Ishihara T. (Ed.) Perovskite Oxide for Solid Oxide Fuel Cells (Fuel Cells and Hydrogen Energy)

Подождите немного. Документ загружается.

essentially based on simulation and experimental work, which has been per-

formed independently in 1995 by Tuckerman et al. and Agmon [7, 8]. Essential

features of the suggested mechanism are that (i) intermolecular proton transfer

occurs through rather short hydrogen bonds; (ii) this transfer is highly coupled

to hydrogen bond-breaking and bond-forming processes in the weaker

bonded second hydration sphere of the protona ted species (H

or H

);

and (iii) the position of the protonic charge follows the center of symmetry of

the hydrogen bond pattern [9]. In other words, proton transfer and structural

reorganization, which together form an uninterrupted trajectory for proton

translocation, take place in different parts of the hydrogen-bonded structure.

In this way, both pro ton transfer and structural reorganization, the two essen-

tial ingredients of ‘‘structure diffusion’’ may take place at very high rates [9].

Later, a similar mechanism has been found for proton conduction in imidazole

[10], an amphoteric molecular compound, which shows high intrinsic proton

conductivity in the liquid state. As in the case of phosphoric and phosphonic

acid, this is also the result of the high degree of self-dissociation, which is low in

the case of water.

Albeit the proton conduction mechanisms in hyd rogen-bonded networks

comprise the interactions of many species, proton conduction mechanisms in

oxides are usually less complex ; this has to do with the fact that protons are

usually present at low concentrations as part of noninteracting [11], positively

charged point defects (generally hydroxide ions residing on oxide ion sites,



). These defects show significant hydrogen bond interaction with neigh-

boring oxide ions, and it is the local dynamics of this hydrogen bonding that

turned out to be the clue to the understanding of proton conduction mechan-

isms in perovskite-type oxides. Only in very few perovskite-type oxides, such as

YSnO

5.5

[12], very high proton content may add some complexity to the

proton conduction mechanism.

In the first part of this short chapter, it is discussed where protons find

energetically favored sites in oxides with the perovskite structure, before the

mechanisms of proton diffusion via these sites are described in detail. While this

discussion is restricted to structure and dynamics of protonic defects in ideal

(cubic) perovskites, the following section addresses complications such as sym-

metry reduction and the effect of the presence of dopants, before, finally, a few

implications for the development of proton-conducting electrolytes for fuel cell

applications are discussed.

13.2 Proton Sites

The possible positions for protons in perovskite-type oxides are strongly con-

strained through the strong binding interaction with the oxygen, i.e., the most

electronegative species in this type of compound, with which protons form

hydroxide ions. In other words, protons are localized within the valence

262 K.D. Kreuer

electron density of the oxygen, which attains some H1s ch aracter. The orienta-

tion of the hydroxide ion is then mainly determined by the hydrogen bonding

with other, neighboring oxide ions, which is an attractive interaction, and the

interaction with the cations of the structure, which, of course, is repulsive.

Although clear signatures of hydrogen bonding were found in the OH-stretch-

ing vibration [13–18], i.e., typica l red-shift ed continua, there have been dif fer-

ent suggestions for the orientation of these bonds. Within the simple cubic

perovskite structure, i.e., a framework of corner-sharing undistorted oxygen

octahedra, each oxygen has eight nearest oxygen neighbors with separations

corresponding to the edge length of the octahedra, and four next nearest

neighbors, which reside on the vertices of the four neighboring octahedra in

thesameplane.

A proton jump width of 170 pm obtained from quasi-elastic neutron scatter-

ing (QNS) spectra of Y-doped SrZrO

was interpreted as the separation

between proton sites on the edges of neighboring octahedra corresponding to

linear hydrogen bonding of the hydroxide ions to their nearest oxygen neigh-

bors [19] Similar positions have been favored by first-principles molecular

dynamics simulations of protons in Sc-doped SrTiO

[20, 21]. From a neutron

and X-ray powder diffraction study on deuterated Ba

1+y

2–y

9–d

, how-

ever, the deuterons were refined to point toward the four next nearest oxygens

[22]. These positions actually do not violate the empirical Badger–Bauer rule

[23], which is excluding hydrogen bonding within the same coordination poly-

hedron. The uncertainty of the proton sites is actually reflecting the locally flat

free energy surfaces, especially in the case of perovskites with small lattice

constants such as SrTiO

. In the very first quantum molecular dynamics

study of protons in BaCeO

[24], the trace of the proton at T ¼ 900 K was

found to be confined to a kind of ‘‘donut’’ with the hydroxide oxygen defining

the center. This ‘‘donut’’ actually contains all the sites just discussed, and from a

careful analysis of the proton probability density function for different perovs-

kite-type oxides, proton sites have been obtained [25, 26]. For perovskites with

large lattice constants such as BaCeO

and BaZrO

, sites close to the octahedra

edges were identified (Fig. 13.1a), whereas in perovskites with small lattice

constants such as SrTiO

and CaTiO

, the proton sites are closer to the plane

formed by the hydroxide ion and its four next nearest oxygen neighbors

(Fig. 13.1b). As illustrated in Fig. 13.1, there are always 8 sites per hydroxide

ion (24 per unit cell), but for large lattice constants the only hydrogen bond

interaction is within octahedra, while for small lattice constants, also hydrogen

bonding to oxygen of neighboring octahedra appea rs to be possible. Hydrogen

bonding is actually indicated by a small contraction (a few pm) of the average

separation between the hydroxide ion and its eight nearest and four next nearest

oxygen neighbors (e.g., [27]); it is most likely the repulsion with the highly charged

B-site cation (Ti

4þ

) and the attraction from the next nearest oxygen, which is

pushing the proton out of the octahedron edge. The resulting degree of hydrogen

bond bending (for the average configuration) surely has severe implications on

the proton conduction mechanisms, which are discussed in the next section.

13 Mechanisms of Proton Conduction in Perovskite-Type Oxides 263

13.3 Mechanisms of Proton Conduction (Undoped, Cubic

Perovskites)

The elementary reactions underlying the mobility of protonic defects have been

investigated in great detail, experimentally as well as by num erical simulations

from quite different points of view. It is near at hand to even consider proton

tunneling, but as in any fast proton conductors tunneling effects are anticipated

to be negligible [1]. Even without taking lattice relaxation (polarization) effects

around the proton into consideration, proton tunneling is estimated to become

significant at extremely low temperature (T < 14 K) only [28].

Apart from the strong localization of the proton within the hydroxide ion,

the proton self-lo calization also involves a slight contraction of the average

OH/O separation (see above) and an expansion of the separation between the

hydroxide ion and the two neighboring B-site cations (e.g., Ce

4þ

; Fig. 13.1)

[24–26]. Any proton conduction mechanism must at least partially compensate

for these relaxation effects; i.e., it must involve some host lattice adjustment in

the transition state configuration. In a first attempt to relate lattice dynamics to

proton transfer between adjacent oxide ions, the coupling of proton transfer to

the O–B–O bending mode was investigated numerically [29]. The rate of proton

Fig. 13.1 Illustration of

proton sites (small spheres )

as obtained from molecular

dynamics simulations for

cubic perovskite-type oxides

with large lattice constants

(a) and small lattice

constants (b) [24, 25]. Only

the oxygens (dark)

interacting with the proton

are shown. Possible

hydrogen bond interactions

are indicated by grey lines

264 K.D. Kreuer

transfer was found to increase, and the H/D isotope effect was decreasing with

increasing anharmonicity of this mode. The influence of the hydrogen bonding

on the local dynamics had been neglected, and only proton transfer had been

treated in this early study, but the importance of the oxide ion dynamics had

already been clearly recognized. Generally speaking, the local lattice relaxation

may be treated as a diffusing small polaron, which the proton is following; this

is nothing but structure diffusion (see Introduction), whi ch was treated analy-

tically by several authors [30, 31, 32]. However, these approaches are either

phenomenological or they make severe assumptions on the relevant modes

(e.g., O–B–O bending). In particular, they do not give any specific information

on chemical interactions and how they determine the evolution of configura-

tions involved in the proton conduction mechanism.

Our current more detailed understanding of proton conduction mechanisms

in perovskite-type oxides actually started to emerge from independent quantum

molecular dynamics simulations [23], quasi-elastic neutron scattering [33], and

m-SR experiments [34, 35]. They all clearly show rapid rotational diffusion of

the proton within a ‘‘donut’’ around the oxygen with which the proton is making

a covalent bond (Fig. 13.2). In this local dynamics the strong covalent bond

remains intact, and it is only the orientation of the hydroxide ion that is

changing. Because of the free energy barriers separating the different orienta-

tions, this dynamics is stochas tic, i.e., diffusional in nature. Of course, it

involves forming and breaking of hydrogen bonds (see red lines in Fig. 13.1),

which obviously costs very little excitation, which is explained later. These early

Fig. 13.2 Proton traces sampled by quantum molecular dynamics simulations involving intra-

octahedron transfer (a) and inter-octahedra proton transfer (b) [24, 25]. The transition state

configurations for proton transfer are shown for both cases: In the case of intra-octahedron

transfer, this is characterized by B–O bond elongations and strong contraction of the OH–O

separation; in the case of inter-octahedra transfer, severe tilting of the participating octahedra

is involved (see text)

13 Mechanisms of Proton Conduction in Perovskite-Type Oxides 265

results also indicate that it is the proton transfer from a hydroxide ion to a

neighboring oxide ion that is rate-limiting, long-range pro ton transpo rt.

Considering the large average oxygen separations (e.g., compared to hydro-

gen bonding in water) of typically 290–320 pm, this appears to be reasonable,

but it seemed to be at odds with the strongly red-shifted OH-stretching absorp-

tions in the IR spectra ([12] and references therein), indicating strong hydrogen

bond interactions, which favor fast proton transfer reactions rather than fast

reorientation processes, the latter requiring the breaking of such bonds. As the

structural oxygen separation is larger than 290 pm in most perovskite-type

oxides and strong hydrogen bonds may only be formed for significantly lower

separations, the free energy the system gains by hydrogen bond formation is

competing with the free energy required for the lattice distortion involved in the

OH/O contraction. A reanalysis of a quantum-MD simulation of a protonic

defect in cubic BaCeO

[25, 36, 37] demonstrates that these two free energy

contributions are almost balanced for a wide range of oxygen separations

(approximately 250–300 pm). In this way, short oxygen separations, which

favor proton transfer, and large oxygen separations, which allow rapid bond

breaking invo lved in rotational diffusion, correspond to similar free energies of

the entire system and, therefore, have similar probabilities of occurring. Indeed,

the simulation finds the protonic defect to form short but transient hydrogen

bonds with all eight nearest oxygen neighbors. In the time-averaged picture seen

in diffraction experiments, this leads only to a slight reduction of the structural

OH/O separations (see Fig. 13.1), but in most instant configurations one of the

eight OH/O separations is reduced to about 280 pm as a result of hydrogen

bonding [24]. Although the hydrogen bond interaction has a stabilizing effect

of about 0.5 eV on this configuration [38], the bond is a soft high-energy

hydrogen bond with extended bond length variations ; this also leads to config-

urations where the protonic defect behaves almost like a free OH with small OH

stretching amplitudes compared to the extended stretching vibrations in the

hydrogen-bonded state.

From the thermodynamics of such ‘‘dynamical hydrogen bonds’’ one may

actually expect an activation enthalpy of long-range proton diffusion not more

than 0.15 eV provided that the configuration O–H...O is linear, for which the

proton transfer barrier vanishes at O/O separations less than about 250 pm.

However, the mobility of protonic defects in cubic perovskite-type oxides has

activation enthalpies of the order of 0.4–0.6 eV [35], which raises the question of

which interactions are controlling the activation enthalpy of proton transfer.

As already pointed out, in the average configuration the hydrogen bonds are

bent to some extent (see Fig. 13.1). This change suppresses proton transfer for

two reasons: there is a remaining proton transfer barrier even at low oxygen

separations, and any proton transfer requires both energy and momentum

transfer in this bent configuration. An analysis of a few transition state config-

urations obtained from molecular dynamics simulations showed that the B–O

bonds are elongated, which reduces the repulsion between the B-site cation and

the proton and allows for the formation of an almost linear, short hydrogen

266 K.D. Kreuer

bond in the transition state complex [Fig. 13.2(a)]. The proton transfer in this

configuration probably occurs over some remaining barrier, as indicated by the

experimentally observed H/D isotope effects [28, 39]. Although the H/B repul-

sion is redu ced in this configuration, major contributions to the activation

enthalpy arise from the B–O bond elongation and the proton transfer barrier.

The importance of the H/B repulsion is also evidenced by the finding that the

activation enthalpies of proton mobility in cubic perovskites with pentavalent

B-site cations (I–V perovskites) are significantly higher than for perovskites

with tetravalent B-site cations (II–IV perovskites) [36]. On this background,

proton mobility in III–III perovskite may be even higher, provided the oxide

shows cubic symmetry.

So far, we have just considered the mechanism of proton conduction in

perovskite-type oxides with large lattice constants, and these are commonly the

ones with potentially high concentrations of protonic defects and therefore also

high proton conductivity [35], but perovskite-type oxides with small lattice con-

stant (e.g., SrTiO

,CaTiO

) may also exhibit high proton mobilities. In these

cases, hydrogen bonding even to the next nearest oxygen, between the vertices of

the octahedra, becomes possible [Fig. 13.1(b)], opening another proton transfer

path as observed in MD simulations [20, 24]. The transition state complexes then

involve a tilting of neighboring octahedra between which the transfer takes place

(Fig. 13.2). While only 25% of the observed proton transfers are occurring

between octahedra (inter-octahedra transfer) in the case of SrTiO

, proton transfer

in CaTiO

is observed to be dominated by inter-octahedra transfers (70%). It

should be noted that for the latter the highest proton mobility has been found in

the simulations [25], but small concentration of protonic defects under common

conditions prevents titanates from showing high proton conductivity [37].

The mechanisms just described provide a qualitative explanation for the

empirical finding that the highest proton conductivities are observed in oxides

with the cubic perovskite structure [40]. The framework of corner-sharing BO

octahedra shows high coordination numbers for both cation sites (12 for the A

site and 6 for the B sit e). There is only 1 oxygen site in the ideal perovskite

structure, with each oxygen surrounded by 8 nearest and 4 next nearest oxygens.

Generally speaking, the high coordination numbers lead to low bond strengths

and low angles between the bonds, which is in favor of the above-described

dynamics. For example, the rotational diffusion of the protonic defect corre-

sponds to a dynamical hydrogen bonding of the OH with the 8 or 12 oxygen s,

forming the ‘‘reaction cages’’ (see Fig. 13.1). The angles between the possible

orientations are small enough that the effective barriers for bond-breaking and

bond-forming processes are usually very low, and the equivalence of all oxygens

guarantees the absence of any extra contribution to the activation energy that

may originate from differences in site energies. The latter is especially important

for the proton trans fer reaction, which may be biased to a particular oxygen

site, which then may act as a kind of proton trap. Also, the high number of

equivalent transfer paths directly enters into the pre-exponential factor of

proton mobility.

13 Mechanisms of Proton Conduction in Perovskite-Type Oxides 267

13.4 Complications (Symmetry Reduction, Doping, Mixed Site

Occupancy)

From these considerations, it is anticipated that any perturbation of the highly

symmetric perovskite structure may lead to a decrease of the rate of long-range

proton transport. This may not only be brought about by the reduction of the

symmetry of the crystallographic (time/space-averaged) structure, but also

locally by the presence of point defects (e.g., acceptor dopants, oxide ion

vacancies), mixed site occupations, or higher dimensional defects, such as

dislocations or grain boundaries.

The effects of the tilting and twisting of the BO

octahedra, i.e., the reduction

of the crystallographic symmetry, has been investiga ted in detail by comparing

structural and dynamical features of protonic defects in Y-doped BaCeO

and

SrCeO

[41]. The large orthorhombic distortion of SrCeO

has tremendous

effects on the arrangement of the lattice oxygen, which leads to the appearance

of shorter and longer O/O separations and also changes the chemical character

of the oxygen. The cubic oxygen site splits into two sites with probabilities of 1/3

(O1) and 2/3 (O2). As a result of different chemical interactions with the cations,

especially the strontium on the A site, the oxygens on these sites show distinctly

different electron densities (basicities) and therefore different binding energies

for the proton. Although in SrCeO

the most basic oxygen is O1, it is O2 in

BaCeO

. Assuming that protons are associated with these sites for most of the

time, they may show long-range proton transport via the more frequent O2 sites

in BaCeO

, wher eas isotropic long-range proton transport in SrCeO

must

involve transfer be tween chemically different O1 and O2 sites. Much more

rapid transport is anticipated to occur between neighboring O1 oxygens. How-

ever, they just form one-dimensional arrays as a result of octahedra tilting

[Fig. 13.3(a)], and transport along these paths is expected to be very sensitive

to any perturbation. Even for the fast rotational diffusion step, the orthorhom-

bic distortion is clearly inducing a biasing of the hydroxide orientation toward

the more basic oxygen neighbors, O1 [Fig. 13.3(b)]. All these observations are

thought to be the reason for the higher activation enthalpy and lower conduc-

tivity in SrCeO

compared to BaCeO

. Similar simulations were later carried

out on orthorhombic CaZrO

[42], and also in this case octahedra tilting is

opening the proton transfer between the vertices of neighboring octahedra,

whereas no intra-octahedron proton transfer was detected in the simulation.

The mobility of protons is not only very sensitiv e tow ard reduction of the

crystallographic symmetry but also toward local structural and chemi cal per-

turbations induced by the acceptor dopant or by mixed occupancy on the B site.

To allow for the formation of protonic defects by dissociative absorption of

water, perovski te-type oxides are commonly doped with aliovalent ions (accep-

tor dopants) matching the ionic radius of the B-site cation (e.g., In

3þ

for Zr

4þ

Indeed, this simple concept has been proven successful, e.g., for oxide ion

conductors (Sc-doped zirconia shows higher oxide ion conductivity than

268 K.D. Kreuer

Y-doped zirconia). But when it comes to pro ton conductivity in oxides, this

approach clearly fails. Although Sc

3þ

and In

3þ

are matching Zr

4þ

with respect

to their ionic radii, BaZrO

shows much lower proton mobility when doped

with Sc or In compared to Y as an acceptor dopant with a significantly higher

ionic radius [39]. Only for the latter, the proton mobility and its activation

enthalpy are virtually independent on the dopant concentration. Although Y on

the Zr site expands the lattice locally and on the average and even leads to

tetragonal dist ortions at concentrations above 5 mol%, it actually leaves the

acid–base properties of the coordinating oxygen almost unchanged [24].

Obviously, this chemical match of the dopant makes it so to say ‘‘invisible’’ to

the diffusing proton. The most common observation, however, is a decreasing

proton mobility and an increasing activation enthalpy with increasing dopant

concentration, as observed in Y-doped BaCeO

, for example [11, 37], for which

the local distortions around the acceptor dopant has recently been studied

experimentally [43].

Fig. 13.3 Proton diffusion

path in orthorhombic

SrCeO

as obtained from a

quantum molecular

dynamics simulations [40].

The proton rotational

diffusion around O1 is

distinctly biased toward the

neighboring O1 (b), which is

part of a one-dimensional

proton diffusion path

formed by O1 oxygen only

(a). Proton transfer between

O1 (dark) and the less basic

and more frequent O2 (grey)

is anticipated to control

long-range, isotropic proton

diffusion (see text)

13 Mechanisms of Proton Conduction in Perovskite-Type Oxides 269

In an attempt to calculate association energies for the formation of clusters

of protoni c defects and different kinds of acceptor dopants in SrCeO

-based

proton conductors [44], the lowest energies have been found for Y and Yb, the

most commonly used dopants for this type of compound. For CaZrO

, the most

favorable dopants are anticipated to be Ga, Sc, and In [45] Even interaction of

protonic defects with residual oxide ion vacancies has been anticipated to

influence the local proton dynamics in SrTiO

[20].

With this background, it is not surprising that also cation ordering on the B

site of complex perovskites may become disadvantageous for proton mobility.

For example, in Ba

1+x

2–x

3–d

, proton conductivity is higher with lower

activation enthalpy when chemically and geometrically different Ca and Nb are

randomly distributed on the B site. Ca/Nb ordering occurring after annealing

significantly reduces the conductivity [46]. This observation is in line with the

results of an investigation of proton conductivity in the system SrTiO

–Ba-

TiO

–SrZrO

–BaZrO

[39]. Although Sr/Ba mixing on the A site actually left

some space for a materials optimization, Zr/Ti mixing on the B site led to a

significant suppression of the proton conductivity. The highest proton mobi-

lities in this system were actually found for the end members SrTiO

and

BaZrO

, both showi ng the simple cubic perovskite structure.

13.5 Implications for the Development of Proton-Conducting

Electrolytes for Fuel Cell Applications

The requirements for the use of proton-conducting oxides as separator material

in solid oxide fuel cells (SOFC) not only comprise a high mobility of protonic

defects but also a high concentration of such defects and stability under fuel

cell operating conditions. At the moment, only electrolytes based on highly

Y-doped BaZrO

seem to combine these properties in a unique way [35]. Surpris-

ingly, the development of this material required very little compromising.

BaZrO

is actually the cubic perovskite with the highest lattice constant. The

high symmetry is essential for the high solubility limit of protonic defects and

for the high isotropic proton mobility. The high lattice constant (a > 420 pm for

Y-doped samples [39]) goes along with a high stability of protonic defects, and the

covalency of the Zr–O bond reduces the Zr/H-repulsive interaction and, there-

fore, the activation enthalpy of the mobility of protonic defects. Only

CaNb

has a similarly favorable lattice constant, but Ca/Nb ordering on

the B site leads to a symmetry reduction. Less densely packed BaCeO

already

shows a significant orthorhombic lattice distortion.

Another key feature is the availability of a nearly perfect acceptor dopant

(i.e., a dopant that leaves the oxygen basicity almost unchanged). Although in

all other reported cases the increase of the acceptor dopant concentration leads

to a reduction of the proton mobility and an entropic destabilization of proto-

nic defects, both the proton mobility and the thermodynamics of hydration are

270 K.D. Kreuer

practically unchanged for dopant levels up to 20% Y in BaZrO

[37]. High

proton mobility and entropically stabilized protoni c defects even at high dopant

concentrations and the high solubility limit lead to the high proton conductivity

of this material. For temperatures below about 7008C and a water partial

pressure of 23 hPa, this exceeds the oxide ion conductivity of the best oxide

ion conductors. Although the bulk conductivity of Y-doped BaZrO

is even

slightly higher than the proton conductivity of BaCeO

-based oxides, the

chemical stability is by far more advantageous, as expected from the higher

electronegativity of Zr compared to Ce and the higher covalency of the Zr–O

bond. For the CO

partial pressure of air (38 Pa corresponding to 380 ppm),

pure BaZrO

is stable above 3008C, which is only slightly higher than for

BaTiO

and SrTiO

, which are known for their superior stabilities [12].

High bulk proton conductivity, high stability, and a wide ionic domain [47]

therefore make Y-doped BaZrO

an interesting parent compound for the

development of proton- conducting electrolytes for SOFC applications. Unfor-

tunately, the unfavorable brit tleness, the grain boundary impedance, and the

increasing phase instability with increasing Y-dopant level remain problems to

be solved. The addition of small amounts of BaCeO

or a compromise in the

choice of the kind of dopant may help to reduce these problems.

In any case, the potentially very high proton mobility in perovskite-type

oxides, which was discussed in detail in this chapter, may become a clue to the

reduction of the unfavorably high ope ration temperature of conventional

SOFCs.

Acknowledgments I thank R. Merkle for reading the proofs and U. Traub for preparing the

figures.

References

1. K.D. Kreuer, Chem. Mater. 8, 610–641 (1996)

2. K.D. Kreuer, W. Weppner, A. Rabenau, Solid State Ionics, 3/4, 353–358 (1981)

3. K.D. Kreuer, A. Rabenau, W. Weppner, Angew. Chem. Int. Ed. Engl. 21, 208 (1982)

4. S.F. Fischer, G.G. Hofacker, M.M. Ratner, J. Chem. Phys. 52, 1934 (1970)

5. R.A. Marcus, J. Chem. Phys. 24, 966 (1956)

6. C.P. Flynn, A.A. Stoneham, Phys. Rev. 1, 3966 (1970)

7. M.E. Tuckerman, D. Marx, M.M. Klein, M. Parrinello, J. Chem. Phys. 103, 45 (1995)

8. N. Agmon, Chem. Phys. Lett. 244, 456 (1995)

9. K.D. Kreuer, Solid State Ionics 136–137, 149–160 (2000)

10. W. Mu

nch, K.K. Kreuer, W. Silvestri, J. Maier, G. Seifert, Solid State Ionics, 145, 437

(2001)

11. K.D. Kreuer, W. Mu

nch, M. Ise, A. Fuchs, U. Traub, J. Maier, Ber. Bunsenges. Phys.

Chem. 101, 1344 (1997)

12. P. Murugaraj, K.K. Kreuer, T. He, T. Schober, J. Maier, Solid State Ionics 98, 1 (1997)

13. K.D. Kreuer, Solid State Ionics 97, 1 (1997)

14. H. Yugami, Y. Shibayama, T. Mattori, M. Ishigame, Solid State Ionics 79, 171 (1995)

15. T. Scherban, Y.Y. Baikov, E.E. Sharkova, Solid State Ionics 66, 159 (1993)

16. S. Shin, H.H. Huang, M. Ishigame, Solid State Ionics 40/41, 910 (1990)

13 Mechanisms of Proton Conduction in Perovskite-Type Oxides 271