Solid State Ionics 128 (2000) 117–130 www.elsevier.com / locate / ssi
Faradaic efficiency and oxygen permeability of Sr 0.97 Ti 0.60 Fe 0.40 O 32d perovskite a, a a b b V.V. Kharton *, A.V. Kovalevsky , A.P. Viskup , F.M. Figueiredo , J.R. Frade , A.A. Yaremchenko a , E.N. Naumovich a a
Institute of Physicochemical Problems, Belarus State University, 14 Leningradskaya Str., 220080 Minsk, Byelorussia b Department of Ceramics and Glass Engineering, University of Aveiro, 3810 Aveiro, Portugal Received 19 July 1999; received in revised form 16 December 1999; accepted 20 December 1999
Abstract Oxygen ionic conduction in the perovskite-type Sr 0.97 Ti 0.60 Fe 0.40 O 32d was studied using oxygen permeability, Faradaic efficiency and total electrical conductivity measurements at 973–1223 K. The ion transference numbers of the strontium titanate–ferrite in air vary from 0.005 to 0.08, decreasing with decreasing temperature. The electron–hole conductivity of the oxide is relatively low but exceeds the ionic conductivity. The activation energy for the electronic conductivity is 3563 kJ / mol at 470–890 K and drops at higher temperatures. Studying the oxygen permeation through dense Sr 0.97 Ti 0.60 Fe 0.40 O 32d ceramic membranes as a function of membrane thickness showed that at temperatures above 1170 K the permeation fluxes are limited by both bulk ionic conductivity and surface exchange rates. Depositing of porous layers of the same material or a mixture of platinum and praseodymium oxide onto the membrane feed-side surface leads to a significant increase in the oxygen permeability. Decreasing temperature results in increasing role of the bulk ionic transport in Sr 0.97 Ti 0.60 Fe 0.40 O 32d as the permeation-determining factor. The oxygen permeation fluxes at 1073 K are limited predominantly by the oxygen ionic conductivity of the ceramics. The thermal expansion coefficients of the ceramic material in air were calculated from dilatometric data to be 11.7 3 10 26 K 21 in the temperature range 300–720 K and 16.6 3 10 26 K 21 at 720–1070 K. 2000 Elsevier Science B.V. All rights reserved. Keywords: Perovskite; Oxygen permeation; Faradaic efficiency; Ionic conductivity; Strontium titanate–ferrite
1. Introduction Mixed ionic–electronic conductors with ABO 3 perovskite structure, derived from ATiO 32d (A 5 Sr, *Corresponding author. Present address: Department of Ceramics and Glass Engineering, University of Aveiro, 3810 Aveiro, Portugal. Tel.: 1351-234-370-263; fax: 1351-234-425300. E-mail address:
[email protected] or:
[email protected] (V.V. Kharton)
Ca, Ba) by partial substitution of titanium with iron or cobalt and by incorporation of rare-earth cations into the A sublattice, are of considerable interest for electrochemical applications such as ceramic membranes for oxygen separation and partial oxidation of hydrocarbons, electrodes of solid oxide fuel cells (SOFCs), and sensors [1–15]. This is associated with a significant oxygen-ion mobility, stability under a wide range of oxygen chemical potentials, significant values of p-type electronic conductivity at high
0167-2738 / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0167-2738( 00 )00275-7
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
118
oxygen partial pressures and n-type conductivity under reducing environments. The electronic contribution to the total conductivity of the titanate-based solid solutions significantly exceeds, as a rule, the ionic contribution [1,3,6–9,12]. Typically, both oxygen ionic and electron–hole conduction increases with increasing concentration of the transition-metal dopant in the B sublattice; a higher oxygen conductivity is due to both greater oxygen deficiency and weaker B–O bonds [1,6,7,9,13]. As a result, oxygen permeation fluxes through SrTi 12x B x O 32d (B 5 Fe, Co) ceramic membranes increase regularly with increasing x [6,9]. Oxygen transport through strontium titanate-based ceramics was found to be limited by both bulk ionic conduction and surface exchange rates at the oxide / gas phase boundaries [6,15]. Among the main disadvantages of the alkaline– earth element-containing perovskites as materials of electrochemical cells, one should mention their reactivity with gas species such as carbon dioxide [3,9,16,17], which are present in atmospheric air and in products of hydrocarbon oxidation. Creation of a moderate cation deficiency in the A sublattice was proposed as a possible method to improve the stability of such materials [3,9,17]. The present work was focused on studying transport and physicochemical properties of Sr 0.97 Ti 0.60 Fe 0.40 O 32d perovskite from the viewpoint of possible applications in high-temperature electrochemical cells.
synthesis route using high-purity SrCO 3 , TiO 2 and FeC 2 O 4 ? 2H 2 O as starting materials. After thermal decomposition of the metal salts, the stoichiometric mixture of the oxides was annealed at 1270–1470 K in air for 35 h with multiple intermediate regrindings. X-ray diffraction (XRD) analysis showed that the obtained powder was single phase. Ceramic samples were pressed (300–500 MPa) in the shape of bars (4 3 4 3 30 mm 3 ) and disks of various thickness (diameter, 10–20 mm). Gas-tight ceramics were sintered in air at 1520–1540 K during 8–20 h. After sintering, the samples were annealed in air at 1170 K for 10–15 h and then slowly cooled in order to obtain oxygen nonstoichiometry values close to the equilibrium ones at room temperature. The density of the ceramics, determined by the standard picnometric technique, was 96.5% of the theoretical density calculated from the XRD data (Table 1). The prepared samples were characterized using XRD, X-ray fluorescence analysis (XFA), emission spectroscopic analysis, differential thermal analysis (DTA), thermal gravimetry (TGA), and infrared (IR) absorption spectroscopy. The experimental procedures used for the characterization as well as techniques of testing gas tightness, measuring electrical conductivity and thermal expansion were described in detail elsewhere [6,8,18–22]. Only gas-tight ceramic samples were used for the oxygen permeation and Faradaic efficiency measurements.
2.2. Oxygen permeability measurements Steady-state oxygen permeation fluxes were measured with an experimental technique based on a yttria-stabilized zirconia (YSZ) solid-electrolyte electrochemical cell, consisting of an oxygen pump and a sensor, as described in detail earlier [6,8,9,18,20]. The oxygen permeability was studied at 973–1223 K in the range of oxygen partial
2. Experimental
2.1. Synthesis and characterization Preparation of powdered Sr 0.97 Ti 0.60 Fe 0.40 O 32d (STF40) was performed by the standard ceramic Table 1 Properties of the Sr 0.97 Ti 0.60 Fe 0.40 O 32d ceramics Lattice parameter A (nm) 0.3898 a
Density rexp (kg / m 3 ) 4895
rexp /rtheor a (%) 96.5
rtheor is the theoretical density of the ceramics calculated from the XRD data.
Thermal expansion coefficients T (K)
a¯ 310 6 (K 21 )
300 – 720 720–1070
11.760.2 16.6060.07
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
pressures at the membrane permeate-side ( p1 ) from 0.1 to 1.5 3 10 4 Pa. For all results presented in this paper, the feed-side oxygen pressure ( p2 ) was maintained to be equal 2.1 3 10 4 Pa (atmospheric air). The thickness of the membrane specimens (d) was varied from 1.0 to 2.0 mm. For analysis of the permeation processes, we use hereafter the quantities of the oxygen permeation flux density (mol s 21 cm 22 ) and specific oxygen permeability (mol s 21 cm 21 ) defined by the formula [23]:
F G
p2 J(O 2 ) 5 jd ln ] p1
21
(1)
The oxygen partial pressure at the membrane permeate side was calculated from e.m.f. (E) values of the electrochemical sensor incorporated into the measuring cell:
S D
p2 RT E 5 ] ln ] 4F p1
(2)
The quantity J(O 2 ) is convenient in order to identify a limiting effect of the surface exchange rate on the oxygen permeation, on the basis of the thickness dependence of the permeation flux [8,24]. As the oxygen permeability is proportional to j 3 d by definition, J(O 2 ) should be independent of thickness if surface limitations to the oxygen permeation flux were negligible. In this case, the physical meaning of the oxygen permeability quantity should refer to the ambipolar conductivity (samb ) of the membrane material, averaged for a given oxygen partial pressure range: so se RT RT ]] J(O 2 ) 5 ]]2 ]] samb 5 ]]2 ]]] 16F 16F so 1 se RT ]]] 5 ]]2 s t o (1 2 t o ) 16F
119
functions of oxygen chemical potential (see, for example, Refs. [6–8,18–20]). Therefore, the values of J(O 2 ) are presented hereafter in combination with the corresponding values of p1 and p2 . Some of the ceramics were characterized by a prolonged process of steady-state attainment. After starting the experiments, the oxygen flux through such membranes decreases, as illustrated by Fig. 1. Taking this transient regime into account, we used time independence of the sensor e.m.f. during 20–25 h (with a drift of 1% under constant current through the oxygen pump), as a criteria for the steady-state attainment. The initial time of stabilization was 50– 150 h during the first application of an oxygen chemical potential gradient across the sample. Subsequent measurements required much smaller transient times (in the range 3–15 h). More detailed studies are necessary in order to identify reasons for such transient behaviour.
2.3. Faradaic efficiency studies The experimental technique of Faradaic efficiency measurements, performed in order to determine oxygen-ion transference numbers, has been described elsewhere [25]. For this purpose, a YSZ electrochemical cell (Fig. 2), similar to the cell for the
(3)
where t o is the oxygen ion transference number, and s, so and se are the total, oxygen ionic and electronic conductivities, respectively. In case interphase oxygen exchange limitations are considerable, J(O 2 ) should increase with increasing membrane thickness due to a decreasing role of the surface exchange, for a given oxygen chemical potential gradient. It should be noted that both ionic conductivity and exchange currents of perovskite-related oxides are
Fig. 1. Time dependence of oxygen permeability of STF40 membranes at 1223 K after placing membranes under oxygen pressure gradient: (1) d51.00 mm, (2) d51.28 mm.
120
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
permeation flux through the membrane was first determined at the required temperature and permeate-side oxygen pressure. A direct current (Iout ) was passed through the pump, removing oxygen from the cell, while the current through the membrane (Iin ) was zero. After the sensor e.m.f. (E) became time-independent, the permeation flux (J, mol / s) for the given conditions (T and p1 ) is proportional to the current through the oxygen pump: J 5 Iout (4F )21
(4)
Then a direct current (Iin ) was passed through the sample in order to pump oxygen into the cell, and Iout was adjusted to provide the same value of E, independent of time. In this case the sum of the oxygen fluxes, driven through membrane by the electrical and chemical potential differences (Jin and J, respectively), is equal to the flux through the pump: Iout (4F )21 5 Jin 1 J 5 t o Iin (4F )21 1 J
(5)
Oxygen ion transference numbers at the given E value can be, therefore, determined as Iout 2 4FJ t o 5 ]]] Iin Fig. 2. Schematic drawing of the cell for the Faradaic efficiency measurements: (1) YSZ electrolyte, (2) electrodes of the oxygen sensor, (3) electrodes of the oxygen pump, (4) specimen under test, (5) high-temperature sealant, (6) porous ceramic insertions, (7) thermocouple, (8) Pt-electrodes applied onto the specimen, (9) furnace.
oxygen permeation studies, was used. Dense ceramic membranes with applied porous Pt-electrodes were sealed onto the cell. Oxygen was pumped through the cell before the sample was sealed in order to avoid diffusion limitations of oxygen transport in the gas phase within the cell, which may appear when significant concentrations of atmospheric nitrogen are present. Once the sensor signal had reached values corresponding to oxygen pressure close to 1 atm, the cell was heated and the ceramic membrane was then sealed. The feed-side oxygen partial pressure during Faradaic efficiency measurements was 2.1310 4 Pa, as in the oxygen permeation studies. In the course of the measurements, the oxygen
(6)
The oxygen ion transference numbers in air were measured under the condition E¯0
(7)
In this case the oxygen fluxes through the membrane and the pump are equivalent Iout (4F )21 5 Jin 5 t o Iin (4F )21
(8)
and t o 5 Iout /Iin
(9)
These measurements were performed in the temperature range 1073–1223 K. The time necessary for steady-state attainment was similar to that for the steady oxygen permeation flux measurements. As mentioned for the permeability measurements, slow surface exchange might also influence Faradaic efficiency measurement results. This effect will be later addressed in this paper.
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
121
3. Results and discussion
3.1. Characterization of the material XRD analysis of the prepared powder and ceramics of Sr 0.97 Ti 0.60 Fe 0.40 O 32d showed that all the samples are single phase. The crystal structure was identified as cubic perovskite, and the unit cell parameter (a) is given in Table 1. Only reflections characteristic for the cubic perovskite phase were observed in the XRD patterns. No considerable changes in the lattice parameter or in the phase composition were found after keeping the samples in the ambient conditions for 20–30 days. Fig. 3 presents SEM micrographs illustrating the microstructures of STF40 ceramics. The fracture micrograph of an as-sintered STF40 sample is given in Fig. 3A. The sample shown in Fig. 3B was fired at 1673 K, polished and then thermally treated at 1570 K. These micrographs demonstrate that the prepared ceramic materials are sufficiently dense and consist of grains of sizes in the range 5–10 mm. Liquid phase formation at the grain boundaries is observed after thermal etching at 1570 K, thus indicating that one should avoid excessively high sintering temperatures. Dilatometric curves of STF40 ceramics can be approximated by two straight lines with a break at approximately 720 K (Fig. 4). Thermal expansion coefficients (TECs) calculated by fitting the dilatometric data, are 11.7310 26 K 21 at 300–720 K and 16.6310 26 K 21 at higher temperatures (Table 1). The relatively high TEC of Sr 0.97 Ti 0.60 Fe 0.40 O 32d at temperatures above 720 K may prevent its combination with solid electrolytes of stabilized ZrO 2 , doped CeO 2 or LaGaO 3 in high-temperature electrochemical cells. The operation temperatures of the cells with these electrolytes vary typically from 870 to 1270 K, whereas TECs of these solid-electrolyte materials are in the range of (9.2–13.5)310 26 K 21 . However, Bi 2 O 3 -based solid electrolytes such as fluorite-type solid solutions Bi(M)O 1.5 (M is the rare-earth or transition metal such as Nb) or BIMEVOX series [26,27], have TEC compatible with STF40. The temperature dependence of the total electrical conductivity of STF40 is presented in Fig. 5. A transition to the metallic-type conductivity was ob-
Fig. 3. SEM micrographs of the fracture of sintered STF40 sample (A), and the ceramic surface after polishing and thermal treatment at 1570 K in air (B).
served at temperatures above 1100 K. This transition, and the conductivity values are in close agreement with preliminary results [3]. The activation energy for the electrical conductivity (Ea ) was calculated using the standard Arrhenius model:
S
A0 Ea s 5 ] exp 2 ] T RT
D
(10)
where A 0 is the pre-exponential factor. Fitting parameters of temperature dependence of the conductivity of STF40 using Eq. (10) as a regression model are listed in Table 2. The activation energy values were found to be 35 kJ / mol at 470–890 K
122
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130 Table 2 Parameters of the regression model Eq. (10) for the temperature dependence of the electrical conductivity of Sr 0.97 Ti 0.60 Fe 0.40 O 32d ceramics in air
Fig. 4. Temperature dependence of relative elongation of STF40 ceramics in air. Solid line corresponds to the fitting results using the linear regression model in two temperature ranges.
T (K)
Ea (kJ / mol)
ln(A 0 ) (S / cm)
470–890 890–1120
3563 1161
11.560.6 8.260.1
and approximately 11 kJ / mol in the temperature range 890–1120 K. It should be noted that the adequacy of the Arrhenius model for describing the conductivity of STF40 as a function of temperature is relatively poor (Fig. 5), suggesting a more complex conduction mechanism. In particular, the regression parameters of this model (Ea and A 0 ) in the high-temperature range where the transition to the metallic conductivity is observed, can be used only for calculations of the electrical conductivity values; no definite physical meaning can be ascribed to these quantities. On analyzing the possible use of Sr 0.97 Ti 0.60 Fe 0.40 O 32d in the electrochemical devices, one should mention that the electrical conductivity of this material is relatively low. In particular, the conductivity values of STF40 are lower than those of other known electrode materials such as lanthanum–strontium cobaltites (La,Sr)CoO 32d or solid solutions SrCo(M)O 32d (M5Fe, Cu) by 10– 1000 times [18,22,24]. Therefore, the perovskite Sr 0.97 Ti 0.60 Fe 0.40 O 32d seems to be unlikely for use in electrodes, and further modifications are necessary in order to increase its conductivity.
3.2. Results of the Faradaic efficiency measurements
Fig. 5. Temperature dependence of the specific electrical conductivity of STF40 ceramics in air. Solid line corresponds to the fitting results using the regression model Eq. (10) in two temperature ranges.
Oxygen ion transference numbers of Sr 0.97 Ti 0.60 Fe 0.40 O 32d determined from the Faradaic efficiency measurements using the cell drawn schematically in Fig. 2 and a membrane with the thickness of 1 mm, are adduced in Table 3. Within the experimental error, the estimated transference numbers were observed to be essentially independent of both current density and permeate-side oxygen partial pressure varied in the range 1310 3 to 2.13
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
123
Table 3 Oxygen ion transference numbers of Sr 0.97 Ti 0.60 Fe 0.40 O 32d determined by the Faradaic efficiency measurements T (K)
E (mV)
Current density through membrane Iin /S a (mA / cm 2 )
to
1223
0 15.5 24.8 35.9 55.9 0 40.3 40.3 61.9 0 54.0 67.8 67.8 0
60.1 52.2 49.1 50.6 58.5 60.1 36.8 52.2 52.2 88.2 52.2 52.2 78.9 153.6
0.041 0.042 0.040 0.044 0.043 0.033 0.031 0.031 0.032 0.0090 0.0074 0.0104 0.0095 0.0050
1173
1073
973 a
S is the membrane surface area.
10 4 Pa. The t o values decrease regularly with decreasing temperature, caused by the higher activation energy for ionic conduction as compared to that for electronic conductivity. In the studied temperature range (973–1223 K), the oxygen ion transference numbers in air vary from 0.005 to 0.04. Thus, the title material represents a mixed ionic–electronic conductor with predominant electronic conductivity. Fig. 6 presents the observed values of the oxygen ionic conductivity of Sr 0.97 Ti 0.60 Fe 0.40 O 32d in air, calculated from the results of Faradaic efficiency and total electrical conductivity measurements
s obs o 5 tos
(11)
as a function of temperature. These results are significantly lower than the ionic conductivity of the well-known Zr 0.92 Y 0.08 O 1.96 solid electrolyte [28]. The activation energy for the ionic conductivity of Sr 0.97 Ti 0.60 Fe 0.40 O 32d in air (98612 kJ / mol) is close to that of ytrria-stabilized zirconia. Note that the values of ionic conductivity of STF40 extracted from Faradaic efficiency experiments are also lower than the estimates of ambipolar conductivity samb 5 s t o (12t o ) obtained from previous permeability experiments [9,15]. Those values of
Fig. 6. Temperature dependence of the oxygen ionic conductivity of Sr 0.97 Ti 0.60 Fe 0.40 O 32d in air: (1) calculated from the results of Faradaic efficiency (membrane thickness of 1.00 mm) and electrical conductivity measurements; (2) calculated from the results on the electrical conductivity and oxygen permeability (d51.00 mm) at 1073 K when no surface exchange limitations to the permeation flux has been found; (3) calculated from the results on the electrical conductivity and oxygen permeability (d52.00 mm) at 1073 K; (4) results of fitting using the Arrhenius model. Curve (5) corresponds to the conductivity data of Zr 0.92 Y 0.08 O 1.96 solid electrolyte from Badwal [28].
samb were close to the ionic conductivity of the YSZ solid electrolyte. Possible reasons for such behavior are discussed below. The ion transference number increases after annealing the STF40 membrane at 970 K under a high oxygen pressure gradient, corresponding to the values of the oxygen sensor e.m.f. of approximately 780 mV (Table 4); this also caused an increase in the resistance, thus showing that the increase in ion transference number is probably due to a decrease of the dominant p-type electronic conductivity. Further pumping of oxygen into the measuring cell and annealing of the membrane in air at 1220 K for 3–4 days did not result in changing transport properties. The observed changes were thus irreversible. However, XRD analysis of this membrane after the Faradaic efficiency experiments showed no decomposition of the perovskite phase, and a more
124
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
Table 4 Effect of keeping under a high oxygen potential gradient on the oxygen transference numbers of Sr 0.97 Ti 0.60 Fe 0.40 O 32d at 1223 K Pre-history
E (mV)
Current density Iin /S (mA / cm 2 )
to
After sealing and attainment of the steady state
0 15.5 24.8 35.9 55.9
60.1 52.2 49.1 50.6 58.5
0.041 0.042 0.040 0.044 0.043
After keeping at E5780620 mV at 973 K during 24 h
36 49.7 50.5
56.9 72.8 88.2
0.075 0.062 0.062
detailed study is necessary in order to clarify reasons for such changes in behaviour.
3.3. Oxygen permeability The values of oxygen permeation fluxes and specific oxygen permeability of Sr 0.97 Ti 0.60 Fe 0.40 O 32d are given in Figs. 7–9 as functions of temperature, oxygen partial pressure and membrane thickness. The effects of thickness on the oxygen permeability showed unambiguously that the permeation at temperatures above 1170 K is limited by both bulk ambipolar conductivity and surface exchange rates. Indeed, the values of J(O 2 ) at 1173– 1223 K increase with increasing membrane thickness, whereas permeation fluxes were observed to decrease with increasing d. As oxygen ionic conductivity is significantly lower than electronic conductivity (Table 3), one can conclude that the ionic conduction is the limiting factor of bulk oxygen transport. Therefore, the permeation fluxes at high temperatures for a membrane thickness below 2 mm are limited by both surface exchange at the gas / oxide boundaries and bulk ionic conductivity of STF40. Such a conclusion is in agreement with previous results [15]. Applying porous layers of either the same composition or a mixture of dispersed platinum and praseodymium oxide onto the membrane feed-side surface leads to a considerable increase in the permeation fluxes (Figs. 7 and 8), caused by higher surface exchange rates. The details on deposition of porous layers of STF40 were reported elsewhere [15]. The layer of a mixture of platinum and praseodymium oxide (sheet density of 11.4 mg / cm 2 )
Fig. 7. Dependence of the oxygen permeation flux density (A) and specific oxygen permeability (B) of the STF40 membranes on the oxygen pressure differential at 1223 K and p2 50.21 atm: (1) d51.00 mm; (2) d52.00 mm; (3) d51.00 mm, membrane with the Pt / PrO x layer applied onto the feed-side surface; (4) d51.75 mm, membrane with the porous STF40 layer applied onto the feed-side surface.
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
Fig. 8. Dependence of the oxygen permeation flux density (A) and specific oxygen permeability (B) of the STF40 membranes on the oxygen pressure differential at 1223 K and p2 50.21 atm: (1) d51.28 mm; (2) d52.00 mm; (3) d51.00 mm, membrane with the Pt / PrO x layer applied onto the feed-side surface; (4) d51.75 mm, membrane with the porous STF40 layer applied onto the feed-side surface.
was made using a paste, containing highly dispersed Pt (about 70 wt%) and Pr 6 O 11 (approximately 30 wt%) with addition of organic binder. The layer was deposited onto the feed-side surface of the STF40 membrane and then annealed at 1270–1320 K for 2 h. Decreasing temperature down to 1073 K leads to a change in the permeation flux-limiting factors. At this temperature, the values of specific oxygen permeability J(O 2 ) are essentially independent of thickness within the experimental error limits, while oxygen fluxes increase regularly with decreasing d (Fig. 9). This suggests that the permeation fluxes at the membrane thickness above 1 mm are limited mainly by the bulk ionic conduction. The most
125
Fig. 9. Dependence of the oxygen permeation flux density (A) and specific oxygen permeability (B) of the STF40 membranes on the oxygen pressure differential at 1223 K and p2 50.21 atm: (1) d51.00 mm; (2) d51.28 mm; (3) d52.00 mm.
probable reason for such behaviour is that the activation energy for the ionic conductivity might be higher than the activation energy for oxygen interphase exchange. The greater values of the activation energy for ionic conduction in comparison with that of oxygen exchange is typical for numerous perovskite-type oxides, such as La 0.3 Sr 0.7 CoO 32d [29] or La 0.6 Sr 0.4 Fe 0.8 CoO 32d [30]. The fact that the oxygen permeation through STF40 membranes at 1073 K is determined by the bulk ionic transport, was confirmed by the estimations of ionic conductivity obtained from oxygen permeability and electrical conductivity measurements. The values of the ionic conductivity at oxygen pressures close to that of atmospheric air (Fig. 6) were obtained using the formula derived from Eqs. (1) and (3) [19,22]:
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
126
S D
d ≠Iout s t o (1 2 t o ) 5 ] ]] S ≠E
E →0
(12)
where S is the membrane surface area. The calculated values of so are sufficiently close to that found from the Faradaic efficiency measurement results (Fig. 6), substantiating thus the conclusion regarding the ionic conduction to be the flux-limiting factor.
3.4. Relationship between parameters determined from the permeation and Faradaic efficiency results The values of the oxygen ion transference numbers in air calculated from the results of the oxygen permeability and Faradaic efficiency measurements are adduced in Fig. 10. To process the data on the Faradaic efficiency, the formula Eq. (9) was used. Estimating the transference numbers on the basis on oxygen permeation data was performed using the quantity of observed ambipolar conductivity:
S D
d ≠Iout ]] s obs amb 5 ] S ≠E
E →0
]]] obs s amb 1 1 t o 5 ] 2 ] 1 2 4]] 2 2 s
œ
(14)
At 1073 K, the values of t o determined by the different techniques are close to each other. Increasing temperature results in greater transference numbers estimated from the oxygen permeation data as compared to the Faradaic efficiency; such difference increases with increasing thickness of the studied membranes. This behaviour can be explained in terms of the effect of the surface exchange limitations, as described below. Schematic representation of the transport processes in a mixed-conducting membrane, placed under oxygen chemical potential gradient, and corresponding equivalent circuit is given in Fig. 11. In a simplified case when all the transport parameters are
(13)
Fig. 10. Temperature dependence of the oxygen ion transference numbers in air, estimated from the results of Faradaic efficiency measurements (curve 1) and oxygen permeation studies [2–6]. Thickness of the membranes: (1,2,5) 1.00 mm; (3) 1.28 mm; (4) 2.00 mm; (6) 1.75 mm. Curves 5 and 6 correspond to the samples with porous layers of Pt / PrO x and STF40, respectively.
Fig. 11. Schematic representation of a mixed-conducting membrane placed under an oxygen chemical potential gradient (A) and the equivalent circuit for the corresponding transference process (B).
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
127
independent of the oxygen pressure, the permeation flux density expressed in electrical units can be written as j4F 5 f R o S 1 R e S 1 R h g 21 ? E
F
d d 1 5 ]1]1] so se Kex
G
21
?E
so se Kex 5 ]]]]]]]E dse Kex 1 dso Kex 1 so se
(15)
where R o and R e are the partial ionic and electronic resistances, respectively, R h is the sum of the areaspecific polarization resistances at the membrane permeate and feed sides (V?cm 2 ), Kex is the reciprocal R h being the coefficient of proportionality between the oxygen flux density through gas / oxide boundaries and the sum of oxygen chemical potential differences across these boundaries. Therefore, the observed transference numbers which can be calculated by Eq. (14) in case of non-negligible surface exchange limitations relate to the true values of t o as t
obs o
1 1 ]]]]] 5 ] 2 ]œ1 2 4t o (1 2 t o )s Z 2 2
(16)
Fig. 12. Schematic representation of a mixed-conducting membrane placed under an electrical potential gradient (A) and the corresponding equivalent circuit (B).
where dKex Z 5 ]]]]]]] dse Kex 1 dso Kex 1 so se
(17)
As a result, true values of the oxygen ion transference numbers can be obtained from the oxygen permeation data under condition, at least, of either negligible surface exchange limitations (Kex 4 so , se ) or sufficiently thick samples. The similar case for the Faradaic efficiency measurements is illustrated in Fig. 12. Here, the equations for the total current (Iin ) and for the oxygen flux expressed in electrical units (Iout ) can be written as
F
1 1 Iin 5 ]]]] 21 1 ] Re Ro 1 RhS
G
?U
tests, and R h corresponds to the classical definition of the polarization resistance. Substituting Eqs. (18) and (19) into Eq. (9), one obtains t obs o 5 tosZ
Thus, the values of transference numbers determined from the Faradaic efficiency data can be also affected by the oxygen surface exchange (polarization resistance). Analogously to the oxygen permeation results, this effect can be negligible only if the partial ionic resistance is much lower than the polarization resistance, or if the thickness of the sample is sufficiently large. In this case Z ¯ s 21
dse Kex 1 dso Kex 1 so se 5 ]]]]]]] ? SU d 2 Kex 1 dso
(18)
so Kex U Iout 5 ]]]] ? SU 21 5 ]]] dK Ro 1 RhS ex 1 so
(19)
where U is the voltage applied to the sample under
(21)
(22)
Note that different oxygen discharge conditions and even different discharge mechanisms in permeation (Fig. 11A) and Faradaic efficiency (Fig. 12B) cells can lead to different oxygen exchange rate currents (expressed by Kex ). With regard to the results presented in Fig. 10, the
128
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
difference between ionic transference numbers determined by the different methods at 1073 K is within the experimental error when oxygen permeability values are independent of the membrane thickness. This fact stresses the accuracy and reproducibility of these data in the low-temperature range (T #1073 K). At higher temperatures, the transference numbers estimated from the two techniques results are close to each other only when the thickness of the samples is the same. Increasing the membrane thickness results in the increase of the t obs o values. In this case, one can only conclude about an approximate order of magnitude of the ionic transference numbers. At 1123 K, the t 0 value was thus estimated to be between 0.04 and 0.08. Higher ionic transference numbers are very unlikely because the e.m.f., measured on the STF40 membrane in the cell shown in Fig. 2 at zero applied voltage, was negligible even at high values of the zirconia sensor e.m.f. (¯100 mV). One should also mention the possible different effects of the ceramic microstructure on the estimates of oxygen ion transference numbers obtained by the different techniques. For instance, the presence of closed porosity in a ceramic sample on the course of a permeation experiment can be interpreted in terms of a reduction of the effective membrane thickness. In this case, the ion transference numbers would be overestimated (see Eqs. (13) and (14)). For the Faradaic efficiency studies, the presence of closed porosity would be similar to a decrease in effective surface area, resulting in increase in both R o and R e ; the effect of the porosity on measured transference numbers should be rather negligible. Therefore, estimates of the transference parameters (ionic conductivity, transference numbers) determined by oxygen permeability and Faradaic efficiency measurements should be verified. The investigation of the dependence of the measured quantities on the membrane thickness and comparison of results obtained by different techniques is suggested for this goal.
3.5. Reactivity with CO2 Preliminary tests demonstrated the stability of Sr 0.97 Ti 0.60 Fe 0.40 O 32d phase in atmospheres containing CO 2 at temperatures above 770 K. For example,
Fig. 13 presents TGA results of STF40 powders in different gas mixtures. Before the tests, all the samples have been annealed in air at 1070–1120 K in the TGA cell. After such pre-treatment, they were slowly cooled to the temperature of the experiments and tested for 10–12 h in CO 2 -containing atmospheres. Within the experimental error limits, no interaction with carbon dioxide was found by TGA at 770–1070 K. DTA results of the powders, kept in an atmosphere of CO 2 for various times, also showed no thermal effects in the heating mode. This suggests that adsorption of CO 2 on the surface of STF40 does not result in considerable decomposition of the perovskite phase. On the other hand, IR absorption studies of the STF40 powders indicated the presence of traces of
Fig. 13. Relative weight changes of STF40 in different atmospheres at 1073 K (A), 885 K (B) and 780 K (C): (1) atmospheric air purified from CO 2 , (2) mixture of CO 2 , O 2 and N 2 . For (A) and (B) the mixture consisted of 14 vol.% O 2 , 18 vol.% CO 2 and 68 vol.% N 2 . For (C) the composition of the mixture was 19 vol.% O 2 , 17 vol.% CO 2 and 64 vol.% N 2 . Before the TGA tests, the samples were annealed in CO 2 -free air at 1073 K (A,B) and at 1120 K (C).
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
SrCO 3 after the powder had been kept in the atmosphere of carbon dioxide at room temperature (Fig. 14). In particular, the IR spectra exhibit an absorption band at |1060 cm 21 , which belongs unambiguously to strontium carbonate [31]. The intensity of this peak increases with increasing time of exposure of STF40 powders in the CO 2 atmosphere. Therefore, surface decomposition of the perovskite phase may still take place at low temperatures, caused by the interaction of Sr 0.97 Ti 0.60 Fe 0.40 O 32d with carbon dioxide. However, bulk samples did not show any evidence of degradation after several months, and a more detailed investigation is thus desirable in order to
Fig. 14. IR absorption spectra of STF40 at room temperature: (1) after annealing in air at 1453 K for 2 h, (2) after being kept in a CO 2 atmosphere at room temperature for 8 h, (3) after being kept in a CO 2 atmosphere at room temperature for 50 h.
129
establish exact limits of the perovskite phase stability.
4. Conclusions Phase-pure samples of perovskite-type Sr 0.97 Ti 0.60 Fe 0.40 O 32d were prepared by the standard ceramic synthesis route with a density of 96.5%, as related to the theoretical density. The thermal expansion coefficient of Sr 0.97 Ti 0.60 Fe 0.40 O 32d ceramics in air was calculated from the dilatometric data to increase from 11.7310 26 K 21 in the temperature range 300–720 K to 16.6310 26 K 21 at 720–1070 K. The electron–hole conductivity of the oxide is relatively low but dominates the ionic conduction. A transition to the metallic-type conductivity was observed at temperatures above 1120 K. The activation energy for the electronic conductivity, calculated using the standard Arrhenius model, is 3563 kJ / mol in the range 470–890 K. Oxygen ion transference numbers of Sr 0.97 Ti 0.60 Fe 0.40 O 32d were determined from the results of the Faradaic efficiency, oxygen permeability and total electrical conductivity measurements. The ion transference numbers in air vary from 0.005 to 0.08 at 973–1223 K, increasing with temperature. The values of the oxygen ionic conductivity of the strontium titanate–ferrite are significantly lower than the conductivity of yttria-stabilized zirconia solid electrolytes at temperatures below 1100 K. The results of studying oxygen permeation through Sr 0.97 Ti 0.60 Fe 0.40 O 32d ceramic membranes as a function of membrane thickness showed that at temperatures above 1170 K the permeation fluxes are limited by both bulk ionic conductivity and surface exchange rates. Deposition of porous layers of the same material or a mixture of platinum and praseodymium oxide onto the membrane feed-side surface leads to a significant increase in the oxygen permeability. Decreasing temperature results in increasing role of the bulk ionic transport as the permeation-determining factor. At 1073 K, the oxygen permeation fluxes are limited predominantly by the oxygen ionic conductivity of the ceramic material. TGA tests in different atmospheres showed that at
130
V.V. Kharton et al. / Solid State Ionics 128 (2000) 117 – 130
temperatures above 770 K the perovskite phase of Sr 0.97 Ti 0.60 Fe 0.40 O 32d is stable with respect to the reaction with carbon dioxide. However, exposure of STF40 powders to a CO 2 atmosphere at lower temperatures may result in a surface decomposition of the perovskite, associated with the formation of strontium carbonate.
Acknowledgements This research was partially supported by the FCT, Portugal (Praxis program and Project P/ CTM / 14170 / 1998), the Belarus Foundation for Basic Research, the Belarus State University, and the company B-2 Limited.
References [1] H. Iwahara, Solid State Ionics 52 (1992) 99. [2] S. Sekido, H. Tachibara, Y. Yamamura, T. Kambara, Solid State Ionics 37 (1990) 253. [3] L.S.M. Traqueia, J.R. Jurado, J.R. Frade, in: J.L. Baptista, J.A. Labrincha, P.M. Vilarinho (Eds.), Proc. Int. Conf. ‘Electroceramics V’, Aveiro, Portugal, 1996, Vol. 2, pp. 151–154. [4] H. Itoh, H. Asano, K. Fukuroi, M. Nagata, H. Iwahara, J. Am. Ceram. Soc. 80 (1997) 1359. [5] I. Denk, F. Noll, J. Maier, J. Am. Ceram. Soc. 80 (1997) 279. [6] V.V. Kharton, L. Shuangbao, A.V. Kovalevsky, E.N. Naumovich, Solid State Ionics 96 (1997) 141. [7] V.P. Gorelov, V.B. Balakireva, Elektrokhimiya 33 (1997) 1450, (in Russian). [8] V.V. Kharton, L. Shuangbao, A.V. Kovalevsky, A.P. Viskup, E.N. Naumovich, A.A. Tonoyan, Mater. Chem. Phys. 53 (1998) 6. [9] J.R. Jurado, F.M. Figueiredo, B. Charbage, J.R. Frade, Solid State Ionics 118 (1999) 89. [10] T. Kawada, T. Watanabe, A. Kaimai, K. Kamamura, Y. Nigara, J. Mizusaki, Solid State Ionics 108 (1998) 391. [11] T.J. Mazanec, T.L. Cable, J.G. Frye, W.R. Kliewer, US Patent 5,744,015 (1998).
[12] S. Xie, W. Liu, K. Wu, P.H. Yang, G.Y. Meng, C.S. Chen, Solid State Ionics 118 (1999) 23. [13] L.A. Dunyushkina, A.K. Demin, B.V. Zhuravlev, Solid State Ionics 116 (1999) 85. [14] S.R. Song, H.-I. Yoo, Solid State Ionics 120 (1999) 141. [15] J.R. Jurado, F.M. Figueiredo, J.R. Frade, Solid State Ionics 122 (1999) 197. [16] M.F. Carolan, P.N. Dyer, J.M. LaBar, R.M. Thorogood, US Patent 5,261,932 (1993). [17] M.F. Carolan, P.N. Dyer, S.A. Motika, P.B. Alba, US Patent 5,712,220 (1998). [18] V.V. Kharton, E.N. Naumovich, A.A. Vecher, A.V. Nikolaev, J. Solid State Chem. 120 (1995) 128. [19] A.A. Yaremchenko, V.V. Kharton, A.P. Viskup, E.N. Naumovich, N.M. Lapchuk, V.N. Tikhonovich, J. Solid State Chem. 142 (1999) 225. [20] V.V. Kharton, V.N. Tikhonovich, L. Shuangbao, E.N. Naumovich, A.V. Kovalevsky, A.P. Viskup, I.A. Bashmakov, A.A. Yaremchenko, J. Electrochem. Soc. 145 (1998) 1363. [21] V.V. Kharton, A.P. Viskup, E.N. Naumovich, A.A. Tonoyan, O.P. Reut, Mater. Res. Bull. 33 (1998) 1087. [22] V.V. Kharton, E.N. Naumovich, A.V. Nikolaev, V.V. Astashko, A.A. Vecher, Russ. J. Electrochem. 29 (1993) 1039. [23] H.-H. Moebius, Extend. Abstr. 37th Meet. ISE, Vilnius, USSR, 1986, Vol. 1, p. 136. [24] A.V. Kovalevsky, V.V. Kharton, V.N. Tikhonovich, E.N. Naumovich, A.A. Tonoyan, O.P. Reut, L.S. Boginsky, Mater. Sci. Eng. B 52 (1998) 105. [25] V.V. Kharton, A.P. Viskup, E.N. Naumovich, N.M. Lapchuk, Solid State Ionics 104 (1997) 67. [26] E.N. Naumovich, V.V. Kharton, V.V. Samokhval, A.V. Kovalevsky, Solid State Ionics 93 (1997) 95. [27] A.A. Yaremchenko, V.V. Kharton, E.N. Naumovich, V.V. Samokhval, Solid State Ionics 111 (1998) 227. [28] S.P. Badwal, Solid State Ionics 52 (1992) 23. [29] R.H.E. van Doorn, I.C. Fullarton, R.A. de Souza, J.A. Kilner, H.J.M. Bouwmeester, A.J. Burggraaf, Solid State Ionics 96 (1997) 1. [30] S.J. Benson, R.J. Chater, J.A. Kilner, in: T.A. Ramanarayanan (Ed.), Ionic and Mixed Conducting Ceramics III, Proceedings Vol. 97–24, The Electrochemical Society, Pennington, NJ, 1998, p. 596. [31] IR spectra of inorganic compounds, in: R.A. Nyquist, R.O. Kagel (Eds.), Handbook of Infrared and Raman Spectra of Inorganic Compounds and Organic Salts, Vol. 4, Academic Press, San Diego, London, Boston, NY, Sydney, Tokyo, Toronto, 1997.
