006) 1997–2003www.elsevier.com/locate/ssi

Solid State Ionics 177 (2

Mn-substituted titanates as efficient anodes for direct methane SOFCs

Alejandro Ovalle a, Juan Carlos Ruiz-Morales a,b, Jesús Canales-Vázquez a,c,David Marrero-López b, John T.S. Irvine a,⁎

a School of Chemistry, Purdie Building, University of St Andrews, St Andrews, Fife KY16 9ST, UKb Departamento de Química Inorgánica, Universidad de La Laguna, 38200 La Laguna, Spain

c Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus UAB, 08193 Bellaterra, Spain

Received 22 August 2005; received in revised form 7 June 2006; accepted 12 June 2006

Abstract

A new family of perovskite titanates with formula La4Sr8Ti12−xMnxO38−δ has been investigated as fuel electrode materials for SOFCs. Thesephases present a rhombohedral (R−3c) unit cell. Mn substitution does not have a large impact on the bulk conductivity of the phases studied,which remains close to the values observed in other related titanates, although the grain boundary contributions are largely improved. Relativelylow polarisation resistances were observed under both hydrogen and methane conditions, e.g., 0.3 and 0.7 Ω cm2 at 950 °C, respectively. Despitethe polarisation resistance in methane being more than twice that in hydrogen, the performances are very similar, which might indicate certainmethane activation at that temperature. Surprisingly, the anodic overpotential was fairly low compared to those reported in the literature for othermaterials and especially for titanate-based anodes, i.e., a value of 55 mVat 0.5A/cm2, at 950 °C, under wet hydrogen was obtained. Additionally, avalue 72 mV was obtained in the same conditions under methane. These values indicate that the use of Mn as dopant for perovskite-relatedtitanates enhanced electrochemical performance of these anodes, especially at high temperatures.© 2006 Elsevier B.V. All rights reserved.

Keywords: SOFC; Anode; Perovskite

1. Introduction

The search of alternative materials to the Ni/YSZ cermets asefficient anodes for SOFCs is one of the main research trends insolid state electrochemistry nowadays. High mixed conductivity,stability under operating conditions, catalytic activity promotingthe oxidation of the fuel (preferably methane) or compatibilitywith the electrolyte (typically YSZ or ceria-based oxides) aresome of the properties targeted. Although the aforementionedcermets meet most of these requirements, they exhibit long-termdegradation due tometal particle coarsening, sulphur poisoning orcarbon build-up when operating with hydrocarbon fuels. Thus,there exists a variety of open research lines ranging from theimprovement of the cermet performance by optimising their mi-crostructure or composition [1], to the use of alternative systems

⁎ Corresponding author.E-mail address: jtsi@st-and.ac.uk (J.T.S. Irvine).

0167-2738/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.ssi.2006.06.014

such as those based on ceria cermets proposed by Gorte et al. [2]or perovskites [3–10].

Some of the perovskite-based anode materials already offerrather competitive performances in hydrogen and methane asreported for several phases in the (La,Sr)(Cr,Mn)O3 system [3–5]. There exists, however, some concern regarding moderateelectronic conductivity under fuel conditions, limiting electrodesupport design concepts. (La,Sr)TiO3+δ phases have been con-sidered as good fuel electrode candidates due to their stabilityunder reducing conditions and also because of their resistance tocarbon build-up and sulphur poisoning, although their perfor-mance under methane was far from competitive in comparisonwith other state-of-the-art anode materials [6–9]. Nevertheless,there are some reports in the literature regarding titanate-basedanodes showing very promising performances, especially usingCeO2 composites [8] or partially substituting Ti by other cationssuch as Sc [10]. The partial substitution of Ti by lower valencecations of similar ionic radii seems to improve the properties ofthese phases.

Fig. 1. (a) Observed, calculated and difference HRPD patterns for La4Sr8Ti11 Mn1O38−δ. (b) Increased splitting of XRD peaks as degree of Mn substitution inLa4Sr8Ti12−xMnxO38−δ increases. (c) Unit cell parameters as a function of the degree of substitution in La4Sr8Ti12−xMnxO38−δ.

1998 A. Ovalle et al. / Solid State Ionics 177 (2006) 1997–2003

Fig. 2. HRTEM image showing the presence of randomly distributed defectsembedded within a perovskite framework in La4Sr8Ti10Mn2O38−δ as arrowed.

Table 1Refined structural parameters for R-3c La4Sr8Ti11MnO38−δ after Rietveldrefinements using combined HRPD and XRD data at room temperature

RT

a (Å) 5.5458 (6)c (Å) 13.5648 (1)V (Å3) 361.31 (2)x (O) 0.4737 (1)χ2 2.488

1999A. Ovalle et al. / Solid State Ionics 177 (2006) 1997–2003

In the present work, the effect of Mn substitution in (La,Sr)TiO3+δ on the structure and performance as an SOFC anode ofthese phases has been evaluated.

2. Experimental

La4Sr8Ti12−xMnxO38−z samples were prepared by traditionalsolid state reaction. Stoichiometric amounts of pre-dried high-purity (>99%) La2O3, SrCO3, TiO2 and Mn2O3 (Aldrich) weremixed and ground in acetone in zirconia ballmills for 30 min.and then calcined at 1200 °C for 6 h. The resulting powder wassubsequently ground, mixed and uniaxially pressed into pellets,which were placed in alumina crucibles on top of sacrificialpowder and then fired at 1300–1500 °C for 2–4 days. Reducedphases were obtained by further reduction at 1000 °C underflowing 5%H2/Ar for 48 h.

X-ray diffraction (XRD) data were collected with a StoeStadiP X-ray diffractometer, using CuK⟨1 radiation in therange 2θ=10–70° in steps of 0.1° at 30 s/step. XRD data wereindexed using a least-squares procedure. Neutron diffractiondata were collected on the High-Resolution Powder Diffractionbeamline (HRPD) at the ISIS facilities, Rutherford AppletonLaboratories, Oxford (Rietveld refinements were carried outusing GSAS).

AC impedance data were acquired using a 1260 Solartronimpedance analyser over the frequency range 1 Hz–1 MHz instatic air. The measurements were performed on dense pelletscoated with a thin layer of organo-platinum paste on each face,which were fired at 900 °C for 1 h. The samples were mountedin a compression jig with Pt contacts.

The anode response was investigated on a three-electrodesetup [5,6] using 20 mm diameter YSZ pellets (2 mm thick) aselectrolyte, LSM (Praxair) as cathode material and La4Sr8Ti12−xMnxO38−z as anode. The anode was prepared by coating severallayers of La4Sr8Ti12−xMnxO38−z-based slurries mixed with YSZ

Table 2Refined structural parameters for R-3c La4Sr8Ti12−nMnnO38−δ after the Rietveldrefinements of X-ray powder diffraction data at room temperature

n a (Å) c (Å) V (Å3)

1a 5.5458 (6) 13.5648 (1) 361.31 (2)2 5.5457 (6) 13.5509 (2) 360.93 (5)3 5.5420 (7) 13.5290 (2) 359.85 (5)4 5.5416 (9) 13.5213 (8) 359.60 (8)5 5.5273 (2) 13.5003 (3) 357.19 (2)6 5.5120 (1) 13.4626 (2) 354.23 (4)a Joint HRPD and XRD refinement.

in different ratios (1:2, 1:1 and 2:1) onto the electrolyte pellets toimprove the adherence between the anode andYSZ.Aumeshwasused as current collector. The resulting specimen was fired then at950 °C for several hours in flowing 5%H2/Ar to achieve fullreduction of the anode material.

The impedance of the electrochemical cell was recorded atopen circuit voltage (OCV) over a range of temperatures andatmospheric conditions. The measurements were carried out inthe 1×105 to 0.01 Hz frequency domain using a 20 mVampli-tude AC signal, rendering stable and reproducible spectra.

Humidified 100% H2 (wet hydrogen) and 100% CH4 (wetmethane) were used as fuels at the working electrode, whilst O2

was supplied at the cathode. The anode gas was bubbledthrough a Dreschel bottle thermostated at 16 °C; thus, the watercontent in each gas was approximately 1.8%. 5%H2 was used tocheck the sealing conditions and to perform in-situ reductionsof the sample, at 950 °C for several hours. The fuel cellperformance was obtained by cyclic voltammetry at a scan rateof 2 mV s−1. All electrochemical tests were performed afterreducing the anode materials in 5%H2 at 950 °C for severalhours to achieve the highest conductivity. The degree ofreduction of the sample was monitored by following the realpart of the impedance at 1 Hz. The measurements started whenno dependence with time was observed.

3. Results and discussion

3.1. Structural characterisation

La4Sr8Ti12O38 is the lowest n XRD cubic phase of theLa4Srn−4TinO3n+2 series; lower values of n show extended defectsin HRTEM and additional peaks in XRD [11]. The cubic phaseshave some extra oxygen that is thought to be accommodated in 5short range linear defects distributed randomly within a

Fig. 3. (a) Nyquist plots showing the temperature dependence of the bulk and grain boundary contributions of a La4Sr8Ti11Mn1O38−δ pellet (b) Temperature variationof the related complex modulus (M″).

2000 A. Ovalle et al. / Solid State Ionics 177 (2006) 1997–2003

perovskite network [6,10,11]. Such linear defects seem to causesuperstructure reflections doubling the ⟨111⟩p, rendering a facecentre cubic unit cell with a=2ap. Previous work has shown thatthe substitution of Ti by lower valence cations (i.e., Sc or Ga )does not significantly affect the structure, which remains as XRDcubic (a≈3.91 Å), which is not surprising as such substitutionreduces the nominal oxygen excess increasing the destabilisationof any extended oxygen defects. For Mn substitution, there aresome questions relating to the oxidation state of Mn in thesephases. Assuming that Mn was likely to be Mn(III), two differentapproaches to solve this particular problem were considered. Inthe first, the La/Sr ratio was maintained giving rise to theLa4Sr8Ti12−x MnxO38−x/2 series, whilst in the latter the oxygenoverstoichiometry was fixed by changing the La/Sr ratio tocompensate the Ti/Mn, i.e., giving rise to the La4+xSr8−xTi12−xMnxO38 series. Therefore, if Mn remains in the 3+ oxidation stateof the starting material, the resulting phases will all lie within thecubic domain. Alternatively, if Mn was in the Mn(IV) oxidationstate, phases in the La4+xSr8−xTi12−xMnxO38+x series will moveinto the layered domain. XRD studies show that for as-preparedsamples, there is Mn(IV) at least to some extent, as the phasesbelonging to this La4+xSr8−xTi12−xMnxO38 series revealed thepresence of reflections at 2θ=28–31°, which are characteristic oflayered domains. As such layers appear to have a large negativeeffect over the electrochemical properties, the present work hasbeen focused on the “non-layered” phases, i.e., La4Sr8Ti12−xMnxO38−δ.

Although these phases still exhibit simple perovskite XRDpatterns, they no longer exhibit the simplest cubic form. Even at lowlevels of Mn substitution, the combined HRPD and XRD patternscould be better refined as a rhombohedral structure (R−3c) (Fig. 1a,Table 1). As Mn substitution increases the rhombohedral splittingbecomes more pronounced (Fig. 1b). Rhombohedral distortionsare common in perovskite manganites and cobaltites and can beascribed to the R−3c space group a the result of equivalent an-tiphase octahedra tilting along all the three crystallographic axes,i.e., a−a−a− followingGlazier's notation [12]. As shown in Fig. 1cand Table 2, the volume of the unit cell gradually decreases withincreasing theMn content followingVegard's law, which indicatesa random distribution of Mn in the B-sites. Clearly the rhombo-hedral distortion may influence how the oxygen excess is accom-modated within the perovskite but this must still involve someshort-range defects. Preliminary HRTEM investigations have re-vealed that some crystallographic shears distributed randomlywithin a perovskite matrix remain in the structure (Fig. 2), whichimplies that the oxygen overstoichiometry is compatible withrhombohedral distortions in the oxygen sublattice. Indeed, La4Srn−4TinO3n+2 (4≤n≤6) phases show very distorted TiO6 octahedraand are also able to accommodate the oxygen excess [13].

4. AC impedance spectroscopy

The AC impedance responses in static air consisted typicallyof two well-defined arcs, one associated with the grain boundary

Fig. 5. Polarisation resistances of La4Sr8Ti11Mn1O38−δ anode, at differenttemperatures and (a) under wet hydrogen and (b) methane. The plots reveal thefairly low polarisations at 950 °C in both gases, although at lower temperaturesthe polarisation increases significantly under methane conditions.

Fig. 4. (a) Arrhenius plots corresponding to LST (La4Sr8Ti12O38−δ) andLa4Sr8Ti11Mn1O38−δ in static air. Although the overall conductivity is quitesimilar for both the non-substituted and the Mn-substituted counterparts, thegrain boundary contribution is reduced for the latter. (b) Arrhenius plotscorresponding to the La4Sr8Ti12−xMnxO38−δ series in static air. The overallconductivity increases as Mn content increases.

Fig. 6. Cross-sectional view of the interface between the La4Sr8Ti11Mn1O38−δ

anode and YSZ. No contact resistance losses can be expected as the interfaceelectrode–electrolyte is very neat. Nevertheless, further optimisation of themicrostructure seems to promise an enhanced catalytic activity towards methaneoxidation at lower temperatures.

2001A. Ovalle et al. / Solid State Ionics 177 (2006) 1997–2003

(3×10− 9 F cm− 1) and the second with bulk processes(1×10−11 F cm−1), which could be fitted as two R-CPE (i.e., aresistor and a constant phase element in parallel) in series. Attemperatures above 400 °C, only the grain boundary responsewasobserved as shown in Fig. 3(a) and (b). No low-frequencyphenomena associated with electrode processes were observed inthe range of temperatures studied. Impedance studies under morereducing conditions were limited due to the very high totalconductivity of these phases, meaning that no elements could beresolved. Four-terminal conductivity measurements indicatedvalues of 7 electronic conductivity under 5%H2 ranging from 2 to0.04 S cm−1 for x=1 to x=4 in La4Sr8Ti12−xMnxO38−δ, stronglypointing to x=1 as being more suitable.

The Arrhenius plots obtained in air (Fig. 4) revealed thatthe bulk conductivity of the Mn-substituted phases iscomparable to the non-substituted [6,7] and Sc-substitutedcounterparts [10], although the grain boundary contributionappears to be diminished. The conductivity was thermallyactivated with an activation energy of 0.84 eV. These resultsindicate that the phases of the present study could have

potential to be used as anodes in fuel cells as discussedbelow.

5. Fuel cell tests

The half-cell tests performed using La4Sr8Ti11MnO38−δ /YSZas anode material revealed rather low Rp in wet H2 and CH4, i.e.,0.3 and 0.7Ω cm2 at 950 °C, respectively (Fig. 5). It is especiallyinteresting to note that the results in methane, at least at 950 °C,are fairly comparable to those in hydrogen. SEM performed onfuel cells after testing revealed good adherence between the anode

Fig. 7. Anodic overpotential of La4Sr8Ti11Mn1O38−δ at different temperaturesand (a) under wet hydrogen and (b) wet methane. Values as low as 55 mV in H2

and 72 mV in CH4, at 0.5 A/cm2, at 950 °C, were obtained. Consequently, good

performances can be expected using these phases as anode material.

2002 A. Ovalle et al. / Solid State Ionics 177 (2006) 1997–2003

and the electrolyte and show good electrode microstructureallowing both connectivity of grains and gas diffusion (Fig. 6),although further microstructural optimisation will be possible.The improved performances over previous work [6] seems to berelated to the presence of Mn on the B-sites. The open circuitvoltages almost matched those predicted thermodynamically inhydrogen atmosphere. The theoretical values are 1.104V, 1.111Vand 1.119 V, at 950, 900 and 850 °C, respectively. The exper-imental values were 1.101V, 1.105 Vand 1.112V, under the sameconditions, indicating good sealing in any case. In methane, theinterpretation is rather complicated because in the presence ofsteam a number of alternative routes are possible for methaneoxidation. The OCV under methane flow decreases with de-creasing temperature giving values 1.080, 0.980 and 0.920 at 950,

8 900 and 850 °C, respectively. This indicates that the fueloxidation probably involves a thermally activated stepwise pro-cess involving internal reforming and cracking. Most of the so-formed carbon is oxidised rapidly as no carbon deposits weredetected after our tests. The use of perovskite-based anodes isgenerally believed to prevent the formation of carbon depositswhen using hydrocarbon fuels, which is one of the major dis-advantages associated with the use of Ni–YSZ cermets.

The anodic overpotentials under hydrogen (Fig. 7) are betterthan the best results reported recently for Sc-substituted titanatesand much better than those corresponding to the non-substitutedcounterpart. For example, the values obtained at 950 °C and900 °C, at 0.5 A/cm2 under wet hydrogen were just 55 mV and116 mV, respectively. These are rather low compared to the250 mVof anodic polarisation obtained for the Sc-doped titanatesunder the same conditions.

In methane, the low values of polarisations are only obtainedjust at 950 °C, i.e., 72 mV at 0.5 A/cm2. At lower temperatures,there is a large increase in the polarisation, i.e., a value of 238 mVat 0.3 A/cm2 is obtained at 900 °C. This could be related to achange in the methane oxidation mechanism, in accord with a∼10% decrease in the OCVobserved on cooling to 900 °C. Thelow polarisation corroborates the low polarisation resistancevalues and suggests the applicability of these materials as anodesfor direct methane fuel cells upon further improvement of themicrostructure.

6. Conclusions

A new family of perovskite titanates with formula La4Sr8Ti12−xMnxO38−δ has been investigated as fuel electrodematerials for SOFCs. These phases present a rhombohedralunit cell (R−3c), although with some short-range defects. Thefuel cell tests performed have revealed a very low polarisationresistance under H2 and CH4 atmospheres and stability underthe most reducing conditions. Because of these results thesematerials are a good potential candidate for anode for SOFCs.

Acknowledgements

We thank EPSRC and the EU Marie Curie Fellowshipscheme (contract no. HPMT-CT-2002-02001) for financialsupport. We also thank EPSRC and RAL ISIS for provisionof neutron beam time.

References

[1] J. Liu, S.A. Barnett, Solid State Ionics 158 (2003) 11;A. Weber, B. Sauer, A.C. Müller, D. Herbstritt, E. Ivers Tiffée, Solid StateIonics 152 (2002) 543.

[2] S.D. Park, J.M. Vohs, R.J. Gorte, Nature 404 (2000) 265;S. McIntosh, R.J. Gorte, Chemical Reviews 104 (10) (2004) 4845.

[3] J. Liu, B.D. Madsen, Z. Ji, S.A. Barnett, Electrochemical and Solid StateLetters 5 (6) (2002) A122.

[4] S.W. Tao, J.T.S. Irvine, Nature Materials 2 (2003) 321.[5] S.W. Tao, J.T.S. Irvine, Journal of Electrochemical Society 151 (2) (2004)

A252.[6] J. Canales Vázquez, S.W. Tao, J.T.S. Irvine, Solid State Ionics 159 (2003)

159.

2003A. Ovalle et al. / Solid State Ionics 177 (2006) 1997–2003

[7] O.A. Marina, N.L. Canfield, J.W. Stevenson, Solid State Ionics 149 (2002)21.

[8] O.A. Marina, L.R. Pederson, Proceedings of the 5th European Solid OxideFuel Cell Forum (Lucerne/Switzerland 2002), 2002, p. 481.

[9] R. Mukundan, E.L. Brosha, F.H. Garzon, Electrochemical and Solid StateLetters 7 (1) (2004) A4.

[10] J. Canales-Vázquez, J.C. Ruiz-Morales, J.T.S. Irvine, W. Zhou, Journal ofElectrochemical Society 152 (7) (2005) 1458.

[11] J. Canales-Vázquez, M.J. Smith, W. Zhou, J.T.S. Irvine, AdvancedFunctional Materials 15 (2005) 1000.

[12] A.M. Glazer, Acta Crystallographica B28 (1972) 3384;A.M. Glazer, Acta Crystallographica A31 (1975) 756.

[13] J. Canales-Vázquez, J.T.S. Irvine, W. Zhou, Journal of Solid StateChemistry 177 (2004) 2039.