[10]electrical Properties In Lasrtio A Potential Anode

Solid State Ionics 159 (2003) 159 – 165 www.elsevier.com/locate/ssi

Electrical properties in La2Sr4Ti6O19
d: a potential anode for high temperature fuel cells J. Canales-Va´zquez, S.W. Tao, J.T.S. Irvine * School of Chemistry, University of St. Andrews, Fife KY16 9ST, UK Received 2 September 2002; received in revised form 25 November 2002; accepted 20 December 2002

Abstract La2Sr4Ti6O19
d has been investigated as a potential anode for fuel cells due to the high total conductivity found under reducing conditions. This mixed oxide is the n = 6 member of the excess oxygen perovskite-related family La2Srn
2TinO3n + 1. The structure of this family of compounds can be described as perovskite slabs joined by crystallographic shears where the characteristic excess oxygen of these mixed oxides is accommodated. Phases such as La2Sr4Ti6O19
d could be considered as a potential oxygen ion or proton conductor due to the significant amount of interstitial oxygen found in both reduced and oxidised forms. Partial removal of the excess oxygen by reduction of Ti4 + might lead to an enhancement of the ionic conductivity together with electronic conductivity due to the presence of Ti3 +. The electrical properties of this material have been studied in a range of oxygen and water partial pressure revealing the important role played by d, i.e. the amount of Ti3 +, on these phases. Under the most reducing conditions, metallic conductivity, e.g. 60 S cm
1, is observed and under slightly higher P(O2), e.g. wet hydrogen, a metal to insulator transition is observed. In addition, initial fuel cell tests were carried out to check the performance of La2Sr4Ti6O19
d as an anode for fuel cells. Using La2Sr4Ti6O19
d as an anode, the polarisation resistance (Rp) varies from 2.97 V cm2 at 900 jC in wet H2 to 8.93 V cm2 at 900 jC operating in wet CH4. A current value of 119 mA cm
2 at 600 mV was found, whereas the maximum power density was 76 mW cm
2 both measured in wet H2 at 900 jC. D 2003 Elsevier Science B.V. All rights reserved. Keywords: La2Sr4Ti6O19
d; Temperature; Anode

1. Introduction Ni-YSZ cermets are currently used as anodes in fuel cells because they meet most of the requirements: mixed conductivity, thermal and chemical stability at operating conditions and also catalytic activity to promote the oxidation of the fuel. There exist several

* Corresponding author. Tel.: +44-1334-463817; fax: +44-1334463808. E-mail address: [email protected] (J.T.S. Irvine).

problems associated with the use of this cermet however, such as carbon build-up when operating with natural gas, sulphur poisoning and long-term degradation related to the agglomeration of Ni particles [1,2]. Mixed oxide materials offer a potential improvement over YSZ cermets, especially due to the ability of transition metals to present multiple oxidation states. The possibility of multiple oxidation states might enhance the electronic conductivity and also facilitates catalytic activity. In addition, mixed oxides are less likely to suffer from carbon build-up or sulphur poisoning than YSZ cermets [3].

0167-2738/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0167-2738(03)00002-X

160

J. Canales-Va´zquez et al. / Solid State Ionics 159 (2003) 159–165

Characteristics such as high electronic conductivity, high ionic conductivity and good catalytic activity, as well as mixed protonic and electronic conduction have been reported in many perovskite-mixed metal oxides [4– 7]. Such characteristics are an essential requirement for high performance fuel cell anode materials. Some of these properties may open up possibilities with respect to fuel cells operating at lower temperature (600 jC) than in current technologies. La2Srn
2TinO3n + 1 is the general formula for a family of layered perovskites, whose main feature is their ability to accommodate excess oxygen in comparison to the parent primitive perovskite [8]. La2Ti2O7 is one of the end members (the n = 2 member) of this family whereas the archetype of the perovskite structure, SrTiO3, is the other end member. The structure of La2Ti2O7 has been widely investigated previously by several authors [9 – 12] and it can be described as perovskite slabs joined by crystallographic shears where the excess oxygen is accommodated (Fig. 1). As n increases, the perovskite slabs become bigger until for n = l (SrTiO3), the structure is just a perovskite framework. According to the structural model proposed by Bowden et al. [13] for La-doped SrTiO3, the structure of La2Sr4Ti6O19
d could be described as La2Ti2O7-like planes randomly distributed within a perovskite framework. Perovskite phases of stoichiometry LaxSr1
xTiO3 have frequently been reported as being prepared in air in the literature [14,15,17,18]. Such compositions are chemically counterintuitive as they necessitate significant Ti3 + formation especially for larger values of x, and this is unlikely in air. In our

experience, the colour of the samples, thermal analysis and the frequent presence of extra diffraction peaks in XRD and especially electron diffraction do not support this stoichiometry. Instead, we suggest that usually such phases contain some excess oxygen as described herein. Reduction of Ti4 + to Ti3 + on these mixed oxides will lead to an improvement in the electronic conductivity together with a loss of oxygen in order to maintain the electroneutrality. Removal of the excess oxygen is likely to occur and it could be considered as a way to enhance oxide ionic or, perhaps, protonic conductivity. In the present work, the electrical characterisation of the n = 6 member (La2Sr4Ti6O19
d) of the La2 Srn
2TinO3n + 1 family was performed by means of ac impedance spectroscopy and dc conductivity measurements in a range of oxygen and water partial pressures. Initial fuel cell tests were also carried out in order to check the performance of the La2Sr4 Ti6O19
d phases as anodes for fuel cells.

2. Experimental La2Sr4Ti6O19
d was synthesized by solid state reaction, grinding stoichiometric amounts of pre-dried powders of >99% purity of La2O3, SrCO3 and TiO2 from Aldrich. This mixture was then calcined at 1300 – 1400 jC for 24 h. The resulting powder was ground and pressed uniaxially at 3.5 –4 tons pressure. These pellets were then fired in air at 1400 –1600 jC

Fig. 1. Schematic representation of the structures of La2Ti2O7 (a) and SrTiO3 (b).

J. Canales-Va´zquez et al. / Solid State Ionics 159 (2003) 159–165

for 48 h. The resultant pellets were ground and pressed again and then fired for further 48 h yielding yellow pellets. Reduced phases were achieved by reduction at 1000 jC in 5% H2/Ar for 48 h yielding black pellets. Phase purity was controlled by X-ray diffraction (XRD) using a Stoe StadiP Transmission X-ray diffractometer, using CuKa1 radiation. Energy dispersive X-ray spectroscopy (EDS) microanalysis using an Oxford Link ISIS system was performed to check the homogeneity of the samples. Ac impedance data were acquired using a 1260 Solartron impedance analyser over the frequency range 0.1 Hz – 6.6 MHz. Measurements were performed on pellets (75% dense) coated with organoplatinum paste on each face and fired afterwards at 900 jC for 1 h. The samples were mounted in a compression jig with Pt wire electrodes. The measurements were performed in a range of water and oxygen partial pressures (static air, wet Ar and dry Ar). Total conductivity was measured over a range of temperatures from 25 to 900 jC in reducing conditions (wet and dry 5% H2/Ar), using the four terminal dc technique also in 75% dense pellets. For fuel cell tests, La2Sr4Ti6O19
d powders obtained at 1500 jC that had been subsequently reduced at 1000 jC for 48 h were used. An ethanol-based slurry formed with the reduced powder was coated onto YSZ pellets and fired at 1300 jC for 4 h to form a thin layer of anode for the fuel cell test. To investigate anode polarisation, a three-probe cell was prepared for ac impedance measurements. The arrangement of working, counter and reference electrode is schematised in Fig. 2. The electrolyte is 20 mm in diameter and 2 mm in thickness. The active area of working electrode (anode) is 1 cm2. A gold current collector was applied

Fig. 2. Schematic drawing of the three-electrode arrangement of the electrochemical cell.

161

onto the working electrode using an Au mesh with a thin partially covering layer of Au paste between the oxide and the mesh to ensure adherence. The fuel cell performance was measured in a normal two-electrode configuration cell. The electrolyte for cell performance measurements is YSZ with 17% alumina addition of 0.3 mm thickness (CeramTec) chosen due to its superior mechanical properties. The cathode used was commercial La0.8Sr0.2MnO3 (Praxair, USA). Platinum paste was used as cathode current collector. For both types of cells, the thickness of anode was about 50 Am, a thickness that has been found to be fairly optimum in previous studies of oxides in our laboratory. The impedance of the electrochemical cell was recorded at 900 jC at open circuit voltage (OCV) and at different atmospheres with a 20-mV ac signal amplitude by which stable and reproducible spectra were observed. Measurements were carried out in the 1  105 – 0.01 Hz frequency domain. Humidified 5% H2 in Ar, 100% H2 and 100% CH4 were used as the fuel at the working electrode, O2 being supplied at the counter electrode. The fuel cell performance was recorded by cyclic-voltammetry at a scan rate of 1 mV s
1. All electrochemical tests were performed after reducing the La2Sr4Ti6O19
d anode in 5% H2 at 850 jC for 2 days to achieve the highest conductivity due to the slow kinetic process of reduction.

3. Results and discussion In order to investigate conductivity as a function of oxygen partial pressure, a series of impedance spectroscopic measurements were performed. In each atmosphere studied, samples were first annealed at 800 jC until equilibrium was achieved, then measurements performed at lower temperatures. A pronounced dependence of the total conductivity (i.e. grain and grain boundary) with the oxygen partial pressure was found, showing features typical of an n-type conductor, i.e. higher conductivity at lower oxygen partial pressure. The activation energy calculated from the corresponding Arrhenius plots (Fig. 3) revealed the same tendency, and it decreased as the P(O2) decreased, from 1.3 eV for the sample measured in air to 0.3 eV for the sample measured in dry argon. The decrease in the activation energy in addition to the increase in conductivity at lower P(O2) must be ex-

J. Canales-Va´zquez et al. / Solid State Ionics 159 (2003) 159–165

162

Fig. 5. Complex impedance plots for La2Sr4Ti6O19-d measured in dry Ar. Fig. 3. Arrhenius plots for La2Sr4Ti6O19-d in air, wet Ar and dry Ar.

4+

3+

plained by Ti being reduced to Ti . The more reduced the conditions, the more Ti3 + is present in the sample, which leads to an enhancement of the electronic conductivity. Nyquist plots were studied to analyse the different contributions to the conductivity due to bulk, grain boundary and electrode responses. In the sample measured in dry Ar at room temperature, it is possible to clearly observe two semicircles: one corresponds to a typical grain boundary response

(capacitance f 10
8 F) and the other was assigned to an electrode response (capacitance f 10
4 F), possibly related to a Schottky barrier (Fig. 4). At higher temperatures, the electrode response is less important (Fig. 5) and above 300 jC only the grain boundary can be observed. Other atmospheres (wet Ar, static air) gave similar responses and sometimes it was possible to distinguish between the responses for the bulk and the grain boundary as can be observed in the ZU and MU = jxZVvs. frequency plots (Fig. 6). The very

Fig. 4. Nyquist plot corresponding to La2Sr4Ti6O19-d measured in dry Ar at room temperature.

J. Canales-Va´zquez et al. / Solid State Ionics 159 (2003) 159–165

Fig. 6. Imaginary impedance (a) and imaginary modulus (b) vs. frequency for La2Sr4Ti6O19-d measured in wet Ar at 631 K.

small value of grain boundary capacitance indicates that narrow, well-ordered grain boundaries are formed, despite the fairly large porosity (25%) of the sample. There is some depression of the semicircle which may relate to porosity; however, the most important role of the porosity is to facilitate gas diffusion and hence reduction of the grain boundaries. Conductivity was also studied in different reducing conditions by the four terminal dc techniques (Fig. 7), again initially annealing at high temperature (950 jC) until equilibrium was achieved. A reduced sample was measured in dry and wet flowing H2/Ar atmospheres. In dry H2/Ar, a metallic behaviour was found with conductivity dropping from 60 S cm
1 at room temperature to 40 S cm
1 at 950 jC. In addition, magnetic measurements on the same sample performed from 4 K to room temperature revealed Pauli paramagnetism which is a typical feature in metals. These conductivity values are somewhat higher than the values reported for the A-site-deficient perovskites Sr1
3x/2LaxTiO3
d [16]. The interstitial oxygen in La2Sr4Ti6O19
d might be easily removed under reducing conditions and therefore the reduction of Ti4 +

163

to Ti3 + is favoured. As a consequence, there is more Ti3 + for the same P(O2), which explains the higher values in conductivity found. These results also revealed a discrepancy with recent studies by Marina et al. [17] that have reported a semiconducting behaviour in similar compositions prepared in air, probably due to the longer time under reducing conditions used to achieve the complete reduction of the sample in our experiments. In addition, our results agree with those reported in previous studies in the La1
xSrxTiO3 system [18]. Furthermore, the sample measured in wet H2/Ar showed the presence of an insulator-metal transition at around 400 jC. This metal-insulator transition could have a number of origins, i.e. that the distribution of Ti3 + changes in a similar fashion as occurs in Ti4O7 (Magne´li phases) [19]. At temperatures below 400 jC, Ti3 + is distributed forming longdistance couples (Ti3 + – O – Ti3 +) that disappear at higher temperatures leading to Ti3 + randomly distributed in the structure. Ordering in couples leads to a partial localisation of electrons avoiding in this way the formation of a metallic band due to the presence of some discrete levels of energy. At temperatures above 400 jC, the Ti3 + are not in couples anymore, facilitating in this way the formation of the metallic band. Alternatively, this metal-insulator transition could be understood in terms of a change from dominant metallic to dominant semiconductor on cooling. The two components could be both in the bulk or it could be a semiconducting grain boundary and a metallic grain. The lower conductivity observed in wet H2/Ar might be explained by the fact of partial oxidation decreasing the amount of Ti3 + leading to a lower conductivity due to the smaller number of charge carriers.

Fig. 7. Plot of conductivity vs. temperature for La2Sr4Ti6O19-d in dry and wet 5% H2/Ar.

164

J. Canales-Va´zquez et al. / Solid State Ionics 159 (2003) 159–165

As stated before, the presence of interstitial oxygen within the perovskite framework might possibly enhance the ionic transport. Nevertheless, no evidence of ionic conduction was found by means of ac impedance or dc measurements due to the fact that the conductivity in these phases is mainly via electrons. Thermogravimetric analysis can be used in order to estimate the concentration of protonic defects, because their creation is mainly caused by dissociative water uptake. However, no evidence of water uptake was observed for La2Sr4Ti6O19
d in any of the atmospheres used. The use of techniques such as electrochemical permeability or studies of Faradaic efficiency could help in determining the existence of ionic transport in La2Sr4Ti6O19
d and this will be investigated in the future.

4. Fuel cell tests The polarisation resistance (Rp) of the La2Sr4Ti6 O19
d anode measured using three electrode configuration in wet 5% H2/Ar is 6.59 V cm2 at 850 jC (Fig. 8). It has been recently reported that the polarisation

Fig. 8. Impedance measurements obtained at OCV and at 850 jC in 4.9% H2 + 2.3% H2O + 92.8% Ar (5), at 900 jC in 97% CH4 + 3% H2O (5), 4.9% H2 + 2.3% H2O + 92.8% Ar (o) and 97% H2 + 3% H2O (4), respectively. Rp = 8.93 V cm2 (CH4), 4.56 V cm2 (5% H2), 2.97 V cm2 (H2) at 900 jC, 6.59 V cm2 (5% H2) at 850 jC.

Fig. 9. Fuel cell performance using La2Sr4Ti6O19-d as anode and La0.8Sr0.2MnO3 as cathode. (a) 97% H2 + 3% H2O at 850 jC; (b) 97% H2 + 3% H2O at 900 jC and (c) 97% CH4 + 3% H2O at 900 jC. Note: current densities at 600 mV are 75, 119, 30 mA cm
2 for (a), (b) and (c), respectively. Maximum power densities are 50 mW cm
2 at 0.45 V, 76 mW cm
2 at 0.47 V and 21 mW cm
2 at 0.45 V for (a), (b) and (c), respectively.

resistance of La0.35Sr0.65TiO3 is 52 V cm2 in wet H2 at 850 jC [20]. This large difference in polarisation resistance may be attributed to the long-time reduction before our tests in order to achieve the highest electrical conductivity or the preparation of materials and anode, as the true chemical composition of the two materials is not significantly different. The Rp decreases with the increase in temperature. It dropped to 4.56 V cm2 at 900 jC in wet 5% H2/Ar. The Rp is also sensitive to hydrogen concentration. A polarisation of 2.97 V cm2 was obtained at 900 jC in wet H2. The relatively high Rp value may be due to the low ionic conductivity, which might be further improved by optimisation of the electrode microstructure and also by modification of the composition. At 900 jC, the polarisation resistance in wet CH4 is almost three times larger than in wet H2 indicating that La2Sr4Ti6 O19
d itself is not a suitable anode material for direct methane fuel cells. The fuel cell performance with La2Sr4Ti6O19
d anode and La0.8Sr0.2MnO3 cathode is shown in Fig. 9. Current densities of 75 and 119 mA cm
2 at 600 mV were achieved at 850 and 900 jC, respectively, when wet H2 was applied as the fuel. The performance of the cell is limited by the anode polarisation. A maximum power density of 76 mW cm
2 is obtained in wet H2 at a potential of 0.47 V. The maximum current density is about 78 mA cm
2 at 900 jC when wet methane was

J. Canales-Va´zquez et al. / Solid State Ionics 159 (2003) 159–165

used as the fuel indicating again that La2Sr4Ti6O19
d itself is not a good anode material for a direct methane fuel cell, which is consistent with the anode polarisation test shown in Fig. 8. In addition, no carbon was observed after the tests, indicating that cracking did not occur.

5. Conclusions The electrical properties of La2Sr4Ti6O19
d were studied by ac impedance spectroscopy and dc conductivity measurements, revealing the potential of this mixed oxide as an anode for fuel cells. The total conductivity revealed a strong dependence on oxygen stoichiometry, i.e. the amount of Ti3 +, showing the typical behaviour of an n-type conductor. High total conductivity (60 S cm
1), mainly electronic, was found for the reduced phases measured under 5% H2/Ar atmospheres, revealing metallic behaviour. A metal-insulator transition was observed when measuring in wet 5% H2/Ar, enhancing the effect of oxygen stoichiometry in the band structure of this material. An anode polarisation resistance of 2.97 V cm2 is achieved in wet H2 at 900 jC using pure La2Sr4Ti6 O19
d as anode. Accordingly, the maximum current and power densities are 321 mA cm
2 and 76 mW cm
2, respectively, in wet H2 at 900 jC. The performance in wet CH4 is not good compared to that in wet H2. With modification of both chemical composition and electrode microstructure, La2Sr4Ti6O19
d could be considered a potential anode material or anode constituent material for SOFCs. Indeed, recent data have shown excellent anode performance for a very similar material composited with ceria [20].

Acknowledgements The authors would like to thank the HiT Proton Project and EPSRC for funding and Martin Smith at

165

St. Andrews for his assistance with magnetic measurements.

References [1] B.C.H. Steele, Solid State Ionics 86 – 88 (1996) 1223 – 1234. [2] G. Pudmich, B.A. Boukamp, M. Gonza´lez-Cuenca, W. Jungen, W. Zipprich, F. Tietz, Solid State Ionics 135 (2000) 433 – 438. [3] T. Norby, Solid State Ionics 125 (1999) 1 – 11. [4] N. Bonanos, Solid State Ionics 53 – 56 (1992) 967 – 974. [5] S. Hamakawa, T. Hibino, H. Iwahara, J. Electrochem. Soc. 141 (1992) 1720. [6] T. Yajima, H. Suzuki, T. Yogo, H. Iwahara, Solid State Ionics 51 (1992) 101. [7] K. Katahira, Y. Kohchi, T. Shimura, H. Iwahara, Solid State Ionics 138 (2000) 91 – 98. [8] J. Canales-Va´zquez, W. Zhou, J.T.S. Irvine, Inst. Phys. Conf. Ser. 168 (7) (2001) 295 – 298. [9] M. Gasperin, Acta Crystallogr., B 31 (1975) 2129 – 2130. [10] N. Ishizawa, F. Marumo, T. Kawamura, M. Kimura, Acta Crystallogr., B 31 (1975) 1912 – 1915. [11] T. Williams, H. Schmalle, A. Reller, F. Lichtenberg, D. Widmer, G. Bednorz, J. Solid State Chem. 93 (1991) 534 – 548. [12] H. Schmalle, T. Williams, A. Reller, A. Linden, G. Bednorz, Acta Crystallogr., B 49 (1993) 235 – 244. [13] M.E. Bowden, D.A. Jefferson, I.W.M. Brown, J. Solid State Chem. 117 (1995) 88 – 96. [14] N.G. Eror, U. Balachandran, J. Solid State Chem. 40 (1981) 85 – 91. [15] R. Moos, T. Bischoff, W. Menesklou, K.H. Ha¨rdtl, J. Mater. Sci. 32 (1997) 4247 – 4252. [16] P.R. Slater, D.P. Fagg, J.T.S. Irvine, J. Mater. Chem. 7 – 12 (1997) 2495 – 2498. [17] O.A. Marina, N.L. Canfield, J.W. Stevenson, Solid State Ionics 149 (2002) 21 – 28. [18] J.E. Sunstrom, S.N. Kauzlarich, P. Klavins, Chem. Mater. 4 (1992) 346 – 353. [19] M. Marezio, D. Tranqui, S. Lakkis, C. Schlenker, Phys. Rev., B 16 (1977) 2811. [20] O.A. Marina, L.R. Pederson, Novel ceramic anode for SOFCs tolerant to oxygen, carbon and sulphur, in: J. Huijsmans (Ed.), Proceedings of the 5th European Solid Oxide Fuel Cell Forum, 1 – 5 July 2002, pp. 481 – 489, Lucerne/Switzerland.