The comparison of SrTi0.98Nb0.02O3–δ-CeO2 and SrTi0.98Nb0.02O3–δ-YSZ composites for use in SOFC anodes
- First online:
Composites of Nb-doped strontium titanate mixed with yttria-stabilized zirconia or cerium oxide in 50:50, 70:30 and 85:15 weight ratios were evaluated as possible anode/electrolyte interface materials for solid oxide fuel cells in terms of chemical compatibility, electrical conductivity and mechanical properties. It has been shown that composite samples prepared by typical powder-mixing methods remain single-phase up to 1400°C. The electrical conductivity of these composites, regardless of their composition and fabrication conditions, is lower than the conductivity of pure SrTi0.98Nb0.02O3–δ, but in most cases sufficient for solid oxide fuel cells anode application. The best properties are found for samples reduced at 1400°C for 10 h in H2 atmosphere. The observations made by scanning electron microscope suggest that the grains of both phases are well-distributed throughout the whole volume of the investigated samples, and that the composites with CeO2 better adhere to the electrolyte surface. The electrical results confirm that composites with at most 30 wt % of YSZ/CeO2 phase fulfill the anode requirements. However, the fuel cell performance tests indicate that the application of composite with CeO2 results in the lower power density than the application of the composite with YSZ.
KeywordsSolid oxide fuel cell Nb-doped SrTiO3 Anode Composite
The dominant anode material for Solid Oxide Fuel Cell (SOFC) application is a Ni-YSZ (metallic-ceramic) composite of Ni and yttria-stabilized zirconia (YSZ). The cermet has many advantages, such as high catalytic activity, thermal expansion coefficient (TEC) compatible with the electrolyte (YSZ)  and high value of electrical conductivity . However, it is also sensitive to deposition of carbon on the surface, which blocks the anode reaction, and can be poisoned by sulphur [2, 3]. These problems limit the possibility of using Ni-YSZ for carbon-containing fuels. Moreover, Ni-YSZ can undergo microstructural changes during redox cycles, which can reduce the three-phase area and thereby reduce the electrode activity [2, 3].
There is a strong need to find an alternative to Ni-YSZ cermets for SOFC applications. The most promising materials are perovskite-related structures, such as SrTiO3. Substitution of a rare earth element (e.g. yttrium) into the A sublattice of SrTiO3, or a transition metal (e.g. niobium) into the B sublattice, increases the electronic conductivity of the material [4, 5]. It has been shown recently that the samples in which 2% of titanium is substituted with niobium: SrTi0.98Nb0.02O3–δ (STNb2) have the highest conductivity in comparison with the other investigated compositions . The conductivity of strontium titanate may also be increased by introducing the strontium vacancies  and by annealing the material in reducing conditions (e.g. the NH3 atmosphere) . Due to its high electronic conductivity, doped SrTiO3 may be competitive to Ni-YSZ as a SOFC anode material. However, at present the ionic conductivity and the catalytic properties of SrTiO3 are not sufficient for anode application . Moreover, the SrTiO3 anode tends to delaminate from the electrolyte surface [9, 10].
The aim of this work is to eliminate these limitations. The idea is based on an analogy to Ni-YSZ anodes. The high temperature firing of the nickel cermet allows the YSZ phase from the anode composite to sinter with the YSZ electrolyte. As a result, a seamless interface between the anode and the electrolyte is formed, and channels, which allow ionic conduction into the electrode, are created . In the case of perovskites, the composite of perovskite phase and YSZ applied as an interface layer is expected to reduce the delamination processes and increase the triple phase boundary (TPB). In consequence, it should improve the mechanical stability of the anode/electrolyte interface and may result in the increase of fuel cell performance.
Another possible composite material for use at the anode/electrolyte interface is a doped-SrTiO3/cerium oxide or doped-SrTiO3/gadolinia-doped ceria (CGO), in which ceria offers the dual benefits of conductivity and catalytic activity. Both doped and undoped ceria are mixed ionic and electronic conductors at low oxygen partial pressure. CeO2-x has high values of oxygen surface exchange and presents an n-type conductivity of about 1 S/cm at 900°C in oxygen partial pressure of 10−18 atm . In this paper only results obtained for undoped ceria will be presented in order to avoid additional complications from possible reactions between Gd2O3 and SrTiO3 phases. Moreover, pure CeO2 seems to fulfil necessary requirements of ionically conducting component of composite.
There are various methods of composite preparation. Samples produced by a typical mixing powders method present both mechanical and electrical advantages. However, Ahn et al.  suggest that the process of YSZ / CeO2 impregnation with STNb2, instead of mixing powders, will provide sufficient electronic conductivity even if conductive phase is lower than 30-vol%. They observed that a non-random structure is formed in which the electronically conductive component (STNb2) coats the YSZ / CeO2 backbone. On the other hand, it is reported  that it is difficult to produce mechanically strong composites based on the porous YSZ / CeO2.
In this paper the properties of the composites of SrTi0.98Nb0.02O3–δ with YSZ or CeO2, prepared by mixing powder method, will be presented and discussed. The possibility of using these composites at the SOFC anode/electrolyte interface has been examined.
Composites of xSrTi0.98Nb0.02O3–δ-(100-x)CeO2 (STNb2-CeO 2 ) and xSrTi0.98Nb0.02O3–δ-(100-x)YSZ (STNb2-YSZ) (where x = 50,70,85 wt. % of STNb2) were fabricated at various temperatures of calcination (1200–1400°C) and reduction (1200–1400°C) in hydrogen for up to 10 h. The 50, 70 and 85 weight % of STNb2 correspond to 58, 77 and 89 vol.% of STNb2 in the case of STNb2-CeO2 composites, whereas in the case of STNb2-YSZ composites such weight ratios correspond to 54, 73 and 87 vol. % of STNb2, respectively.
The SrTi0.98Nb0.02O3–δ ceramic was prepared using a conventional solid-state reaction method. SrCO3 (Sigma Aldrich), TiO2 (Sigma Aldrich) and Nb2O5 (Fluka) powders were mixed in the suitable stoichiometric ratios and ball milled with ethanol for 24 h. After drying the mixture was pressed into bars (~280 MPa) and calcined in air at 1200°C for 12 h. Then, the pellets were reground to obtain better samples quality, uniaxially pressed at 560 MPa and sintered at 1400°C for 12 h in air. The SrTi0.98Nb0.02O3–δ samples prepared in this way were ball milled again to obtain a fine-grained anode powder. Then the powder was mixed in a proper weight ratio either with YSZ (8-mol% yttria-stabilized zirconia, Daiichi Kigenso HSY-8) or with CeO2 (Fluka) powder in a ball mill for 12 h. This final ball-milling step is necessary to disperse the two phases of the composite. Indeed, we find that if we use a mortar, rather than the ball mill, the grains of electronically conductive (STNb2) phase agglomerate, creating large islands that are isolated from each other. To obtain porous samples the composite powder was mixed with starch (5 wt. %).
To improve the electrical properties all of the samples after the synthesis were reduced in dry hydrogen at various temperatures (1200°C–1400°C). For comparison, some samples of each composition, before the reduction, were sintered at 1400°C for 3 h in air. Such processes of initial heat treatment in air strongly influence the porosity. The values of porosity vary in the range of 10% to 35% and are presented together with the conductivity plots in the further part of this paper. The thickness of the bulk samples was approximately 1 mm, while of the thin layers was approximately 50 μm.
To examine the materials in the operating fuel cell the powders were mixed with 5 wt % of carbon black and the organic binder. One layer of composite paste (approximately 15 μm thick) and 2 layers of STNb2 anode paste (approximately 30 μm thick) were painted onto the surface of dense 0.8 mm thick YSZ electrolyte. In order to burn out the organic binder, the cells were calcined at 700°C for 3 h in air. Then they were reduced at 1400°C for 10 h in dry hydrogen. Finally, the La(Ni,Fe)O3 cathode material was deposited on the cells. Cathode was sintered during the operation of the cell. All the cell measurements were taken at 800°C using humidified (3% H20) hydrogen as a fuel and air as an oxidant.
The composites were examined using various methods. The phase composition of the bulk samples was analyzed by an X-ray diffraction method by the X’Pert Pro MPD Philips diffractometer using Cu Kα (1.542 Å) radiation at room temperature. The electrical conductivity of bulk samples was measured using the four terminal DC method over temperature range of 400–950°C in humidified hydrogen. The measurements were performed at constant heating and cooling rates (2°C/min) on bar samples with parallel 9907 ESL silver electrodes. The current–voltage (i-V) curves of operating fuel cells were obtained by a Solartron 1260 Frequency Response Analyzer coupled with Solartron 1294 Impedance Interface. The morphology of the samples was characterized by the Philips XL30 scanning electron microscope (SEM), using secondary electron (SE) and back-scattered electron (BSE) detectors. To recognize the elements and their distribution in composite samples the Energy-Dispersive X-ray spectroscopy (EDX) was performed by the EDAX Sapphire® spectrometer in particular points of each sample. The density of the samples was measured with the Archimedes method assisted by vacuum saturation.
3 Results and discussion
The SEM observations also suggest that the phase distribution is not influenced by heat treatment. It appears that the heat treatment changes only the density of the samples. Specifically, density increases with the temperature of initial calcination / reduction.
For the STNb2-CeO2 composites (Fig. 9), in which the addition of CeO2 phase was to improve the catalytic properties of the material, the same tendency is observed. However, the conductivity of the 85STNb2–15CeO2 sample is considerably higher than that of 70STNb2–30CeO2.
In order to mix the advantages of both STNb2-YSZ and STNb2-CeO2 composites there is an idea to infiltrate the STNb2 or STNb2-YSZ anode support with CeO2 solution instead of mixing powders. Such process may prevent the creation of YSZ (Ce,Ti) dense layer responsible for the fuel cell performance’s decrease and should be investigated further.
It should be also noticed that a higher cell performance for all samples is expected when reducing the thickness of the electrolyte and using better cathode material. Moreover, the catalytic activity of anode can be additionally increased by infiltrating it with small amounts of a catalyst, such as Ce, Ni, Co, Ru, Pt and Pd. These catalysts decrease the polarisation resistance of anode and some of them (e.g. Ce, Ru) improve the anode tolerance to sulphur containing fuels [15, 16].
Another fact that should be considered, is the mechanical misfit between the composite components. This appears to be especially important for STNb2-CeO2, because the resistance to layer cracking and delamination during reduction-oxidation decreases with the increasing CeO2 concentration [8, 12]. Koutcheiko et al.  reports that it is caused by the contraction of the CeO2 lattice parameter in less reducing atmospheres. The thermal expansion coefficient (TEC) of composites consisting of yttrium-doped strontium titanate and cerium oxide is very similar to the TEC of YSZ (10.3 × 10−6 K−1 [8, 17]), at least in the temperature range 50–1000°C and in air . However, the value of the TEC increases in a reducing atmosphere, and the higher the concentration of CeO2 in the composite, the higher the TEC of composite material . We observed, in the present work, that this mismatch results in the cracking of bulk pellets and in the delamination of composite layers from the YSZ electrolyte support. Mixing with starch to increase porosity minimizes this problem for samples with 15 and 30 wt. % of CeO2. However, even with the increased porosity, composites with 50 wt. % of CeO2 are still mechanically unstable and not suitable for SOFC application. The problems with TEC misfit between the composite and the YSZ electrolyte do not take place for STNb2-YSZ.
The electrical and structural properties of xSrTi0.98Nb0.02O3–δ-(100-x)CeO2 and xSrTi0.98Nb0.02O3–δ-(100-x)YSZ (where x = 50,70,85 wt. % of SrTi0.98Nb0.02O3–δ) have been studied, with particular attention given to their potential for use at anode/electrolyte interfaces in SOFC. The XRD analyses demonstrate, that in all investigated composites there is no reaction between the components. The required conductivity level is obtained for at least 70 weight % of perovskite phase content. The best properties were observed for samples reduced at 1400°C for 10 h in H2 atmosphere. The SEM analysis confirms a good dispersion of both phases in the investigated composites. Although the STNb2-CeO2 composite material (with at most 30 wt. % of CeO2) presents very good mechanical and electrical properties from the fuel cell point of view, it results in a slightly lower fuel cell performance than STNb2-YSZ composite. To conclude, it can be stated that both composites are noteworthy.
This project is partially supported by Ministry of Higher Education under the grant No. N511 376135 and National Science Center under the grant No. NCN DEC-2011/01/N/ST5/05579.
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.