Secondary Phase Formation in Ba0.5Sr0.5Co0.8Fe0.2O3–d Studied

Jan 23, 2013 - Ba0.5Sr0.5Co0.8Fe0.2O3–d (BSCF) compacts were prepared and annealed under application-relevant temperatures between 700 and 1000 ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/cm

Secondary Phase Formation in Ba0.5Sr0.5Co0.8Fe0.2O3−d Studied by Electron Microscopy Philipp Müller,†,* Heike Störmer,† Matthias Meffert,† Levin Dieterle,†,∥ Christian Niedrig,‡ Stefan F. Wagner,‡ Ellen Ivers-Tiffée,‡,§ and Dagmar Gerthsen†,§ †

Laboratorium für Elektronenmikroskopie, Karlsruher Institut für Technologie (KIT), Engesserstraße 7, 76131 Karlsruhe, Germany Institut für Werkstoffe der Elektrotechnik, Karlsruher Institut für Technologie (KIT), Adenauerring 20b, 76131 Karlsruhe, Germany § DFG-Center for Functional Nanostructures (CFN), Karlsruher Institut für Technologie (KIT), 76131 Karlsruhe, Germany ‡

ABSTRACT: Ba0.5Sr0.5Co0.8Fe0.2O3−d (BSCF) compacts were prepared and annealed under application-relevant temperatures between 700 and 1000 °C for 100 h. The microstructure and chemical composition was investigated by electron microscopic techniques and energy dispersive X-ray spectroscopy (EDXS) to obtain a detailed insight into the formation of secondary phases which are involved in the degradation of the ionic conductivity of BSCF. Secondary phases are precipitated from the cubic BSCF phase at temperatures ≤900 °C. In addition to the well-known hexagonal phase, another secondary phase was identified which has the same crystal structure as Ban+1ConO3n+3(Co8O8) with n ≥ 2 which is denoted as a BCO-type phase. Regions with plate-like morphology are formed which contain thin lamellae of the cubic, hexagonal, and BCO-type phase. The negative impact of the secondary phases on the material performance is discussed. KEYWORDS: BSCF, transmission electron microscopy, perovskite, phase stability, mixed conductor



Al,25 or Cu.26 Nevertheless, further improvement of the thermodynamic stability of the material requires a detailed understanding of the degradation processes on a microscopic scale. Understanding of the degradation of the ionic conductivity in BSCF mainly stems from studies where XRD and conductivity measurements were correlated. However, XRD only yields information on the phases without further details on microstructure. Only few electron microscopic investigations which directly unveil the change of the microstructure and its correlation with the degradation processes were published. Arnold et al.11 and Mueller et al.11,14 investigated the formation of the hexagonal phase in detail. However, it was found by our group and others13,15,27 that at least one further phase with a plate-like morphology is involved in the degradation process. Efimov et al.15 found a new ordered lamellar phase with a 15R structure and described the similarity to Ba3Co2O9(Co8O8) reported by Sun et al.28 It was also already shown that the formation kinetics of the plate-like regions is fast compared to the formation of the hexagonal phase.13 This suggests that the plate-like regions are at least partially responsible for the degradation of the ionic conductivity which was reported to be faster at the beginning of the long-term conductivity tests.7,29 This influence on material performance motivates further investigation of the plate-like regions.

INTRODUCTION In recent years, much attention has been paid to mixed ionic and electronic ceramic conductors as their improvement promises advancement in the fields of solid oxide fuel cells and oxygen permeation membranes.1−6 Among several candidates, Ba0.5Sr0.5Co0.8Fe0.2O3−d (BSCF) showed the best oxygen-conduction performance at intermediate temperatures (600−800 °C).7,8 However, as the material was tailored to achieve a high ionic conductivity, it is characterized by a high oxygen deficit d with respect to the perovskite structure ABO3. The high performance was achieved by incorporation of multivalent cations like cobalt which tends to destabilize the cubic perovskite phase and leads to formation of secondary phases. Previous studies by X-ray diffractometry (XRD) showed that the formation of a hexagonal phase at temperatures below 840 °C9−15 is responsible for degradation of the ionic conductivity. As any deviation from the cubic perovskite phase of BSCF causes a significant reduction of oxygen-ion conductivity,16 the understanding of the decomposition processes and stabilization of the material has highest priority for further material developments, especially in regard to producing dense BSCF thin films which showed the highest oxyen permeation reported so far.17 Several approaches for the stabilization of the cubic BSCF phase, all resulting in limited success, were reported. Doping with single-valent cations like Zr and Y was observed to improve stability to some extent.18−20 Furthermore, the search for cobalt-free high-performing membranes yielded new promising material systems by exchange of Co with Zn,21−24 © 2013 American Chemical Society

Received: November 13, 2012 Revised: January 23, 2013 Published: January 23, 2013 564

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

Figure 1. Plot of the logarithm of the Cliff−Lorimer factors as a function of the X-ray intensity of the element with the weakest X-ray absorption at the energy of the characteristic X-ray line for (a) BaFe12O19, (b) SrTiO3, (c) CoTiO3, and (d) FeTiO3. Subsequently, detailed microstructure characterization of the annealed samples was carried out. Samples for scanning electron microscopy (SEM) were polished and chemically etched using a colloidal silicon dioxide solution (Oxid Polier Suspension, Oberflächentechnologien Dieter Ziesmer, Germany). SEM images were taken with a Zeiss 1530 Gemini microscope using the in-lens detector. Samples for transmission electron microscopy (TEM) were prepared by conventional preparation methods involving grinding, dimpling, polishing, and Ar+-ion etching. TEM and STEM analyses were performed using a Philips CM200FEG/ST and an FEI Titan3 80-300 microscope operated at 200 and 300 keV, respectively. High-angle annular dark-field (HAADF) STEM images were taken to directly visualize different phase regions by exploiting the chemical sensitivity (atomic-number contrast30−32) of this imaging mode. The chemical composition of different phases was determined by EDXS using the 30 mm2 EDAX Si(Li) detector system of the FEI Titan3 80-300 microscope with an ultrathin window and an energy resolution of 136 eV. Composition quantification of the spectra was carried out by using the FEI “TEM imaging and analysis” (TIA) software (Version: 4.3 build 904) on the basis of experimentally determined Cliff−Lorimer factors. The Cliff−Lorimer method33 converts the observed integrated X-ray intensity ratio for elements A and B into the atomic fraction ratio by the equation

This study reports on the results of detailed phase analyses of plate-like regions in BSCF by combining high-resolution (scanning) transmission electron microscopy (TEM, STEM) and chemical analyses by energy dispersive X-ray spectroscopy (EDXS). On the basis of crystal structure and composition determination it will be shown that considerable structure and composition variations occur in plate-like regions. This requires a more precise definition of what is considered to be the platelike phase. On the basis of the phase analyses, a detailed understanding of the formation mechanism of the plate-like regions and their influence on the degradation of BSCF is achieved.



SAMPLE PREPARATION AND EXPERIMENTAL TECHNIQUES

BSCF bulk samples (2 × 5 × 5 mm3) were prepared from commercially available powder with an average grain size of 2.4 μm by isostatic pressing at about 100 MPa. The compacts were then sintered at 1000 °C for 12 h. In a second step the samples were annealed at the desired temperature for 100 h in ambient air and quenched in water. The samples were homogenized at 1000 °C for 24 h prior to the final annealing treatment to ensure that the samples contain only the cubic BSCF phase. 565

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials cA I = kAB A cB IB



(1)

kATi kBTi

(2)

where A, B denotes the elements Sr, Co, Ba, or Fe. To take any thickness dependence into account, absorption-free Cliff−Lorimer factors were derived by the method described by Horita et al.34,35 with data for the X-ray mass absorption coefficients taken from Chantler.36 The intensities of the respective X-ray lines were determined after subtraction of the background signal which was performed by defining background windows in spectrum regions without characteristic X-ray lines. The background subtraction was carried out by fitting a fifth order polynomial to the windows and by extrapolation under the characteristic peaks. Peak fitting was performed standardless by fitting Gaussians to the characteristic peaks of the Kα1 and Lα1 lines. Absorption-free Cliff−Lorimer factors were determined by plotting the logarithm of experimentally determined Cliff−Lorimer factors as a function of the TEM sample thickness, i.e., kBaFe for BaFe12O19 (Figure 1a), kSrTi for SrTiO3 (Figure 1b), kCoTi for CoTiO3 (Figure 1c), and kFeTi for FeTiO3 (Figure 1d). The TEM sample thickness cannot be directly determined. However, the intensity of the X-ray lines with the lowest mass-absorption coefficient can be taken as a good measure of the sample thickness if the same probe current and acquisition time is used. The intersection of the fitted straight line (red curves in Figure 1) with the y-coordinate gives the logarithm of the absorption-free Cliff−Lorimer factor. The resulting absorption-free Cliff−Lorimer factors for the cations are given in Table 1. They are calculated for the use in the TIA

Table 1. Absorption-Free Cliff−Lorimer Factors for Cations in BSCF cation

kATi (this work)

kASi (TIA)

Ba Sr Co Fe

3.17 3.10 1.27 1.08

3.08 3.03 1.44 1.36

EXPERIMENTAL RESULTS

Figure 2a−d shows typical SEM images of BSCF after annealing at temperatures between 700 and 1000 °C for 100 h. The assignment of different phases in the SEM images is based on the image intensity. It was previously shown by a combination of SEM imaging and TEM electron crystallography13 that the cubic phase is characterized by an intermediate intensity, while the hexagonal phase is located in the regions with a darker contrast. Plate-like regions are distinguished by their bright contrast and plate-like shape. Figure 2a shows a sample after annealing at 1000 °C which contains only the cubic BSCF phase. The two large particles with bright contrast correspond to CoO precipitates. As not all microstructural features are visible in the presented SEM images Figure 2a−d, the microstructure is schematically illustrated in Figure 2e−h, which summarizes the evaluation of numerous SEM images and the investigation of grain boundaries by TEM. After annealing at 900 °C (Figure 2b,f), large plate-like regions with sizes of several micrometers are formed at grain boundaries. After annealing at 800 °C (Figure 2c,g), the hexagonal phase is formed, in agreement with ref 9, in addition to the plate-like regions. The precipitation of the hexagonal phase typically occurs at triple points of grain boundaries. A typical hexagonal region is marked in Figure 2c. Smaller plate-like regions nucleate at CoO particles and at grain boundaries. At 700 °C (Figure 2d,h), the formation of plate-like regions is even more pronounced by the presence of fishbonelike structures. Furthermore, the hexagonal phase may completely decorate grain boundaries and occupies a large volume fraction in the vicinity of CoO particles and plate-like regions. Figure 3 shows TEM images (Figure 3a,c) and a scheme (Figure 3b) of a typical plate-like region after annealing at 700 °C. The overview bright-field TEM image Figure 3a contains plate-like regions (indicated by dashed white boxes) and the hexagonal phase. The assignment of phases is based on electron diffraction as reported earlier13 and Fourier analysis of highresolution TEM images (Figure 3c,d). The plate-like regions consist of a lamellar substructure (inside white frames in Figure 3a) which is schematically visualized in the scheme in Figure 3b. The plates with a lamellar substructure are often embedded in the hexagonal phase as shown in Figure 2d. The color scheme in Figure 3b is chosen according to the image intensities of the phases in SEM images. The HRTEM image in Figure 3c reveals the structure of a plate-like region in more detail. It shows the interface between two lamellae which are embedded in a larger region containing exclusively the hexagonal phase. The identification of the phases is accomplished by calculating the Fourier transform of the HRTEM image denoted as a diffractogram in the following. The diffractogram of each lamella is given in the insets of Figure 3c which can be assigned to the cubic phase (left) and the hexagonal phase (right). Figure 3d presents the diffractogram of the whole interface region in Figure 3c with the cubic lattice oriented in [101]- and the hexagonal phase in [21̅1̅0]zone axis. The corresponding indices are given in blue for the cubic phase and red for the hexagonal phase. The orientation relationship between the two phases can be derived from Figure 3d which shows that the (111)cubic and (0001)hexagonal planes are oriented parallel to each other. The projected elementary cells are rotated by an angle of 34° as indicated by the colored boxes for the cubic (blue) and hexagonal lattice (red). The

where kAB is the so-called Cliff−Lorimer k-factor. cAB are the atomic fractions and IAB the measured integrated characteristic X-ray intensities of two elements within the analyzed volume. Element specific Cliff−Lorimer factors kAB33 were determined for all cations in BSCF using different material standards with well-known composition. Stoichiometric titanates of all cations are available. BaTiO3 and SrTiO3 single crystals, as well as CoTiO3 and FeTiO3 powders with a purity of 99.8% were prepared for TEM. EDX spectra of the material standards were recorded at different TEM sample thicknesses and evaluated. The Kα1 lines were used for the lighter elements (Sr, Co, and Fe) and the Lα1 line for the heavier Ba. As the X-ray peaks for Ba (Lα) and Ti (Kα) were found to be too close for an adequate separation, an additional BaFe12O19 sample was prepared. The Ba content of this standard material was analyzed by a gravimetric assay with H2SO4 to be 11.8 wt % compared to calculated 12.4 wt % for the stoichiometric material. All Cliff−Lorimer factors were calculated with respect to Ti with the relation

kAB =

Article

program yielding wt % concentrations. For comparison the Cliff− Lorimer factors of the TIA program are listed which are given with respect to Si. According to eq 2 the influence of the experimental Cliff−Lorimer factors can be considered by calculating kBaSr and kCoFe for both sets of factors. While the effect on the quantification of the elemental composition for the A-site cations Ba and Sr is negligible (kBaSr(this work) = kBaSr(TIA) = 1.02), there is a significant change for the B-site cations Co and Fe (kCoFe(this work) = 1.18 vs kCoFe(TIA) = 1.06). 566

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

Figure 2. SEM images with corresponding schemes which illustrate the formation of secondary phases in BSCF after 100 h annealing at (a,e) 1000 °C, (b,f) 900 °C, (c,g) 800 °C, and (d,h) 700 °C in ambient air.

diffractogram of the hexagonal phase contains streaks along the [0001]-direction, which are related to the planar defects (stacking faults) at the interface between the cubic and hexagonal phase. Additional reflections are visible in the diffractogram of the hexagonal phase (Figure 3c), which cannot be assigned to the defects at the interface. The additional reflections were also observed in other hexagonal regions and may occur due to damage of the phase during the TEM sample preparation or by the electron beam. A detailed analysis of the crystal structure of thin lamellae in a plate-like region is performed on the basis of the highresolution HAADF STEM image Figure 4a. It is noted that the clarity of the contrast is impaired by the amorphization of the sample due to electron-beam irradiation and damage in the course of the Ar+-ion etching process. The columns of the heavier A-site cations Ba and Sr show bright contrast in the image and are marked by large green dots. The lighter B-site cations (smaller blue dots) are not resolved due to the bad signal-to-noise ratio. Two-dimensional projections of the atomic structure in the imaged region are given in Figure 4b. In the perovskite structure, the oxygen atoms form octahedrons (marked in red). If one considers the three-dimensional arrangement of the atoms, the O-octahedrons are aligned corner-sharing in the cubic phase and face-sharing in the hexagonal phase. In the scheme of Figure 4b the alignment of the O-octahedrons therefore allows the distinction between cubic stacking viewed along the [101]cubic-direction and hexagonal stacking in [21̅1̅0]hexagonal-direction. As indicated in Figure 4b, the cubic (010)-planes and hexagonal (011̅0)-planes form an angle of 35.26° at the interface. This matches well with the rotation of 34° measured in the diffractogram in Figure 3d. Some regions can be assigned neither to the cubic nor to the hexagonal phase, which are located inside the white dashed boxes in Figure 4a. In the course of detailed investigations, another phase with a different structure and composition was identified. Figure 5

shows high-resolution HAADF STEM images of a plate-like region in the sample, which was annealed at 800 °C, with the corresponding atomic models in two different zone axes ([211̅ 0̅ ]-zone axis in Figure 5a−c and [1010̅ ]-zone axis in Figure 5d−f). Heavy A-cation columns appear with high intensity in the HAADF STEM images, while B-columns show a weaker contrast. The projected structures along the two zone axes are presented in Figure 5c,f where A-columns are marked in green, B-columns in blue, and oxygen columns in red. The structure models shown in Figures 5c,f are derived from a phase described by Sun et al.28 and David et al.37 which is denoted in the following as BCO-type phase. The phase is also similar to the R15 phase described by Efimov et al.15 The structural details of the phase will be explained in context with Figure 6. In the center of Figure 5a,d, a defect is observed which consists of 17 (111) planes with cubic stacking order. The structure models displayed in Figure 5c,f perfectly describe the experimental images apart from a 3° tilt which can be seen by comparison of Figure 5e,f. The tilt can be explained either by drift during the image acquisition (the scan direction is tilted in respect to the presented image orientation) or by distortions due to a different chemical composition compared to the phase used for the simulation.28 Figure 6 contains the projected structure of the BCO-type phase described by Sun et al.28 along the [21̅1̅0]-direction. It consists of a periodic arrangement of two (111)-planes of Ooctahedrons with cubic stacking order and three planes with a CdI2-type structure and two interface planes. The cubic planes are separated from the other planes by dashed black lines in Figure 6. The CdI2-type structure is characterized by an edgesharing configuration of O-octahedrons. The overall composition of the BCO-type phase according to Sun et al.28 is Ba3Co10O17. The two cubic unit cells are composed of Ba2Co2O6, leaving a composition of BaCo8O11 for the two interface planes and the CdI2-plane. In contrast to the cubic and hexagonal structure, the CdI2-plane violates the ABO3 567

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

Figure 3. (a) Overview bright-field TEM image of a sample after annealing at 700 °C, (b) scheme of the microstructure of plate-like regions, (c) high-resolution TEM image of an interface in a plate-like region containing a cubic and a hexagonal lamella with corresponding diffractograms (insets), and (d) diffractogram of the complete image (c). The cubic phase is observed along the [101]-zone axis with reflections indices denoted in blue. The hexagonal phase is viewed in [21̅1̅0] orientation (red). The rotation of 34° between the two lattices is emphasized by colored boxes.

results from imprecise background subtraction, peak-fitting, and absorption effects, as well as potential artifacts introduced by Ar+-ion milling. Overall the remaining systematic error is estimated to be ±3 atom %. All phases can be clearly distinguished by their chemical composition. The composition of the cubic phase only shows small deviations from the intended composition of Ba0.5Sr0.5Co0.8Fe0.2O3−d (first line in Table 2). The hexagonal phase is characterized by the absence of Fe. However, the chemical composition is close to the ABO3 stoichiometry of the perovskite structure. The composition of the BCO-type phase strongly deviates from the other phases with respect to the A/ B-ratio due to a strongly enriched Co content and the absence of Sr. Plate-like regions consisting of lamellae of all three detected phases (cubic, hexagonal, and BCO-type phase) show average chemical compositions which depend on the volume fraction of the different phases and show therefore significant composition variations. The absence of Sr and the different A/B-ratios (compared to the perowskite structure) in the BCO-type phase agree well with the series of barium cobaltate phases described by Sun et al.28 More precisely, it is a series of structures with the chemical formula Ban+1ConO3n+3(Co8O8) (BCO). For n = 2 the simulated cation positions well explain the HAADF STEM images in Figure 5. Nevertheless the phase observed in this

perovskite stoichiometry since it does not contain A-site cations. As the O-octahedrons are packed more densely in the CdI2-plane, the corresponding B-site columns show a brighter contrast in Figure 5a,b,d,e compared to the other B-sites. Furthermore, the interface planes introduce additional B-sites (blue atomic sites beyond the octahedrons) located in tetrahedral gaps. We note that the BCO-type phase can also occur as lamella beside cubic and hexagonal lamellae in platelike regions. Moreover, single CdI2-type planes are occasionally found in plate-like regions with nonperiodic arrangements of cubic and hexagonal stacking sequences, which explain regions with unidentified stacking in Figure 4b. To analyze the composition of the secondary phases in BSCF (plate-like regions with a random arrangement of cubic, hexagonal,and BCO-type lamellae after annealing temperatures ≤900 °C and hexagonal phase below 840 °C),12,13 extensive standard-based chemical analyses were conducted by EDXS. Line scans with one full EDX spectrum per pixel over regions with different phases were carried out in the FEI Titan3 operated in STEM mode. The spectra were quantified with the Cliff−Lorimer factors listed in Table 1. The chemical compositions of the different phases are given in Table 2 in atom % which were obtained after averaging up to 200 spectra. The errors correspond to the standard deviations of the average values. An additional systematic error has to be considered. It 568

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

Figure 4. (a) HAADF STEM image of a thin lamella in a plate-like region and (b) corresponding structure models of cubic stacking (viewed along the [101]-projection) and hexagonal stacking (viewed along the [21̅1̅0]-projection). The cubic phase consists of oxygen octahedrons (red) in corner-sharing configuration, while the hexagonal phase contains face-sharing octahedrons. Regions which cannot be assigned to any of the two phases are located inside the dashed boxes in (a).

work does not exactly agree with the phase described by Sun et al.28 because Fe is present. According to Sun et al. and David et al.28,37 the valence state of the Co-cations in BCO strongly differs on different Co-sites. As there are positions with low Covalence states (+2.15 on average), these sites can also accommodate larger Fe-cations. The phase reported in this work therefore carries Fe and Co on B-sites. Figures 5e,f reveal a 3° tilt of rows of A- and B-cation sites compared to the originally proposed structure. This distortion may either result from the addition of iron compared to pure BCO or from sample drift during the acquisition of the STEM image. The A/B-ratio for the BCO phase for n = 2 is expected to be 1:3.3 compared to 1:3 determined in this work. This may be partially caused by inaccuracies of the EDXS analyses but also by an intermixing of different members of the BCO series. If more than two cubic perovskite planes (n > 2) are present in the stacking sequence, the A/B-ratio may be shifted toward 1:3. Such a region with larger number of cubic perovskite planes is imaged in Figure 5a,d, where n = 17 planes with cubic stacking are observed in the center of the image. The complex chemical composition of a plate-like region which consists of lamellae of cubic, hexagonal, and BCO-type phases is demonstrated in Figure 7. The HAADF STEM images in Figure 7a,c were taken from the sample that was annealed at 700 °C. The positions and acquisition directions of the EDXS line-scans are indicated by red dashed arrows. Figure 7b shows an EDXS line-scan across a plate-like region embedded in a larger hexagonal region, which itself is bounded by the cubic phase. The hexagonal phase is characterized by the absence of iron (red curve in Figure 7b). The composition of the region in the center of the line-scan is determined by an average value of thin lamellae of the different phases, which are not resolved. The line-scan also demonstrates that the surrounding cubic phase accumulates the cations which are present in low concentration in the hexagonal phase (Fe) and BCO-type

phase (Sr). Corresponding composition gradients are observed in the cubic phase in the regions inside the black lines in Figure 7b. These regions are also depleted of Co cations which are enriched in the hexagonal and BCO-type phase. Figure 7d shows an EDXS line-scan with high spatial resolution inside a plate-like region. While clear identification of the phases in the thin lamellae is still not possible, the strong oscillation of the Co-concentration and the overall low Fe-content indicates that the lamellae consist of the cubic, hexagonal, and the BCO-type phases.



DISCUSSION Structural and chemical analyses of BSCF samples annealed at temperatures between 700 and 1000 °C for 100 h have revealed a complex microstructure with several secondary phases. In addition to the cubic BSCF phase and CoO precipitates which are present in all samples, the well-known hexagonal phase was found after annealing at 800 and 700 °C. In addition, regions with plate-like morphology are formed at temperatures between 700 and 900 °C. Focusing on the regions with plate-like morphology, we observe a considerable variation of the crystal structures and composition in plate-like regions. They consist of thin lamellae of the cubic and hexagonal phase. In addition, we have identified the BCO-type phase in the plate-like regions. This phase contains planes with CdI2-type structure with a different A/B-ratio violating the ABO3 perovskite stoichiometry. The overall chemical composition of the plate-like regions is determined by the volume fractions of the cubic, hexagonal and BCO-type phases. The BCO-type phase is expected to form the main portion of the plate-like regions at 900 °C because the formation of the hexagonal phase was shown to be confined to temperatures below 840 °C. This explains the strong Sr depletion in the plate-like regions reported for samples annealed above 840 °C13 and the different microstructure of the plates-like regions 569

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

Figure 6. Projected structure of the BCO-type phase along the [21̅1̅0]-direction containing a periodic arrangement of two (111)planes with cubic stacking and three planes consisting of a plane with CdI2-structure and two interface planes. The cubic planes are separated from the new stacking sequence by black dashed lines.

Moreover, Co-cations from the vicinity of the hexagonal phase diffuse in and occupy the vacancies left by the Fe-cations causing Co-depletion in the vicinity of the hexagonal phase. Because of the required cation demixing, the hexagonal phase forms at a slow rate in BSCF. Mueller et al.14 as well as Niedrig et al.9 reported a t90-time of about 1120 and 1090 h for the formation of the hexagonal phase. The increased B-site valence state also causes a reduction of the oxygen deficit d which results in the reduction of the oxygen conductivity. Plate-like regions containing thin lamellae with cubic, hexagonal, and BCO-type phase occur in significant volume fractions in BSCF after annealing at temperatures between 700 and 900 °C. As plate-like regions of cubic and hexagonal lamellae do not induce additional reflections apart from the reflections of the cubic and hexagonal bulk phase in XRD, the influence of these regions on the degradation is not well investigated yet. They form at a fast rate compared to the bulk hexagonal phase.13 This suggests that the formation of platelike regions influences the early stages of degradation where a faster degradation was reported.7,29 In long-term conductivity measurements with reduced time resolution the influence of the plate-like regions may only be detected as an overall lower performance. As the plate-like regions contain lamellae with hexagonal stacking and correspondingly reduced oxygen conductivity, the plate-like regions can be considered as barriers for oxygen conduction. Furthermore, according to this work and other studies27 the plate-like regions are preferentially formed at grain boundaries which are believed to contribute significantly to oxygen diffusion.27 Plate-like regions therefore constitute a diffusion barrier for grainboundary- as well as bulk-diffusion. The influence of the BCO-type phase on the oxygen conductivity is expected to be degrading as well, because the Co-valence state is larger (+2.12 up to +3.20)28 compared to cubic BSCF which lowers the oxygen deficit. Moreover, the

Figure 5. HAADF STEM micrographs and projected structure models of the sample after annealing at 800 °C along the (a,b,c) [21̅1̅0]-zone axis and (d,e,f) [101̅0]-zone axis of the BCO-type phase. A-site cations are marked in green, B-site cations in blue, and oxygen in red in (c,f).

at 900 °C (few large plate-like regions) and below 840 °C (many small plate-like regions). At temperatures below the formation threshold of the hexagonal phase a mixture of all three phases is expected in plate-like regions and observed in this study. The influence of the plate-like regions on the degradation of the ionic conductivity may differ significantly from the hexagonal bulk phase which is discussed in the following. The influence of the hexagonal phase on the degradation was already discussed in earlier studies.10,11,14,15,38 However, the different chemical composition of the hexagonal phase compared to the cubic phase did not attract any particular attention. The two-dimensional projection of the crystal structure in Figure 4b shows that the B-sites in the facesharing octahedrons of hexagonal phase are more densely packed compared to the cubic BSCF phase. Therefore the Bsite cations have to be smaller in the hexagonal phase to fit into the structure. As Co is able to decrease its size by changing its valence state from 2+ to 3+,15 Co remains in the hexagonal phase. In contrast, as demonstrated by Harvey et al.39 and Arnold et al.,40 Fe does not form smaller ions in BSCF and is therefore expelled from the hexagonal phase. This explanation is consistent with the EDXS-linescan (Figure 7b) where Fecation excess is observed in the vicinity of the hexagonal phase. 570

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

Table 2. Concentrations of Ba, Sr, Co and Fe in atom % for the Different Observed BSCF Phases Determined by EDXS Compared to the Nominal (Intended) Composition of Cubic BSCF (first line)a

a

phase

Ba [atom %]

Sr [atom %]

Co [atom %]

Fe [atom %]

Ba0.5Sr0.5Co0.8Fe0.2O3−d cubic phase hexagonal phase BCO-type phase

25 26 ± 1 30 ± 1 21.5 ± 3

25 25.5 ± 1.5 27 ± 1.5 3±1

40 40 ± 1.5 42 ± 1 67.5 ± 2.5

10 8.5 ± 1 1 ± 0.5 8±1

The errors are the standard deviations of the average values of up to 200 spectra recorded for the same phase.

Figure 7. (a,c) HAADF STEM images and (b,d) corresponding EDXS line-scans at positions indicated by dashed red arrows in the HAADF STEM images. Regions of the cubic phase inside the black lines in (b) contain composition gradients.

CdI2-type planes consist of densely packed cations which narrow the oxygen diffusion paths through the material perpendicular to the CdI2-type planes. However, for BCOtype lamellae with numerous planes with cubic stacking (n > 2), oxygen diffusion parallel to the (111)-planes might be unaffected. Overall, we expect that the BCO-type phase contributes to the degradation of the ionic conductivity in BSCF because it is characterized by a lower concentration of oxygen vacancies and introduces a diffusion barrier.

and EDXS were applied to gain an overview of the formation of secondary phases apart from the cubic BSCF phase, which are correlated with the degradation of the ionic conductivity. After annealing at 1000 °C only the cubic BSCF phase and CoO precipitates are found. Large plate-like regions are formed at grain boundaries after 100 h of annealing at 900 °C. At 800 °C, the hexagonal phase is present, typically at triple points of the grain boundaries. Furthermore, smaller plate-like regions occur at grain boundaries and CoO precipitates with a higher density compared to 900 °C. At 700 °C the growth of plate-like regions is even more pronounced. The hexagonal phase often decorates grain boundaries completely. It also often surrounds plate-like regions.



SUMMARY Ba0.5Sr0.5Co0.8Fe0.2O3−d was annealed at temperatures between 700 and 1000 °C for 100 h. Electron microscopic techniques 571

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

dense polycrystalline BSCF bulk samples. Furthermore we thank Prof. A. Feldhoff (Institut für Physikalische Chemie und Elektrochemie, Universität Hannover, Germany) for fruitful discussion.

A detailed investigation reveals the lamellar substructure of the plate-like regions. Beside lamellae of cubic and hexagonal phase an additional phase contributing to the plate-like regions could be assigned to the Ban+1ConO3n+3(Co8O8) (BCO)-type barium cobaltites with n ≥ 2. The hexagonal phase is strongly depleted in Fe because the cation radius of Fe is too large for the B-sites of the hexagonal phase. The required interdiffusion of iron and cobalt in hexagonal regions therefore limits the formation rate of the hexagonal phase. The composition of the BCO-type phase is Co-rich and varies depending on the number of planes with cubic stacking order n. In addition Fe occupies also B-sites of the BCO-structure which can accommodate larger cations. Plate-like regions with lamellae of all three discussed phases show an average chemical composition, depending on the volume fraction of each phase. All secondary phases are discussed to have a negative impact on oxygen-ion conduction in BSCF. Especially the thin lamellae of the hexagonal phase in plate-like regions may have been underestimated severely in earlier studies, as they cannot be distinguished from a mixture of cubic phase and bulk hexagonal phase by XRD. Moreover, the formation rate of the plate-like regions is fast in comparison to the bulk hexagonal phase. This sheds new light on the two-step degradation in electrical longterm conductivity measurements at 800 °C previously shown in ref 9, Figure 10, where a fast decrease was observed during the first 100 h, now attributed to the formation of the plate-like regions. At 700 and 800 °C, grain boundaries are often decorated by plate-like regions and the hexagonal phase. Since significant oxygen transport can be expected to occur along grain boundaries, secondary phase formation at grain boundaries is expected to introduce additional barriers for oxygen transport.





ABBREVIATIONS BCO, Ban+1ConO3n+3(Co8O8); BSCF, Ba0.5Sr0.5Co0.8Fe0.2O3−d; EDXS, energy dispersive X-ray spectroscopy; HAADF, high angle annular dark field; SEM, scanning electron microscopy; STEM, scanning transmission electron microscopy; TEM, transmission electron microscopy; XRD, X-ray diffractometry



REFERENCES

(1) Bouwmeester, H. J. M.; Burggraaf, A. J. The CRC handbook of solid state electrochemistry; CRC Press: 1997; Chapter 14, p 481−553. (2) Czyperek, M.; Zapp, P.; Bouwmeester, H. J. M.; Modigell, M.; Ebert, K.; Voigt, I.; Meulenberg, W. A.; Singheiser, L.; Stöver, D. J. Membr. Sci. 2010, 359 (1−2), 149−159. (3) Hashim, S. M.; Mohamed, A. R.; Bhatia, S. Adv. Colloid Interface Sci. 2010, 160 (1−2), 88−100. (4) Leo, A.; Liu, S.; da Costa, J. C. D. Int. J. Greenhouse Gas Control 2009, 3, 357−367. (5) Liu, Y.; Tan, X.; Li, K. Catal. Rev.: Sci. Eng. 2006, 48, 145−198. (6) Sunarso, J.; Baumann, S.; Serra, J. M.; Meulenberg, W. A.; Liu, S.; Lin, Y. S.; da Costa, J. C. D. J. Membr. Sci. 2008, 320 (1−2), 13−41. (7) Shao, Z.; Yang, W.; Cong, Y.; Dong, H.; Tong, J.; Xiong, G. J. Membr. Sci. 2000, 172, 177−188. (8) Shao, Z.; Haile, S. M. Nature 2004, 431 (7005), 170−173. (9) Niedrig, C.; Taufall, S.; Burriel, M.; Menesklou, W.; Wagner, S. F.; Baumann, S.; Ivers-Tiffée, E. Solid State Ionics 2011, 197 (1), 25− 31. (10) Svarcová, S.; Wiik, K.; Tolchard, J.; Bouwmeester, H. J. M.; Grande, T. Solid State Ionics 2008, 178 (35−36), 1787−1791. (11) Arnold, M.; Gesing, T. M.; Martynczuk, J.; Feldhoff, A. Chem. Mater. 2008, 20 (18), 5851−5858. (12) Müller, P.; Dieterle, L.; Müller, E.; Störmer, H.; Gerthsen, D.; Niedrig, C.; Taufall, S.; Wagner, S. F.; Ivers-Tiffee, E. ECS Trans. 2010, 28 (11), 309−314. (13) Müller, P.; Störmer, H.; Dieterle, L.; Niedrig, C.; Ivers-Tiffée, E.; Gerthsen, D. Solid State Ionics 2012, 206, 57−66. (14) Mueller, D. N.; De Souza, R. A.; Weirich, T. E.; Roehrens, D.; Mayer, J.; Martin, M. Phys. Chem. Chem. Phys. 2010, 12, 10320− 10328. (15) Efimov, K.; Xu, Q.; Feldhoff, A. Chem. Mater. 2010, 22 (21), 5866−5875. (16) Mogensen, M.; Lybye, D.; Bonanos, N.; Hendriksen, P. V.; Poulsen, F. W. Solid State Ionics 2004, 174 (1−4), 279−286. (17) Baumann, S.; Serra, J. M.; Lobera, M. P.; Escolástico, S.; Schulze-Küppers, F.; Meulenberg, W. A. J. Membr. Sci. 2011, 377 (1− 2), 198−205. (18) Yakovlev, S.; Yoo, C.-Y.; Fang, S.; Bouwmeester, H. J. M. Appl. Phys. Lett. 2010, 96 (25), 254101. (19) Haworth, P.; Smart, S.; Glasscock, J.; da Costa, J. C. D. Sep. Purif. Technol. 2011, 81 (1), 88−93. (20) Haworth, P.; Smart, S.; Glasscock, J.; da Costa, J. C. D. Sep. Purif. Technol. 2012, 94 (0), 16−22. (21) Wang, H.; Tablet, C.; Feldhoff, A.; Caro, J. Adv. Mater. 2005, 17, 1785−1788. (22) Feldhoff, A.; Martynczuk, J.; Wang, H. Prog. Solid State Chem. 2007, 35 (2−4), 339−353. (23) Feldhoff, A.; Martynczuk, J.; Arnold, M.; Myndyk, M.; Bergmann, I.; Sepelák, V.; Gruner, W.; Vogt, U.; Hähnel, A.; Woltersdorf, J. J. Solid State Chem. 2009, 182 (11), 2961−2971. (24) Martynczuk, J.; Efimov, K.; Feldhoff, A. J. Membr. Sci. 2009, 344, 62−70. (25) Martynczuk, J.; Liang, F.; Arnold, M.; Sepelák, V.; Feldhoff, A. Chem. Mater. 2009, 21 (8), 1586−1594.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address ∥

Institut für Hochfrequenztechnik, Technische Universität Braunschweig, Schleinitzstrasse 22, 38106 Braunschweig, Germany. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work has been performed within the project F.2 of the DFG Research Center for Functional Nanostructures (CFN). It has been further supported by a grant from the Ministry of Science, Research and the Arts of Baden-Württemberg (Az: 7713.14-300). Financial support from the Helmholtz Association of German Research Centres through the portfolio topic MEM-BRAIN and the German Federal Ministry of Economics and Technology (BMWi Grant 0327803F) is also gratefully acknowledged. The authors thank the Fraunhofer Institute for Ceramic Technologies and Systems (IKTS) and Hermsdorf/ Germany for supply with commercially available BSCF powder, as well as Dr. S. Baumann and Mr. S. Heinz (Institut für Energie- und Klimaforschung (IEK-1), Forschungszentrum Jülich/Germany) for technical support in the preparation of 572

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573

Chemistry of Materials

Article

(26) Zhao, L.; He, B.; Zhang, X.; Peng, R.; Meng, G.; Liu, X. J. Power Sources 2010, 195 (7), 1859−1861. (27) Liang, F.; Jiang, H.; Luo, H.; Caro, J.; Feldhoff, A. Chem. Mater. 2011, 23 (21), 4765−4772. (28) Sun, J.; Yang, M.; Li, G.; Yang, T.; Liao, F.; Wang, Y.; Xiong, M.; Lin, J. Inorg. Chem. 2006, 45 (23), 9151−9153. (29) Arnold, M.; Wang, H.; Feldhoff, A. J. Membr. Sci. 2007, 293, 44−52. (30) Crewe, A. V. Science 1966, 154 (3750), 729−738. (31) Crewe, A. V.; Wall, J.; Welter, L. M. J. Appl. Phys. 1968, 39 (13), 5861−5868. (32) Wall, J.; Langmore, J.; Isaacson, M.; Crewe, A. V. Proc. Natl. Acad. Sci. 1974, 71 (1), 1−5. (33) Cliff, G.; Lorimer, G. W. J. Microsc. 1975, 103 (2), 203−207. (34) Horita, Z.; Sano, T.; Nemoto, M. J. Microsc. 1986, 143 (3), 215−231. (35) Horita, Z.; Sano, T.; Nemoto, M. Ultramicroscopy 1987, 21 (3), 271−276. (36) Chantler, C. T. J. Phys. Chem. Ref. Data 1995, 24 (1), 71−643. (37) David, R.; Pautrat, A.; Kabbour, H.; Sturza, M.; Curelea, S.; André, G.; Pelloquin, D.; Mentré, O. Chem. Mater. 2011, 23 (23), 5191−5199. (38) Rutkowski, B.; Malzbender, J.; Steinbrech, R. W.; Beck, T.; Bouwmeester, H. J. M. J. Membr. Sci. 2011, 381 (1−2), 221−225. (39) Harvey, A. S.; Litterst, F. J.; Yang, Z.; Rupp, J. L. M.; Infortuna, A.; Gauckler, L. J. Phys. Chem. Chem. Phys. 2009, 11 (17), 3090−3098. (40) Arnold, M.; Xu, Q.; Tichelaar, F. D.; Feldhoff, A. Chem. Mater. 2009, 21 (4), 635−640.

573

dx.doi.org/10.1021/cm303670m | Chem. Mater. 2013, 25, 564−573