CO2-Resistant Hydrogen Permeation Membranes Based on Doped

May 28, 2010 - Influence of fabrication process of Ni–BaCe0.7Zr0.1Y0.2O3−δ cermet .... Slapping a 25% tariff on $200 billion worth of Chinese imp...
0 downloads 0 Views 1MB Size
10986

J. Phys. Chem. C 2010, 114, 10986–10991

CO2-Resistant Hydrogen Permeation Membranes Based on Doped Ceria and Nickel Shumin Fang, Lei Bi, Litao Yan, Wenping Sun, Chusheng Chen, and Wei Liu* CAS Key Laboratory of Energy ConVersion Materials, UniVersity of Science and Technology of China, 96 Jinzhai Road, Hefei, Anhui 230026, P.R. China ReceiVed: March 12, 2010; ReVised Manuscript ReceiVed: May 10, 2010

Composite membranes consisting of a proton-conducting ceramic and an electronic conductor are promising in reducing the cost in the separation of hydrogen from CO2. However, the lack of a stable ceramic in CO2 with sufficient proton conductivity remains a great hurdle. In this study, we investigated the hydrogen permeation performance and chemical stability of composite membranes based on doped ceria and Ni. Doped ceria used to be considered as a very poor proton conductor for a long time. However, our results show that ceria heavily doped with rare earth element possesses significant proton conductivity. Compared with membranes based on perovskite-type oxides, hydrogen separation membranes based on fluorite-type ceria show much higher stability in H2O and CO2. 1. Introduction Hydrogen fuel cell technology seems to be an ideal solution to energy and environmental issues caused by vast combustion of fossil fuels.1 Cost-effective approaches for hydrogen production and utilization will bring the realization of a sustainable hydrogen economy closer.2 Hydrogen is often accompanied by other gaseous compounds, mainly CO2 in industrial chemical reactions such as steam reforming of natural gas.

CH4(g) + 2H2O(g) ) CO2(g) + 4H2(g)

(1)

The separation of hydrogen from CO2 is a crucial step in the mass production of pure hydrogen. Membranes composed of a solid-state proton conductor and an electronic conductor (e.g., nickel) can be used to extract pure hydrogen from gas mixtures without the need for electrodes or external electrical circuit,3 and thus the separation cost can be reduced. Proton conductors have become promising electrolyte candidates for intermediatetemperature solid oxide fuel cells because of their high proton conductivity and low activation energy.4-7 Therefore, the development of high-temperature proton conductors is of exceptional interest in the hydrogen technology. Because Iwahara et al. found that some perovskite oxides possess excellent proton conductivity at high temperature,8 numerous proton conductors have been developed to meet the requirements of practical applications, such as high proton conductivity, adequate chemical stability in H2O and CO2. Acceptor-doped BaCeO3 shows one of the highest proton conductivities, but it suffers from poor chemical stability in H2O and CO2.9-11 The chemical stability can be improved by the doping of Zr at the Ce site, but this strategy also causes the reduction of proton conductivity and sintering activity.12,13 Nonperovskite-type oxides usually possess high chemical stability but too low proton conductivities. Haugsrud and Norby investigated the proton conductivities of rare-earth orthoniobates and ortho-tantalates and found that Ca-doped LaNbO4 had the highest proton conductivity (∼10-3 S cm-1 at 800 * Corresponding author. Tel: +86 551 3606929. Fax: +86 551 3601592. E-mail: [email protected].

°C).14,15 This level of proton conductivity is still too low to be applicable in high-drain applications. Similar proton conductivity is found in pyrochlore-type oxides, such as doped La2Zr2O7, which possesses a fluorite-related structure,16 suggesting that appreciable proton conductivity may be found in fluorite-related oxides. In fact, proton conductivity in a typical fluorite-type oxide (doped ceria) has been confirmed. Nigara et al. measured the hydrogen permeability of low-level doped CeO2 and found that they possessed very low proton conductivities (e.g., Ce0.9Gd0.1O2-δ, 1.2 × 10-6 S cm-1 at 797 °C).17-20 Because high-level acceptor doping in ceria will cause the association of oxygen vacancies and the reduction in oxygen ionic conductivity,21 most researches are focused on low-level doped ceria (e.g., Gd0.2Ce0.8O2-δ and Sm0.2Ce0.8O2-δ). It is noticeable that Wang et al. synthesized ammonia from N2 and H2 at atmosphere pressure in a solid-state proton-conducting cell-reactor using La1.95Ca0.05Ce2O7-δ as electrolyte; the performance of reactor using La1.95Ca0.05Ce2O7-δ as electrolyte was comparable to that of doped BaCeO3,22 suggesting it might possess considerable proton conductivity. La1.95Ca0.05Ce2O7-δ was once considered to be pyrochlore-type, but it actually is fluorite-type.23 Because most fluorite oxides have a molecular formula of RO2, La1.95Ca0.05Ce2O7-δ can be rewritten as Ca0.0125La0.4875Ce0.5O2-δ. It seems that the proton conductivity of ceria can be greatly improved by high-level doping of rare-earth element, and thus high proton conductivity may be found in RExCe1-xO2-δ (RE ) rare earth element). It should be pointed out that the process of ammonia synthesis using proton conductor is very complex (e.g., ammonia also decomposes at high temperature), and thus Wang’s results cannot be used to estimate the level of proton conductivity in La1.95Ca0.05Ce2O7-δ. They also reported that La1.95Ca0.05Ce2O7-δ (Ca0.0125La0.4875Ce0.5O2-δ) showed slightly higher conductivity in wet H2 than dry air, which was also found in other proton conductors. It is known that partial Ce4+ can be reduced to Ce3+ in fluorite-type CeO2 resulting in n-type conduction.24 Therefore, higher conductivity in wet H2 than dry air could be the reflection of n-type conduction but not proton conduction. Neither ammonia synthesis nor conductivity measurement can be directly used to assess the proton conductivity of RExCe1-xO2-δ, and thus we perform electrochemical hydrogen

10.1021/jp102271v  2010 American Chemical Society Published on Web 05/28/2010

CO2-Resistant Hydrogen Permeation Membranes permeation tests on dense membranes consisting of RExCe1-xO2-δ and an electronic conductor (Ni). If RExCe1-xO2-δ is a proton conductor when a hydrogen concentration gradient is applied between two surfaces of membrane, then hydrogen will permeate through the membrane in the form of protons. The level of hydrogen permeation flux represents the proton conductivity of RExCe1-xO2-δ. For the primary study, we select La as the main doping element. The proton conductivity of La2-xCaxZr2O7-δ (x ) 0.015, 0.03, and 0.05) in wet H2 increases with Ca2+ doping level,17 and thus the introduction of Ca may also improve the proton conductivity of ceria. In this work, the hydrogen permeation fluxes of Ni-LaxCe1-xO2-δ (x ) 0.2, 0.4, 0.5, 0.6, and 0.7) and Ni-CayLa0.5-yCe0.5O2-δ (y ) 0.0125, and 0.025) were investigated at 700-900 °C. The chemical stability and performance of membranes with optimal composition in H2O and CO2 were also studied. 2. Experimental Section A series of ceramic powders with the general formula of LaxCe1-xO2-δ (x ) 0.2, 0.4, 0.5, 0.6, 0.7) and CayLa0.5-yCe0.5O2-δ (y ) 0.0125, 0.025) were synthesized using a complexation-combustion method. Appropriate amounts of La2O3 (A.R.) and CaCO3 (A.R.) were dissolved in dilute nitric acid solution, after which Ce(NO3)3 · 6H2O (A.R.) was dissolved in the solution at the stoichiometric ratio. Citric acid was added to the solution in a molar ratio of citric acid/metal 2:1. The pH value of the solution was adjusted to ∼8 by ammonia under continuous stirring. The solution was then evaporated until a viscous gel was formed. The gel was heated in a ceramic pan on an electric furnace until it self-ignited resulting in white ash. The ash was calcined at 850 °C for 3 h to form ceramic powder. After required phase composition was confirmed by X-ray diffraction (XRD) analysis, the powders were mixed with Ni powder at a volume ratio of 60:40. The obtained powders were ball-milled in alcohol for 24 h, dried, ground, and then pressed uniaxially into pellets at 200 MPa with a diameter of 15 mm. The green pellets were sintered at 1430 °C for 5 h in 4% H2 balanced with N2. The densities of acquired pellets were measured using Archimedes method in mercury and were ∼95% of the theoretical density. The electrical resistance measured between two surfaces of the pellets is 0.6 > 0.4 > 0.7 > 0.2. The proton conductivity increases with increasing lanthanum doping level before x < 0.5 and decreases when x > 0.5, which is probably due to the blocking effect of La2O3. A plausible mechanism for proton conduction in doped ceria is suggested on the basis of defect chemistry. The doping of CeO2 with trivalent cation is compensated by the formation of oxygen vacancy. In Kro¨ger-Vink notation, the reaction is given as CeO2

′ + 3O×o + V..o Ln2O3 98 2LnCe (Ln ) La, Y, Sm, Pr......) (2)

Fang et al.

Figure 4. Hydrogen fluxes of (9) Ni-La0.2Ce0.8O2-δ, (b) NiLa0.4Ce0.6O2-δ, (2) Ni-La0.5Ce0.5O2-δ, (1) Ni-La0.6Ce0.4O2-δ, (0) Ni-La0.7Ce0.3O2-δ with a thickness of 0.6 mm at 700-900 °C. The feed gas consists of 20 mL/min H2, 3 mL/min H2O, and 77 mL/min N2. The sweep gas is 20 mL/min high purity Ar. The effective permeation area is about 0.7 cm2.

Figure 5. Plausible mechanism of proton conduction in La-doped ceria.

On the basis of results of quantum mechanical calculation, Andersson et al. suggest that oxygen vacancy formation energy in next nearest neighbor position (NNN) of La3+ is lower than that in nearest neighbor (NN) position in ceria.26 Therefore, the vacancy generated by La-doping is in NNN position (close to Ce4+, shown in Figure 5). In wet atmosphere, water fills the oxygen vacancy forming protonic defects.4 The reaction is as follows

H2O + V••o + O×o f 2OH•o

(3)

Because the oxygen vacancy generated by La-doping occupies the NNN position of La ion, these protonic defects are also located in this position. These protons suffer electrostatic repulsion from central cation (Ce4+ or La3+) forming O-H-O bonds and can hop between two adjacent oxygen ions at high temperature.27 Considering that Ce4+ has a higher valence and smaller radius than La3+, the repulsion between La3+ and proton is lower than that between Ce4+ and proton. Therefore, these protons have a higher affinity to La ions and move toward the oxygen ions in NN positions. At high temperatures, these protons can hop between two oxygen ions close to La3+. When these processes are repeated, proton conduction can happen (Figure 5). This mechanism can explain the results well. When the La content in ceria increases, the concentration of oxygen vacancy also increases, which is beneficial for the incorporation of water and proton conduction. Therefore, the performance of

CO2-Resistant Hydrogen Permeation Membranes

Figure 6. Hydrogen fluxes of Ni-La0.5Ce0.5O2-δ (2), NiCa0.0125La0.4875Ce0.5O2-δ (1), Ni-Ca0.025La0.475Ce0.5O2-δ (b), and NiBaCe0.8Y0.2O3-δ (9) with a thickness of 0.6 mm at 700-900 °C. The other conditions are the same as those in Figure 4.

Ni-LaxCe1-xO2-δ membranes is improved with higher La content. However, this mechanism still needs to be verified by more results. Figure 6 shows the effect of Ca-doping on the hydrogen permeation performances. The hydrogen fluxes of Ni-CayLa0.5-yCe0.5O2-δ (y ) 0, 0.0125, and 0.025) with a thickness of 0.6 mm at 900 °C are 1.6, 1.9, and 1.4 × 10-8 mol cm-2 s-1, respectively. An appropriate Ca-doping level benefits the proton conductivity of La0.5Ce0.5O2-δ, whereas excessive doping of Ca decreases it. To compare the performance of ceria-based membranes with that based on BaCeO3, we measured the performance of Ni-BaCe0.8Y0.2O3-δ (fabricated via the solidreaction method)25 under the same condition. The results are also shown in Figure 6. The performance of Ni-Ca0.0125La0.4875Ce0.5O2-δ (Ni-CLC125) is 19-28% of that of NiBaCe0.8Y0.2O3-δ at 700-900 °C. Among all samples studied, Ni-CLC125 shows the highest hydrogen fluxes. Hydrogen permeation through the membranes involves the exchange of hydrogen across the gas/solid interface, transport of protons through the doped ceria phase, and electrons through the Ni phase. If the performance is controlled by the latter process, then it can be improved by reducing the membrane thickness. Therefore, we also investigated the relationship between membrane thickness and performance of Ni-CLC125. Figure 7 shows the hydrogen permeation flux (JH2) of the composite membrane with different thicknesses. The fluxes of the membranes with thicknesses of 0.32, 0.51, 0.63, and 0.82 mm in 20% H2 and 3% H2O at 900 °C are ∼2.5, 2.1, 1.9, and 1.4 × 10-8 mol cm-2 s-1, respectively. We can improve its performance by fabricating thinner membrane, for example, dense Ni-CLC125 thin film on porous substrate. It should be pointed out that the effect of microstructure (e.g., the grain size, distribution of grains) or composition (e.g., the volume ratio of Ni and CLC, other doping elements) on the membrane performance has not been studied in this work. By optimizing these conditions, membranes with higher performance can be expected. 3.3. Chemical Stability in H2O and CO2. The chemical stability of hydrogen permeation membranes in H2O and CO2 is essential for their applications; therefore, we further studied the chemical stability of Ni-CLC125. XRD patterns obtained from the surface of Ni-CLC125 pellets after exposure to H2O and CO2 are shown in Figure 8. Apparently, the surfaces of all

J. Phys. Chem. C, Vol. 114, No. 24, 2010 10989

Figure 7. Hydrogen fluxes of Ni-CLC125 with different thicknesses in 20% H2 and 3% H2O at 700-900 °C. The other conditions are the same as those in Figure 4.

Figure 8. XRD patterns of Ni-CLC125 obtained from (a) surface of sintered pellet, (b) surface after treatment in water at 180 °C for 50 h, and (c) surface after test in 40% CO2 at 900 °C.

pellets only consist of Ni and Ca0.0125La0.4875Ce0.5O2-δ (CLC125), suggesting that both phases are stable under these conditions. Figure 8b shows the XRD pattern obtained from the surface of sample after exposure to H2O at 180 °C for 50 h. It is remarkable that the relative intensity of CLC125 peaks is much lower than that in sintered pellet. Figure 9a shows the backscattered SEM micrograph obtained from the surface of a sintered pellet. It can be seen that the sample is composed of two phases with different gray levels. A phase with higher average atomic weight is lighter in backscattered micrographs, and thus the light phase is CLC125, and the dark phase is Ni. The surface is very dense with only a few small pores on the ceramic surface. Figure 9b shows the backscattered micrograph from the surface after exposure to H2O at 180 °C for 50 h. Because the water pressure was 10 times higher than the standard atmosphere pressure and ceramic was easier to fracture than metal under mechanical stress, many CLC125 particles on the surface were washed away, whereas Ni particles were reserved, which was consistent with the decrease in relative intensity of peaks corresponding to CLC125 in Figure 8b. EDX analysis performed on the light phase suggests that the CLC phase still consists of La, Ca, Ce, and O with an atomic ratio of around 19.6:1.2:26.1:53.1. Considering that EDX is a semiquantitative technique in the determination of element composition, the change in the element ratios is not serious. The XRD and SEM/EDX results demonstrate that the chemical composition of the membrane is not changed after the treatment in hot water. Chen et al. found that

10990

J. Phys. Chem. C, Vol. 114, No. 24, 2010

Fang et al.

Figure 10. Time dependence of hydrogen permeation fluxes of NiCLC125 at 700 and 900 °C in the presence of CO2.

Figure 9. SEM micrographs of Ni-CLC125 obtained from (a) surface of fresh sample, (b) surface after treatment in water at 180 °C for 50 h, and (c) surface after test in 40% CO2 at 900 °C.

sintered pellets of BaCe0.9Nd0.1O3-δ disintegrated into a mixed powder of BaCO3, CeO2 and a little BaCeO3 after boiling in water for 6 h.28 Because the condition in our experiment is much more aggressive than that of theirs, the stability of CLC125 in water is much higher than that of BaCeO3. Figure 9c shows the micrograph obtained from the feed side surface after test in 40% CO2 at 900 °C. No significant change is observed on the surface. As a result, the composite membrane shows high tolerance to H2O and CO2. The hydrogen permeation performance of Ni-CLC125 membrane with CO2 in the feed gas at 700 and 900 °C is shown in

Figure 10. The hydrogen permeation performance kept stable after the introduction of 40% CO2 into the feed gas at 700 °C but decreased by ∼4% when 20% CO2 was introduced and continued to fall by 6% with 40% CO2 at 900 °C. Traditional Ni-BaZr0.8-xCexY0.2O3-δ (0.4 e x e 0.8) membranes suffered a performance loss of ∼35 and ∼100% in 30% CO2 at 900 °C,3 whereas Ni-CLC125 membrane had a performance loss of only 10% in 40% CO2 at the same temperature. The performance loss of Ni-BaZr0.8-xCexY0.2O3-δ (0.4 e x e 0.8) membranes in CO2 at lower temperature has not been studied yet but should be even worse than that at 900 °C because the carbonation reaction is thermodynamically favored at lower temperature.9 However, the Ni-CLC125 membrane had no performance loss at 700 °C. These results strongly suggest that Ni-CLC125 membrane possessed a much higher stability in the presence of high concentration CO2 than traditional membranes based on BaCeO3. The performance degradation at 900 °C may be due to the interaction between the membrane and CO2 or the reduction in hydrogen partial pressure in the feed gas caused by the reverse water-gas shift reaction between H2 and CO2.9 The performance degradation observed in Ni-BaCe0.8Y0.2O3-δ and Ni-BaZr0.8-xCexY0.2O3-δ (0.4 e x e 0.8) is due to the reaction between the BaCeO3 and CO2 forming insulating BaCO3.3 Because of the limited gas/solid contact zone, the gas-solid reaction takes a long time to complete (20-80 h). Thermogravimetric analysis results suggest that the reaction is thermodynamically favored at lower temperatures.9 However, in our case, no new phase is found by XRD after the test in CO2 (Figure 8c). It is also observed that the degradation is rather quick (in 2 h) and severe at higher temperature. Therefore, we think that the degradation in performance is not caused by the reaction between the membrane and CO2. The composite membrane contains very low amounts of alkaline-earth metal ions, which makes it highly stable in CO2. The reverse water-gas shift reaction is as follows

H2(g) + CO2(g) ) H2O(g) + CO(g)

(4)

To determine whether reaction 4 took place, we analyzed the content of CO and H2 in the effluent feed gas by the gas chromatograph. The CO content in the effluent gas was ∼0.06% at 700 °C when 40% CO2 was introduced, suggesting that the reaction is negligible at this temperature. When 20 and 40% CO2 was introduced into the feed gas at 900 °C, the effluent gases contained 2.6, 4.6% CO and 17.4, 15.4% H2, respectively.

CO2-Resistant Hydrogen Permeation Membranes

J. Phys. Chem. C, Vol. 114, No. 24, 2010 10991 only 10% with 40% CO2 in the feed gas at 900 °C, suggesting that the chemical stability of Ni-CLC125 in CO2 was much higher than that of traditional membrane based on BaCeO3. The performance loss was probably caused by the reverse water-gas shift reaction. This work provides unambiguous evidence of considerable proton conductivity in ceria doped with high concentration of rare earth element with a fluorite structure. Potential applications such as hydrogen permeation membranes, fuel cells, and hydrogen sensors may be developed based on these materials. Acknowledgment. This work is supported by the Key Program of Chinese Academy of Sciences (grant no. KJCX1.YW07) and the National High-tech R&D Program of China (grant no.: 2007AA05Z157). References and Notes

Figure 11. Hydrogen permeation fluxes of Ni-CLC125 after the introduction and removal of 40% CO2 in the feed gas at 900 °C.

The flux loss due to reduction in H2 concentration was calculated to be about 3 and 5%, respectively, through the Wagner equation.29 The calculated flux loss was lower than that in our experiment (3 and 5% compared with 4 and 10%). This should be caused by the hydrogen concentration change during the cooling of the feed gas from 900 °C to room temperature. Reaction 4 is slightly endothermic and tends to shift toward reactants as temperature decreases. Therefore, H2 concentration on the membrane surface was lower than that measured in effluent gases. The lower H2 concentration resulted in higher performance loss. The reverse water-shift reaction caused the decrease in hydrogen partial pressure in feed gas, and thus the performance degraded. If CO2 was removed from the feed gas, then the hydrogen permeation flux should regenerate quickly and completely. Therefore, we measured the regeneration behavior of Ni-CLC125 in 40% CO2 at 900 °C, which is shown in Figure 11. Obviously, the hydrogen permeation fluxes decreased by ∼10% after the introduction of CO2 and recovered completely in 1 h after the removal of CO2. This also suggested that the cause of the performance loss in CO2 was the reverse watershift reaction. 4. Conclusions Composite membranes based on Ni-LaxCe1-xO2-δ (x ) 0.2, 0.4, 0.5, 0.6, and 0.7) showed considerable hydrogen permeability at 700-900 °C. The hydrogen permeation fluxes increased with lanthanum doping level (x) when x e 0.5 but decreased with x when x > 0.5 because of the formation of La2O3 impurity. Ca doping at a much lower level had an analogous effect on the fluxes. The performance of Ni-CLC125 membrane was 19-28% of traditional Ni-BaCe0.8Y0.2O3-δ at 700-900 °C and could be improved by using thinner membranes. The hydrogen permeation flux of the composite membrane decreased

(1) Marban, G.; Vales-Solis, T. Int. J. Hydrogen Energy 2007, 32, 1625–1637. (2) Penner, S. S. Energy 2006, 31, 33–43. (3) Zuo, C. D.; Dorris, S. E.; Balachandran, U.; Liu, M. L. Chem. Mater. 2006, 18, 4647–4650. (4) Kreuer, K. D. Annu. ReV. Mater. Res. 2003, 33, 333–359. (5) Zuo, C. D.; Zha, S. W.; Liu, M. L.; Hatano, M.; Uchiyama, M. AdV. Mater. 2006, 18, 3318–3320. (6) Boehm, E.; McEvoy, A. J. Fuel Cells 2006, 6, 54–58. (7) Bi, L.; Zhang, S.; Fang, S.; Tao, Z.; Peng, R.; Liu, W. Electrochem. Commun. 2008, 10, 1598–1601. (8) Iwahara, H.; Esaka, T.; Uchida, H.; Maeda, N. Solid State Ionics 1981, 3-4, 359–363. (9) Zakowsky, N.; Williamson, S.; Irvine, J. T. S. Solid State Ionics 2005, 176, 3019–3026. (10) Wu, Z. L.; Liu, M. L. J. Electrochem. Soc. 1997, 144, 2170–2175. (11) Matsumoto, H.; Kawasaki, Y.; Ito, N.; Enoki, M.; Ishihara, T. Electrochem. Solid-State Lett. 2007, 10, B77–B80. (12) Zhong, Z. M. Solid State Ionics 2007, 178, 213–220. (13) Ryu, K. H.; Haile, S. M. Solid State Ionics 1999, 125, 355–367. (14) Haugsrud, R.; Norby, T. Nat. Mater. 2006, 5, 193–196. (15) Haugsrud, R.; Norby, T. Solid State Ionics 2006, 177, 1129–1135. (16) Omata, T.; Otsuka-Yao-Matsuo, S. J. Electrochem. Soc. 2001, 148, E252–E261. (17) Nigara, Y.; Yashiro, K.; Kawada, T.; Mizusaki, J. Solid State Ionics 2001, 145, 365–370. (18) Nigara, Y.; Yashiro, K.; Kawada, I.; Mizusaki, J. Solid State Ionics 2003, 159, 135–141. (19) Nigara, Y.; Mizusaki, J.; Kawamura, K.; Kawada, T.; Ishigame, M. Solid State Ionics 1998, 115, 347–354. (20) Nigara, Y.; Kawamura, K.; Kawada, T.; Mizusaki, J. Solid State Ionics 2000, 136, 215–221. (21) Ou, D. R.; Mori, T.; Ye, F.; Kobayashi, T.; Zou, J.; Auchterlonie, G.; Drennan, J. Appl. Phys. Lett. 2006, 89, 171911–3. (22) Wang, J. D.; Xie, Y. H.; Zhang, Z. F.; Liu, R. Q.; Li, Z. H. Mater. Res. Bull. 2005, 40, 1294–1302. (23) Yamamura, H.; Nishino, H.; Kakinuma, K.; Nomura, K. J. Ceram. Soc. Jpn. 2003, 111, 902–906. (24) Shimonosono, T.; Hirata, Y.; Ehira, Y.; Sameshima, S.; Horita, T.; Yokokawa, H. Solid State Ionics 2004, 174, 27–33. (25) Fang, S. M.; Bi, L.; Wu, X. S.; Gao, H. Y.; Chen, C. S.; Liu, W. J. Power Sources 2008, 183, 126–132. (26) Andersson, D. A.; Simak, S. I.; Skorodumova, N. V.; Abrikosov, I. A.; Johansson, B. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 3518–3521. (27) Sata, N.; Hiramoto, K.; Ishigame, M.; Hosoya, S.; Niimura, N.; Shin, S. Phys. ReV. B 1996, 54, 15795. (28) Chen, F. L.; Sorensen, O. T.; Meng, G. Y.; Peng, D. K. J. Mater. Chem. 1997, 7, 481–485. (29) Norby, T.; Larring, Y. Solid State Ionics 2000, 136, 139–148.

JP102271V