The microstructural instability of Ce1–xAxNbO4±δ under reducing atmosphere is
suggested to be associated with variations of the cerium valence and subsequent volume expansion similar to CeO2; the degree of which, was shown to reduce with
increasing dopant additions. However, even after 40% and 50% Gd and Er doping respectively, the increase in cell volume upon reduction was reported to be approx- imately 1% and not completely effective at preventing microstructural degradation [50]. This might be expected given that virtually all cerium sites in CeO2 occupy the tetravalent state under ambient conditions, so reductive annealing has the po- tential to significantly alter the net valence and oxygen stoichiometry even after heavy doping. More significant microcracking might be expected for CeNbO4+δ based materials after successive heating and cooling cycles due to fluctuations of the cerium valence as the material alters its stoichiometry as a function of temperature and polymorphic forms. However the loss of structural integrity was quite signif- icant for Ce1–xAxNbO4±δ heated at 800 °C under 5% H2/N2(g) after just several
hours. Chapter 5 showed that at this temperature, the sample reduces leaving only compensatory Ce4+, the level of which is much lower than CeO2.
This implies another factor is significant in determining the stability. For example Ce4+ may have a tendency to also segregate to the grain boundary core or sur- rounding region so that upon reduction, perhaps the combined effects of volume expansion, anisotropic thermal expansion and hydrogen diffusion act to cleave the boundary. Substitution of the A site with a redox stable species is expected to re- duce cell expansion and improve stability, but would naturally alter the properties of Ce1–xAxNbO4±δ. As shall be discussed in more detail in 6.4, previous electrical
measurements performed on LaNbO4 under reducing conditions have not indicated
phase or microstructural instability. Therefore isovalent lanthanum substitutions may help stabilise Ce1–xAxNbO4±δ whilst retaining levels of electronic conductivity
(Chapter 7). Unlike CeO2 this is not expected to alter the extrinsic defect concen-
tration (or defect interactions), but may reduce dopant solubility, recalling the work of Norby et al who suggested LaNbO4 has a solubility limit of ≈2% [231].
X-ray diffraction patterns with corresponding lattice parameters for Ce0.9–xLaxSr0.1NbO4±δ doped with 20% and 70% lanthanum on the A site
are presented in Figure 6.14 and Table 6.7. Ce0.7La0.2Sr0.1NbO4±δ could be
attributable to Sr2Nb2O7 and which is clearly seen in the pattern of as-prepared Ce0.2La0.7Sr0.1NbO4±δ showing that the solubility limit reduces with lanthanum
additions.
The cell volume also increases with dopant concentration, consistent with the re- moval of both smaller Ce3+ and compensatory Ce4+. As expected, exposure of Ce0.9–xLaxSr0.1NbO4 ± δ to 5% H2/N2(g) for 12 hours lead to an increase of cell
volume, which is again greater for Ce0.7La0.2Sr0.1NbO4±δ due to the higher solu-
bility of the dopant. For Ce0.2La0.7Sr0.1NbO4±δ this is accompanied by an increase in the approximate phase fraction of Sr2Nb2O7 phase, and which was confirmed
by comparison of the estimated integrated intensities. The patterns post reduction show preferential orientation of the (040) reflection, the top of which has been cut for clarity. Reductive annealing may have changed the particle surface energy and influenced particle packing in the flat plate. Further experimentation with Rietveld analysis is required to understand the defect chemistry of Ce0.9–xLaxSr0.1NbO4 ± δ,
however it is feasible that vacancies may be stable under oxidising conditions and mixed compensation mechanisms are occurring.
Figure 6.14: X-ray diffraction patterns of as-prepared Ce0.9–xLaxSr0.1NbO4 ± δ and after
reductive annealing in 5% H2/N2(g)at 800°C for 12 hours. Asterix highlights a reflection corresponding to secondary Sr2Nb2O7 phase.
Ce0.7La0.2Sr0.1NbO4±δ a(˚A) b(˚A) c(˚A) β (°) Vol(˚A3)
As-prepared 7.2656(1) 11.3996(2) 5.1640(1) 130.61(1) 324.70(9) Reduced 7.3105(2) 11.4499(3) 5.1854(2) 130.998(1) 327.59(2)
Ce0.2La0.7Sr0.1NbO4±δ a(˚A) b(˚A) c(˚A) β (°) Vol(˚A3)
As-prepared 7.3123(1) 11.4708(2) 5.1874(1) 130.80(1) 329.40(9) Reduced 7.3441(3) 11.5042(4) 5.2027(2) 131.06(1) 331.47(2)
Table 6.10: Lattice parameters of as-prepared Ce0.9–xLaxSr0.1NbO4±δ and after 12 hours
exposure to 5% H2/N2(g) obtained by Le-Bail refinement.
SEM micrographs of Ce0.9–xLaxSr0.1NbO4 ± δ (Figure 6.15) confirm the presence of
secondary phases for both compositions. The atomic percentages of each constituent determined by EDXS (Table 6.8) are in reasonable agreement with that of the tar- geted composition Ce0.7La0.2Sr0.1NbO4±δ However small portions of perhaps liquid
secondary phases were identified at the grain boundaries. The strontium content at the grain interior of Ce0.2La0.7Sr0.1NbO4±δ is more depleted and is accompanied by notable secondary phase formation, the composition of which is similar to that found in as-sintered Ce0.9Sr0.1NbO4±δ
Figure 6.15: SEM micrographs of sintered Ce0.9–xLaxSr0.1NbO4 ± δ (18 hours 1500°C) reduced
Composition Ce (at%) La (at%) Sr (at %) Nb (at%) O (at%) La=0.2 11.67 3.33 1.66 16.67 66.67 GI 13.43 ± 0.51 3.84 ± 0.16 1.81 ± 0.05 18.81 ± 0.39 62.11 ± 1.04 Spectrum 1 2.17 0.75 9.46 22.42 65.20 La=0.7 3.33 11.67 1.66 16.67 66.67 GI 4.76 ± 0.43 16.51 ± 1.61 1.13 ± 0.09 20.3 ± 0.91 57.25 ± 2.97 Spectrum 1 0.85 3.50 12.40 16.92 66.33
Table 6.11: Average atomic percentage of each constituent calculated by EDXS at the grain interior in Ce0.9–xLaxSr0.1NbO4±δ sintered under air (18 hours 1500°C) then reduced under 5%
H2/N2(g) at 800°C for 24 hours. Example compositions of the secondary phases corresponding
to Figure 6.15 are also provided.
Whilst it is very difficult to quantify the level of microstructural degradation, exposure of Ce0.9–xLaxSr0.1NbO4±δ to 5% H2/N2(g) for 24 hours did not re-
sult in significant inter or intra-granular fracturing (Figure 6.15). As-prepared Ce0.7La0.2Sr0.1NbO4±δ showed some fracturing, but no notable change was appar- ent for either composition after reduction. However to quantify this more accurately, perhaps a modified grain intercept or similar counting method may be adopted to record the number of intercepting grains that are fractured between different sam- ples. Certainly the incorporation of both lanthanum and cerium on the A site may be a viable strategy for combining both electronic conductivity and higher dopant concentrations relative to LaNbO4 without compromising microstructural stability under reducing conditions. For example the 5% dopant level is probably closer to the solubility limit which is still twice that of LaNbO4.