Surface-limited oxygen transport and electrode properties of La2Ni0.8Cu0.2O4+δ

Solid State Ionics 166 (2004) 327 – 337 www.elsevier.com/locate/ssi

Surface-limited oxygen transport and electrode properties of La2Ni0.8Cu0.2O4+d V.V. Kharton a,b,*, E.V. Tsipis a, A.A. Yaremchenko a, J.R. Frade a a

b

Department of Ceramics and Glass Engineering, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal Institute of Physicochemical Problems, Belarus State University, 14 Leningradskaya Str., 220050 Minsk, Belarus Received 19 June 2003; received in revised form 20 November 2003; accepted 1 December 2003

Abstract The submicron powder of La2Ni0.8Cu0.2O4 + d with K2NiF4-type structure, having grain size of 30 – 60 nm, was synthesized via glycine – nitrate process (GNP) and used for the preparation of porous cathode layers applied onto (La0.9Sr0.1)0.98Ga0.8Mg0.2O3
d (LSGM) solid electrolyte. In air, dense ceramics of La2Ni0.8Cu0.2O4 + d possess thermal expansion coefficient of 13.3  10
6 K
1 at 400 – 1240 K, p-type electronic conductivity of 50 – 85 S/cm at 800 – 1300 K and relatively high oxygen permeability limited by the surface exchange. These properties provide a substantially high performance of porous electrodes, exhibiting cathodic overpotential lower than 50 mV at 1073 K and current density of 200 mA/cm2. As for the oxygen transport through dense membranes, the results on electrode behavior, including the overpotential – microstructure relationships and the p(O2) dependence of polarization resistance, suggest that the cathodic reaction rate is affected by surface-related processes. Due to this, electrode performance can be considerably enhanced by surface activation, particularly via impregnation with Pr-containing solutions, and also by decreasing fabrication temperature. At 873 K, the surface modification with praseodymium oxide decreases overpotential of La2Ni0.8Cu0.2O4 + d cathode, screen-printed onto LSGM and annealed at 1473 K, from 330 down to approximately 175 mV at 50 mA/cm2. D 2003 Elsevier B.V. All rights reserved. Keywords: Lanthanum nickelate; IT SOFC cathode; Oxygen permeation; Cathodic overpotential; Polarization resistance

1. Introduction High level of oxygen ionic conductivity in solid electrolytes based on lanthanum gallate, in particular La(Sr) Ga(Mg)O3
d, enables their use for intermediate-temperature solid oxide fuel cells (IT SOFCs) operating at 870 – 1070 K [1– 8]. Reducing of SOFC operation temperature is associated, however, with increasing role of electrode polarization as a performance-limiting factor, since the apparent activation energy for the polarization resistance is typically higher than that for ionic transport in solid electrolytes. The electrochemical activity of conventional electrode materials, such as lanthanum – strontium manganites, at temperatures below 1050 K is insufficient. Further developments of IT SOFCs require a search for new cathode compositions highly active * Corresponding author. Present address: Department of Ceramics and Glass Engineering, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal. Tel.: +351-234-370263; fax: +351-234-425300. E-mail address: [email protected] (V.V. Kharton). 0167-2738/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2003.11.020

in contact with La(Sr)Ga(Mg)O3
d. A high performance could be achieved when using materials derived from Srsubstituted rare-earth cobaltites, Ln(Sr)CoO3
d (Ln = La – Sm) [4,6 – 9]. However, the values of thermal expansion coefficients (TECs) of cobaltite phases [3,7,9] are 50 – 300% higher with respect to possible electrolyte candidates, La(Sr)Ga(Mg)O3
d and Ce(Gd)O2
d [7,10,11]. Due to thermomechanical incompatibility, long-term stability of the cells with LnCoO3-based cathode layers appears very problematic. Another group of materials promising for IT SOFC cathodes refers to La2NiO4-based phases with K2NiF4-type structure [10]. These compounds exhibit a relatively high oxygen-ion diffusivity, TECs compatible with solid electrolytes, and predominant p-type electronic conductivity in whole p(O2) range where the K2NiF4-type phases exist [10,12 –18]. The ionic conductivity of oxygen-hyperstoichiometric La2NiO4 + d, being 5 –10 times lower than that of oxygen-deficient perovskites SrCo1
xFexO3
d (x = 0.2 – 0.5) and Sr1
xLaxCoO3
d (x = 0.2 –0.4), is nevertheless

328

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

remarkably higher as compared to lanthanum – strontium ferrites and manganites [10,14,17]. For La2NiO4-based phases, the bulk ionic transport occurs via diffusion of interstitial ions in the rock-salt-type layers and vacancies in the perovskite layers of K2NiF4-type structure; the oxygen permeation through dense nickelate membranes is, as a rule, limited by the surface exchange rate [12,16 – 18]. Continuing our research on electrode materials for intermediate-temperature electrochemical cells [9,10,16 –20], the present work is focused on the electrochemical and transport properties of La2Ni0.8Cu0.2O4 + d, evaluated as cathode material in contact with (La0.9 Sr 0.1 ) 0.98 Ga 0.8 Mg 0.2 O 3
d (LSGM) solid electrolyte studied earlier [21]. In order to preserve electrode microstructure and to prevent grain growth and interaction with electrolyte, the temperature of electrode fabrication should be minimum while still providing necessary mechanical strength. While the melting point of La2NiO4 is rather high, lanthanum cuprate forms liquid phase at approximately 1323 F 10 K [22]. The existence of continuous solid solution in pseudobinary La 2NiO 4 – La2CuO4 system [15,23] makes it possible to expect that incorporation of moderate amounts of Cu in the lattice of La2NiO4 + d should promote sintering without a significant deteriorating effect on the transport properties. A similar approach was earlier exploited to improve sinterability of SrCo(Fe,Cu)O 3
d perovskites [19]. The composition La2Ni0.8Cu0.2O4 + d, studied in this work as a model electrode material, was selected on the basis of preliminary studies on adhesion, interaction with LSGM, and sintering of La2Ni1
xCuxO4 + d (x = 0– 1) layers and ceramics; detailed results of these investigations will be summarized in a separate publication.

2. Experimental

The sintering conditions and properties of dense LSGM ceramics, prepared from a commercial powder (PSC, Seattle), were described elsewhere [21]. Formation of single K2NiF4-type phase of La2Ni0.8Cu0.2O4 + d during sintering of electrode layers and ceramics was confirmed by X-ray diffraction (XRD) analysis (Fig. 1). The crystal structure of powders obtained from sintered cathode layers followed by ultrasonic dispersion was also examined using selected area electron diffraction (SAED). Transmission electron microscopy (TEM, Hitachi H-9000, 300 kV) and scanning electron microscopy (SEM, Hitachi S-4100) combined with energy dispersive spectroscopy (EDS) were used for microstructural studies. The sintering behavior of green compacts and thermal expansion of the ceramics were studied using an alumina Linseis L70/2001 dilatometer (heating rate of 5 K/min). Description of experimental procedures and equipment used for the materials characterization, including the measurements of total conductivity (four-probe DC), Seebeck coefficient and steady-state oxygen permeation fluxes, is found elsewhere [9,16 – 21]. The conductivity and Seebeck coefficient were measured at 300 –1300 K in the oxygen partial pressure range from 2 Pa to 50 kPa. The data on oxygen permeability were obtained at 973 – 1223 K and the feed-side oxygen pressure ( p2) fixed at 21 kPa (air); the permeate-side oxygen pressure ( p1) varied from 0.2 to 20 kPa. The electrode polarization measurements were carried out by the three-electrode technique in cells with porous Pt counter and reference electrodes (CE and RE, respectively); the cell geometry was chosen according to Refs. [25,26]. The experiments were performed in the galvanostatic mode using an AUTOLAB PGSTAT20 instrument at 873– 1073 K in flowing air and O2 – Ar mixtures (flow rate of 25 cm3/min). The composition of gas mixtures was controlled by Bronkhorst mass-flow controllers; the oxygen partial pressure additionally measured by an electrochemical oxygen sensor,

Submicron powder of single-phase La2Ni0.8Cu0.2O4 + d was synthesized using the glycine – nitrate process (GNP), a self-combustion method with glycine as fuel and nitrates of the metal components as oxidant. GNP is well known as appropriate for multicomponent systems, where the use of a standard ceramic route may be hampered due to kinetic reasons, and enables to prepare fine homogeneous powders with a large specific surface area [24]. In the course of GNP, the stoichiometric amounts of high-purity La(NO3)36H2O, Ni(NO3)26H2O and Cu(NO3)26H2O were dissolved in an aqueous solution of nitric acid with subsequent addition of glycine; the molar glycine/nitrate ratio, calculated assuming the only gaseous products of the reaction to be H2O, CO2 and N2, was 1:1. After drying and firing, the resultant powder was calcined in air at 1073 K for 2 h. Dense ceramic samples, used for the measurements of transport properties and thermal expansion, were uniaxially pressed at 250 –400 MPa and then sintered in air at 1503 K for 2 h. Porous cathode layers with the sheet density of 11.3 F 0.5 mg/cm2 were screen-printed onto LSGM electrolyte and annealed at 1473 –1523 K for 2 h.

Fig. 1. Powder XRD patterns of La2Ni0.8Cu0.2O4 + d after synthesis and annealing in air at 1073 K, and after sintering at 1503 K. Arrow in the pattern of as-synthesized powder indicates minor impurity peak, unidentified due to low intensity.

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

329

annealing in air at 1073 K for 2 h; then the overpotential – current dependencies were re-measured. In all cases, the absence of fast degradation processes was verified by the measurements of time dependencies of the overpotential at fixed currents during 50 –200 h. As for the as-prepared layers, after electrochemical measurements the electrodes were examined by XRD, SEM/EDS and TEM/EDS in order to reveal possible structural and microstructural changes.

3. Results and discussion Fig. 2. Typical impedance spectra of La2Ni0.8Cu0.2O4 + d cathode, asprepared and surface-activated with PrOx. Spectra obtained in the course of polarization measurements at 973 K in air.

varied in the range from 40 Pa to 101 kPa. The overpotential (g) was calculated as: g ¼ U
IR

ð1Þ

where U is the steady-state potential difference between the working electrode (WE) and RE, I is the current between WE and CE, and I  R is the ohmic contribution to the total potential drop. The values of the ohmic resistance, R, were determined from the corresponding impedance spectra collected in the frequency range from 10 mHz to 50 kHz. The impedance spectroscopy data were also used to calculate polarization resistance, Rg, which can be derived as the difference between the low- and high-frequency intercepts (Fig. 2). In this work, the frequency range was chosen minimum, necessary to provide R and Rg values. The behavior at frequencies higher than 50 kHz, simultaneously studied by a HP4284A precision LCR meter, will be reported in a separate paper. The current density and overpotential values varied in the ranges 0– 220 mA/cm2 and 0– 350 mV, respectively; the time necessary to attain steady-state conditions was 0.5 – 10 h. The reproducibility of results was separately checked after each measurement cycle; one example is shown in Fig. 3. After the measurements, selected electrode layers were surface-modified with praseodymium oxide, according to previous results [27]. The activation included impregnation with a Pr(NO3)35H2O solution in ethanol, drying and

Fig. 3. Reproducibility of overpotential vs. current dependence for La2Ni0.8Cu0.2O4 + d cathode, fabricated at 1473 K, in the course of the measurement cycles repeated several times (see text).

3.1. Phase development, microstructure and sintering The X-ray and electron diffraction studies demonstrated that complete formation of crystalline La2Ni0.8Cu0.2O4 + d with K2NiF4-type lattice can be achieved at temperatures around 1200 K. For example, XRD pattern of the GNPsynthesized powder annealed at 1073 K shows traces of phase impurity, which cannot be identified due to very low

Fig. 4. Selected area electron diffraction pattern of the cathode layer sintered at 1473 K for 2 h (top), and dark- and bright-field TEM images of La2Ni0.8Cu0.2O4 + d powder prepared using GNP followed by annealing in air at 1073 K (bottom).

330

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

Table 1 Properties of La2Ni0.8Cu0.2O4 + d ceramics Unit cell parameters, nm

dexp, g/cm3

dexp/dtheor, %

a¯ 106, K
1 (400 – 1240 K)

Activation energy Process

T, K

Ea, kJ/mol

a = 0.3858(0) c = 1.2787(3)

6.87

98.1

13.27 F 0.01

Total conduction Oxygen permeation

320 – 760 973 – 1223

8.6 F 0.1 171 F 6

dexp and dtheor are the experimental and theoretical density values, respectively; a¯ is the average linear thermal expansion coefficient; activation energies (Ea) and A0 were calculated using the standard Arrhenius model; Ea for oxygen permeability was calculated for oxygen partial pressure gradient of 21/2.1 kPa.

intensity of the corresponding peaks (Fig. 1); the presence of amorphous components was also indicated by the analysis of electron diffraction patterns [28]. After sintering of ceramics and electrodes at 1473– 1523 K, no secondary phases are observed. One example of SAED pattern typical for the crystallites in electrode layers is given in Fig. 4; the XRD pattern of ceramic material sintered at 1503 K is presented in Fig. 1. The unit cell parameters (Table 1) are in a good agreement with literature [23]. TEM inspection showed that the particle size in the powder, prepared by GNP with subsequent annealing at 1073 K and used for screen-printing, is of nano-scale, 30– 60 nm (Fig. 4). Due to agglomeration, this powder has a foam-like microstructure with a large surface area (Fig. 5A). The active sintering accompanied with complete crystallization of La2Ni0.8Cu0.2O4 + d phase starts at approximately

Fig. 5. SEM micrographs of La2Ni0.8Cu0.2O4 + d powder synthesized via GNP and annealed at 1073 for 2 h (A), and La2Ni0.8Cu0.2O4 + d ceramics sintered at 1503 K for 2 h (B).

1100 –1200 K (Fig. 6). The density of ceramics higher than 98% of theoretical (Table 1) is obtained after sintering at 1503 K; a sufficient mechanical stability of the electrode layers applied onto LSGM was achieved only at temperatures above 1470 K [28]. The sintering at 1473 –1523 K resulted in grain growth up to 0.2– 4 Am (Figs. 5B and 7). The EDS analysis of ceramics and electrode layers confirmed homogeneous cation distribution, within the limits of experimental uncertainty; no segregation at the grain boundaries was detected. 3.2. Thermal expansion and electrical properties Thermal expansion of La2Ni0.8Cu0.2O4 + d ceramics is linear within the studied temperature range (Fig. 8); the TEC value averaged at 400– 1240 K is 13.3  10
6 K
1 (Table 1). The relatively low thermal expansion coefficient makes La2Ni0.8Cu0.2O4 + d compatible with IT SOFC solid electrolytes, such as LSGM or Ce(Gd)O2
d, having slightly lower but still very similar TEC values, (10.5 – 12.8)  10
6 K
1 [7,10,11]. The total conductivity of La2Ni0.8Cu0.2O4 + d exhibits a semiconductor-like behavior with activation energy of 8.6 kJ/ mol at 300 – 750 K (Fig. 8). At higher temperatures, an apparent transition to pseudometallic behavior is observed. Similar trends are typical for most La2NiO4-based materials [13,15 – 18]. The pseudometallic dependencies of the conductivity on temperature result from oxygen losses on heating, resulting in decreasing p-type charge carrier concen-

Fig. 6. Shrinkage of green compact of GNP-synthesized La2Ni0.8Cu0.2O4 + d in air (ramp rate of 5 K/min).

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

331

Fig. 7. SEM micrographs of La2Ni0.8Cu0.2O4 + d cathode layers applied onto the surface of LSGM solid electrolyte, before (A) and after (B, C) polarization measurements, and following measurements after modification with PrOx (D). Fabrication temperatures were 1473 K (A, B, D) and 1523 K (C).

tration; the electron-hole transport mechanism remains smallpolaronic up to, at least, 1250 K [15,18]. The total conductivity values of La2Ni0.8Cu0.2O4 + d, varying in the range 50 – 85 S/cm at 800– 1300 K in air, are sufficiently high for SOFC cathode applications. It should be separately mentioned that, as for other nickelates [15 – 18], the conductivity of La 2 Ni 0.8Cu 0.2O 4 + d is predominantly p-type electronic [15,29]; the oxygen ion transference numbers estimated using the oxygen permeation data are lower than 10
3. Reducing oxygen partial pressure leads to a linear decrease of the conductivity, whilst the Seebeck coefficient has positive sign and increases, thus confirming dominant electron-hole transport in oxidizing conditions (Fig. 9). Since this paper is mainly focused on the electrochemical behavior of La2Ni0.8Cu0.2O4 + d cathodes, the analysis of

Fig. 8. Temperature dependencies of relative elongation and total conductivity of La2Ni0.8Cu0.2O4 + d ceramics in air.

Fig. 9. Oxygen partial pressure dependence of total conductivity (A) and Seebeck coefficient (B) of La2Ni0.8Cu0.2O4 + d. Solid lines show fitting results using the power regression models (Table 2).

332

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

electrical properties as function of the oxygen pressure is to be published separately [29]. One should mention, however, that the slope of log r – log p(O2) and a –(k/ e)log p(O2) curves is lower than 1/10 (Table 2). The slope of p(O2) dependence of the polarization resistance, discussed below, is considerably higher; this provides an additional evidence that, although the conductivity of La2Ni0.8Cu0.2O4 + d is 10 – 20 times lower than that of Ln(Sr)CoO3
d [5,9], the cathode performance is unaffected by electronic transport. 3.3. Surface-limited oxygen permeability of La2Ni0.8Cu0.2O4+d The oxygen permeation fluxes ( j) through La2Ni0.8 Cu0.2O4 + d ceramics are plotted in Fig. 10 as function of the oxygen partial pressure gradient and membrane thickness (d). At 1223 K, the permeation fluxes increase with decreasing thickness; these variations are, however, lower than it might be expected from the integral form of Wagner law, which predicts that the product j  d should be thickness independent if the flux is determined by the bulk ambipolar conductivity. Taking into account that the p-type electronic conductivity of La2Ni0.8Cu0.2O4 + d is high (Fig. 9), one can conclude that the oxygen transport is determined by two factors, namely the bulk ionic conduction and surface exchange. When temperature decreases down to 1123– 1173 K, the oxygen exchange rate becomes major permeation-limiting factor as indicated by the oxygen fluxes independent of the membrane thickness. Such a behavior is quite typical for La2NiO4based ceramics [12,16 – 18]. This behavior suggests that the cathodic performance of La2Ni0.8Cu0.2O4 + d may be limited by the surface exchange processes. In the case of porous electrode layers where the specific surface area is much larger (Fig. 7), the overall exchange rate should be substantially higher compared to the dense ceramics. However, the surface exchange may still affect electrode performance, especially at moderate temperatures. For instance, oxygen permeation fluxes through La(Sr)MnO3
d ceramics are also surfaceTable 2 The slope parameter (m) of the oxygen partial pressure dependencies of total conductivitya and Seebeck coefficientb of La2Ni0.8Cu0.2O4 + d ceramics at p(O2) = 1.5 – 4.7  104 Pa T, K

Total conductivity

Seebeck coefficient

973 1023 1073 1123 1173 1223

20.2 F 0.9 16.1 F 0.3 13.4 F 0.4 11.6 F 0.5 10.3 F 0.5 9.5 F 0.4

31.5 F 0.8 24 F 1 19 F 1 16 F 1 14.4 F 0.9 13.0 F 0.8

a Regression model for conductivity: r = r0p(O2)1/m, with m and r0 being constants. b Model for Seebeck coefficient: a =
k/e ln[ p(O)2]1/m + C, where C is constant.

Fig. 10. Oxygen permeation fluxes through La2Ni0.8Cu0.2O4 + d membranes vs. oxygen partial pressure gradient.

limited [27,30]; the oxygen reduction kinetics on La(Sr)MnO3-based cathodes is determined by exchangerelated factors, such as the concentration of oxygen vacancies on electrode surface, the role of which increases when temperature decreases [27,31,32]. 3.4. Oxygen transport in La2Ni0.8Cu0.2O4+d: comparison with other electrode materials Fig. 11 compares oxygen permeation through La2Ni0.8 Cu0.2O4 + d and other prospective cathode materials having TECs compatible with LSGM, under a fixed oxygen chemical potential gradient. The membrane thickness was 1.0 mm in all cases, except for La0.7Sr0.3MnO3
d where the permeation flux was measured for the ceramics with d = 0.6 mm and then normalized. Detailed information on the alternative electrode materials, including thermal expansion, oxygen permeability and total conductivity, is found elsewhere [10,33 – 35]. In general, LaMnO3- and SrMnO3-based phases exhibit a low ionic transport inappropriate for IT SOFC cathodes. Oxygen permeability of LaFe(Ni)O3
d perovskites is determined by the vacancy concentration and achieves sufficient level only for LaFe0.5Ni0.5O3
d, the composition close to solid solution formation limit [34]. The permeability of La2Ni0.8Cu0.2O4 + d ceramics is similar to that of LaGa0.65Mg0.15Ni0.20O3
d; the TECs of the latter composition, (17.0 – 18.4)  10
6 K
1 at 773 – 1273 K [33], are excessively high to provide stable operation in contact with LSGM. Thus, due to relatively high oxygen permeability and moderate thermal expansion, La2Ni0.8 Cu0.2O4 + d can be considered as a promising candidate for IT SOFC cathode materials. 3.5. Electrochemical behavior of La2Ni0.8Cu0.2O4+d electrodes The oxygen partial pressure dependence of electrode polarization resistance, under zero applied DC voltage, is

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

333

Fig. 11. Comparison of oxygen permeation fluxes through various ceramic materials, having TECs compatible with LSGM, under fixed oxygen pressure gradient. The data for La0.7Sr0.3MnO3, measured for membrane thickness of 0.60 mm, is normalized to 1.00 mm.

presented in Fig. 12. The slope parameter (n) of the simplest power model Rg = Rg0  p(O2)1/n is equal to (
3 F 0.3), being very different from the slope of p(O2) dependencies of the total conductivity and Seebeck coefficient (Table 2). This clearly indicates that the transport of electronic charge

carriers cannot be considered as a rate-limiting step of the electrode reaction. Taking into account that the ionic conduction in La2Ni(Cu)O4 + d is relatively high (Fig. 11) and that oxygen permeation through La2Ni(Cu)O4 + d ceramics at temperatures below 1150 K is primarily determined by

Fig. 12. Oxygen partial pressure dependence of the polarization resistance of La2Ni0.8Cu0.2O4 + d electrode fabricated at 1473 K. Inset shows impedance spectra at various oxygen pressures, after correction for the electrode area and subtraction of the displacement from the origin of the Z V-axis resulting from ohmic losses.

334

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

the surface exchange kinetics (Fig. 10), one may expect a significant role of exchange rate at the electrode/gas interface [36]. In particular, the electrochemical reaction might be located not only at the triple-phase boundary (TPB) but also at the electrode surface, with subsequent oxygen ion diffusion through the bulk and, possibly, along the surface. Studying of La2Ni0.8Cu0.2O4 + d electrodes under high current densities (Figs. 13 and 14) demonstrated a relatively high electrochemical activity. For example, at 1073 K and current density of 200 mA/cm2, the cathodic overpotential of electrode layer annealed at 1473 K was 46 F 2 mV. The overpotential vs. current dependencies follow Tafel equation without any indication of a limiting current density, thus implying an absence of gas diffusion limitations. At 1073 K, the anodic performance is worse than cathodic; at lower temperatures the anodic and cathodic overpotentials become similar. This may suggest that at 1073 K the exchange currents are affected by the surface concentration of active sites, such as oxygen vacancies, the coverage of which with oxygen species significantly depends on the current direction. High cathodic polarization is supposed to decrease oxygen content in the lattice of La2Ni0.8Cu0.2O4 + d electrode near surface, thus increasing the number of active sites participating in the reaction; this may decrease electrode polarization at a given current density. Under high anodic polarization, the situation is opposite. A similar behavior is characteristic of La(Sr)MnO3
d electrodes, where the electrochemical reaction mechanism involves oxygen vacancies on the surface [31,32]. Decreasing temperature seems to change the exchange mechanism, presumably to increase role of adsorption processes, in agreement with oxygen permeation data (Fig. 10).

Fig. 14. Current dependence of cathodic overpotential for one La2Ni0.8 Cu0.2O4 + d electrode layer, as-prepared at 1473 K and surface-activated with PrOx. Solid lines show fitting results according to Tafel equation.

If the overall rate is determined by surface-related processes, with no essential effect on the ionic charge carrier concentration in the electrode bulk under low polarization, a simplified model based on the assumptions [37] may include three consecutive steps: 1 O2 ðgÞ ! Oad 2 Oad þ e¯ !

O
ad

1

r1 ¼ k1 pðO2 Þ 2
k1V½Oad  1 r2 ¼ k2 ½Oad exp
2  1
k2V½O
ad exp 2

¯ ! O O
ad þ VO þ e O SS

Fig. 13. Comparison of anodic and cathodic polarization curves for La2Ni0.8Cu0.2O4 + d electrode layer, sintered at 1473 K, in contact with LSGM solid electrolyte in air. Inset compares overpotential vs. current dependencies for two La2Ni0.8Cu0.2O4 + d cathodes sintered at different temperatures, measured at 1073 K in air. Solid lines are for visual guidance.

FE RT FE RT

ð2Þ   ð3Þ

  1 FE SS r3 ¼ k3 ½O
½V exp
O ad 2 RT   1 FE
k3Vexp ð4Þ 2 RT

where E is the electrode potential, ri, ki and kiVare the total rate and the rate constants for the forward and backward reactions, respectively. The charge transfer coefficients (transport factors) in the kinetic equations (Eqs. (3) and (4)) are taken equal to 1/2 [37]. On assuming that cathodic process on oxygen-hyperstoichiometric La2Ni0.8Cu0.2O4 + d electrodes is associated with direct incorporation of interstitial oxygen ions, one should replace Eq. (4) by the
reaction Oad + e¯ ! OiW. However, the higher anodic overpotentials (Fig. 13) are more consistent with Eq. (4). If the second elementary step (Eq. (4)) is considered as rate-limiting with two other steps being in virtual equilibri-

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

um, the concentrations of surface oxygen species can be expressed from Eqs. (2) and (4). Since the current (i) is equal to (
2Fr), under high polarization the simplified expressions for anodic and cathodic currents can be derived from Eq. (3):   k 2Vk 3V SS
1 3 FE anodic i2 ¼ 2F ½VO exp ð5Þ k3 2 RT   1 k2 k1 1 FE 2 exp
pðO
icathodic ¼ 2F Þ ð6Þ 2 2 2 RT k1V Comparison of Eqs. (5) and (6) may qualitatively explain the observed difference in anodic and cathodic performance (Fig. 13). Under equilibrium conditions, the oxygen vacancy concentration is determined by intrinsic Frenkel defect formation and oxygen exchange with gas phase O O V

SS

VO þ OiW

S O O þ 2h V

SS

Kpl ¼ ½VO ½OiW

1 SS O2 þ VO 2

ð7Þ

pðO2 Þ1=2  ½VO p2 SS

Kp2 ¼

ð8Þ

SS and the electroneutrality condition p + 2[VO ] = 2[OiW], where SS p is the electron-hole concentration. If [VO ]b[OiW], the total oxygen hyperstoichiometry should be proportional to p(O 2 ) 1/6. This is confirmed by experimental data on La2NiO4 + d [13] and La2CuO4 + d [38]. In these conditions SS [VO ] c Kvp(O2)
1/6. The equilibrium electrode potential, E0 = const+(RT/mF)ln p(O2), can be calculated under conditions when ianodic = icathodic u i0, with i0 being the exchange current density. The overpotential is defined as g = E
E0. When the polarization is low, the surface concentration of oxygen vacancies should be dependent of the oxygen pressure and essentially independent of current. SS Substituting E and [VO ] into Eqs. (5) and (6) and then inserting them into the steady-state current i = i anodic + icathodic, one can obtain after simple transformations: 

1  3 5 k 2Vk 3V 4 k2 k1 4 i2 ¼ 2F pðO2 Þ 12 k3 K v k1V      3 Fg 1 Fg  exp
exp
2 RT 2 RT

ð9Þ

Under low polarization (gbRT/F), one can calculate polarization resistance as Rg = g/I by taking linear part of the exponential members. For the model expressed by Eq. (9), Rg is proportional to p(O2)
5/12. The slope of the experimental dependence (Fig. 12), equal to
0.33, is quite close to the model value of
0.41. Another approximation, based on the assumption that the concentration of surface active centers is p(O2)-independent, gives the slope of
0.375. In general, by selecting a more complex exchange mechanism or by considering another rate-determining steps, one can easily obtain the values of (1/n) in the range

335

from (
1/12) to (
1/2); representative examples derived in a similar manner can be found in Refs. [37,39]. The data on Rg vs. p(O2) dependence cannot, therefore, be used to unambiguously identify a single process limiting exchange currents. Nevertheless, these results suggest that the ratelimiting steps are related to the gas/electrode interface; if the oxygen exchange kinetics would be limited by ion diffusion in the electrode bulk, a slope close to (
1/6) could be supposed [17]. 3.6. Cathode performance: effects of microstructure and surface activation As expected, increasing the electrode fabrication temperature from 1473 to 1523 K resulted in grain growth and agglomeration (Fig. 7B and C). This leads to considerably worse electrochemical activity (inset in Fig. 13). Although such a behavior might be partly ascribed to decreasing triple-phase boundary (TPB) length, the most likely reason relates to decreasing specific surface area and, thus, overall exchange currents. Therefore, the performance of La2Ni0.8 Cu0.2O4 + d cathodes can be improved by reducing fabrication temperature, by further decrease of the grain size, and by the incorporation of electrocatalytically active components onto electrode and electrolyte surfaces. Fig. 14 compares overpotential vs. current dependencies for one La2Ni0.8Cu0.2O4 + d layer after preparation at 1473 K and after surface modification with praseodymium oxide; the corresponding microstructures are shown in Fig. 7B and D. SEM/EDS, TEM/EDS and XRD analyses, performed before and after electrochemical tests, confirmed that praseodymia is distributed mainly in pores and on the surface of electrode and electrolyte; the interaction between La2Ni0.8Cu0.2O4 + d and PrOx phases was below the detection limits. In particular, the lattice parameters of La2Ni0.8Cu0.2O4 + d remain essentially unchanged after surface activation and testing during 150– 200 h. At the same time, surface interaction between these phases cannot be excluded. As for La(Sr)MnO3-based cathodes [27], the activation drastically increases electrochemical activity, especially at reduced temperatures. For instance, at 873 K the cathodic overpotential is decreased by approximately two times, from 330 down to 175 mV at 50 mA/cm2. Possible nature of such an enhancement in electrode performance was discussed earlier [27]. In fact, similar to other electrochemical data on mixed-conducting cathodes, this phenomenon cannot be unambiguously attributed to a single factor; the activation may lead to a faster surface exchange of both solid electrolyte and electrode, to an enlargement of TPB and electrode surface area, and even to an improvement of transport properties of the porous electrode layer due to significant mixed conductivity of praseodymium oxide (Refs. [27,40] and references cited). The latter is indicated, in particular, by the substantial decrease in the ohmic resistance after the activation as illustrated by Fig. 2. Most likely, a combi-

336

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337

is limited by the surface exchange rate. The significant mixed conductivity leads to a relatively high performance of porous La2Ni0.8Cu0.2O4 + d cathodes, screen-printed onto (La0.9Sr0.1)0.98Ga0.8Mg0.2O3
d (LSGM) solid electrolyte and annealed at 1473 –1523 K. The cathodic overpotential of La2Ni0.8Cu0.2O4 + d layers at 1073 K and current density of 200 mA/cm2 was lower than 50 mV. As for the oxygen transport through dense ceramics, the results on the electrode behavior suggest a considerable role of surface-related processes, such as oxygen sorption or discharge of oxygen ad-atoms on the electrode surface, with the active site concentration dependent on current. Due to this, the performance of La2Ni0.8Cu0.2O4 + d electrodes can be substantially enhanced by surface modification via impregnation with Prcontaining solutions, and also by decreasing fabrication temperature leading to greater surface area. Fig. 15. Dependence of the polarization resistance of La2Ni0.8Cu0.2O4 + d electrode, as-prepared at 1473 K and after surface modification with PrOx, on the direct current passed through electrode. Solid lines are for visual guidance.

nation of all these factors is responsible for the enhanced electrochemical activity. Nonetheless, the data on surfacemodified La2Ni0.8Cu0.2O4 + d layers are in agreement with the assumption that their performance is influenced by surface-related processes. This hypothesis is also supported by the dependence of polarization resistance on the cathodic current density (Fig. 15). For non-activated La2Ni0.8Cu0.2O4 + d layer, the values of Rg calculated from impedance spectra decrease with increasing current, which well corresponds to the relationship between cathodic and anodic overpotentials (Fig. 13). Again, this behavior is very similar to literature data [27,31,32] on La(Sr)MnO3
d and La(Sr)CoO3
d electrodes, where cathodic polarization leads to formation of oxygen vacancies on the surface, contributing to the electrochemical activity enhancement. After the surface activation of La2Ni0.8Cu0.2O4 + d layer, the polarization resistance becomes current-independent (Fig. 15). One may suggest that, due to an increase in the concentration of active centers on the electrode surface, the role of this factor decreased.

4. Conclusions The submicron powder of Cu-substituted lanthanum nickelate, La2Ni0.8Cu0.2O4 + d, with K2NiF4-type structure and grain size of 30 –60 nm, was synthesized using glycine –nitrate process and used for preparation of porous electrode layers and ceramics with 98% density. In air, La2Ni0.8Cu0.2O4 + d exhibits predominant p-type electronic conductivity varying in the range of 50– 85 S/cm at 800 – 1300 K, thermal expansion coefficient of 13.3  10
6 K
1 at 400 – 1240 K, and oxygen permeability higher than that of another group of prospective cathode materials, La(Sr)Fe (Co)O3
d. At temperatures below 1173 K, the permeation

Acknowledgements This work was partially supported by FCT, Portugal (projects BD/6827/2001 and BPD/11606/2002), INTAS (project 00276), and NATO Science for Peace program (project 978002). The authors are sincerely grateful to Aliaksandr Shaula for experimental assistance. References [1] T. Ishihara, H. Matsuda, Y. Takita, J. Am. Chem. Soc. 116 (1994) 3801. [2] N. Maffei, A.K. Kuriakose, J. Power Sources 75 (1996) 162. [3] F. Chen, M. Liu, J. Solid State Electrochem. 3 (1998) 7. [4] K. Huang, J. Wan, J.B. Goodenough, J. Mater. Sci. 36 (2001) 1093. [5] T. Ishihara, M. Honda, T. Shibayama, H. Minami, H. Nishiguchi, Yu. Takita, J. Electrochem. Soc. 145 (1998) 3177. [6] K. Huang, M. Feng, J.B. Goodenough, C. Milliken, J. Electrochem. Soc. 144 (1997) 3620. [7] K. Huang, M. Feng, J.B. Goodenough, M. Schmerling, J. Electrochem. Soc. 143 (1996) 3630. [8] R. Maric, S. Ohara, T. Fukui, H. Yoshida, M. Nishimura, T. Inagaki, K. Miura, J. Electrochem. Soc. 146 (1999) 2006. [9] V.V. Kharton, E.N. Naumovich, P.P. Zhuk, A.K. Demin, A.V. Nikolaev, Russ. J. Electrochem. 28 (1992) 1376. [10] 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. [11] H. Hayashi, M. Suzuki, H. Inaba, Solid State Ionics 128 (2000) 131. [12] B. Vigeland, R. Glenne, T. Breivik, S. Julsrud, Int. Patent Application PCT WO 99/59702 (1999).. [13] V.V. Vashook, I.I. Yushkevich, L.V. Kokhanovsky, L.V. Makhnach, S.P. Tolochko, I.F. Kononyuk, H. Ullmann, H. Altenburg, Solid State Ionics 119 (1999) 23. [14] S.J. Skinner, J.A. Kilner, Solid State Ionics 135 (2000) 709. [15] F. Mauvy, J.M. Bassat, E. Boehm, P. Dordor, J.P. Loup, Solid State Ionics 158 (2003) 395. [16] V.V. Kharton, A.P. Viskup, E.N. Naumovich, F.M.B. Marques, J. Mater. Chem. 9 (1999) 2623. [17] V.V. Kharton, A.P. Viskup, A.V. Kovalevsky, E.N. Naumovich, F.M.B. Marques, Solid State Ionics 143 (2001) 337. [18] A.A. Yaremchenko, V.V. Kharton, M.V. Patrakeev, J.R. Frade, J. Mater. Chem. 13 (2003) 1136.

V.V. Kharton et al. / Solid State Ionics 166 (2004) 327–337 [19] V.V. Kharton, E.N. Naumovich, A.V. Nikolaev, V.V. Astashko, A.A. Vecher, Russ. J. Electrochem. 29 (1993) 1039. [20] E.N. Naumovich, V.V. Kharton, V.V. Samokhval, A.V. Kovalevsky, Solid State Ionics 93 (1997) 95. [21] V.V. Kharton, A.L. Shaula, N.P. Vyshatko, F.M.B. Marques, Electrochim. Acta 48 (2003) 1817. [22] P. Schenck (Ed.), Phase Equilibria Diagrams. CD-Rom Database, The American Ceramic Society, Westerville, OH, 1998. [23] K.K. Singh, P. Ganguly, J.B. Goodenough, J. Solid State Chem. 52 (1984) 254. [24] L.A. Chick, L.R. Pederson, G.D. Maupin, J.L. Bates, L.E. Thomas, G.J. Exarhos, Mater. Lett. 10 (1990) 6. [25] F.M. Figueiredo, J. Frade, F.M.B. Marques, Bol. Soc. Esp. Ceram. Vidr. 38 (1999) 639. [26] J. Mizusaki, H. Tagawa, J. Electrochem. Soc. 141 (1994) 1674. [27] V.N. Tikhonovich, V.V. Kharton, E.N. Naumovich, A.A. Savitsky, Solid State Ionics 106 (1998) 197. [28] V.V. Kharton, A.A. Yaremchenko, E.V. Tsipis, J.R. Frade, in: S.C. Singhal, M. Dokiya (Eds.), SOFC-VIII, The Electrochemical Society, Pennington, NJ, 2003, pp. 561 – 570, PV2003-07. [29] V.V. Kharton, A.A. Yaremchenko, A.L. Shaula, M.V. Patrakeev, E.N. Logvinovich, D.I. Logvinovich, J.R. Frade, F.M.B. Marques, J. Solid State Chem. (2003) (in press).

337

[30] V.V. Kharton, A.V. Nikolaev, E.N. Naumovich, A.A. Vecher, Solid State Ionics 81 (1995) 201. [31] H.Y. Lee, W.S. Cho, S.M. Oh, H.-D. Wiemho¨fer, W. Go¨pel, J. Electrochem. Soc. 142 (1995) 2659. [32] G. Reinhardt, V. Baitinger, W. Go¨pel, Ionics 1 (1995) 504. [33] A.L. Shaula, A.P. Viskup, V.V. Kharton, D.I. Logvinovich, E.N. Naumovich, J.R. Frade, F.M.B. Marques, Mater. Res. Bull. 38 (2003) 353. [34] V.V. Kharton, A.P. Viskup, E.N. Naumovich, V.N. Tikhonovich, Mater. Res. Bull. 34 (1999) 1311. [35] A.A. Yaremchenko, V.V. Kharton, A.P. Viskup, E.N. Naumovich, N.M. Lapchuk, V.N. Tikhonovich, J. Solid State Chem. 142 (1999) 325. [36] S.B. Adler, J.A. Lane, B.C.H. Steele, J. Electrochem. Soc. 143 (1996) 3554. [37] F.H. van Heuveln, H.J.M. Boumeester, J. Electrochem. Soc. 144 (1997) 134. [38] H. Kanai, J. Mizusaki, H. Tagawa, S. Hoshiyama, K. Hirano, K. Fujita, M. Tezuka, T. Hashimoto, J. Solid State Chem. 131 (1997) 150. [39] M.V. Perfiliev, A.K. Demin, B.L. Kuzin, A.S. Lipilin, High-Temperature Electrolysis of Gases, Nauka, Moscow, 1988. In Russian. [40] V.V. Kharton, E.N. Naumovich, A.A. Vecher, J. Solid State Electrochem. 3 (1999) 61.