Acta mater. 48 (2000) 4709–4714 www.elsevier.com/locate/actamat

GADOLINIA-DOPED CERIA AND YTTRIA STABILIZED ZIRCONIA INTERFACES: REGARDING THEIR APPLICATION FOR SOFC TECHNOLOGY
A. TSOGA1*, A. GUPTA1, A. NAOUMIDIS1 and P. NIKOLOPOULOS2
Institut fur Werkstoffe und Verfahren der Energietechnik (IWV1), Forschungszentrum Julich, D-52425 ¨ ¨ Julich, Germany and 2Chemical Engineering Department, University of Patras, GR 265 00 Patras, Greece ¨
1

Abstract—For solid oxide fuel cells (SOFCs) operating at intermediate temperatures the adjacency of the state-of-the-art yttria-stabilized zirconia (YSZ) electrolyte with ceria-based materials to both anodic and cathodic sides is regarded as crucial for the effectiveness of the cell. Solid-state reaction, however, and interdiffusion phenomena between YSZ and ceria-based materials can cause degradation of the electrolyte. When a gadolinia-doped-ceria (GDC) layer is used to protect YSZ against interaction with Co-containing cathodes, an unfavorable solid state reaction at the YSZ–GDC interface can be efficiently suppressed when a thin ( 1 µm thick) interlayer with nominal composition of Ce0.43Zr0.43Gd0.10Y0.04O1.93 is incorporated at the interface. When ceria is to be employed at the electrolyte–anode interface to reduce polarization losses, use of a ceria–40% vol Ni cermet is recommended, since suppression of the reactivity between YSZ and ceria can also be achieved in the presence of Ni. © 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved. Keywords: Solid oxide fuel cells; Interface; Diffusion; Microstructure

1. INTRODUCTION

Reduction of the operation temperature of solid oxide fuel cells (SOFCs) from 900–1000°C to 700–800°C is of great importance because it means both a prolonged stack lifetime and a cost reduction, since the use of low-cost metallic components as separator materials is then possible. However, for the operation of SOFCs at intermediate temperature to be technically feasible two parameters should be considered: the development of high-performance electrodes, because the electrode reaction rates decrease at such temperatures, and minimization of cell resistance. The latter means minimization of both the ohmic losses inside the electrolyte and the polarization losses at the electrolyte–electrode interfaces. It is known that La(Sr)CoO3-based perovskites (LSC), when sputtered on the yttrium-stabilized zirconia (YSZ) electrolyte, exhibit higher cathodic performance than state-of-the-art La0.85Sr0.15MnO3 (LSM) cathode material [1, 2] and lower polarization

* To whom all correspondence should be addressed at: CERECO, 72nd km of the national road, Athens-Lamia, PO Box 146, 34 100 Chalkida, Greece. Tel.: 30-262-71226; Fax: 30-262-71461. E-mail address: a.tsoga@chemeng.upatras.gr (A. Tsoga)

values also at intermediate temperatures. However, LSC tends to react with YSZ, forming isolating reaction products such as La2Zr2O7 or SrZrO3 [3, 4]. The only materials chemically compatible with LSC are those based on CeO2 [5] which, although they possess a higher ionic conductivity than YSZ, cannot be used as electrolytes because under a fuel gas atmosphere they are prone to develop electronic conductivity, resulting from the reduction of Ce4 to Ce3 . As a solution to this problem, consideration is being given to the use of a CeO2-based interlayer between a thin YSZ electrolyte and LSC electrode. From the anodic side, when CeO2-based materials are employed at the electrolyte–anode interface they significantly decrease polarization losses and enhance the performance of the cell [6, 7]. From these findings it is obvious how beneficial the presence of CeO2based materials is on both sides of the YSZ electrolyte. However, the chemical compatibility between YSZ and CeO2-based materials is not without problems, since the two materials react and diffuse into each other during the sintering process at 1200°C [8, 9]. Figure 1(a) shows the microstructure of the YSZ– GDC interface after sintering at 1400°C for 4 h in air, conditions usually used for the sintering of the electrolyte in SOFC technology. The atomic distributions of Zr, Y, Ce and Gd across the interface [Fig.

1359-6454/00/$20.00 © 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 9 - 6 4 5 4 ( 0 0 ) 0 0 2 6 1 - 5

4710

TSOGA et al.: CERIA AND YTTRIA STABILIZED ZIRCONIA

and elemental distribution across the interfaces under consideration are examined with the aid of scanning electron microscopy and electronic probe microanalysis.
2. EXPERIMENTAL PROCEDURE

Fig. 1. (a) Microstructure of the YSZ–GDC interface after sintering in air at 1400°C for 4 h. (b) Elemental distributions of Zr, Ce, Y, and Gd across the interface shown in (a).

1(b)] suggest the involvement of enhanced solid state reaction and interdiffusion phenomena, resulting in the formation of an interaction zone between the materials, which is most extensive in the zirconia part of the sample. According to the results of previous work [8, 9], the composition of the reaction zone formed at the interface is governed by the attainment of a chemical equilibrium, which imposes the formation of a reaction product enriched in Gd with a nominal composition of Ce0.37Zr0.38Gd0.18Y0.07O1.87, exhibiting at 800°C ionic conductivity lower by two orders of magnitudes than YSZ. Emerging porosity can also be recognized on the ceria side, resulting from the differences in diffusivity of the counter-diffusing cations (Kirkendahl effect). The aim of the present work is to examine how solid state reaction and interdiffusion phenomena at the interfaces of YSZ in contact with both gadolinia doped ceria (GDC) and a ceria-doped YSZ–Ni cermet can be suppressed or even avoided. For the YSZ– GDC interface the contribution of a compositiongraded microstructure is studied. The interaction between YSZ and the ceria-doped anode cermet is studied as a function of ceria content in the anode cermet for a Ni content of 40% vol. Microstructure

The GDC powder used, with a composition of Ce0.8Gd0.2O1.9, was synthesized by a strike co-precipitation method using 0.05 M oxalic acid as the precipitant and crystallized in methanol at 200°C [10]. The YSZ powder used was the commercially available 8 mol% Y2O3-stabilized ZrO2 powder (TZ-8Y, Tosoh) with a purity >99.9%, ball-milled for 120 h. The solid-solution phase with nominal composition Ce0.43Zr0.43Gd0.10Y0.04O1.93, used as interlayer between YSZ and GDC, was synthesized by the glycine-nitrate combustion technique, a combustion synthesis method that is particularly useful for synthesizing ultrafine, multicomponent oxide powders [11]. The samples used to investigate solid-state reaction and interdiffusion phenomena at the interfaces under consideration were prepared by subsequent screenprinting of the corresponding layers on a YSZ–40 vol% Ni cermet substrate. Sintering of the samples was performed in a conventional resistance furnace in air and at 1400°C for 4 h. A JEOL JSM 6300 scanning electron microscope (SEM) equipped with an Oxford Link pentafet energy dispersive X-ray analyzer (EDS) as well as with a Microspec WDX 600 wave dispersive X-ray analyzer (WDS) was used for overall microstructural and compositional analysis. Samples for SEM and EDS were mounted in epoxy in cross-sectional orientation, polished using standard metallographic techniques and coated with a thin film of carbon to provide electrical conductivity. AC impedance was measured by an impedance analyzer (SI 1260 Solatron) from 1 mHz to 100 kHz and 10 mV amplitude. The measured AC impedance spectra were fit to appropriate equivalent circuit models using a pattern fitting software program (EQUIVCRT, by B. Boukamp, University of Twente).
3. RESULTS

3.1. YSZ–GDC interface To avoid extended solid-state reaction and interdiffusion phenomena at the YSZ–GDC interface, the concept of a microstructure with a graded composition was tested. As an interlayer between YSZ and GDC a solid solution of the single materials was used with the nominal composition Ce0.43Zr0.43Gd0.10 Y0.04O1.93. Its composition was optimized towards ionic conductivity, for a Ce/Zr atomic ratio equal to unity [11], in order to act as a diffusion barrier between YSZ and GDC. Figure 2 illustrates the microstructure across the interface after sintering at 1400°C in air for 4 h [Fig. 2(a)] as well as the atomic

TSOGA et al.: CERIA AND YTTRIA STABILIZED ZIRCONIA

4711

Chemical compatibility was directly correlated with the Ce atomic fraction, expressed as Ce/(Ce Zr), across the electrolyte–anode interface, as illustrated in Fig. 3. In the presence of Ni the migration of the Ce component into the YSZ lattice appeared to be effectively suppressed. After sintering at 1400°C in air for 4 h and reduction at 900°C in Ar 4 vol% H2, the Ce atomic fractions measured inside the first micrometers of the YSZ electrolyte were lower than 0.2. Even lower Ce atomic fractions were found with decreasing ceria content and thickness of the ceriacontaining anode function layer.
4. DISCUSSION

4.1. GDC as a protective layer between YSZ electrolyte and LSC cathode In order for the YSZ electrolyte to be effectively protected against reaction with LSC-based cathode electrodes, it should be coated with a dense GDC thin layer. However, extensive solid-state reaction and interdiffusion phenomena occurring during sintering at the GDC–YSZ interface cause the performance of the electrolyte to deteriorate for two reasons. These are formation of solid solutions at the interface, which exhibit up to two orders of magnitude lower ionic conductivity than YSZ [8], and formation of porosity, which results from the higher diffusion rates of Ce and Gd cations inside the YSZ lattice compared with that of the counter-diffusing Zr and Y cations. Taking into account the Ce atomic fractions, expressed as Ce/(Ce Zr), across the interface shown in Fig. 1(a), the electrical conductivity profile inside the electrolyte can be estimated, based on published values of the electrical conductivity of solid solutions between GDC and YSZ [11] with the same Ce atomic fraction

Fig. 2. (a) Microstructure of the YSZ–Ce0.43Zr0.43Gd0.10Y0.04 O1.93–GDC interface after sintering in air at 1400°C for 4 h. (b) Elemental distributions of Zr, Ce, Y, and Gd across the interface shown in (a).

distributions of Zr, Ce, Y, and Gd across the interface [Fig. 2(b)]. Compared with the elemental distribution across the YSZ–GDC interface shown in Fig. 1(b), the gradation in composition appears to succeed in effectively suppressing the solid state reaction and interdiffusion phenomena at the interface of the single materials, avoiding the formation of the reaction product observed in the case of the YSZ–GDC interface and additionally eliminating the effect of pore formation. 3.2. YSZ–GDC interface in the presence of Ni Both YSZ and ceria are known to exhibit good chemical compatibility with Ni. Therefore, solid state reaction and interdiffusion phenomena between YSZ and ceria are expected to be suppressed in the vicinity of Ni, depending on the quantity of Ni present. For a Ni-content of 40 vol%, which is the usual amount used for the anode electrode in SOFC technology, the chemical compatibility between the YSZ electrolyte and the ceria-doped (YSZ–Ni) cermet, used as an anode function layer, was studied as a function of the ceria doping content and compared with that when pure ceria is employed as an anode function layer.

Fig. 3. Variation of the Ce atomic fraction, expressed as Ce/(Ce Zr) atomic ratio, across the YSZ electrolyte–anode function layer interface, where the following compositions were used as anode function layers: 5 µm pure ceria, 5 µm ceria–40 vol% Ni cermet, 5 µm (30 wt% ceria–70 wt% YSZ)– 40 vol% Ni cermet, 2 µm (30 wt% ceria–70 wt% YSZ)–40 vol% Ni cermet and 5 µm (10 wt% ceria–90 wt% YSZ)–40 vol% Ni cermet, indicated as “pure ceria 5µm”, “Ce100 5µm”, “Ce30 5µm”, “Ce30 2µm” and “Ce10 5µm”, respectively.

4712

TSOGA et al.: CERIA AND YTTRIA STABILIZED ZIRCONIA

Fig. 4. Variation of the Ce atomic fraction, expressed as Ce/(Ce Zr) atomic ratio and of the total electrical conductivity in air at 800°C across the interface shown in Fig. 1(a).

Fig. 6. Variation of the Ce atomic fraction, expressed as Ce/(Ce Zr) atomic ratio, and of the total electrical conductivity in air at 800°C across the interface shown in Fig. 2(a).

(Fig. 4). The decrease observed in the electrical conductivity corresponds to the formation of a reaction product from YSZ and GDC. If we consider a reaction product zone of 1 µm, a significant deterioration of the performance of the cell can be observed (line 3 in Fig. 5), compared with that when there is no interaction between YSZ and GDC (line 1 in Fig. 5). The I–V curves shown in Fig. 5 were calculated for a temperature of 800°C, considering only the ohmic resistance of a layered electrolyte, regarded as a mixed conductor. The calculation procedure was performed on the assumption that the same current passes through each of the electrolyte’s layers, taking into account both the ionic and the electronic conductivity of each layer of the electrolyte. A more detailed description of the calculation procedure is given in Appendix A. Following the same procedure as before, the electrical conductivity profile inside the electrolyte of Fig.

2(a) with the 1 µm Ce0.43Zr0.43Gd0.10Y0.04O1.93 interlayer is illustrated in Fig. 6. Although the interlayer used exhibits a lower electrical conductivity than the single YSZ and GDC layers, it is still one order of magnitude higher than that of the reaction product formed at the YSZ–GDC interface when no protective interlayer is involved. Further, because its thickness is rigorously controlled and specified during sample preparation, a better electrolyte performance can be achieved (line 2 in Fig. 5). Verification of the above theoretically determined behavior of a cell using a composite electrolyte was performed by measuring the AC impedance of thick YSZ (100 µm) as well as YSZ (75 µm)–GDC (25 µm) and YSZ (75 µm)-interlayer (5 µm)–GDC (20 µm) composite electrolytes. Figure 7 shows the electrolyte area-specific resistance R, obtained from impedance spectra plotted against 1/T. Also shown for comparison are the R-values calculated for the electrolytes under consideration based on the reported conductivity data. The measured values for the single YSZ electrolyte are very close to the calculated ones. However, this is not the case for the bilayer YSZ– GDC electrolyte, where the measured area-specific resistance values were about three to five times higher

Fig. 5. Variation of the terminal voltage of a cell based on a composite electrolyte with total thickness of 25 µm as a function of current density when no reaction is considered between YSZ (20 µm) and GDC (5 µm) (line 1), when a 1 µm Ce0.43Zr0.43Gd0.10Y0.04O1.93 interlayer is incorporated between YSZ (19.5 µm) and GDC (4.5 µm) (line 2) and when a 1 µm reaction product zone has been formed between YSZ (19.5 µm) and GDC (4.5 µm) (line 3). Calculation conditions: T 800 °C; fuel, H2 97 vol%–H2O 3 vol%; oxidant, air.

Fig. 7. The measured electrolyte area-specific resistance R of thick YSZ (100 µm) as well as YSZ (75 µm)–GDC (25 µm) and YSZ (75 µm)-interlayer (5 µm)–GDC (20 µm) composite electrolytes plotted against 1/T and compared with the theoretically expected ones, predicted based on the conductivity data reported in Table 1.

TSOGA et al.: CERIA AND YTTRIA STABILIZED ZIRCONIA
Table 1. Materials and their electrical properties at 800°C, where sel Material YSZ GDC Reaction product Interlayer Nominal composition Zr0.85Y0.15O1.93 Ce0.80Gd0.20O1.90 Ce0.37Zr0.38Gd0.18Y0.07O1.87 Ce0.43Zr0.43Gd0.10Y0.04O1.93 0.054 0.087 0.00125 0.00603 soP(O2) n
1/4

4713
[11] sno (S/cm) 7.29 8.18 3.99 3.88 10 10 10 10
13 6 6 6

sion (S/cm)

than expected. According to the above analysis, this results from the formation of the low-conductive reaction product at the YSZ–GDC interface. Use of an interlayer between YSZ and GDC produced lower area-specific resistance values than did the bilayer YSZ–GDC electrolyte, proving its beneficial contribution. More pronounced improvement is expected using a thin-layer electrolyte. 4.2. Presence of ceria at the electrolyte–anode interface The use of ceria in the vicinity of the electrolyte– anode interface has been shown to reduce polarization losses. This is because under low oxygen partial pressures ceria exhibits a mixed conductivity, thus promoting charge-transfer reactions over the whole area of the electrode–gas interface. The scope of the present work was to examine chemical compatibility between the YSZ electrolyte and an anode function layer consisting of either pure ceria or ceria-doped YSZ–Ni cermets with different ceria doping contents. The problems associated with a possible chemical incompatibility between the YSZ electrolyte and the ceria-containing anode function layer are mostly related to the migration of ceria into the YSZ lattice. If this is the case, reduction of Ce4 to Ce3 in the fuel gas atmosphere can cause expansion of the YSZ crystal lattice, and consequently thermal expansion mismatch inside the electrolyte, owing ˚ to the higher ionic radius of Ce3 (1.143 A) compared 4 ˚ ) [12]. Based on the Ce atomic diswith Zr (0.84 A tributions presented in Fig. 3, this can occur when pure ceria is used as an anode function layer. In the presence of Ni, the Ce atomic fractions inside the electrolyte were suppressed to a level lower than 0.2, even when a CeO2–Ni cermet was used.
5. CONCLUSIONS

Acknowledgements—AT is a TMR grant holder.

APPENDIX A

Based on Wagner’s theory [13] the local charge fluxes Iion and Iel are functions of the corresponding partial conductivity si, the electrochemical potential gradient hi and the valence Zi according to the relationship: Ii A(si/ZiF) hi, i:ion, el (A1)

where F is Faraday’s constant and A is the cross-sectional area of the electrolyte. Choudhury and Patterson [14] showed that for a mixed conductive electrolyte under steady-state conditions, the local charge fluxes Iion due to ions and Iel due to electrons should be individually independent of the location and therefore the ratio: r Iion/Iel (A2)

which is used to characterize steady-state conditions inside the electrolyte because of fixed P(O2)1 and P(O2)2 values and constant temperature, with r 1 for open circuit condition and |r|>1 at SOFC operation condition. When every basic equation is then written as a function of the oxygen partial pressure, the ionic current density is then given as: Iion RTsion∂h(O2) 4F ∂x
P2(O2

rIel

(A3)

) )

RT 4FL
P1(O2

rsionsel d ln P(O2) (rsel sion)

Solid-state reaction and interdiffusion phenomena at the YSZ–GDC interface can be efficiently suppressed when a thin (1 µm thick) interlayer with nominal composition Ce0.43Zr0.43Gd0.10Y0.04O1.93, acting as a diffusion barrier, is incorporated at the interface between the single materials. Suppression of the reactivity between YSZ and ceria can also be achieved in the presence of Ni. When ceria is to be employed at the electrolyte–anode interface to reduce polarization losses it is recommended that it should be used as a ceria–Ni cermet.

where L is the thickness of the electrolyte. Each value of the parameter r corresponds to a value of external current density Iext, and terminal voltage Vt, whose quantitative formulae may be derived by appropriate integration as functions of the partial conductivity, the parameter r and the oxygen partial pressures on both sides of the mixed conductor, assuming steady-state conditions and negligible polarization outside the electrolyte:

4714

TSOGA et al.: CERIA AND YTTRIA STABILIZED ZIRCONIA

Iext (1 r)RT 4FL

Iion
P2(O2

Iel )

(A4) The ionic and electronic conductivities of the materials involved at 800°C are summarized in Table 1.

sionsel d ln P(O2) (rsel sion) )

P1(O2

P2(O2

)

REFERENCES 1. Ohno, Y., Nagata, S. and Sato, H., Solid State Ionics, 1981, 3/4, 439. 2. Yamamoto, O., Takeda, Y., Kanno, R. and Noda, M., Solid State Ionics, 1987, 22, 241. 3. Kawada, T., Sakai, N., Yokokawa, H., Dokiya, M. and Ansai, I., Solid State Ionics, 1989, 50, 189. 4. Syskakis, E., Stochniol, G., Naoumidis, A., Nickel, H., in Proceedings of the 2nd International Conference on Ceramics in Energy Applications, The Institute of Energy, London, UK, 1996, p. 91. 5. Godickermeier, M. and Gauckler, L. J., J. Electrochem. Soc., 1998, 145, 414. 6. Uchida, H., Arisaka, S. and Watanabe, M., Electrochem. Solid-State Lett., 1999, 2, 428. 7. Marina, O. A., Primdahl, S., Bagger, C. and Mogensen, M., in Proceedings of the 5th International SOFC Symposium, Vol. 40, ed. U. Stimming. The Electrochemical Society, 1997, p. 540. 8. Tsoga, A., Gupta, A., Naoumidis, A., Skarmoutsos, D. and Nikolopoulos, P., Ionics, 1999, 4, 234. 9. Tsoga, A., Naoumidis, A., Gupta, A. and Stover, D., ¨ Materials Science Forum 1999, 308–311, 794. 10. Tsoga, A., Naoumidis, A., Jungen, W. and Stover, D., ¨ Journal of the European Ceramic Society, 1999, 19, 907. 11. Tsoga, A., Naoumidis, A., and Stover, D., Solid State Ion¨ ics, 1999, 5, 175. 12. Shannon, R. D., Acta Cryst., 1976, A32, 751. 13. Wagner, C., Z. Physik. Chem., 1933, B21, 25. 14. Choudhury, N. S. and Patterson, J. W., J. Electrochem. Soc., 1971, 118, 1398.

sion Vt d ln P(O2) (A5) (rsel sion) P1(O2) When considering a multilayer composite electrolyte, the current density through the external circuit, IL, is equal to the current density of each of the layers, calculated for each one using Equation (A4): RT 4F IL Iext,i (A6)

Considering equation (A6) for all the layers of the electrolyte the equation system we use is then solved for fixed values of IL on condition that oxygen partial pressure inside the electrolyte is continuously reduced from P1(O2) 0.21 atm on the oxidant side to the oxygen partial pressure on the fuel side, corresponding to that of an H2 gas with 3% H2O. The calculations were performed at 800°C. The oxygen partial pressures at the interface are then used to determine the terminal voltage VTOT of the cell given by the sum of the partial voltage at the ends of each layer: n VTOT
1

Vt,i

(A7)