- Email: [email protected]
Inter-diffusion in lanthanum tungsten oxide
Solid State Ionics 244 (2013) 57–62
Contents lists available at SciVerse ScienceDirect
Solid State Ionics journal homepage: www.elsevier.com/locate/ssi
Inter-diffusion in lanthanum tungsten oxide Einar Vøllestad, Truls Norby, Reidar Haugsrud ⁎ University of Oslo, Department of Chemistry, Centre for Materials Science and Nanotechnology, FERMiO, Gaustadalleen 21, 0349 Oslo, Norway
a r t i c l e
i n f o
Article history: Received 27 February 2013 Received in revised form 19 April 2013 Accepted 23 April 2013 Keywords: Cation diffusion Lanthanum tungstate Proton conductor Hydrogen membrane Demixing Walkout
a b s t r a c t Cation diffusion in lanthanum tungsten oxide (LaWO) has been studied on La27W5O55.5–Nd27W5O55.5 and La27W5O55.5–La27Mo1.5W3.5O55 inter-diffusion couples at temperatures between 1150 and 1350 °C in air and 5% H2 in Ar. Inter-diffusion coefficients were derived by fitting concentration profiles of Nd, La, W, and Mo from Electron Probe Micro Analysis to the appropriate solution of Fick's second law. The diffusivity in LaWO follows Arrhenius type behaviour with almost identical bulk diffusion coefficients for La and W site transport, and similar activation energies; 410 ± 30 kJmol-1 (La) and 450 ± 30 kJmol-1 (W). This behaviour has been interpreted to reflect that migration occurs via a common migration mechanism for both species, facilitated by vacancies on the lanthanum sublattice. Enhanced grain boundary diffusion was only observed for La site diffusion in the Nd27W5O55.5 phase, with much lower activation energy (170 ± 50 kJmol-1). The similar bulk diffusivities and relatively slow migration kinetics indicate that LaWO is more stable towards cation diffusion-related degradation than many of the most promising oxygen transport membrane materials. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Rare earth tungsten oxides with nominal formula close to Ln6WO12 exhibit considerable mixed ionic and electronic conductivity at elevated temperatures [1–4]. Below 800 °C under wet conditions, the ionic conductivity is for a large part protonic. The most promising member, lanthanum tungsten oxide (LaWO), exhibits essentially pure proton conductivity at 600 °C in the order of 0.001 Scm − 1, while oxide ion and—under reducing and oxidizing conditions—electronic conductivity become significant above 700 °C. Despite the high contents of rare earth cations, the rare earth tungsten oxides (LnWO) are stable in acidic gases such as CO2, contrary to most alkaline earth-based perovskite structured proton conductors [5]. These characteristics make LaWO an interesting candidate material for components in high temperature electrochemical energy conversion devices and hydrogen separation membranes. LaWO, with a La/W-ratio of 5.3–5.7, can be described as a defective fluorite-type structure, crystallizing in the F43m space group [4]. Based on recent investigations of the structural and functional characteristics [4,6,7] it has been postulated that the defective (partially occupied) oxygen sublattice is inherently disordered and can be described with an effective partial negative charge on the oxygen ions, charge compensated by correspondingly positive oxygen vacancies and an excess of tungsten residing on La sites acting as donors [7]. Hence, the stoichiometry of the LaWO structure can more correctly be represented as La28−xW4 + xO54 + 3x/2v2−3x/2 (v is a vacant oxygen site), with x ⁎ Corresponding author at: FERMiO, Forskningspraken, Gaustadalleen 21, 0349 Oslo, Norway. Tel.: +47 22840659; fax: +47 22840651. E-mail address: [email protected] (R. Haugsrud). 0167-2738/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ssi.2013.04.021
typically ranging from 0.78 to 1.08. For a more detailed description of the defect chemistry in LaWO, see Ref. [6]. Even though LaWO is chemically stable towards acidic gases, there may still be issues regarding long-term stability under electrolyte and gas separation membrane operating conditions due to cation diffusion. Gradients in the chemical potential of hydrogen will in many cases, depending on water vapour pressures, impose gradients in the chemical potential of the constituent metal cations according to the Gibbs– Duhem relation. These gradients will then act as driving forces for cation transport, and the constituent metal cations of the membrane thus migrate towards the high pO2 side [8,9]. If there is significant difference in the diffusivity of the cations, accumulation of the fastest and slowest moving species may additionally occur on the high and low pO2 side of the membrane, respectively. Thus, the material becomes inhomogeneous, potentially deteriorating the performance of the membrane. When the accumulation exceeds the cation solid solubility in the material during demixing, decomposition will occur on one or both sides of the membrane. Over time, these processes may lead to walkout, loss of sealing or decreased membrane performance, which ultimately would result in breakdown of the membrane. These degradation processes underline the importance of investigations on cation transport to evaluate the durability of fuel cell and hydrogen transport membrane candidate materials. Despite this, there are so far no data on cation transport for proton conducting oxides, presumably due to the relatively immature state of technologies based on such functional ceramics. Clearly, increased emphasis on long-term stability is required when moving towards commercialization of proton ceramic fuel cells, proton ceramic electrolyser cells, and hydrogen transport membranes. In this work, diffusion couples of La27W5O55.5–Nd27W5O55.5 (LaWO–NdWO) and La27W5O55.5–La27Mo1.5W3.5O55.5 (LaWO–LWMo)
58
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
were studied to determine the diffusivity of La and W represented by La–Nd and W–Mo inter-diffusion coefficients. Concentration profiles of the respective cations that develop during annealing under different temperatures and oxygen partial pressures were determined by means of Electron Probe Micro Analysis (EPMA). Inter-diffusion coefficients were derived by analysing these profiles according to Fick's second law. The functional dependencies of the diffusion data on the external conditions are discussed in light of the complex structure and defect chemistry of the LaWO-based material.
a
2. Experimental NdWO powder was synthesized using a wet chemical approach: Stoichiometric amounts of pre-dried Nd2O3 and WO3 powders were dissolved in aqueous HNO3 and NH3, respectively, before EDTA was added to each mixture in a 1:1 molar ratio between the sum of the metal cations and EDTA. The separate solutions were brought to pH = 7 and mixed, followed by drying at 200 °C. The resulting xerogel was calcined at 600 °C, before the powder finally was annealed at 1000 °C for 5 h. The powder was subsequently milled at 200 rpm in an all-agate planetary ball-mill for 1 h to promote a narrow particle size distribution. The LaWO and LWMo powders were available commercially (Cerpotech, Norway). Green bodies of diameter 13 mm were made by uniaxial pressing and sintered at 1500 °C for 3 h yielding dense pellets (>96% of relative density) with 8–30 μm large grains, except for NdWO which had grain sizes of 5–12 μm. To promote proper contact between the materials during the diffusion annealing, the pellets were first ground and then polished down to a 0.25 μm surface finish with diamond abrasives. Mounted samples were held in contact by an alumina spring load. The inter-diffusion annealing experiments were performed for 180– 400 h at temperatures between 1150 and 1350 °C in an atmosphere of air, argon or 5% H2 in argon (all dry). After annealing, the inter-diffusion couples were mounted in epoxy resin, cut perpendicular to the original interface and polished down to a 0.25 μm surface finish with diamond abrasive. Concentration profiles of Nd, La, W, and Mo were determined by EPMA (Cameca SX 100) using an acceleration voltage and beam current of 15 kV and 20 nA, respectively. The detection limit of EPMA is 100– 1000 ppm depending on the specific element. Each cross-section was analysed by determining the concentration of the relevant species every 1 μm along 5 different lines normal to the phase boundary. Typical scans penetrated 25–30 μm into each phase to ensure that 2/3 of the concentration profiles originated from outside of the diffusion zone.
50 µm
b
20 µm
3+
La
c
3. Results 3.1. Microstructure Fig. 1a shows a backscattered electron (BSE) image of a cross-section from an inter-diffusion couple between NdWO (top) and LaWO (bottom) annealed in dry air at 1350 °C for 300 h. The original boundary is no longer visible and can only be recognized by a small change in microstructure between the two phases. Small dark La-rich needles (c.f. Fig. 1a) formed along the grain boundaries during the annealing. Dark spots represent porosity, whereas the dark line perpendicular to the phase boundary originates from one of the EPMA concentration profiles. Locations of the EPMA profiles were chosen so as to minimize the influence of secondary phases and porosity. Fig. 1b and c show X-ray maps of La and Nd reflecting their concentrations. The mosaic shape of the La distribution in the deeper end of the diffusion zone in Fig. 1b is indicative of enhanced transport along the grain boundaries in NdWO. This behaviour is representative of what according to Harrison's classification is termed type-B kinetics [10], where both bulk and grain boundary diffusion contribute significantly to the transport in the intermediate regime of the diffusion zone. For
20 µm
3+
Nd
Fig. 1. a) BSE image and concentration maps of b) La and c) Nd over a NdWO–LaWO inter-diffusion couple after annealing at 1350 °C in dry air.
diffusion of Nd into LaWO, the mapping (cf. Fig. 1c) reveals a homogeneous distribution of the elements parallel to the phase boundary. Fig. 2a shows a similar BSE image of an LWMo–LaWO diffusion couple annealed at 1350 °C in air for 300 h. Similar as for the LaWO–NdWO diffusion couple the original phase boundary is no longer visible, but here there is no clear difference in microstructure. Fig. 2b and c show maps representing the Mo and W concentrations. There is no indication of enhanced transport along the grain boundaries in either of the phases. However, due to a smaller difference in W and Mo concentration between the phases—and thus less contrast—distinct features such as the mosaic
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
shapes seen in Fig. 1b are more difficult to detect than for the LaWO– NdWO inter-diffusion couple. Enhanced grain boundary diffusion for the W site cations can therefore not be discarded based on Fig. 1c alone. 3.2. Evaluation of concentration profiles
59
constant across the entire concentration profile. It is therefore necessary to calculate the concentration dependent diffusion coefficient ˜ ðC Þ using a full Boltzmann–Matano analysis [11–13]. D C
A representative example of the concentration profiles of La and Nd from EPMA analysis of an LaWO–NdWO inter-diffusion couple is shown in Fig. 3. As the cation transport in this case occurs in a concentration gradient, one cannot assume directly that the diffusivity is
a
x ˜ ðCx Þ ¼ − 1 · dx ∫ xdC D 2t dC C L
ð1Þ
In Eq. (1), x = 0 has been defined at the Matano-plane (xM), i.e. the plane across which an equal number of atoms have crossed in both directions, which can be determined using the requirement of conservation of mass (Eq. (2)) [13]: xM
∞
∫ ½C x −C L dx ¼ ∫ ½C R −C x dx −∞
ð2Þ
xM
CL and CR represent the initial concentration on the left and right sides of the diffusion couple, respectively, and Cx represents the concentration at position x. To evaluate the integral and the derivative in Eq. (1), the data were fitted to a cubic spline function that reproduces the concentration profile by which the original concentration profile can be repopulated with a higher point density [14]. Thus, it was possible to calculate the inter-diffusion coefficient as a function of concentration of the diffusing species, as shown in Fig. 4 for Nd (a) and Mo (b) diffusion for the different temperatures. The inter-diffusion coefficients are essentially independent of concentration, which indicates that they effectively reflect both species and both phases. On this basis, the inter-diffusion coefficient can be obtained directly from the concentration profile, using the thick film solution to Fick's second law with a constant source approximation [15];
50 µm
b
Cx;bulk ¼
6+
20 µm
W
20 µm
Mo
C R −C L x erfc qffiffiffiffiffiffiffiffiffiffiffi þ C L 2 e 2 D bulk t
ð3Þ
under the assumption that CL b CR. The concentration profiles were fitted to Eq. (3) to obtain the bulk inter-diffusion coefficient (Dbulk), generally with good agreement between the model and the experimental behaviour (cf. Fig. 3). The inter-diffusion coefficients derived from Eq. (3) are included in Fig. 4 as dotted lines and are evidently in accordance with the values from the Boltzmann–Matano analysis. Thereby we concluded that the thick film approximation yielded satisfactory results and could be applied to derive the bulk inter-diffusion coefficients.
c
6+
Fig. 2. a) BSE image and concentration maps of b) W and c) Mo over a LaWO–LWMo inter-diffusion couple after annealing at 1350 °C in dry air.
Fig. 3. Concentration profile for the LaWO–NdWO inter-diffusion couple after annealing at 1300 °C in air for 300 hours. The lines correspond to fitting of Eq. (4) to the data.
60
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
Fig. 4. Diffusion coefficient vs. Nd (a) and Mo (b) site fraction over the entire temperature range. Dotted lines show the values obtained through the thick film constant diffusivity approximation.
3.3. Grain boundary contribution The contribution from grain boundary transport, evident from the mapping of La in NdWO (cf. Fig. 1b), can be estimated by applying the approximate Whipple–Le Claire solution for in-diffusion from a constant source [16,17]:
wDgb
!−5 = rffiffiffiffiffiffiffiffiffiffi 3 Dbulk dlogC ¼ 0:3292 6 =5 t dx
ð4Þ
Here, w and Dgb are the grain boundary width and the grain boundary diffusion coefficient, respectively. Fig. 5 presents the logarithm of the La and Nd concentration vs. x 6/5 across the diffusion couple. Grain boundary transport can be recognized by the linear slope in the deeper parts of the La profile. The grain boundary contribution to the concentration profile can thus be expressed as; 6=5
logC x;gb ¼ a þ zx
ð5Þ 6=5 ;
where z is the factor dlogC=dx from Eq. (4). At any point, the concentration of the diffusing species is the sum of the bulk and grain boundary contribution: logC x ¼ log C x;bulk þ C x;gb
ð6Þ
log (Cx / Cbulk)
0
Calculations based on either pure bulk diffusion (Eq. (3)) or a combination of bulk and grain boundary diffusion (Eq. (6)) were conducted to evaluate the importance of grain boundary contributions (cf. Fig. 5). The La profile is better reproduced using the combined solution, while the Nd profile is reproduced equally well using both solutions, as was also the case for both W and Mo diffusion on the W site. This coincides well with the concentration maps in Figs. 1 and 2. Thus, it was concluded that enhanced grain boundary transport is only evident in the NdWO phase, while the concentration profiles in the three remaining phases are dominated by bulk diffusion. 3.4. Inter-diffusion coefficients Fig. 6 shows bulk and grain boundary inter-diffusion coefficients vs. 1/T as measured in dry air. The diffusivity in LaWO follows Arrhenius type behaviour with almost identical bulk diffusion coefficients for A and B site transport with activation energies in the order of 400 to 450 kJmol-1. As outlined above, enhanced grain boundary diffusion was only observed for A site diffusion, and restricted to diffusion of La3+ into NdWO. An effective grain boundary width of 1 nm has been assumed for the calculation of Dgb in NdWO. The grain boundary diffusivity is 5–6 orders of magnitude higher than the bulk diffusivity. Accordingly, the activation energy for grain boundary transport is—as expected—significantly lower than for bulk (170 ± 50 kJmol-1 [18]. Fig. 7a and b show bulk inter-diffusion coefficients measured in air and 5% H2 in argon for the LaWO–LWMo and LaWO–NdWO interdiffusion couples, respectively. The diffusivity under reducing conditions is higher than in air, with slightly lower apparent activation energy. The values for the pre-exponential and activation energy of the diffusion coefficients are summarized in Table 1. 4. Discussion
-1
4.1. Bulk diffusion -2
La Nd Bulk diffusion Bulk +gb diffusion
-3
-4
-30
-20
-10
0
x6/5
10
20
30
[µm6/5]
Fig. 5. Log (Cx/Cbulk) vs. x6/5 for a LaWO-NdWO inter-diffusion couple after annealing at 1300 °C in dry air for 300 h.
Considering the large difference in size and charge between La and W, it is intriguing to note their almost identical diffusion coefficients, independent of temperature. Similar behaviour is observed in the literature for some perovskites where the A and B site cations have substantially different coordination, size and charge, interpreted to reflect that the cations migrate via a common diffusion mechanism [19–21]. To evaluate possible diffusion mechanisms for La3+ and W6+, we first have to examine the LnWO structure more closely. The materials used in this work crystallize essentially in the same defective fluorite-type structure, with similar ionic radii for La3+ and Nd3+ on La site and W6+ and Mo6+ on W site. Thus, the results should effectively represent diffusion
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
61
Table 1 Pre-exponential factors and activation energies for the calculated inter-diffusion coefficients, according to D = D0*exp(−Ea/RT). Diffusion coefficient
Dbulk, La Dbulk, W Dgb,La
Fig. 6. Bulk and grain boundary diffusivity vs. 1/T for La and W site diffusion in dry air.
of La3+ and W6+ in the LnWO structure. There are two different La sites in the structure, La1(4a) and La2(24 g). The La1 sites are generally 8-fold coordinated in regular cubes and the La2 sites are 7-coordinated in distorted cubes. Hence, the La2 site is more flexible for substitution of the normally 6-fold coordinated W 6+ ion. It is also reasonable to assume that cation vacancies would preferentially form on the distorted La2 sites as opposed to the more stable La1 and W sites. The low fraction of W sites in the structure furthermore implies that all nearest-neighbours to a particular W site are La sites. Consequently, cation transport can only occur via interstitial sites or—more importantly here—via vacancies on the La2 sublattice, which is not only occupied by La but also by 1/5 of the W cations. Migration via interstitial sites has been shown to occur in fluorites with cation excess [22]. However, the enthalpy was very large (400– 500 kJmol-1), due to migration through a saddle point in the cubic oxygen sublattice. Thus, it would be expected that the activation energy of interstitial diffusion in LnWO should be considerably higher than what was
a
b
Fig. 7. Bulk diffusivity vs. 1/T for a) La and b) W site in air and 5% H2 in Ar (all dry).
Air
5% H2 in Ar
D 0/ cm2s−1
Ea/ kJmol-1
D0/ cm2s−1
Ea/ kJmol-1
100.2 ± 0.8 101.7 ± 1 10−2 ± 1.6
410 ± 20 450 ± 30 170 ± 50
10−3 ± 1 101.6 ± 0.5 –
300 ± 30 430 ± 15 –
observed in this work, considering an additional contribution from the formation enthalpy of cation interstitials. Furthermore, formation and diffusion of W6+ interstitials should be more favourable than for La3+ interstitials due to size considerations, and one would expect higher diffusivity and lower activation energy for the smaller cation if both were diffusing as interstitials. Thus, the similarity in diffusion coefficients and activation energies suggests that La3+ and W6+ diffuse via vacancies on the La2 sublattice, where both cations are present. Table 1 indicates that the pre-exponential factor for tungsten inter-diffusion is about one order of magnitude larger than for lanthanum inter-diffusion. However, estimating pre-exponential factors from logarithmic data introduces large uncertainties and the observed variation between the two values is not conclusive. From Fig. 7, it is evident that both La and W diffusion is higher in reducing compared to oxidizing atmospheres, with a larger increase in diffusivity for W migration. Additionally, the diffusivity was essentially constant going from air to argon at 1200 and 1300 °C. Generally, a change in diffusivity due to changes in pO2 reflects a change in concentration of the defect facilitating the transport. However, the observed increase in diffusivity does not correspond to any specific limiting case of defect chemistry and may rather be attributed to a change in valence state on some of the B site cations under reducing conditions. Such a change in valency could alter the migration properties in the material. This interpretation is supported by the larger increase in diffusivity for the LWMo–LaWO diffusion couple, wherein the Mo6+ is more readily reduced than W 6+ yielding a higher degree of distortion due to changes in valence state for the B site cations during inter-diffusion experiments under reducing conditions. 4.2. Grain boundary diffusion Evidence of enhanced grain boundary transport was only observed in the NdWO phase. The fact that grain boundary transport is limited to one particular phase, and not to one of the diffusing species, indicates that the behaviour is phase-specific and does not reflect a more favourable dynamic for La site diffusion along the grain boundaries. It has previously been shown that the Nd-containing rare earth tungsten oxides have a crystal structure which is less cubic than lanthanum-based tungsten oxides [23]. Thus, one may speculate whether there is a higher degree of misfit and rearrangement of the lattices along the grain boundaries in NdWO facilitating grain boundary diffusion. The observed differences in grain boundary transport could also simply have a micro-structural origin. From Figs. 1 and 2 it can be seen that the grain size in NdWO is significantly smaller than the other phases, which yields a higher concentration of grain boundaries. As the concentration profiles in this work were collected by means of line scans, a higher number of grain boundaries would influence the measurement profiles, making the contribution from grain boundary diffusion more visible than for the phases with larger grain sizes. Thus, the experimental approach in this work is not sufficient to conclude regarding the mechanism for grain boundary diffusion, or to exclude the possibility of enhanced grain boundary transport in the other lanthanum-based phases. Other experimental approaches, such as tracer diffusion studies using SIMS analysis, are called for to further clarify the mechanism and origin of grain boundary diffusion in the rare earth tungsten oxides.
62
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
H2 in Ar. Bulk diffusivities for La and W site were found to be similar at all temperatures, and lower than most oxide ion conducting membranes. It is proposed that both cations migrate via vacancies on the La2 sublattice. Evidence of enhanced grain boundary diffusion was only observed in the NdWO phase. The similar bulk diffusivities and relatively slow migration kinetics measured in this work indicate that the durability of a device using a LaWO membrane will not be limited by cation migration in the membrane itself.
Acknowledgments We acknowledge Muriel Marie Laure Erambert from the Department of Geology at the University of Oslo for acquisition of EPMA data. This publication has been produced with support from the BIGCCS Centre, performed under the Norwegian research program Centres for Environment-friendly Energy Research (FME). The authors acknowledge the following partners for their contributions: Aker Solutions, ConocoPhilips, Det Norske Veritas, Gassco, Hydro, Shell, Statkraft, Statoil, TOTAL, GDF SUEZ and the Research Council of Norway (193816/S60).
References Fig. 8. Compilation of available data on bulk inter-diffusion coefficients, as compared with tracer diffusion of 96Zr in ZrO2.
[1] [2] [3] [4]
4.3. Stability of LnWO-type materials in chemical gradients
[5] [6]
As outlined in the introduction, long term stability of fuel cell and gas separation membranes depends on cation diffusion in the membrane material. From the inter-diffusion coefficients in LaWO, it can be concluded that kinetic demixing and decomposition based on bulk diffusion is unlikely since the cations exhibits similar diffusivities. The bulk diffusivities in LaWO are lower than those reported for some commercial oxygen transport membranes, except for yttria-stabilized zirconia (see Fig. 8) [24–27]. These results indicate that LaWO-based membranes will be more stable towards degradation due to cation migration as compared to most state-of-the-art oxide ion conducting oxides. Based on a first approximation, applying the bulk diffusivities obtained in this work, it would in fact take 8 years for a 10 μm LaWO membrane to displace 100 nm under a gradient between hydrogen and air at 800 °C. Consequently one may anticipate that the life-time of LnWO membranes would not be limited by cation diffusion. One should bear in mind though that these are only the first reported results on cation transport in rare earth tungsten oxides. More research is needed to enhance our understanding of the transport mechanisms and to further clarify effects of grain boundary diffusion. 5. Conclusions The La and W site migration properties in LaWO have been studied by inter-diffusion measurements on LaWO–NdWO and LaWO–LWMo diffusion couples at temperatures from 1150 to 1350 °C in air and 5%
[7]
[8] [9] [10] [11] [12] [13] [14]
[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
T. Shimura, S. Fujimoto, H. Iwahara, Solid State Ionics 143 (1) (2001) 117. R. Haugsrud, Solid State Ionics 178 (7–10) (2007) 555. R. Haugsrud, C. Kjølseth, J. Phys. Chem. Solids 69 (7) (2008) 1758. A. Magraso, C. Frontera, D. Marrero-Lopez, P. Nunez, Dalton Trans. (46) (2009) 10273. S. Escolástico, C. Solís, J.M. Serra, Int. J. Hydrogen Energy 36 (18) (2011) 11946. S. Erdal, L.E.Q. Kalland, R. Hancke, J. Polfus, R. Haugsrud, T. Norby, A. Magrasó, Int. J. Hydrogen Energy 37 (2012) 8051. A. Magraso, J.M. Polfus, C. Frontera, J. Canales-Vazquez, L.-E. Kalland, C.H. Hervoches, S. Erdal, R. Hancke, M.S. Islam, T. Norby, R. Haugsrud, J. Mater. Chem. 22 (5) (2012) 1762. M. Martin, J. Chem. Thermodyn. 35 (8) (2003) 1291. H. Schmalzried, W. Laqua, Oxid. Met. 15 (3) (1981) 339. L.G. Harrison, Trans. Faraday Soc. 57 (1961) 1191. H. Mehrer, Diffusion in Solids: Fundamentals, Methods, Materials, Diffusion-Controlled Processes, Springer, 2007. L. Boltzmann, Wiedemanns Ann. Phys. 5 (1894) 959. C. Matano, Jpn. J. Phys. 8 (1933) 109. C. Holzapfel, Fe-Mg interdiffusion at high pressures in mineral phases relevant for the Earth's mantle, Mathematisch-Naturwissenschaftlichen Fakultät, University of Cologne, 2004. 11. J. Crank, The mathematics of diffusion, Clarendon Press, 1979. A.D.L. Claire, Br. J. Appl. Phys. 14 (6) (1963) 351. R.T. Whipple, Philos. Mag. A 45 (1954) 1225. P. Kofstad, Nonstoichiometry, diffusion, and electrical conductivity in binary metal oxides, Wiley-Interscience, 1972. S. Koerfer, R.A. De Souza, H.-I. Yoo, M. Martin, Solid State Sci. 10 (6) (2008) 725. O. Schulz, M. Martin, C. Argirusis, G. Borchardt, Phys. Chem. Chem. Phys. 5 (11) (2003) 2308. H.-I. Yoo, C.-E. Lee, R.A. De Souza, M. Martin, Appl. Phys. Lett. 92 (25) (2008) 252103. P.-L. Chen, I.W. Chen, J. Am. Ceram. Soc. 77 (9) (1994) 2289. M. Amsif, A. Magrasó, D. Marrero-López, J.C. Ruiz-Morales, J. Canales-Vázquez, P. Núñez, Chem. Mater. 24 (20) (2012) 3868. J.B. Smith, T. Norby, A. Fossdal, J. Am. Ceram. Soc. 89 (2) (2006) 582. M. Palcut, R. Knibbe, K. Wiik, T. Grande, Solid State Ionics 202 (1) (2011) 6. M. Kilo, M.A. Taylor, C. Argirusis, G. Borchardt, B. Lesage, S. Weber, S. Scherrer, H. Scherrer, M. Schroeder, M. Martin, J. Appl. Phys. 94 (12) (2003) 7547. N. Cebasek, Cation Diffusion and Related Degradation Phenomena in La2NiO4+d for Oxygen Separation Membranes, Thesis at Department of Chemistry, University of Oslo, 2012.
Contents lists available at SciVerse ScienceDirect
Solid State Ionics journal homepage: www.elsevier.com/locate/ssi
Inter-diffusion in lanthanum tungsten oxide Einar Vøllestad, Truls Norby, Reidar Haugsrud ⁎ University of Oslo, Department of Chemistry, Centre for Materials Science and Nanotechnology, FERMiO, Gaustadalleen 21, 0349 Oslo, Norway
a r t i c l e
i n f o
Article history: Received 27 February 2013 Received in revised form 19 April 2013 Accepted 23 April 2013 Keywords: Cation diffusion Lanthanum tungstate Proton conductor Hydrogen membrane Demixing Walkout
a b s t r a c t Cation diffusion in lanthanum tungsten oxide (LaWO) has been studied on La27W5O55.5–Nd27W5O55.5 and La27W5O55.5–La27Mo1.5W3.5O55 inter-diffusion couples at temperatures between 1150 and 1350 °C in air and 5% H2 in Ar. Inter-diffusion coefficients were derived by fitting concentration profiles of Nd, La, W, and Mo from Electron Probe Micro Analysis to the appropriate solution of Fick's second law. The diffusivity in LaWO follows Arrhenius type behaviour with almost identical bulk diffusion coefficients for La and W site transport, and similar activation energies; 410 ± 30 kJmol-1 (La) and 450 ± 30 kJmol-1 (W). This behaviour has been interpreted to reflect that migration occurs via a common migration mechanism for both species, facilitated by vacancies on the lanthanum sublattice. Enhanced grain boundary diffusion was only observed for La site diffusion in the Nd27W5O55.5 phase, with much lower activation energy (170 ± 50 kJmol-1). The similar bulk diffusivities and relatively slow migration kinetics indicate that LaWO is more stable towards cation diffusion-related degradation than many of the most promising oxygen transport membrane materials. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Rare earth tungsten oxides with nominal formula close to Ln6WO12 exhibit considerable mixed ionic and electronic conductivity at elevated temperatures [1–4]. Below 800 °C under wet conditions, the ionic conductivity is for a large part protonic. The most promising member, lanthanum tungsten oxide (LaWO), exhibits essentially pure proton conductivity at 600 °C in the order of 0.001 Scm − 1, while oxide ion and—under reducing and oxidizing conditions—electronic conductivity become significant above 700 °C. Despite the high contents of rare earth cations, the rare earth tungsten oxides (LnWO) are stable in acidic gases such as CO2, contrary to most alkaline earth-based perovskite structured proton conductors [5]. These characteristics make LaWO an interesting candidate material for components in high temperature electrochemical energy conversion devices and hydrogen separation membranes. LaWO, with a La/W-ratio of 5.3–5.7, can be described as a defective fluorite-type structure, crystallizing in the F43m space group [4]. Based on recent investigations of the structural and functional characteristics [4,6,7] it has been postulated that the defective (partially occupied) oxygen sublattice is inherently disordered and can be described with an effective partial negative charge on the oxygen ions, charge compensated by correspondingly positive oxygen vacancies and an excess of tungsten residing on La sites acting as donors [7]. Hence, the stoichiometry of the LaWO structure can more correctly be represented as La28−xW4 + xO54 + 3x/2v2−3x/2 (v is a vacant oxygen site), with x ⁎ Corresponding author at: FERMiO, Forskningspraken, Gaustadalleen 21, 0349 Oslo, Norway. Tel.: +47 22840659; fax: +47 22840651. E-mail address: [email protected] (R. Haugsrud). 0167-2738/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ssi.2013.04.021
typically ranging from 0.78 to 1.08. For a more detailed description of the defect chemistry in LaWO, see Ref. [6]. Even though LaWO is chemically stable towards acidic gases, there may still be issues regarding long-term stability under electrolyte and gas separation membrane operating conditions due to cation diffusion. Gradients in the chemical potential of hydrogen will in many cases, depending on water vapour pressures, impose gradients in the chemical potential of the constituent metal cations according to the Gibbs– Duhem relation. These gradients will then act as driving forces for cation transport, and the constituent metal cations of the membrane thus migrate towards the high pO2 side [8,9]. If there is significant difference in the diffusivity of the cations, accumulation of the fastest and slowest moving species may additionally occur on the high and low pO2 side of the membrane, respectively. Thus, the material becomes inhomogeneous, potentially deteriorating the performance of the membrane. When the accumulation exceeds the cation solid solubility in the material during demixing, decomposition will occur on one or both sides of the membrane. Over time, these processes may lead to walkout, loss of sealing or decreased membrane performance, which ultimately would result in breakdown of the membrane. These degradation processes underline the importance of investigations on cation transport to evaluate the durability of fuel cell and hydrogen transport membrane candidate materials. Despite this, there are so far no data on cation transport for proton conducting oxides, presumably due to the relatively immature state of technologies based on such functional ceramics. Clearly, increased emphasis on long-term stability is required when moving towards commercialization of proton ceramic fuel cells, proton ceramic electrolyser cells, and hydrogen transport membranes. In this work, diffusion couples of La27W5O55.5–Nd27W5O55.5 (LaWO–NdWO) and La27W5O55.5–La27Mo1.5W3.5O55.5 (LaWO–LWMo)
58
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
were studied to determine the diffusivity of La and W represented by La–Nd and W–Mo inter-diffusion coefficients. Concentration profiles of the respective cations that develop during annealing under different temperatures and oxygen partial pressures were determined by means of Electron Probe Micro Analysis (EPMA). Inter-diffusion coefficients were derived by analysing these profiles according to Fick's second law. The functional dependencies of the diffusion data on the external conditions are discussed in light of the complex structure and defect chemistry of the LaWO-based material.
a
2. Experimental NdWO powder was synthesized using a wet chemical approach: Stoichiometric amounts of pre-dried Nd2O3 and WO3 powders were dissolved in aqueous HNO3 and NH3, respectively, before EDTA was added to each mixture in a 1:1 molar ratio between the sum of the metal cations and EDTA. The separate solutions were brought to pH = 7 and mixed, followed by drying at 200 °C. The resulting xerogel was calcined at 600 °C, before the powder finally was annealed at 1000 °C for 5 h. The powder was subsequently milled at 200 rpm in an all-agate planetary ball-mill for 1 h to promote a narrow particle size distribution. The LaWO and LWMo powders were available commercially (Cerpotech, Norway). Green bodies of diameter 13 mm were made by uniaxial pressing and sintered at 1500 °C for 3 h yielding dense pellets (>96% of relative density) with 8–30 μm large grains, except for NdWO which had grain sizes of 5–12 μm. To promote proper contact between the materials during the diffusion annealing, the pellets were first ground and then polished down to a 0.25 μm surface finish with diamond abrasives. Mounted samples were held in contact by an alumina spring load. The inter-diffusion annealing experiments were performed for 180– 400 h at temperatures between 1150 and 1350 °C in an atmosphere of air, argon or 5% H2 in argon (all dry). After annealing, the inter-diffusion couples were mounted in epoxy resin, cut perpendicular to the original interface and polished down to a 0.25 μm surface finish with diamond abrasive. Concentration profiles of Nd, La, W, and Mo were determined by EPMA (Cameca SX 100) using an acceleration voltage and beam current of 15 kV and 20 nA, respectively. The detection limit of EPMA is 100– 1000 ppm depending on the specific element. Each cross-section was analysed by determining the concentration of the relevant species every 1 μm along 5 different lines normal to the phase boundary. Typical scans penetrated 25–30 μm into each phase to ensure that 2/3 of the concentration profiles originated from outside of the diffusion zone.
50 µm
b
20 µm
3+
La
c
3. Results 3.1. Microstructure Fig. 1a shows a backscattered electron (BSE) image of a cross-section from an inter-diffusion couple between NdWO (top) and LaWO (bottom) annealed in dry air at 1350 °C for 300 h. The original boundary is no longer visible and can only be recognized by a small change in microstructure between the two phases. Small dark La-rich needles (c.f. Fig. 1a) formed along the grain boundaries during the annealing. Dark spots represent porosity, whereas the dark line perpendicular to the phase boundary originates from one of the EPMA concentration profiles. Locations of the EPMA profiles were chosen so as to minimize the influence of secondary phases and porosity. Fig. 1b and c show X-ray maps of La and Nd reflecting their concentrations. The mosaic shape of the La distribution in the deeper end of the diffusion zone in Fig. 1b is indicative of enhanced transport along the grain boundaries in NdWO. This behaviour is representative of what according to Harrison's classification is termed type-B kinetics [10], where both bulk and grain boundary diffusion contribute significantly to the transport in the intermediate regime of the diffusion zone. For
20 µm
3+
Nd
Fig. 1. a) BSE image and concentration maps of b) La and c) Nd over a NdWO–LaWO inter-diffusion couple after annealing at 1350 °C in dry air.
diffusion of Nd into LaWO, the mapping (cf. Fig. 1c) reveals a homogeneous distribution of the elements parallel to the phase boundary. Fig. 2a shows a similar BSE image of an LWMo–LaWO diffusion couple annealed at 1350 °C in air for 300 h. Similar as for the LaWO–NdWO diffusion couple the original phase boundary is no longer visible, but here there is no clear difference in microstructure. Fig. 2b and c show maps representing the Mo and W concentrations. There is no indication of enhanced transport along the grain boundaries in either of the phases. However, due to a smaller difference in W and Mo concentration between the phases—and thus less contrast—distinct features such as the mosaic
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
shapes seen in Fig. 1b are more difficult to detect than for the LaWO– NdWO inter-diffusion couple. Enhanced grain boundary diffusion for the W site cations can therefore not be discarded based on Fig. 1c alone. 3.2. Evaluation of concentration profiles
59
constant across the entire concentration profile. It is therefore necessary to calculate the concentration dependent diffusion coefficient ˜ ðC Þ using a full Boltzmann–Matano analysis [11–13]. D C
A representative example of the concentration profiles of La and Nd from EPMA analysis of an LaWO–NdWO inter-diffusion couple is shown in Fig. 3. As the cation transport in this case occurs in a concentration gradient, one cannot assume directly that the diffusivity is
a
x ˜ ðCx Þ ¼ − 1 · dx ∫ xdC D 2t dC C L
ð1Þ
In Eq. (1), x = 0 has been defined at the Matano-plane (xM), i.e. the plane across which an equal number of atoms have crossed in both directions, which can be determined using the requirement of conservation of mass (Eq. (2)) [13]: xM
∞
∫ ½C x −C L dx ¼ ∫ ½C R −C x dx −∞
ð2Þ
xM
CL and CR represent the initial concentration on the left and right sides of the diffusion couple, respectively, and Cx represents the concentration at position x. To evaluate the integral and the derivative in Eq. (1), the data were fitted to a cubic spline function that reproduces the concentration profile by which the original concentration profile can be repopulated with a higher point density [14]. Thus, it was possible to calculate the inter-diffusion coefficient as a function of concentration of the diffusing species, as shown in Fig. 4 for Nd (a) and Mo (b) diffusion for the different temperatures. The inter-diffusion coefficients are essentially independent of concentration, which indicates that they effectively reflect both species and both phases. On this basis, the inter-diffusion coefficient can be obtained directly from the concentration profile, using the thick film solution to Fick's second law with a constant source approximation [15];
50 µm
b
Cx;bulk ¼
6+
20 µm
W
20 µm
Mo
C R −C L x erfc qffiffiffiffiffiffiffiffiffiffiffi þ C L 2 e 2 D bulk t
ð3Þ
under the assumption that CL b CR. The concentration profiles were fitted to Eq. (3) to obtain the bulk inter-diffusion coefficient (Dbulk), generally with good agreement between the model and the experimental behaviour (cf. Fig. 3). The inter-diffusion coefficients derived from Eq. (3) are included in Fig. 4 as dotted lines and are evidently in accordance with the values from the Boltzmann–Matano analysis. Thereby we concluded that the thick film approximation yielded satisfactory results and could be applied to derive the bulk inter-diffusion coefficients.
c
6+
Fig. 2. a) BSE image and concentration maps of b) W and c) Mo over a LaWO–LWMo inter-diffusion couple after annealing at 1350 °C in dry air.
Fig. 3. Concentration profile for the LaWO–NdWO inter-diffusion couple after annealing at 1300 °C in air for 300 hours. The lines correspond to fitting of Eq. (4) to the data.
60
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
Fig. 4. Diffusion coefficient vs. Nd (a) and Mo (b) site fraction over the entire temperature range. Dotted lines show the values obtained through the thick film constant diffusivity approximation.
3.3. Grain boundary contribution The contribution from grain boundary transport, evident from the mapping of La in NdWO (cf. Fig. 1b), can be estimated by applying the approximate Whipple–Le Claire solution for in-diffusion from a constant source [16,17]:
wDgb
!−5 = rffiffiffiffiffiffiffiffiffiffi 3 Dbulk dlogC ¼ 0:3292 6 =5 t dx
ð4Þ
Here, w and Dgb are the grain boundary width and the grain boundary diffusion coefficient, respectively. Fig. 5 presents the logarithm of the La and Nd concentration vs. x 6/5 across the diffusion couple. Grain boundary transport can be recognized by the linear slope in the deeper parts of the La profile. The grain boundary contribution to the concentration profile can thus be expressed as; 6=5
logC x;gb ¼ a þ zx
ð5Þ 6=5 ;
where z is the factor dlogC=dx from Eq. (4). At any point, the concentration of the diffusing species is the sum of the bulk and grain boundary contribution: logC x ¼ log C x;bulk þ C x;gb
ð6Þ
log (Cx / Cbulk)
0
Calculations based on either pure bulk diffusion (Eq. (3)) or a combination of bulk and grain boundary diffusion (Eq. (6)) were conducted to evaluate the importance of grain boundary contributions (cf. Fig. 5). The La profile is better reproduced using the combined solution, while the Nd profile is reproduced equally well using both solutions, as was also the case for both W and Mo diffusion on the W site. This coincides well with the concentration maps in Figs. 1 and 2. Thus, it was concluded that enhanced grain boundary transport is only evident in the NdWO phase, while the concentration profiles in the three remaining phases are dominated by bulk diffusion. 3.4. Inter-diffusion coefficients Fig. 6 shows bulk and grain boundary inter-diffusion coefficients vs. 1/T as measured in dry air. The diffusivity in LaWO follows Arrhenius type behaviour with almost identical bulk diffusion coefficients for A and B site transport with activation energies in the order of 400 to 450 kJmol-1. As outlined above, enhanced grain boundary diffusion was only observed for A site diffusion, and restricted to diffusion of La3+ into NdWO. An effective grain boundary width of 1 nm has been assumed for the calculation of Dgb in NdWO. The grain boundary diffusivity is 5–6 orders of magnitude higher than the bulk diffusivity. Accordingly, the activation energy for grain boundary transport is—as expected—significantly lower than for bulk (170 ± 50 kJmol-1 [18]. Fig. 7a and b show bulk inter-diffusion coefficients measured in air and 5% H2 in argon for the LaWO–LWMo and LaWO–NdWO interdiffusion couples, respectively. The diffusivity under reducing conditions is higher than in air, with slightly lower apparent activation energy. The values for the pre-exponential and activation energy of the diffusion coefficients are summarized in Table 1. 4. Discussion
-1
4.1. Bulk diffusion -2
La Nd Bulk diffusion Bulk +gb diffusion
-3
-4
-30
-20
-10
0
x6/5
10
20
30
[µm6/5]
Fig. 5. Log (Cx/Cbulk) vs. x6/5 for a LaWO-NdWO inter-diffusion couple after annealing at 1300 °C in dry air for 300 h.
Considering the large difference in size and charge between La and W, it is intriguing to note their almost identical diffusion coefficients, independent of temperature. Similar behaviour is observed in the literature for some perovskites where the A and B site cations have substantially different coordination, size and charge, interpreted to reflect that the cations migrate via a common diffusion mechanism [19–21]. To evaluate possible diffusion mechanisms for La3+ and W6+, we first have to examine the LnWO structure more closely. The materials used in this work crystallize essentially in the same defective fluorite-type structure, with similar ionic radii for La3+ and Nd3+ on La site and W6+ and Mo6+ on W site. Thus, the results should effectively represent diffusion
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
61
Table 1 Pre-exponential factors and activation energies for the calculated inter-diffusion coefficients, according to D = D0*exp(−Ea/RT). Diffusion coefficient
Dbulk, La Dbulk, W Dgb,La
Fig. 6. Bulk and grain boundary diffusivity vs. 1/T for La and W site diffusion in dry air.
of La3+ and W6+ in the LnWO structure. There are two different La sites in the structure, La1(4a) and La2(24 g). The La1 sites are generally 8-fold coordinated in regular cubes and the La2 sites are 7-coordinated in distorted cubes. Hence, the La2 site is more flexible for substitution of the normally 6-fold coordinated W 6+ ion. It is also reasonable to assume that cation vacancies would preferentially form on the distorted La2 sites as opposed to the more stable La1 and W sites. The low fraction of W sites in the structure furthermore implies that all nearest-neighbours to a particular W site are La sites. Consequently, cation transport can only occur via interstitial sites or—more importantly here—via vacancies on the La2 sublattice, which is not only occupied by La but also by 1/5 of the W cations. Migration via interstitial sites has been shown to occur in fluorites with cation excess [22]. However, the enthalpy was very large (400– 500 kJmol-1), due to migration through a saddle point in the cubic oxygen sublattice. Thus, it would be expected that the activation energy of interstitial diffusion in LnWO should be considerably higher than what was
a
b
Fig. 7. Bulk diffusivity vs. 1/T for a) La and b) W site in air and 5% H2 in Ar (all dry).
Air
5% H2 in Ar
D 0/ cm2s−1
Ea/ kJmol-1
D0/ cm2s−1
Ea/ kJmol-1
100.2 ± 0.8 101.7 ± 1 10−2 ± 1.6
410 ± 20 450 ± 30 170 ± 50
10−3 ± 1 101.6 ± 0.5 –
300 ± 30 430 ± 15 –
observed in this work, considering an additional contribution from the formation enthalpy of cation interstitials. Furthermore, formation and diffusion of W6+ interstitials should be more favourable than for La3+ interstitials due to size considerations, and one would expect higher diffusivity and lower activation energy for the smaller cation if both were diffusing as interstitials. Thus, the similarity in diffusion coefficients and activation energies suggests that La3+ and W6+ diffuse via vacancies on the La2 sublattice, where both cations are present. Table 1 indicates that the pre-exponential factor for tungsten inter-diffusion is about one order of magnitude larger than for lanthanum inter-diffusion. However, estimating pre-exponential factors from logarithmic data introduces large uncertainties and the observed variation between the two values is not conclusive. From Fig. 7, it is evident that both La and W diffusion is higher in reducing compared to oxidizing atmospheres, with a larger increase in diffusivity for W migration. Additionally, the diffusivity was essentially constant going from air to argon at 1200 and 1300 °C. Generally, a change in diffusivity due to changes in pO2 reflects a change in concentration of the defect facilitating the transport. However, the observed increase in diffusivity does not correspond to any specific limiting case of defect chemistry and may rather be attributed to a change in valence state on some of the B site cations under reducing conditions. Such a change in valency could alter the migration properties in the material. This interpretation is supported by the larger increase in diffusivity for the LWMo–LaWO diffusion couple, wherein the Mo6+ is more readily reduced than W 6+ yielding a higher degree of distortion due to changes in valence state for the B site cations during inter-diffusion experiments under reducing conditions. 4.2. Grain boundary diffusion Evidence of enhanced grain boundary transport was only observed in the NdWO phase. The fact that grain boundary transport is limited to one particular phase, and not to one of the diffusing species, indicates that the behaviour is phase-specific and does not reflect a more favourable dynamic for La site diffusion along the grain boundaries. It has previously been shown that the Nd-containing rare earth tungsten oxides have a crystal structure which is less cubic than lanthanum-based tungsten oxides [23]. Thus, one may speculate whether there is a higher degree of misfit and rearrangement of the lattices along the grain boundaries in NdWO facilitating grain boundary diffusion. The observed differences in grain boundary transport could also simply have a micro-structural origin. From Figs. 1 and 2 it can be seen that the grain size in NdWO is significantly smaller than the other phases, which yields a higher concentration of grain boundaries. As the concentration profiles in this work were collected by means of line scans, a higher number of grain boundaries would influence the measurement profiles, making the contribution from grain boundary diffusion more visible than for the phases with larger grain sizes. Thus, the experimental approach in this work is not sufficient to conclude regarding the mechanism for grain boundary diffusion, or to exclude the possibility of enhanced grain boundary transport in the other lanthanum-based phases. Other experimental approaches, such as tracer diffusion studies using SIMS analysis, are called for to further clarify the mechanism and origin of grain boundary diffusion in the rare earth tungsten oxides.
62
E. Vøllestad et al. / Solid State Ionics 244 (2013) 57–62
H2 in Ar. Bulk diffusivities for La and W site were found to be similar at all temperatures, and lower than most oxide ion conducting membranes. It is proposed that both cations migrate via vacancies on the La2 sublattice. Evidence of enhanced grain boundary diffusion was only observed in the NdWO phase. The similar bulk diffusivities and relatively slow migration kinetics measured in this work indicate that the durability of a device using a LaWO membrane will not be limited by cation migration in the membrane itself.
Acknowledgments We acknowledge Muriel Marie Laure Erambert from the Department of Geology at the University of Oslo for acquisition of EPMA data. This publication has been produced with support from the BIGCCS Centre, performed under the Norwegian research program Centres for Environment-friendly Energy Research (FME). The authors acknowledge the following partners for their contributions: Aker Solutions, ConocoPhilips, Det Norske Veritas, Gassco, Hydro, Shell, Statkraft, Statoil, TOTAL, GDF SUEZ and the Research Council of Norway (193816/S60).
References Fig. 8. Compilation of available data on bulk inter-diffusion coefficients, as compared with tracer diffusion of 96Zr in ZrO2.
[1] [2] [3] [4]
4.3. Stability of LnWO-type materials in chemical gradients
[5] [6]
As outlined in the introduction, long term stability of fuel cell and gas separation membranes depends on cation diffusion in the membrane material. From the inter-diffusion coefficients in LaWO, it can be concluded that kinetic demixing and decomposition based on bulk diffusion is unlikely since the cations exhibits similar diffusivities. The bulk diffusivities in LaWO are lower than those reported for some commercial oxygen transport membranes, except for yttria-stabilized zirconia (see Fig. 8) [24–27]. These results indicate that LaWO-based membranes will be more stable towards degradation due to cation migration as compared to most state-of-the-art oxide ion conducting oxides. Based on a first approximation, applying the bulk diffusivities obtained in this work, it would in fact take 8 years for a 10 μm LaWO membrane to displace 100 nm under a gradient between hydrogen and air at 800 °C. Consequently one may anticipate that the life-time of LnWO membranes would not be limited by cation diffusion. One should bear in mind though that these are only the first reported results on cation transport in rare earth tungsten oxides. More research is needed to enhance our understanding of the transport mechanisms and to further clarify effects of grain boundary diffusion. 5. Conclusions The La and W site migration properties in LaWO have been studied by inter-diffusion measurements on LaWO–NdWO and LaWO–LWMo diffusion couples at temperatures from 1150 to 1350 °C in air and 5%
[7]
[8] [9] [10] [11] [12] [13] [14]
[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
T. Shimura, S. Fujimoto, H. Iwahara, Solid State Ionics 143 (1) (2001) 117. R. Haugsrud, Solid State Ionics 178 (7–10) (2007) 555. R. Haugsrud, C. Kjølseth, J. Phys. Chem. Solids 69 (7) (2008) 1758. A. Magraso, C. Frontera, D. Marrero-Lopez, P. Nunez, Dalton Trans. (46) (2009) 10273. S. Escolástico, C. Solís, J.M. Serra, Int. J. Hydrogen Energy 36 (18) (2011) 11946. S. Erdal, L.E.Q. Kalland, R. Hancke, J. Polfus, R. Haugsrud, T. Norby, A. Magrasó, Int. J. Hydrogen Energy 37 (2012) 8051. A. Magraso, J.M. Polfus, C. Frontera, J. Canales-Vazquez, L.-E. Kalland, C.H. Hervoches, S. Erdal, R. Hancke, M.S. Islam, T. Norby, R. Haugsrud, J. Mater. Chem. 22 (5) (2012) 1762. M. Martin, J. Chem. Thermodyn. 35 (8) (2003) 1291. H. Schmalzried, W. Laqua, Oxid. Met. 15 (3) (1981) 339. L.G. Harrison, Trans. Faraday Soc. 57 (1961) 1191. H. Mehrer, Diffusion in Solids: Fundamentals, Methods, Materials, Diffusion-Controlled Processes, Springer, 2007. L. Boltzmann, Wiedemanns Ann. Phys. 5 (1894) 959. C. Matano, Jpn. J. Phys. 8 (1933) 109. C. Holzapfel, Fe-Mg interdiffusion at high pressures in mineral phases relevant for the Earth's mantle, Mathematisch-Naturwissenschaftlichen Fakultät, University of Cologne, 2004. 11. J. Crank, The mathematics of diffusion, Clarendon Press, 1979. A.D.L. Claire, Br. J. Appl. Phys. 14 (6) (1963) 351. R.T. Whipple, Philos. Mag. A 45 (1954) 1225. P. Kofstad, Nonstoichiometry, diffusion, and electrical conductivity in binary metal oxides, Wiley-Interscience, 1972. S. Koerfer, R.A. De Souza, H.-I. Yoo, M. Martin, Solid State Sci. 10 (6) (2008) 725. O. Schulz, M. Martin, C. Argirusis, G. Borchardt, Phys. Chem. Chem. Phys. 5 (11) (2003) 2308. H.-I. Yoo, C.-E. Lee, R.A. De Souza, M. Martin, Appl. Phys. Lett. 92 (25) (2008) 252103. P.-L. Chen, I.W. Chen, J. Am. Ceram. Soc. 77 (9) (1994) 2289. M. Amsif, A. Magrasó, D. Marrero-López, J.C. Ruiz-Morales, J. Canales-Vázquez, P. Núñez, Chem. Mater. 24 (20) (2012) 3868. J.B. Smith, T. Norby, A. Fossdal, J. Am. Ceram. Soc. 89 (2) (2006) 582. M. Palcut, R. Knibbe, K. Wiik, T. Grande, Solid State Ionics 202 (1) (2011) 6. M. Kilo, M.A. Taylor, C. Argirusis, G. Borchardt, B. Lesage, S. Weber, S. Scherrer, H. Scherrer, M. Schroeder, M. Martin, J. Appl. Phys. 94 (12) (2003) 7547. N. Cebasek, Cation Diffusion and Related Degradation Phenomena in La2NiO4+d for Oxygen Separation Membranes, Thesis at Department of Chemistry, University of Oslo, 2012.