two-electrode symmetric cells with GDC pellets as the substrate were fabricated for impedance studies. The dense electrolyte pellets were prepared by dry pressing at 30 MPa for 10 min with the commercial Ce0.9Gd0.1O1.9 (GDC10-N, Fuel Cell Materials, USA) and subsequently sintered at 1500  C for 10 h in air. The synthesized cathode powder with an average grain size of 4 mm (see Fig. S1 in supporting information) was dispersed in a pre-mixed solution (6 wt% ethyl cellulose in terpineol solution) to prepare the cathode slurry, which was then screen-printed onto both sides of a GDC pellet in a symmetric configuration and subsequently sintered at 950  C for 2 h in air. To fabricate complete electrochemical cells, NiO-YSZ (NiO, High Purity Chemicals, Japan; YSZ, Tosoh, Japan) anode supporting electrode and YSZ electrolyte was fabricated by the co-tape casting and co-firing technique which was reported in our previous work [21]. GDC was used as the barrier layer prepared by screen-printing and then sinter at 1300  C for 2 h to prevent the chemical reaction between BBCNy cathodes and YSZ electrolyte during high temperature operations. The cathode slurry was screen-printed onto the barrier layer and calcined at 950  C for 2 h in air. The diluted silver paste (DAD-87, Shanghai Research Institute of Synthetic Resins, China) and filamentary silver were used as the current collector and conductor for symmetric cells and single cells, respectively. 2.2. Characterizations The crystal structures of BBCNy samples were identified by Xray diffraction (XRD, PANalytical X'Pert PRO X-ray diffractometer) with Cu Ka radiation (l ¼ 1.5418 Å) over the 2q range of 20e80 with an interval of 0.02 . The obtained XRD data was analysed by Rietveld method with GSAS-EXPGUI software [22] and Powder4 was used for data preparation. The Pm-3m cubic perovskite structure was employed as the initial model for BBCNy (y ¼ 0.10e0.20) samples, where Ba was located at (0, 0, 0) site, Bi/ Co/Nb at (0.5, 0.5, 0.5) site, and O at (0.5, 0.5, 0) site, respectively. During the refinement process, lattice parameters, oxygen occupation, peak shape (Pseudo-Voigt function), background and viso parameter were refined. The oxygen nonstoichiometry of BBCNy powders at room temperature was measured by iodometric titration method. Approximately 1 g of KI powder and 0.05 g of the sample were dissolved in about 5 mL HCl solution (~6 mol L1). The Na2S2O3 solution, whose concentration was titrated by a standard K2Cr2O7 solution, was used for the titration. About 2 mL starch solution was added into the solution as a titration indicator before the titration terminal point. At the titration terminal point, the colour of the solution abruptly changed from blue to yellow. The chemical compatibility between BBCNy cathodes and GDC electrolyte is examined by XRD. In a 1:1 weight ratio, the mixture of BBCN0.15 and GDC was sintered at 950  C for 10 h in air. Then XRD was used to detect the phase reaction result. Thermal expansion coefficient (TEC) was measured in air using a Netzsch DIL 402PC dilatometer from 30 to 900  C at a heating rate of 5  C min1. The electrical conductivity was measured in air by the four-terminal technique using a Keithley 2400 multimeter with an

S. He et al. / Journal of Power Sources 295 (2015) 33e40

35

interval of 50  C over a temperature range of 100e800  C. The dense BBCNy bars with the dimensions of 2 mm  5 mm  10 mm were calcined at 1000  C for 5 h in air. The density of the sintered bars was measured by the Archimedes method and only the samples with the relative density higher than 95% were selected for TEC and electrical conductivity tests. The electrochemical impedance spectra (EIS) were typically investigated on symmetric cells under open-circuit conditions. To better elucidate the ORR mechanism, EIS measurements were performed on the single cell of Ni-YSZjYSZjGDCjBBCN0.15 as a function of the dc bias at 700  C. The dc bias was applied between 0.1 V and 0.6 V with reference to the open-circuit voltage [23,24]. The EIS of single cells with different cathodes at different temperatures were also measured under open-circuit conditions. All the EIS were investigated using an electrochemical workstation (PARSTAT 2273) with the frequency range of 0.01 Hze100 kHz under a signal amplitude of 10 mV. The single-cell performances were measured by Arbin instruments using humidified (3% H2O) hydrogen as fuel with a flow rate of 80 mL min1 and ambient air as oxidant. To test the long-term stability of the cathode material, the voltage under a constant current density of the single cell with BBCN0.15 as the cathode was monitored for 140 h. The morphology of the single cell after electrochemical tests was observed using scanning electron microscopy (SEM, SU8000). 3. Results and discussion 3.1. Phase structure and chemical compatibility Fig. 1(a) shows the typical XRD patterns of BBCNy (y ¼ 0.00, 0.05, 0.10, 0.15 and 0.20) and magnified XRD patterns of (110) peak of BBCNy (y ¼ 0.10e0.20) calcined at 1100  C for 10 h in air. As can be seen from Fig. 1(a), BBCNy oxides crystallize in a single-phase cubic perovskite structure without any detectable impurity phases within the limits of the XRD measurement from y ¼ 0.10 to y ¼ 0.20. However, BaBi0.05Co0.95O3d and BaBi0.05Co0.9Nb0.05O3d are mainly in a hexagonal polymorph phase BaCoO2.7 (JCPDS No. 47-0211) [25]. This result indicates that Nb is successfully incorporated into the Co-site of BaBi0.05Co0.95O3d and exchanging Co by proper Nb (y ¼ 0.10e0.20) in B-site can stabilize the cubic perovskite structure to room temperature. Because the oxygen vacancy in the hexagonal phase is associated and/or ordered, which is severely detrimental for the cathode performance, only the BBCNy samples in cubic perovskite structure will be studied systematically in this manuscript [26]. It can also be seen that with the increase of Nb replacing content, the corresponding (110) peaks of BBCNy samples (y ¼ 0.10e0.20) shift towards a smaller angle first and then shift towards a larger angle, indicating that the lattice parameter increases with Nb substituting from y ¼ 0.10 to y ¼ 0.15 and then decreases with the further increase of Nb replacing level. To obtain the detail information about the crystal structure of BBCNy (y ¼ 0.10e0.20) samples, Rietveld refinement was applied on the diffraction profiles using GSAS-EXPGUI software. The calculated lattice parameters (a, b and c) and oxygen nonstoichiometry coefficient (dR) are listed in Table 1. The small values of c2, Rwp and Rp demonstrate the good agreement between calculated and observed data. Fig. 1(b) shows the calculated and experimental XRD profiles of BBCN0.15 as an example. With the increase of Nb substituting content, dR decreases, while the lattice parameter increases first and then decreases. During the refinement, the valence state of bismuth doped in B-site was defined as þ5. The good quality of the simulation demonstrates that in BBCNy oxides, Bi mainly takes a valence state of þ5. In other cubic perovskite oxides with Bi doped in B-site, the valence state of Bi is also assumed as þ5 [14,27e29]. According to Bhalla [30], in the

Fig. 1. (a) Typical and magnified X-ray diffraction patterns of BBCNy samples sintered at 1100  C for 10 h in air and (b) Rietveld refinement profile of BBCN0.15.

Table 1 Lattice parameters (a, b and c), oxygen nonstoichiometry coefficient (dRa and dIb) and average Co valence state (nIb) of BBCNy (y ¼ 0.10e0.20) oxides calculated by Rietveld refinement and iodometric titration method, respectively. Samples

a ¼ b ¼ c (Å)

c2

Rwp (%)

Rp (%)

dR

dI

nI

y ¼ 0.10 y ¼ 0.15 y ¼ 0.20

4.0706 4.0719 4.0714

1.590 1.409 1.488

7.03 6.84 6.94

5.61 5.39 5.51

0.327 0.321 0.252

0.376 0.345 0.293

2.939 2.889 2.885

a b

dR is calculated by Rietvelt refinement. dI and nI are measured by iodometric titration method.

perovskite phase with a general formula of ABO3, the A-site cation has a large ionic radius (1.10e1.80 Å) while the B-site cation has a medium ionic radius (0.62e1.00 Å). For Bi3þ and Biþ5 in sixcoordination, the ionic radius is 1.03 and 0.76 Å, respectively. Therefore, it is reasonable to conclude that the bismuth in BBCNy oxides has the valence state of þ5. The common valence states of Co cations are þ2, þ3 and þ 4, while Nb ion has the fixed valence state of þ5 [20]. The substitution of Co2þ, Co3þ or Co4þ by Nb5þ in B-site may lead to the decrease of oxygen vacancy concentration and the reduction of Co ions from high valence state Co4þ/Co3þ to low valence state Co3þ/Co2þ for the electrostatic neutrality consideration [31]. The former will result in the decrease of lattice parameters due to the larger interaction between A/B cations and oxygen ions, whereas the latter will lead to the lattice expansion due to the larger ionic radius of Co ions with lower valence state. Oxygen

36

S. He et al. / Journal of Power Sources 295 (2015) 33e40

nonstoichiometry coefficient (dI) and average Co valence state (nI) of BBCNy (y ¼ 0.10e0.20) oxides measured by iodometric titration method are also listed in Table 1. When Nb exchanging content increases from y ¼ 0.10 to y ¼ 0.15, dI approximately remains the same while nI decreases obviously. However, with further increase of Nb replacing concentration from y ¼ 0.15 to y ¼ 0.20, dI decreases clearly while nI almost stay the same. Therefore, the electrostatic neutrality is mainly realized through the reduction of Co ions instead of the decrease of oxygen vacancy concentration with Nb substituting from y ¼ 0.10 to y ¼ 0.15, while it is on the contrary for Nb replacing from y ¼ 0.15 to y ¼ 0.20. This may be associated with the low average valence state of Co ions (2.889) in BBCN0.15, which is difficult to be further reduced to achieve the electrostatic neutrality when Nb exchanging content increases from y ¼ 0.15 to y ¼ 0.20. The interfacial reaction between electrolyte and cathode is highly detrimental for the cell performance due to the increase of the interfacial polarization resistance. Fig. 2 shows XRD patterns of BBCN0.15, GDC and BBCN0.15-GDC mixture sintered at 950  C for 10 h in air, respectively. As can be seen from Fig. 2, no additional phases and any peak shifts are identified within the limits of the XRD measurement, demonstrating the good chemical compatibility between BBCNy cathodes and GDC electrolyte. 3.2. Thermal expansion behavior Besides achieving the requirement of chemical compatibility, a satisfactory cathode material must meet other requirements such as the mechanical compatibility, i.e. the matching TECs. Fig. 3 illustrates the thermal expansion curves of BBCNy (y ¼ 0.10e0.20) over a temperature range of 30e900  C. It can be seen that all the curves are nonlinear with inflections occurring in the temperature range of 250e550  C. To explain the difference among the thermal expansion behavior of different samples more clearly, differential curves of DL/L0 vs. temperature were calculated and the results are separately presented in Fig. S2 in the supporting information. For all the samples, there are peaks in the temperature range of 250e550  C, which are corresponding to the inflection points in thermal expansion curves. This can be attributed to the thermal reduction of cobalt cations from Co4þ/Co3þ to Co3þ/Co2þ and the loss of lattice oxygen. In addition, the intensity of the peak decreases with the increase of Nb replacing level, indicating that the

Fig. 2. XRD patterns of BBCN0.15, GDC and BBCN0.15-GDC sintered at 950  C for 10 h in air.

Fig. 3. Thermal expansion curves of BBCNy samples.

amount of Co ions carrying out the thermal reduction reaction decreases with Nb substituting. It can be attributed to the decrease of average Co valence state with Nb replacing due to the electrostatic neutrality. The peaks shift towards a slightly lower temperature with Nb substituting. According to the previous report, the lower average valence state of cobalt cations, the lower temperature thermal reduction happens [32]. Therefore, this may be associated with the decrease of average Co valence state with the increase of Nb exchanging content. The TEC values of BBCNy samples are 22.0  106, 21.5  106 and 20.8  106 K1 for y ¼ 0.10, 0.15 and 0.20 over the temperature range of 100e800  C, respectively, nearly twice those of the most commonly used electrolyte materials YSZ (~10.5  106 K1) [33e35] and GDC (~12.5  106 K1) [35e38]. Even though the TEC of BBCN0.2 is comparable to those of the novel cathode materials containing cobalt cations for IT-SOFCs such as Ba0.5Sr0.5Co0.8Fe0.2O3d (20.9  106 K1) [39], BaCo0.7Fe0.2Nb0.1O3d (18.2  106 K1) [40] and SrCoNb0.1O3d (24.2  106 K1) [41], the high TEC can become a major obstacle for use of BBCNy as the cathode material for IT-SOFCs. 3.3. Electrical conductivity Electrical conductivity is an important performance indicator. The mixed ionic and electronic conductors exhibit simultaneously electronic and ionic conductivities due to the coexistence of electron holes and oxygen vacancies [40]. Because the electronic conductivity is always much higher than the ionic conductivity, the measured electrical conductivity is approximately equal to the electronic conductivity. Fig. 4 shows the temperature dependence of electrical conductivity of BBCNy samples measured in air. As can be seen from Fig. 4, the electrical conductivity increases gradually with temperature, which exhibits a semiconductor-like behavior. However, there are fluctuations in the temperature range of 250e550  C, which can be attributed to the formation of oxygen vacancy as a result of the thermally induced loss of lattice oxygen [42]. The formation of oxygen vacancies is accompanied by the reduction of Co ions from Co4þ/Co3þ to Co3þ/Co2þ, resulting in the decrease of the charge carrier (mainly electron holes) concentration, and hence the decrement of the increment rate [43]. Furthermore, the decrement of the increment rate at the inflection temperature is weakened by the substitution of Nb for Co in B-site. It can be attributed to the decrease in the amount of Co ions reduced from high valence state to low valence state with Nb

S. He et al. / Journal of Power Sources 295 (2015) 33e40

37

Fig. 4. The electrical conductivity of BBCNy as a function of temperature measured in air.

substituting, which is consistent with the thermal expansion behavior. It can also be seen that the electrical conductivity decreases with the increase of Nb exchanging concentration at temperatures higher than 400  C. The electron conduction in perovskite oxide proceeds via the Bnþ-O2--B(n1)þ network due to the strong overlapping between B:3d and O:2p orbitals by a small polaron hopping conduction mechanism similar to the Zener double exchange process [44,45]. As demonstrated in Table 1, the donor doping of Nb in B-site results in the partial reduction of Co4þ/ Co3þ to Co3þ/Co2þ as charge compensation and consequently decreases the concentration of electron holes, hence the lower electrical conductivity. In addition, with Nb substituting, the nonconducting NbeO bond increases, which obstructs the polaron transport and as a result the electrical conductivity decreases [41].

▫

Fig. 5. Electrochemical impedance spectra of BBCNy cathodes ( , D and B denote the measured data of y ¼ 0.10, 0.15 and 0.20, respectively, while the lines represent the fitting results) measured in air at different temperatures: (a) 600  C, (b) 650  C and (c) 700  C.

3.4. Polarization resistance To examine the electrochemical properties of BBCNy cathodes, EIS were measured in symmetric half cells (BBCNyjGDCjBBCNy) under open-circuit conditions in air. The Nyquist plots and fitting results of BBCNy (y ¼ 0.10, 0.15 and 0.20) at 600e700  C are presented in Fig. 5(a)e(c). In order to clearly and easily compare the cathode polarization resistance, the ohmic resistance from leading wires and electrolyte are normalized to zero while the interfacial polarization resistance is divided by two due to the contributions of two symmetric electrodes. According to the simulated results, the impedance spectra can be separated to one high-frequency arc and one low-frequency arc, indicating the existence of at least two different electrode processes during the ORR process. The high frequency arc can be attributed to charge transfer steps, which include the electron transfer at the cathode/current collector interface and the ion transfer at the cathode/electrolyte interface. The low frequency arc can be related to nonchemical processes including oxygen adsorption/desorption, oxygen diffusion within the cathode surface and oxygen ions diffusion in the cathode bulk [46e49]. By measuring the EIS at different dc voltages, it is possible to examine the role of different ORR processes [50,51]. Fig. 6 shows the Nyquist plots and fitting results of the single cell NiYSZjYSZjGDCjBBCN0.15 under different dc biases at 700  C in air. It can be seen that the calculated curves are in good agreement with the measured data. According to the simulated results, there are three arcs in the impedance spectra: one high-frequency arc, one middle-frequency arc and one low-frequency arc. The RH, RM and RL

Fig. 6. Electrochemical impedance spectra of the single cell Ni-YSZjYSZjGDCjBBCN0.15 under different dc biases at 700  C in air. Inset: RH, RM and RL obtained rom the impedance measurement as the function of the negative applied dc voltages.

values

derived

from

the

equivalent circuit Rohthe negative applied dc bias are shown in the inset in Fig. 6. The RH and RM are not significantly affected by the dc bias, whereas RL decreases remarkably with the increase of applied dc voltage. In fact, the applied dc bias can create a higher oxygen vacancy concentration near the cathode/current collector interface, favouring oxygen adsorption/desorption over the surface lattice defects [52], oxygen diffusion within the cathode surface and oxygen ions transportation [53]. Even though the high oxygen vacancy concentration is beneficial for the ion transfer at the cathode/electrolyte interface, as demonstrated by the electrical

mic(QHRH)(QMRM)(QLRL) as a function of

38

S. He et al. / Journal of Power Sources 295 (2015) 33e40

conductivity analysis, the high oxygen vacancy concentration is accompanied with the decrease of electron hole concentration, which is detrimental for the electron transfer at the cathode/current collector interface. Therefore, the RH has a weaker dependence on the overpotential than RL. This result confirms, at least in part, that the high frequency arc is associated with charge transfer steps, whereas the low-frequency arc is related to nonchemical processes and it is not the rate-limiting step of ORR, which is consistent to the reported phenomenon that the low-frequency arc is normally associated with gas diffusion [54]. Polarization resistances (RH, RL and Rp) obtained from the equivalent circuit Ro(QHRH)(QLRL) at 600e700  C for symmetric cells BBCNyjGDCjBBCNy are illustrated in Table 2. When Nb replacing concentration increases up to y ¼ 0.15, Rp reaches a minimum, i.e. 0.311 U cm2 at 600  C, 0.152 U cm2 at 650  C and 0.079 U cm2 at 700  C, respectively. Compared with the polarization resistances of BBCN0.1, the RH of BBCN0.15 is approximately equal to that of BBCN0.1 while the RL of BBCN0.15 is lower and responsible for the lower Rp, indicating that the charge transfer steps are similar for BBCN0.1 and BBCN0.15 whereas the nonchemical steps are more convenient in BBCN0.15. This can be attributed to the fact that the oxygen vacancy concentration and electrical conductivity for BBCN0.15 are only a little lower than that for BBCN0.1, which may not impede the charge transfer steps obviously. However, with Nb substituting from y ¼ 0.1 to y ¼ 0.15, the lattice parameter increases, which is beneficial for the oxygen ions transportation in the bulk. As demonstrated by iodometric titration method, the charge compensation is mainly realized by the decrease of oxygen vacancy concentration with Nb replacing from y ¼ 0.15 to y ¼ 0.20, which can hinder the charge transfer steps and hence the largest RH which is responsible for the largest Rp of BBCN0.2 [13]. The low Rp values demonstrate that the BBCNy cathodes have superior electrocatalytic activity for ORR at intermediate temperatures. Arrhenius plots of RH and RL for BBCNy, accompanying with activation energy (Ea) and standard deviation for Ea are shown in Fig. S3(a) and (b) in supporting information, respectively. All the resistances (RH and RL) decrease remarkably with increasing temperature, indicative of thermal activation behavior of the ORR processes, which can be attributed to the increase of oxygen vacancy concentration as a result of the thermally induced loss of lattice oxygen. In addition, the Ea values for RL are much higher than those for RH, indicative of more significant influence of working temperature on nonchemical processes than charge transfer steps, which is similar to the influence of the applied dc bias on polarization resistances. Therefore, although RH dominates Rp demonstrated by data in Table 2, the higher Ea for RL indicates that RL will become more significant at lower temperatures. Fig. 7 shows the Arrhenius plots of Rp for BBCNy samples, accompanying with Ea

Fig. 7. Arrhenius plots of Rp for BBCNy samples.

Table 3 Ea of the superior reported cathode materials and BBCN0.2 prepared in this work. Composition

Ea (kJ mol1)

Reference

SrCo0.9Nb0.1O3d SrCo0.95Sb0.05O3d SrCo0.95Mo0.05O3d SrCo0.95Ti0.05O3d La0.3Sr0.7Ti0.4Co0.6O3d La0.6Sr0.4Co0.2Fe0.8O3d BaCo0.7Fe0.2Nb0.1O3d Ba0.5Sr0.5Co0.8Fe0.2O3d BaBi0.05Co0.8Nb0.2O3d

104.4 93.4 93.4 132.0 110.7 157.0 100.0 106.0 87.7

[41] [55] [56] [57] [47] [58] [59] [60] This work

and standard deviation for Ea. In fact, the Ea is an activation energy for at least two compared processes, RH and RL for the symmetric cell. The high standard deviation for Ea of Rp demonstrates that there is a true difference in Ea between samples, which can be attributed to the difference in intrinsic properties of the samples. The activation energies for RH, RL and Rp are also summarized in Table 2. It can be seen that the corresponding activation energies for Rp of y ¼ 0.10, 0.15 and 0.20 cathodes are 100.8, 96.7 and 87.7 kJ mol1, respectively. As shown in Table 3, the Ea of BBCN0.2 cathode is superior to and comparable with those of reported novel cathode materials, indicating a potentially better oxygen reduction activity at low temperatures.

Table 2 Polarization resistances obtained from the equivalent circuit R0(QHRH)(QLRL) at 600e700  C for symmetric cells BBCNyjGDCjBBCNy and the corresponding activation energy (Ea). Temperature ( C) y ¼ 0.10

y ¼ 0.15

y ¼ 0.20

RH (U cm2) RL (U cm2) Rp (U cm2) chi2 (  106) RH (U cm2) RL (U cm2) Rp (U cm2) chi2 (  106) RH (U cm2) RL (U cm2) Rp (U cm2) chi2 (  106)

600

650

700

Ea (kJ mol1)

0.246 0.112 0.358 7.77 0.247 0.064 0.311 5.62 0.306 0.065 0.371 6.36

0.133 0.025 0.158 4.13 0.134 0.018 0.152 5.79 0.172 0.023 0.195 4.34

0.079 0.007 0.086 5.56 0.075 0.004 0.079 5.13 0.101 0.006 0.107 6.45

80.2 195.8 100.8 e 84.1 195.2 96.7 e 78.2 167.6 87.7 e

S. He et al. / Journal of Power Sources 295 (2015) 33e40

3.5. Single-cell performance The electrochemical performances of BBCNy cathodes were further evaluated in single cells with the configuration of NiYSZjYSZjGDCjBBCNy at the temperature range of 650e750  C. Fig. 8(a) shows the curves of power density and cell potential as a function of current density for Ni-YSZjYSZjGDCjBBCN0.15, using humidified (3% H2O) H2 as the fuel and air as the oxidant. The open-

39

circuit voltages of the cells reach 1.07, 1.11 and 1.11 V (see Fig. S4(a) and (b) in supporting information) for y ¼ 0.10, 0.15 and 0.20 cathodes at 650  C, respectively. This result indicates that the electrolyte membranes are sufficiently dense. The highest maximum power densities for y ¼ 0.10, 0.15 and 0.20 cathodes are obtained at 750  C, i.e. 1.18, 1.23 and 1.08 W cm2 (see Fig. S4(a) and (b) in supporting information), respectively, which are comparable with that of the similar fuel cell with BaCo0.6Fe0.3Sn0.1O3d (BCFSn631) as the cathode (1.1 W cm2 at 750  C) [9]. The high power densities demonstrate that the BBCNy with cubic perovskite structure are promising cathode materials for IT-SOFCs. The electrocatalytic activity of BBCN0.15 is the best, which is consistent with the impedance analysis. This result indicates that BBCN0.15 cathode is preferred for application in IT-SOFCs. The impedance spectra of single cells Ni-YSZjYSZjGDCjBBCNy at 700  C under open-circuit conditions are shown in Fig. S4(c) in supporting information. The measured data is well fitted with the simulated curves based on the equivalent circuit Rohmic(QHRH)(QMRM)(QLRL). The calculated results are shown in the inset in Fig. S4(c). The ohmic resistances (Rohmic) for all the cells are approximately the same (~0.8 U cm2), and the variation of RH and RL for single cells are similar to that for symmetric cells with different cathodes. The total cell resistances (Rtotal) are 0.382, 0.361 and 0.423 U cm2 for BBCN0.1, BBCN0.15 and BBCN0.2, respectively, which are in accordance with discharge performances of the single cells. Fig. 8(b) shows the EIS of Ni-YSZjYSZjGDCjBBCN0.15 at different temperatures from 650 to 750  C. The calculated results from Rohmic(QHRH)(QMRM)(QLRL) are shown in the inset in Fig. 8(b). The Rtotal, Rohmic and RL decrease gradually with increasing temperatures, whereas RH decreases significantly with the increase of temperature. However, there is no regular change of RM with the increase of temperature and the reason is not clear. The Rtotal are 0.567, 0.361 and 0.213 U cm2 for the single cell with BBCN0.15 cathode at 650, 700 and 750  C, respectively. The long-term stability test of the single cell with BBCN0.15 as the cathode at 650  C was performed and the result is shown in Fig. 8(c). Under a current density of 280 mA cm2, the single cell maintains a stable voltage at around 0.8 V over a test period of 140 h. A low degradation rate of 0.217 mV h1 is observed, which is comparable with the similar fuel cell with BCFSn631 as the cathode, whose degradation rate is 0.197 mV h1 [9]. The good stability can be attributed to the stable cubic phase structure, which is obtained by replacing Co by Nb in the B-site of BaBi0.05Co0.95O3d. However, compared to the traditional cathode La1xSrxMnO3 without apparent voltage degradation under cathodic polarization conditions (~0.03 mV h1) [34], the long-term stability for BBCN0.15 cathode is weak and needs to be enhanced for application in ITSOFCs. The SEM image of the cross-section of Ni-YSZjYSZjGDCjBBCN0.15 after cell testing is shown in Fig. S5 in supporting information. The YSZ membrane, GDC barrier layer and BBCN0.15 cathode layer are approximately 50, 10 and 20 mm in thickness, respectively. There are no cracks and delaminations, indicating that the GDC interlayer is suitable to prevent the chemical reaction between BBCNy cathodes and YSZ electrolyte. 4. Conclusions

Fig. 8. (a) Current density dependence of potential and powder density for the single cell Ni-YSZjYSZjGDCjBBCN0.15 measured at 650e750  C, (b) EIS of the single cell measured at different temperatures in air and (c) long-term stability test of the single cell measured at 650  C.

Perovskite oxides BBCNy (y ¼ 0.00e0.20) were successfully synthesized by the solid-state reaction and selected compositions were systematically investigated as the cathode materials for ITSOFCs. The Nb doping content at B-site had a significant effect on crystal structure, oxygen vacancy concentration, electrical conductivity, thermal expansion behavior and electrochemical performance. The cubic perovskite structure was stabilized to room

40

S. He et al. / Journal of Power Sources 295 (2015) 33e40

temperature for BBCNy oxides with Nb doping content y  0.10. Among the studied compositions, BBCN0.15 presented the biggest lattice parameter 4.719 Å. At 950  C for 10 h in air, BBCN0.15 cathode was chemically compatible with GDC electrolyte. With the increase of Nb doping content, the electrical conductivities and TEC values gradually decreased. For BBCNy samples, the electrical conductivities were sufficiently high in the intermediate temperature range. The TEC values of BBCNy were nearly twice those of YSZ and GDC, which can become a major drawback for application of BBCNy as the cathode material for IT-SOFCs. The Rp for BBCNy with y ¼ 0.10, 0.15 and 0.20 on GDC electrolyte were 0.086, 0.079 and 0.107 U cm2 at 700  C, respectively. For y ¼ 0.10, 0.15 and 0.20, the maximum power densities of the single-cells (Ni-YSZjYSZjGDCjBBCNy) with BBCNy as the cathodes reached 1.18, 1.23 and 1.08 W cm2 at 750  C, respectively. The Rtotal are 0.567, 0.361 and 0.213 U cm2 for the single cell with BBCN0.15 cathode at 650, 700 and 750  C, respectively. The good long-term stability of the single cell with BBCN0.15 as the cathode was demonstrated. All these results indicated that the BBCNy oxides with y ¼ 0.1e0.2 were very promising cathode materials for IT-SOFCs. In view of the best electrochemical performances, BBCN0.15 was preferred for application in IT-SOFCs. Acknowledgements The authors gratefully acknowledge financial support from Postdoctoral Science Research Foundation (contract No. LBHQ11112) and National Natural Science Foundation of China (contract No. 21006016). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2015.06.134. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]

S.B. Adler, Chem. Rev. 104 (2004) 4791e4843. S. McIntosh, R.J. Gorte, Chem. Rev. 104 (2004) 4845e4865. Z. Shao, S.M. Haile, Nature 431 (2004) 170e173. D.J. Brett, A. Atkinson, N.P. Brandon, S.J. Skinner, Chem. Soc. Rev. 37 (2008) 1568e1578. , S. Liu, Z. Shao, J. Mater. Chem. A 1 F. Dong, Y. Chen, R. Ran, D. Chen, M.O. Tade (2013) 9781e9791. S.B. Adler, J.A. Lane, B.C.H. Steele, J. Electrochem. Soc. 143 (1996) 3554e3564. S. Pang, X. Jiang, X. Li, Q. Wang, Z. Su, J. Power Sources 204 (2012) 53e59. C. Xia, W. Rauch, F. Chen, M. Liu, Solid State Ion. 149 (2002) 11e19. B. Qian, Y. Chen, M.O. Tade, Z. Shao, J. Mater. Chem. A 2 (2014) 15078e15086. H. Hayashi, H. Inaba, M. Matsuyama, N.G. Lan, M. Dokiya, H. Tagawa, Solid State Ion. 122 (1999) 1e15. S. McIntosh, J. Vente, W. Haije, D. Blank, H. Bouwmeester, Solid State Ion. 177 (2006) 1737e1742. P. Zeng, Z. Chen, W. Zhou, H. Gu, Z. Shao, S. Liu, J. Membr. Sci. 291 (2007) 148e156. C. Zhu, X. Liu, C. Yi, L. Pei, D. Yan, J. Niu, D. Wang, W. Su, Electrochem. Commun. 11 (2009) 958e961. Z. Shao, W. Yang, Y. Cong, H. Dong, J. Tong, G. Xiong, J. Membr. Sci. 172 (2000) 177e188. Y. Niu, W. Zhou, J. Sunarso, L. Ge, Z. Zhu, Z. Shao, J. Mater. Chem. 20 (2010) 9619e9622.

[16] Y. Niu, J. Sunarso, W. Zhou, F. Liang, L. Ge, Z. Zhu, Z. Shao, Int. J. Hydrogen Energy 36 (2011) 3179e3186. [17] W. Zhou, J. Sunarso, Z.G. Chen, L. Ge, J. Motuzas, J. Zou, G. Wang, A. Julbe, Z. Zhu, Energy Environ. Sci. 4 (2011) 872e875. [18] T. Nagai, W. Ito, T. Sakon, Solid State Ion. 177 (2007) 3433e3444. [19] W. Zhou, J. Sunarso, M. Zhao, F. Liang, T. Klande, A. Feldhoff, Angew. Chem. Int. Ed. 52 (2013) 14036e14040. [20] Z. Wang, Y. Kathiraser, S. Kawi, J. Membr. Sci. 431 (2013) 180e186. [21] S. Le, K.N. Sun, N. Zhang, X. Zhu, H. Sun, Y.X. Yuan, X. Zhou, J. Power Sources 195 (2010) 2644e2648. [22] B.H. Toby, J. Appl. Cryst. 34 (2001) 210e213. [23] J.R. Macdonald, D.R. Franceschetti, J. Chem. Phys. 68 (1978) 1614e1637. [24] E. Magnone, M. Miyayama, E. Traversa, J. Electrochem. Soc. 157 (2010) B357eB364. [25] T. Nagai, W. Ito, Solid State Ion. 262 (2014) 650e653. [26] W. Yang, T. Hong, S. Li, Z. Ma, C. Sun, C. Xia, L. Chen, ACS Appl. Mater. Interfaces 5 (2013) 1143e1148. [27] J. Sunarso, J. Motuzas, S. Liu, J.C.D.d. Costa, J. Membr. Sci. 361 (2010) 120e125. [28] P. Zeng, Z. Shao, S. Liu, Z.P. Xu, Sep. Purif. Technol. 67 (2009) 304e311. [29] J. Yang, H. Zhao, X. Liu, Y. Shen, L. Xu, Int. J. Hydrogen Energy 37 (2012) 12694e12699. [30] A.S. Bhalla, R. Guo, R. Roy, Mat. Res. Innov. 4 (2000) 3e26. [31] Y. Cheng, H. Zhao, D. Teng, F. Li, X. Lu, W. Ding, J. Membr. Sci. 322 (2008) 484e490. [32] L. Zhang, W. Long, F. Jin, T. He, Int. J. Hydrogen Energy 38 (2013) 7947e7956. [33] O. Yamamoto, Electrochim. Acta 45 (2000) 2423e2435. [34] S.P. Jiang, J. Mater. Sci. 43 (2008) 6799e6833. [35] M. Mogensen, T. Lindegaard, U.R. Hansen, G. Mogensen, J. Electrochem. Soc. 141 (1994) 2122e2128. [36] C. Sun, R. Hui, J. Roller, J. Solid State Electrochem 14 (2009) 1125e1144. [37] I. Burmistrov, O.A. Drozhzhin, S.Y. Istomin, V.V. Sinitsyn, E.V. Antipov, S.I. Bredikhin, J. Electrochem. Soc. 156 (2009) B1212eB1217. [38] X. Meng, S. Lü, Y. Ji, T. Wei, Y. Zhang, J. Power Sources 183 (2008) 581e585. [39] S. McIntosh, J.F. Vente, W.G. Haije, D.H.A. Blank, H.J.M. Bouwmeester, Chem. Mater. 18 (2006) 2187e2193. [40] Z. Yang, M. Han, P. Zhu, F. Zhao, F. Chen, Int. J. Hydrogen Energy 36 (2011) 9162e9168. [41] F. Wang, Q. Zhou, T. He, G. Li, H. Ding, J. Power Sources 195 (2010) 3772e3778. [42] K.T. Lee, D.M. Bierschenk, A. Manthiram, J. Electrochem. Soc. 153 (2006) A1255eA1260. [43] S. Park, S. Choi, J. Shin, G. Kim, J. Power Sources 210 (2012) 172e177. [44] C. Zener, Phys. Rev. 82 (1951) 403e405. [45] C. Zener, Phys. Rev. 81 (1951) 440e444. [46] M. Li, Y. Wang, Y. Wang, F. Chen, C. Xia, ACS Appl. Mater. Interfaces 6 (2014) 11286e11294.  [47] Z. Du, H. Zhao, Y. Shen, L. Wang, M. Fang, K. Swierczek, K. Zheng, J. Mater. Chem. A 2 (2014) 10290e10299. [48] W. Zhou, R. Ran, Z. Shao, W. Zhuang, J. Jia, H. Gu, W. Jin, N. Xu, Acta Mater. 56 (2008) 2687e2698. [49] J.W. Yin, Y.M. Yin, J. Lu, C. Zhang, N.Q. Minh, Z.-F. Ma, J. Phys. Chem. C 118 (2014) 13357e13368. [50] B. Liu, Y. Zhang, L. Zhang, Int. J. Hydrogen Energy 34 (2009) 1008e1014. [51] E. Siebert, A. Hammouche, M. Kleitz, Electrochim. Acta 40 (1995) 1741e1753. [52] A. Esquirol, N.P. Brandon, J.A. Kilner, M. Mogensen, J. Electrochem. Soc. 151 (2004) A1847eA1855. [53] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, A. Weber, E. Ivers-Tiffee, Electrochim. Acta 46 (2001) 1837e1845. [54] Z. Yang, C. Yang, C. Jin, M. Han, F. Chen, Electrochem. Commun. 13 (2011) 882e885. rez-Coll, C.d.l. Calle, M.T. Ferna ndez-Díaz, [55] A. Aguadero, J.A. Alonso, D. Pe J.B. Goodenough, Chem. Mater. 22 (2010) 789e798. rez-Coll, J.A. Alonso, S.J. Skinner, J. Kilner, Chem. Mater. 24 [56] A. Aguadero, D. Pe (2012) 2655e2663. [57] Y. Shen, F. Wang, X. Ma, T. He, J. Power Sources 196 (2011) 7420e7425. [58] E.P. Murray, M.J. Sever, S.A. Barnett, Solid State Ion. 148 (2002) 27e34. [59] Z. Yang, C. Yang, B. Xiong, M. Han, F. Chen, J. Power Sources 196 (2011) 9164e9168. [60] Z. Chen, R. Ran, W. Zhou, Z. Shao, S. Liu, Electrochim. Acta 52 (2007) 7343e7351.