(Fe,Ni)O3−δ perovskites - CiteSeerX

Solid State Ionics 179 (2008) 2170–2180

Contents lists available at ScienceDirect

Solid State Ionics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s s i

Mixed conductivity, Mössbauer spectra and thermal expansion of (La,Sr)(Fe,Ni)O3 − δ perovskites E.V. Tsipis a,b, E.A. Kiselev a,c, V.A. Kolotygin a, J.C. Waerenborgh b, V.A. Cherepanov c, V.V. Kharton a,⁎ a b c

Department of Ceramics and Glass Engineering, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal Chemistry Department, Instituto Tecnológico e Nuclear, CFMC-UL, EN 10, 2686-953 Sacavém, Portugal Chemistry Department, Ural State University, Lenin Av. 51, Ekaterinburg 620083, Russia

a r t i c l e

i n f o

Article history: Received 6 May 2008 Received in revised form 26 July 2008 Accepted 27 July 2008 Keywords: Perovskites Mixed conductors Ferrites–nickelates Oxygen permeability Phase relationships p-Type electronic conductivity Seebeck coefficient Dilatometry

a b s t r a c t The incorporation of Sr2+ in perovskite-type La1 − ySryFe1 − xNixO3 − δ increases oxygen deficiency and ionic transport at elevated temperatures, but leads to a lower thermodynamic stability as reflected by narrowing the solid-solution domain at 1373 K down to x ≈ 0.25 at y = 0.10 and x ≈ 0.12 at y = 0.20 in air. The average thermal expansion coefficients of La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.4, y = 0.1–0.2) ceramics vary in the ranges (12.4–13.4) × 10− 6 K− 1 at 700–1150 K and (14.2–18.0) × 10− 6 K− 1 at 1150–1370 K, rising with strontium and nickel contents. These additives lower also the temperature of orthorhombic → rhombohedral phase transition visible in the dilatometric curves, promote hole delocalization, increase total conductivity and decrease Seebeck coefficient studied in the oxygen partial pressure range from 10− 5 to 0.4 atm at 973–1223 K. The steady-state oxygen permeation fluxes through single-phase La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.2, y = 0.1–0.2) membranes, with an activation energy of 206–235 kJ/mol, are determined by both surface-exchange and bulk ionic conduction limited by the oxygen-vacancy concentration. Doping with nickel has a weak negative effect on the oxygen transport, probably due to defect cluster formation involving the vacancies and Ni2+. On the contrary, at low temperatures when (La,Sr)(Fe,Ni)O3 − δ becomes almost oxygen-stoichiometric in air, the Mössbauer spectroscopy and thermogravimetric analysis showed that the average oxidation state of nickel is higher than +3, similar to iron cations. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Mixed-conducting materials derived from the Ruddlesden–Popper (RP) nickelates, (LaO)(LaNiO3)n, are of considerable interest for hightemperature electrochemical applications, such as cathodes of solid oxide fuel cells (SOFCs) and ceramic membranes for oxygen separation and natural gas conversion [1–6]. As a general rule, increasing the number of perovskite layers (n) in the RP nickelates enhances their electrochemical activity and partial oxygen-ionic and electronic conductivities, but lowers thermodynamic stability [1,2,6,7]. At atmospheric oxygen pressure, perovskite-like LaNiO3 − δ is only stable at temperatures below 1230–1270 K; further heating leads to decomposition into La2NiO4 + δ with K2NiF4-type structure and NiO, via the separation of RP-type La4Ni3O10 − δ and La3Ni2O7 − δ phases at intermediate stages [6–8]. These transformations limiting practical use of LaNiO3-based materials can be partly suppressed by extensive doping into the nickel sublattice. The solid solutions La(Ni,M)O3 − δ (M = Ti, Cr, Mn, Fe, Co, Ga, Zr) containing 50–60% Ni cations retain perovskite-like structure at elevated temperatures, although the substitution decreases hole mobility and conductivity; opposite effects including ⁎ Corresponding author. Tel.: +351 234 370263; fax: +351 234 425300. E-mail address: [email protected] (V.V. Kharton). 0167-2738/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2008.07.017

an improvement of the p-type electronic conduction and perovskite lattice destabilization, are observed on acceptor-type doping [6,8–15]. An attractive combination of the electronic transport properties, moderate thermal expansion and phase stability in air is known for LaFe1 − xNixO3 − δ (x = 0.4–0.6) and La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.2; y = 0.1–0.3) series, which are widely considered for the SOFC applications [8–10,13,16–19]. In these systems, the average linear thermal expansion coefficients (TECs) at atmospheric p(O2) were reported to vary in the ranges (9.7–13.4) × 10− 6 K− 1 at 300–1270 K for the strontium-free perovskites and (11.7–12.8) × 10− 6 K− 1 at 300–1120 K for the Sr-containing compositions [8,9,13,16,18]. Such TECs are low enough to provide thermomechanical compatibility with solid oxide electrolyte ceramics in the intermediate-temperature range. Another situation can be however expected on heating up to the temperatures necessary for electrode sintering, when progressive oxygen losses and the resultant chemical expansion typical for ferrite-based phases [20,21] may induce additional mechanical stresses. Moreover, literature data on thermal expansion of the ferrite–nickelate ceramics [8,9,13,16,18,21] display a significant scatter, which may partly originate from kinetically stagnated oxygen-nonstoichiometry variations, and require careful re-examination. Also, information on the oxygen nonstoichiometry and ionic transport in (La,Sr)(Ni,Fe)O3 − δ is still scarce.

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

2171

Continuing our research on La(Ni,Fe)O3-based systems [9,11,12,17], this work focuses on the thermal expansion, oxygen permeability and electronic conduction mechanisms in perovskite-type La1 − ySryFe1 − xNixO3 − δ. Particular attention was also centered on the phase relations at elevated temperatures, relevant for the development of SOFC materials stable in the whole range of the fuel cell fabrication and operation conditions. In order to assess the states of nickel and iron cations at low temperatures, the Mössbauer spectroscopy and thermogravimetric analysis were used. 2. Experimental Submicron powders of La1 − ySryFe1 − xNixO3 − δ (x = 0–0.4, y = 0.1–0.4) were synthesized by the glycine–nitrate process [22], annealed in air at 1273 K for 2 h to remove organic residues, ball-milled and then pressed into discs at 150–200 MPa. The gas-tight ceramic samples with relative density ≥95% were sintered in air at 1643 K for 10 h, followed by slow (1–2 K/min) cooling down to room temperature in order to achieve equilibrium oxygen content. The scanning electron microscopy combined with energy dispersive spectroscopy (SEM/EDS, Hitachi S4100) confirmed a very minor closed porosity and an absence of grainboundary anomalies in the sintered ceramics. Two typical SEM micrographs are presented in Fig. 1. For all compositions, the average grain size varied in the range 1–4 μm; no traces of liquid phase formation were observed. The powdered samples used for X-ray diffraction (XRD), Mössbauer spectroscopy and thermogravimetric analysis (TGA), were obtained by grinding of sintered ceramics. Additional series of ceramic samples used for the studies of phase composition at elevated temperatures, were annealed at 1173–1373 K in various atmospheres, quenched and reground. The phase relationships discussed below were also verified using two alternative synthesis methods, namely the standard solid-state and citrate routes, and subsequent XRD analysis. In

Fig. 2. TGA curve collected on reduction of La0.9Sr0.1Fe0.8Ni0.2O3 − δ at 1373 K, illustrating the determination of total oxygen content. Inset shows one example of the relative weight changes of La0.9Sr0.1Fe0.8Ni0.2O3 − δ in the course of isothermal TGA steps in air.

all cases, phase composition of (La,Sr)(Fe,Ni)O3 − δ was unaffected by the processing method. The crystal structure was refined by the full-profile Rietveld method [23] using room-temperature XRD patterns (Rigaku D/ Max-B diffractometer, CuKα radiation, 2θ = 10–100°, step 0.02°, 2–10 s/ step). The cation composition of selected samples was confirmed by the inductively-coupled plasma (ICP) spectroscopic analysis using a Jobin Yvon instrument, model JY 70 plus. In all cases, the deviations from nominal cation composition were lower than 0.1 at.%, comparable to the experimental error. The concentration of impurity cations (Co, Ca, Ba, Mn, Al, Si) detected by the ICP spectroscopy was lower than 0.02 at.%. The total oxygen content at atmospheric oxygen pressure was calculated from the TGA data collected using a Setaram SetSys 16/18 instrument. The TGA procedure included temperature cycling in flowing dry air in the range 298–1373 K with equilibration at each temperature during 1– 3 h, and subsequent isothermal reduction in a dry 10% H2–90% N2 flow at 1373–1423 K; these measurement regimes are illustrated by Fig. 2. The transmission Mössbauer spectroscopy analysis was performed on a conventional constant-acceleration spectrometer with a 25 mCi 57Co (Rh) source; the velocity scale was calibrated to pure α-Fe. The lowtemperature spectra were collected using a JANIS SVT-400 bath cryostat where the sample was immersed in liquid He. The Mössbauer spectra were fitted to Lorentzian lines by a nonlinear least-squares method and to distributions of magnetic splittings according to the histogram method [24,25]. General characterization of the ceramic materials included also dilatometric analysis in air (Linseis L75V/1250, heating/ cooling rate of 3 K/min), and measurements of total electrical conductivity (4-probe DC), Seebeck coefficient and steady-state oxygen permeation fluxes as function of temperature and oxygen partial pressure. All data on oxygen permeability presented in this work correspond to the membrane feed-side oxygen partial pressure (p2) maintained at 0.21 atm (atmospheric air); the permeate-side oxygen partial pressure (p1) varied in the range 0.01–0.20 atm. Detailed description of the equipment and techniques used for materials characterization was published earlier (see [9,17,20,24,26–28] and references cited). 3. Results and discussion 3.1. Phase relationships

Fig. 1. SEM micrographs of La0.9Sr0.1Fe0.9Ni0.1O3 − δ (A) and La0.8Sr0.2Fe0.9Ni0.1O3 − δ (B) ceramics.

In the strontium-containing La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.4, y = 0.1–0.4) series, formation of single perovskite phases was only

2172

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

Fig. 3. XRD patterns of La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.4, y = 0.1–0.2), sintered at 1643 K and equilibrated with atmospheric oxygen at low temperatures. The calculated patterns are plotted on the experimental data points (empty circles) as solid lines. The difference lines are shown below each XRD pattern. The calculated peak positions at the bottom are given for orthorhombic La0.8Sr0.2Fe0.9Ni0.1O3 − δ. Black triangles show the La2NiO4based phase impurity peaks.

observed for La0.9Sr0.1Fe0.9Ni0.1O3 − δ, La0.9Sr0.1Fe0.8Ni0.2O3 − δ and La0.8Sr0.2Fe0.9Ni0.1O3 − δ where the structure at room temperature was identified as orthorhombic (space group Pbnm). Fig. 3 presents XRD

patterns of La0.9Sr0.1Fe1 − xNixO3 − δ (x =0.1–0.4) and La0.8Sr0.2Fe0.9Ni0.1O3 − δ ceramics equilibrated with atmospheric oxygen at low temperatures, including the calculated and difference patterns according to the structure refinement results. The space groups, lattice parameters and relative amount of secondary phases determined by the Rietveld refinement are listed in Table 1; other refinement results will be summarized in a separate publication. For x = 0.3–0.4, the orthorhombic phase was found to coexist with a rhombohedrally distorted perovskite (S.G. R3̄ c) and minor amounts of tetragonal La2NiO4based solid solution (S.G. I4/mmm); the peaks of NiO segregated due to the perovskite decomposition were essentially indistinguishable in the XRD patterns and were thus excluded from the refinement procedure. The amounts of K2NiF4-type phase and NiO increase with further strontium and nickel additions, as expected. Attempts to sinter La0.7Sr0.3Fe0.9Ni0.1O3 − δ and La0.8Sr0.2Fe0.8Ni0.2O3 − δ ceramics with major perovskite phase, and even to synthesize single-phase powders at 1170–1370 K and atmospheric oxygen pressures, failed. Consequently, the solubility of nickel in La1 − ySryFe1 − xNixO3 − δ at temperatures above 1300 K was estimated as x ≈ 0.25 at y = 0.10 and x b 0.15 at y = 0.20. The results are in a good agreement with the literature data [6,9] showing that Ni doping in La(Sr)FeO3 − δ gradually increases the perovskite lattice symmetry and decreases phase stability. Very similar conclusions can also be drawn considering the fragment of preliminary phase diagram of the pseudo-quaternary system LaFeO3 − δ–SrFeO3 − δ–“SrNiO3”–“LaNiO3” at atmospheric oxygen pressure and 1373 K (Fig. 4), which was constructed on the basis of XRD analysis of quenched La1 − ySryFe1 − xNixO3 − δ samples in combination with previous results [9,11,20,27]. The maximum attention in this work was hence focused on single-phase materials, namely La0.9Sr0.1Fe0.9Ni0.1O3 − δ, La0.9Sr0.1Fe0.8Ni0.2O3 − δ and La0.8Sr0.2Fe0.9Ni0.1O3 − δ, covering almost the whole solid-solution domain (Fig. 4). Due to the lattice-destabilizing effect of Sr doping [6,9,15], these perovskites exist however in a relatively narrow range of the oxygen partial pressure. In particular, annealing of the ceramic samples at 1223 K in flowing Ar, p(O2) ≈ 10− 5 atm, resulted in a slow separation of the La2NiO4-based phase. As an example, in the case of La0.8Sr0.2Fe0.9Ni0.1O3 − δ ceramics annealed in Ar flow for 24 h, the amount of K2NiF4-type impurity estimated by XRD analysis was approximately 1.5 mol%; the remaining perovskite-type solid solution had an orthorhombic structure with a = 5.5453 Å, b = 5.5113 Å, and c = 7.8026 Å. Exact determination of the low-p(O2) stability

Table 1 Properties of (La,Sr)(Fe,Ni)O3 ceramics at atmospheric oxygen pressure Composition

La0.9Sr0.1Fe0.9Ni0.1O3

La0.9Sr0.1Fe0.8Ni0.2O3

La0.9Sr0.1Fe0.7Ni0.3O3

La0.9Sr0.1Fe0.6Ni0.4O3

La0.8Sr0.2Fe0.9Ni0.1O3

Major phase and unit-cell parameters Pbnm a = 5.5520 Å b = 5.5230 Å c = 7.8129 Å Pbnm a = 5.5487 Å b = 5.5106 Å c = 7.7835 Å R3̄c a = 5.4956 Å c = 13.5253 Å R3̄c a = 5.5183 Å c = 13.3175 Å Pbnm a = 5.5455 Å b = 5.5082 Å c = 7.7965 Å

Secondary phases

Relative density, %

Average linear thermal expansion coefficients T, K

TEC × 106, K− 1

Activation energy for the total conductivity T, K

Ea, kJ/mol

–

97

370–650 750–1150 1150–1300

10.49 ± 0.01 12.45 ± 0.01 14.18 ± 0.01

440–720

29 ± 1

–

95

370–480 580–1150 1150–1370

9.63 ± 0.01 12.74 ± 0.01 15.34 ± 0.01

430–780

23.2 ± 0.5

Pbnm (19%) I4/mmm (3%) Pbnm (15%) I4/mmm (5%) –

97

420–700 700–1150 1150–1370

11.92 ± 0.01 13.02 ± 0.01 16.13 ± 0.02

660–1040

13.4 ± 0.1

97

420–700 700–1150 1150–1370

12.17 ± 0.01 13.06 ± 0.01 16.32 ± 0.02

460–1080

9.67 ± 0.03

99

440–700 700–1150 1150–1370

11.99 ± 0.01 13.38 ± 0.01 18.00 ± 0.02

430–730

16.1 ± 0.6

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

2173

Fig. 4. Schematic diagram showing phase relationships in the La1 − ySryFe1 − xNixO3 − δ system at 1373 K in air. The shaded area corresponds to the narrow perovskite-type solid-solution domain studied in this work. O, R, C and T correspond to the orthorhombic, rhombohedral and cubic perovskites and tetragonal K2NiF4-type phase, respectively.

boundaries of (La,Sr)(Ni,Fe)O3 − δ by TGA or coulometric titration techniques was impossible owing to the low concentrations of nickel and, thus, small amounts of the segregated secondary phases. Nevertheless, the observed behavior made it necessary to limit the p(O2) range studied in this work, in most cases at ~ 10− 5 atm. Another necessary comment is related to the first-order phase transition at atmospheric oxygen pressure, accompanied by a drastic volume contraction visible in the dilatometric curves at temperatures below 750 K (Figs. 5 and 6). This behavior is well reproducible on heating and cooling (Fig. 6) and was ascribed to the transformation of orthorhombic lattice into the rhombohedral polymorph, typical for numerous Ni- and Fe-containing perovskites such as La0.95Ni0.5Ti0.5O3 − δ [28] and La1 − ySryFeO3 − δ [29]. Note that in the latter case, increasing y was reported to lower the transition temperature [29]. Analogously, strontium and nickel additions in La1 − ySryFe1 − xNixO3 − δ shift the phase transformation temperatures from ~700 K (x = 0.1, y = 0.1) down to 523 K (x = 0.2, y = 0.1) and 393 K (x = 0.1, y = 0.2).

Fig. 5. Dilatometric curves of La1 − ySryFe1 − xNixO3 − δ ceramics in air, collected on heating (3 K/min).

Fig. 6. Temperature dependencies of the linear thermal expansion coefficients of La1 − ySryFe1 − xNixO3 − δ ceramics, measured on heating and cooling in air (3 K/min). The minima are indicative of first-order phase transition.

3.2. Mössbauer spectra, oxygen nonstoichiometry and thermal expansion The room-temperature Mössbauer spectra of La1 − ySryFe1 − xNixO3 − δ equilibrated with atmospheric oxygen (Fig. 7) consist of the sextets with

Fig. 7. Room-temperature Mössbauer spectra of La1 − ySryFe1 − xNixO3 − δ equilibrated with atmospheric oxygen at low temperatures.

2174

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

Fig. 8. Mössbauer spectra of La1 − ySryFe1 − xNixO3 − δ at 4.2 K. The curves plotted on experimental data points are the sum of two distributions of magnetic splittings corresponding to octahedrally-coordinated Fe3+ and Fe5+, shown slightly shifted for clarity. The probability distributions (P) of the magnetic hyperfine fields (Bhf) are presented near the corresponding spectra.

broad peaks and relatively small splittings, indicating that magnetic ordering in these perovskites occurs at temperatures not far above 295 K and, probably, accompanies the structural transition. The relaxation broadening ordinary for perovskite-related ferrites is caused by spin fluctuations and dominates over possible distribution effects [24,27,30– 32]. Therefore, precise analysis of the iron oxidation states becomes only possible for the spectra collected far below the magnetic ordering temperature. The 4.2 K spectrum of La0.8Sr0.2Fe0.9Ni0.1O3 − δ (Fig. 8) consists of at least two magnetic sextets where the peaks of lower intensity appear at the velocities close to the corresponding values in La (Sr)Fe(Al)O3 [27,30,32]. In the case of La0.9Sr0.1Fe1 − xNixO3 − δ where this signal is small, the fitting quality was also enhanced considering two components similar to La0.8Sr0.2Fe0.9Ni0.1O3 − δ. In addition, irrespective of the improved resolution achieved at 4.2 K, the best fitting results were obtained when analyzing narrow distributions of the magnetic hyperfine fields (Bhf) associated, most likely, with the disorder arising from different nearest-neighbor configurations of iron cations. The estimated isomer shifts (IS) and Bhf values (Table 2) are typical for octahedrally-coordinated Fe3+ and Fe5+ in perovskite-type oxides where low-temperature Fe4+ diproportionation takes place [27,30,32]. Although the shape of the lower-field distribution for La0.8Sr0.2Fe0.9-

Ni0.1O3 − δ might suggest the presence of a third minor contribution characteristic of localized Fe4+, which may coexist with the itinerant states [27,31] and has larger IS and Bhf values compared to Fe5+, detailed analysis of the relevant phenomena will be presented in a separate publication. Whatever the diproportionation mechanisms at low temperatures, the Fe3+ distributions display very small fractions of the cations with low IS/Bhf values (Fig. 8), thus revealing that pentacoordinated iron and oxygen vacancies are almost absent. In agreement with the Mössbauer spectroscopy, thermogravimetric analysis showed that the oxygen deficiency (δ) of La1 − ySryFe1 − xNixO3 − δ at room temperature is negligibly low, ≤0.002 atoms per formula unit (Fig. 9). The incorporation of Sr2+ is, therefore, charge-compensated by increasing oxidation states of the transition metal cations. The approximate fractions of tetravalent cations, estimated combining the Mössbauer spectroscopy and TGA data, are given in Table 2 which lists also the data on Ni-free (La,Sr)FeO3 for the sake of comparison. The average oxidation states of nickel are essentially similar for all singlephase compositions and vary in the narrow range +3.18 to +3.24. At y = 0.1, these values seem even slightly higher than those of iron cations, but the difference lies within the limits of experimental error. The error is primarily determined by possible hydration of the samples in the

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180 Table 2 Parameters estimated from the Mössbauer spectra, collected at 4.2 K, for (La,Sr)(Fe,Ni) O3 − δ perovskites equilibrated with atmospheric oxygen at low temperatures Composition

Cation IS, mm/s ε, mm/s Bhf, T I, %

½Ni4þ  ½Btotal

½Fe4þ  ½Btotal

½B4þ  ½Btotal

La0.9Sr0.1Fe0.9Ni0.1O3

Fe3+ Fe5+ Fe3+ Fe5+ Fe3+ Fe5+ Fe3+ Fe5+ Fe3+ Fe5+

0.024

0.072

0.096

0.048

0.048

0.096

–

0.10

0.10

0.018

0.180

0.198

–

0.20

0.20

La0.9Sr0.1Fe0.8Ni0.2O3 La0.9Sr0.1FeO3 La0.8Sr0.2Fe0.9Ni0.1O3 La0.8Sr0.2FeO3

0.46 − 0.04 0.46 − 0.04 0.47 − 0.06 0.46 0.01 0.46 − 0.03

−0.06 −0.09 −0.04 −0.39 −0.07 0.09 −0.06 −0.06 −0.05 0.11

54.8 25.4 52.7 24.9 55.5 26.1 53.7 27.3 54.6 26.4

96 4 97 3 95 5 90 10 90 10

Notes: IS, Bhf, ε and I are the average isomer shift relative to metallic α-Fe at 295 K, average magnetic hyperfine field, quadrupole shift and relative area, respectively. Estimated standard deviations are b2% for I, b0.2 T for Bhf and b 0.02 mm/s for the other ½B4þ  parameters. ½B corresponds to the fraction of nominal B4+ cations per formula unit, total calculated combining the Mössbauer spectroscopy and TGA data. Data on (La,Sr)FeO3 − δ are presented for comparison.

2175

Furthermore, the oxidation state of cobalt cations in perovskite-type La0.5Ba0.5Fe1 − xCoxO3 − δ prepared in air is higher than that of iron at low temperatures [36]. Another situation is however expected on heating when the redox equilibria should shift towards the formation of Fe4+ and Ni3+ and, then, Ni2+ [13,16,33]. Increasing temperature above 700–900 K leads to progressive oxygen losses from (La,Sr)(Ni,Fe)O3 − δ lattice (Fig. 9) and to decreasing oxidation states of the transition metal cations. As expected, this effect becomes more pronounced with strontium and nickel additions. The resultant chemical contribution to the lattice expansion causes nonlinearities in the dilatometric curves at 1000–1150 K when oxygen deficiency increases up to 0.01–0.02 atoms per formula unit (Figs. 5 and 9). At 1150–1370 K, the apparent TECs of La1 − ySryFe1 − xNixO3 − δ ceramics in air are as high as (14.2–18.0) × 10− 6 K− 1, increasing with the dopant content (Table 1). These values are considerably larger compared to the TECs of solid oxide electrolytes, (10–13) × 10− 6 K− 1, which limits the range of potential applications of (La,Sr)(Ni,Fe)O3 − δ perovskites and makes it necessary to carefully optimize the electrode sintering protocols. 3.3. Total conductivity and Seebeck coefficient

course of their equilibration during slow cooling, which may lead to a slightly over-estimated oxygen content calculated from TGA data, and also by the statistical errors of the Mössbauer spectra analysis. The corresponding values are estimated as 1% and 2%, respectively. Taking into account these uncertainties, one may only conclude that the average oxidation states of nickel and iron cations at low temperatures, when the perovskite lattice of (La,Sr)(Ni,Fe)O3 − δ is oxygen-stoichiometric and the level of hole delocalization is significant, become substantially similar to each other. In other words, irrespective of the electron interaction mechanisms in these conditions, the sites occupied by iron and nickel are essentially indistinguishable from the energetic point of view. Notice that for both cations, the 4+ oxidation state is quite unusual and exists mainly in perovskite-like oxides. Whilst the formation of Fe4+ is welldocumented in literature (e.g. [14,20,24,27,30–32] and references cited), information on Ni4+-contaning compounds is scarce. The presence of 10% Ni4+ was evidenced for La0.9Sr0.1NiO3 prepared at 1273 K under the oxygen pressure of 200 atm [33]. The Ni oxidation states higher than 3+ at 300 K were also identified for RNi0.98Fe0.02O3, where R is rare-earth cation and nickel disproportionates as Ni3+ → Ni3 + λ + Ni3 − γ [34]. For LaGa0.65Mg0.15Ni0.20O3 − δ, the average state of Ni cations remains higher than 3+ up to 1173 K under moderately oxidizing conditions [35].

Fig. 9. Temperature dependencies of the equilibrium oxygen nonstoichiometry of La1 − ySryFe1 − xNixO3 − δ in air, and the dilatometric curve of La0.8Sr0.2Fe0.9Ni0.1O3 − δ ceramics illustrating the effect of oxygen losses.

The total conductivity (σ) of La1 − ySryFe1 − xNixO3 − δ in air exhibits a semiconductor-like behavior at least up to 1100 K (Fig. 10). The activation energies (Ea, Table 1) vary in the range 9–29 kJ/mol, decreasing when x and y increase. At higher temperatures, increasing oxygen nonstoichiometry results in decreasing hole concentration and in an apparent transition to pseudometallic conduction, wellknown for numerous ferrites and nickelates [1,2,10,16, 21,26,27,35]. One should stress that the conductivity is predominantly electronic; the oxygen ion transference numbers estimated from the permeation data, discussed below, are lower than 0.0001 at 1123–1223 K in the entire perovskite phase-stability domain. In combination with the positive sign of Seebeck coefficient (α, Fig. 11), this unambiguously confirms the prevailing role of p-type electronic transport. Fig. 12 displays the hole mobility (μp) in air, calculated from the total conductivity and TGA data: σ≈μ p
e
p
Nfu =Vuc

ð1Þ

where p is the hole concentration per unit formula as evaluated from the oxygen content by the electroneutrality condition, Nfu is the

Fig. 10. Temperature dependencies of the total electrical conductivity of La1 −ySryFe1 −xNixO3 −δ ceramics in air.

2176

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

Fig. 11. Oxygen partial pressure dependencies of the total conductivity (A,C,E) and Seebeck coefficient (B,D,F) of La0.9Sr0.1Fe0.9Ni0.1O3 − δ (A,B), La0.9Sr0.1Fe0.8Ni0.2O3 − δ (C,D) and La0.8Sr0.2Fe0.9Ni0.1O3 − δ (E,F) at 973–1223 K.

number of formula units per unit cell, and Vuc is the unit-cell volume. Despite the relatively high values exceeding an approximate threshold of 0.1 cm2 × V− 1 × s− 1 used to separate small-polaron and broad-band conduction, the mobility in La0.9Sr0.1Fe1 − xNixO3 − δ has a temperatureactivated character. This suggests that the hole transport mechanism may still be considered as polaronic. Further Sr addition reduces the mobility activation energy down to 1.3 kJ/mol, thus ascertaining progressive delocalization of the p-type electronic charge carriers; at the same time, similar μp values for La0.8Sr0.2Fe0.9Ni0.1O3 − δ and La0.9Sr0.1Fe0.9Ni0.1O3 − δ point out that the higher conductivity of the former (Figs. 10 and 13) results from a greater hole concentration determined by Sr2+ content, in the entire phase-stability domain. The tendency to growing p-type charge-carrier delocalization is also observed on nickel doping in La0.9Sr0.1Fe1 − xNixO3 − δ series, where the conductivity increases and Seebeck coefficient decreases down to zero level at x = 0.4 (Fig. 13). These trends are in accordance with a hopping mechanism in La1 − ySryFeO3 − δ at y ≤ 0.4 (e.g. [37]), and metallic conductivity known for (La,Sr)NiO3 − δ in a wide temperature range [15].

The linearity of isothermal α vs. ln σ dependencies illustrated by Fig. 14 indicates that the density of states in La1 − ySryFe1 − xNixO3 − δ is essentially p(O2)-independent. Such an observation is consistent with both broad-band and polaron conduction when the charge-carrier concentration is small. However, the slopes of these dependencies are relatively low, from −0.40 × (k/e) to −0.61 × (k/e), implying invalidity of the simplest Heikes formula [38] in the given case. The origin may be associated, first of all, with significant charge-carrier interactions. For instance, attractive on-site interaction leading to the electron pair formation would give a slope of −1/2 × (k/e) [38]; in the case of siteexclusion phenomena near holes blocking neighboring positions, selfscreening may yield −1/2 × (k/e) or −2/3 × (k/e) [39]. Whatever the electronic interaction mechanisms, inspection of Figs. 12 and 14 shows clearly that the density of states decreases with Ni additions, whereas Sr doping has rather a minor effect. These variations can be explained by local lattice distortions around the large Ni2+ cations and/or by defect clustering processes involving Ni2+ and oxygen vacancies, in agreement with the oxygen permeation data discussed below. The association of divalent B-site cations and anion vacancies in the ABO3

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

Fig. 12. Temperature dependencies of the electron–hole mobility in La1 − ySryFe1 − xNixO3 − δ at atmospheric oxygen pressure, calculated from the conductivity and oxygen-nonstoichiometry data.

perovskite structure is, again, well established for both Ni- and Fecontaining compounds [6,32,35,40], and occurs much easier with respect to clustering involving A2+ cations as the B−O bonds are shorter than the interatomic A−O distances. Finally, Fig. 15 compares the total conductivity variations in La0.8Sr0.2Fe0.9Ni0.1O3 − δ, measured on cycling oxygen pressure in a wide p(O2) range where the perovskite decomposition occurs, and

2177

Fig. 14. Comparison of the Jonker-type plots of La1 − ySryFe1 − xNixO3 − δ at 1223 K.

selected phase boundaries in the La–Ni–O [7] and La–Fe–O [41] systems. Due to the low nickel content, the conductivity seems essentially insensitive to the segregation of Ni-containing phases and exhibits a good reproducibility on redox cycling. The variations observed at oxygen pressures lower than 10− 5 atm, typical for the perovskite-type ferrites [27,37,42], are governed by the major LaFeO3-based phase. Moreover, the plateau-like behavior and subsequent conductivity drop at p(O2) = 10− 10–10− 6 atm result from the stagnated equilibration kinetics leading to the measurement errors [42]. In combination with the XRD data, this justifies selection of the relatively narrow p(O2) range, 10− 5 to 0.4 atm, used for the electrical measurements. 3.4. Oxygen-ionic transport Fig. 16 compares the steady-state oxygen permeation fluxes (j) through dense La0.9Sr0.1Fe0.9Ni0.1O3 − δ, La0.9Sr0.1Fe0.8Ni0.2O3 − δ and La0.8Sr0.2Fe0.9Ni0.1O3 − δ membranes with a fixed thickness, d = 1.00 ± 0.02 mm. As the ceramic membrane materials are single-phase, an excellent level of the oxygen flux reproducibility was achieved

Fig.13. Comparison of the total conductivity and Seebeck coefficient of La1 − ySryFe1 − xNixO3 − δ (x=0.1–0.4, y=0.1–0.2) under oxidizing conditions, at 1123 K.

Fig. 15. Oxygen partial pressure dependencies of the total conductivity of La0.8Sr0.2Fe0.9Ni0.1O3 − δ, collected on cycling p(O2). The equilibrium oxygen pressures over threephase mixtures in the La–Ni–O and La–Fe–O systems [7,41] are shown for comparison.

2178

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

Fig. 18. Oxygen permeation fluxes (A) and specific oxygen permeability (B) of dense La0.9Sr0.1Fe0.9Ni0.1O3 − δ ceramics with various thicknesses vs. oxygen partial pressure gradient.

Fig. 16. Oxygen permeation fluxes through dense La0.9Sr0.1Fe0.9Ni0.1O3 − δ, La0.9Sr0.1Fe0.8Ni0.2O3 − δ and La0.8Sr0.2Fe0.9Ni0.1O3 − δ membranes vs. temperature (A) and oxygen partial pressure gradient at 1173 K (B). The membrane thickness is 1.00 ± 0.02 mm.

(Fig. 17). Due to rising oxygen deficiency (Fig. 9), increasing Sr content up to 20% leads to considerably higher oxygen permeability (Fig. 16). On the contrary, increasing nickel concentration has a modest but negative impact. The decrease of the oxygen-ionic transport in La0.9Sr0.1Fe1 − xNixO3 − δ series may be associated with either local distortions of the perovskite lattice around nickel cations or with the vacancy-association processes involving Ni2+. Fig. 18(A) shows the oxygen fluxes through La0.9Sr0.1Fe0.9Ni0.1O3 − δ membranes with various thicknesses. The specific permeability J(O2) (Fig. 19B), used to identify the permeation-limiting factors [9,27,28,43], was calculated as   −1 p2 : J ðO2 Þ ¼ j
d
ln p1

ð2Þ

Since this term is proportional to j × d, its value is thicknessindependent when surface-exchange limitations are negligible but increases with d when they become significant. In the case of La0.9Sr0.1Fe0.9Ni0.1O3 − δ, increasing membrane thickness results in lower oxygen fluxes and higher J(O2), Fig. 18, thus displaying that the overall oxygen transport is governed by both bulk ambipolar conductivity (σ̄amb) and exchange kinetics. The influence of surfacerelated processes seems to increase on cooling, as for (La,Sr)FeO3 − δ [27]. The relative roles of these factors may be estimated using an approximate relationship [43]:   1 p2 16F 2 d 16F 2 −1 ¼ þ k−1
ðd þ dc Þ
ln 2 þ k1 ¼ p1 RTσ amb j RTσ amb Fig. 17. Comparison of the oxygen permeation fluxes through two different La0.8Sr0.2Fe0.9Ni0.1O3 − δ membranes (d = 1.00 mm), illustrating reproducibility of the permeation data.

ð3Þ

where k2 and k1 are the exchange coefficients at the feed and permeate sides, respectively, and the critical thickness dc corresponds

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

to the transition from surface-exchange to bulk diffusion control. The results of such estimations are summarized in Fig. 19. The calculated ambipolar conductivity is 104–105 times lower than the corresponding σ values (Fig. 11) and is, therefore, nearly equal to the partial oxygenionic conductivity (σO). Reducing p(O2) leads to a higher ambipolar conduction, confirming that the transport is limited by the oxygenvacancy concentration. Although the surface-exchange coefficients are found essentially p(O2)-independent in the narrow range of oxygen chemical potentials studied in this work, the critical thickness increases on reducing permeate-side oxygen pressure, as an increasing ionic conduction rises relative contribution of the interfacial kinetics. At 1123–1223 K, the dc values are 0.2–1.1 mm and decrease when temperature increases. The oxygen-vacancy diffusion coefficient (DV, Table 3) of La0.9Sr0.1Fe0.9Ni0.1O3 − δ was calculated from the ionic conductivity and oxygen-nonstoichiometry data using the Nernst–Einstein equation as [37,44] DV ¼

kT σ O Vuc
: 4e2 δNfu

ð4Þ

The estimated DV values are slightly lower compared to (La,Sr) FeO3 − δ, whilst the vacancy diffusion activation energy is very close to that for La0.75Sr0.25FeO3 [44] and (La0.6Sr0.4)0.99FeO3 − δ [37]. The ionic conductivity activation energy of La0.9Sr0.1Fe0.9Ni0.1O3 − δ, 155 kJ/mol, is also very similar to the Ea value reported for (La0.6Sr0.4)0.99FeO3 − δ, 156 kJ/mol [37]. This suggests that Ni doping decreases the concentration of mobile oxygen vacancies, but has no substantial effect on the diffusion mechanism, again validating the above

2179

Table 3 Comparison of the oxygen-vacancy diffusion coefficients in (La,Sr)FeO3-based perovskites Composition

T, K

Method and p(O2)

DV × 106, cm2/s

Ea, kJ/mol

Ref.

La0.9Sr0.1Fe0.9Ni0.1O3

1123 1173 1223 1173 1223

Oxygen permeation (0.21 atm) 18 O/16O isotopic exchange (0.06 atm) 18 O/16O isotopic exchange (0.06 atm) Conductivity relaxation (0.05 atm)

1.0 2.2 3.2 3.2–4.5 4.6

134 (1123–1223 K)

This work

79 (1173–1373 K)

[44]

6.1 6.2

114 (1173–1323 K)

[44]

1.5 4.4 8.2

129 (1073–1273 K)

[37]

La0.9Sr0.1FeO3

La0.75Sr0.25FeO3

1173 1223

(La0.6Sr0.4)0.99FeO3

1073 1173 1223

hypothesis regarding the formation of stable defect clusters which involve anion vacancies and nickel cations. 4. Conclusions The total electrical conductivity, Seebeck coefficient, oxygen nonstoichiometry and steady-state oxygen permeability of perovskite-type La1 − ySryFe1 − xNixO3 − δ were studied in the oxygen partial pressure range from 10− 5 to 0.4 atm at 973–1223 K. Doping with strontium was found to enhance the oxygen-vacancy concentration and ionic transport at elevated temperatures, simultaneously decreasing thermodynamic stability. At 1373 K and atmospheric oxygen pressure, the compositional boundary of the narrow solid-solution domain corresponds to approximately x = 0.25 at y = 0.10 and x = 0.12 at y = 0.20. Reducing oxygen pressure down to ~10− 5 atm also leads to the segregation of K2NiF4type lanthanum nickelate. At low temperatures when the orthorhombic (La,Sr)(Fe,Ni)O3 − δ perovskites are essentially oxygen-stoichiometric, the Mössbauer spectroscopy and thermogravimetric analysis revealed that the average oxidation state of nickel is higher than 3+, comparable to iron cations. Heating up to 340–700 K results in the orthorhombic → rhombohedral transition accompanied with a volume contraction. The average TECs of La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.4, y = 0.1–0.2) ceramics vary in the range (12.4–13.4)× 10− 6 K− 1 at 700– 1150 K, and increase up to (14.2–18.0) × 10− 6 K− 1 at 1150–1370 K when the oxygen nonstoichiometry becomes significant. The Sr and Ni additions increase thermal expansion, hole delocalization and total conductivity, which is predominantly p-type electronic in the entire phase-stability domain. The steady-state oxygen permeation fluxes through single-phase La1 − ySryFe1 − xNixO3 − δ (x = 0.1–0.2, y = 0.1–0.2) membranes, with an activation energy of 206–235 kJ/mol, are determined by both surface-exchange and bulk ionic conduction limited by the oxygen-vacancy concentration. The substitution of nickel for iron has a modest negative effect on the ionic transport, probably due to oxygen-vacancy association with Ni2+ cations, in agreement with the conductivity and Seebeck coefficient variations. Acknowledgements This work was supported by the FCT, Portugal (projects PTDC/CTM/ 64357/2006 and SFRH/BPD/28629/2006), and the Russian Federal Agency on Science and Innovations (state contract No. 02.444.11.7020). Experimental assistance and helpful discussions made by S. Yakovlev, A. Yaremchenko and A. Kovalevsky, are gratefully acknowledged. References

Fig. 19. Estimated ambipolar conductivity (A), surface-exchange coefficients (B) and critical thickness of La0.9Sr0.1Fe0.9Ni0.1O3 − δ membranes vs. oxygen partial pressure gradient (see text).

[1] G. Amow, S.J. Skinner, J. Solid State Electrochem. 10 (2006) 538. [2] M. Al Daroukh, V.V. Vashook, H. Ullmann, F. Tietz, I.A. Raj, Solid State Ionics 158 (2003) 141.

2180

E.V. Tsipis et al. / Solid State Ionics 179 (2008) 2170–2180

[3] D.C. Zhu, X.Y. Xu, S.J. Feng, W. Liu, C.S. Chen, Catal. Today 82 (2003) 151. [4] G. Kim, A.J. Jacobson, Mater. Res. Soc. Symp. Proc. 972 (2007) 175. [5] S. Castillo, R.F. Cienfuegos, M.L. Fontaine, P. Lenormand, P. Bacchin, F. Ansart, Mater. Res. Bull. 42 (2007) 2125. [6] V.V. Kharton, A.A. Yaremchenko, E.N. Naumovich, J. Solid State Electrochem. 3 (1999) 303. [7] M. Zinkevich, F. Aldinger, J. Alloys Compd. 375 (2004) 147. [8] R. Chiba, F. Yoshimura, Y. Sakurai, Solid State Ionics 124 (1999) 281. [9] V.V. Kharton, A.P. Viskup, E.N. Naumovich, V.N. Tikhonovich, Mater. Res. Bull. 34 (1999) 1311. [10] R. Chiba, F. Yoshimura, Y. Sakurai, Solid State Ionics 152–153 (2002) 575. [11] E.A. Kiselev, N.V. Proskurnina, V.I. Voronin, V.A. Cherepanov, Inorg. Mater. 43 (2007) 167. [12] L.Y. Gavrilova, T.V. Aksenova, L.A. Bannykh, Y.V. Teslenko, V.A. Cherepanov, J. Struct. Chem. 44 (2003) 248. [13] K. Kammer, L. Mikkelsen, J.B. Bilde-Sorensen, J. Solid State Electrochem. 10 (2006) 934. [14] M.A. Penã, J.L.G. Fierro, Chem. Rev. 101 (2001) 1981. [15] H. Obayashi, T. Kudo, Jap. J. Appl. Phys. 14 (1975) 330. [16] K. Huang, H.Y. Lee, J.B. Goodenough, J. Electrochem. Soc. 145 (1998) 3220. [17] V.V. Kharton, F.M. Figueiredo, L. Navarro, E.N. Naumovich, A.V. Kovalevsky, A.A. Yaremchenko, A.P. Viskup, A. Carneiro, F.M.B. Marques, J.R. Frade, J. Mater. Sci. 36 (2001) 1105. [18] S.P. Simner, J.F. Bonnett, N.L. Canfield, K.D. Meinhardt, V.L. Sprenkle, J.W. Stevenson, Electrochem. Solid State Lett. 5 (2002) A173. [19] Y.D. Zhen, A.I.Y. Tok, S.P. Jiang, F.Y.C. Boey, J. Power Sources 170 (2007) 61. [20] V.V. Kharton, A.A. Yaremchenko, M.V. Patrakeev, E.N. Naumovich, F.M.B. Marques, J. Eur. Ceram. Soc. 23 (2003) 1417. [21] S. Hashimoto, K. Kammer, P.H. Larsen, F.W. Poulsen, M. Mogensen, Solid State Ionics 176 (2005) 1013. [22] L.A. Chick, L.R. Pederson, G.D. Maupin, J.L. Bates, L.E. Thomas, G.J. Exarhos, Mater. Lett. 10 (1990) 6. [23] J. Rodriguez-Carvajal, Physica B 192 (1993) 55. [24] J.C. Waerenborgh, D.P. Rojas, N.P. Vyshatko, A.L. Shaula, V.V. Kharton, I.P. Marozau, E.N. Naumovich, Mater. Lett. 57 (2003) 4388.

[25] J. Hesse, A. Rübartsch, J. Phys. E 7 (1974) 526. [26] V.V. Kharton, A.A. Yaremchenko, A.L. Shaula, M.V. Patrakeev, E.N. Naumovich, D.I. Logvinovich, J.R. Frade, F.M.B. Marques, J. Solid State Chem. 177 (2004) 26. [27] V.V. Kharton, J.C. Waerenborgh, A.P. Viskup, S. Yakovlev, M.V. Patrakeev, P. Gaczyński, I.P. Marozau, A.A. Yaremchenko, A.L. Shaula, V.V. Samakhval, J. Solid State Chem. 179 (2006) 1273. [28] S.O. Yakovlev, V.V. Kharton, E.N. Naumovich, J. Zekonyte, V. Zaporojtchenko, A.V. Kovalevsky, A.A. Yaremchenko, J.R. Frade, Solid State Sci., 8 (2006) 1302. [29] A. Fossdal, M. Menon, I. Wærnhus, K. Wiik, M.-A. Einarsrud, T. Grande, J. Am. Ceram. Soc. 87 (2004) 1952. [30] S.E. Dann, D.B. Currie, M.T. Weller, M.F. Thomas, A.D. Al-Rawwas, J. Solid State Chem. 109 (1994) 134. [31] P. Adler, S. Ghosh, Solid State Sci. 5 (2003) 445. [32] P.D. Battle, T.C. Gibb, S. Nixon, J. Solid State Chem. 77 (1988) 124. [33] J.A. Alonso, M.J. Martínez-Lope, H. Falcόn, R.E. Carbonio, Phys. Chem. Chem. Phys. 1 (1999) 3025. [34] S.-J. Kim, I. Presniakov, G. Demazeau, K. Pokholok, A. Baranov, A. Sobolev, D. Pankratov, N. Ovanesyanz, J. Solid State Chem. 168 (2002) 126. [35] E.N. Naumovich, V.V. Kharton, A.A. Yaremchenko, M.V. Patrakeev, D.G. Kellerman, D.I. Logvinovich, V.L. Kozhevnikov, Phys. Rev. B 74 (2006) 064105. [36] Y. Minet, V. Lefranc, N. Nguyen, B. Domenges, A. Maignan, B. Raveau, J. Solid State Chem. 121 (1996) 158. [37] M. Søgaard, P.V. Hendriksen, M. Mogensen, J. Solid State Chem. 180 (2007) 1489. [38] P.M. Chaikin, G. Beni, Phys. Rev. B 13 (1976) 647. [39] M.S. Seltzer, A.Z. Hed, J. Electrochem. Soc. 117 (1970) 815. [40] A.A. Yaremchenko, A.L. Shaula, D.I. Logvinovich, V.V. Kharton, A.V. Kovalevsky, E.N. Naumovich, J.R. Frade, F.M.B. Marques, Mater. Chem. Phys. 82 (2003) 684. [41] T. Nakamura, G. Petzow, L.J. Gauckler, Mater. Res. Bull. 14 (1979) 649. [42] C.Y. Park, A.J. Jacobson, J. Electrochem. Soc. 152 (2005) J65. [43] I.P. Marozau, V.V. Kharton, A.P. Viskup, J.R. Frade, V.V. Samakhval, J. Eur. Ceram. Soc. 26 (2006) 1371. [44] T. Ishigaki, S. Yamauchi, K. Kishio, J. Mizusaki, K. Fueki, J. Solid State Chem. 73 (1988) 179.