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

Подождите немного. Документ загружается.

and protons are the representative conducting ions in perovskite oxide. Also,

the lithium ion is mobile in some perovskite-type oxides. In addition, several

non-oxide perovskite compounds such as halides are known to be ionic

conductors. Furthermore, some kinds of antiperovskite non-oxide compounds

are good silver ion conductors. Regarding oxide ion conduction and proton

conduction, detailed descriptions are given in Chapters 4 and 11, respectively.

In this chapter, an outline of ionic conduction in perovskite compounds and a

brief history of the resear ch in this field are described , introducing some

examples of the studies.

3.2 Conduction Behavior of Perovskite-Type Compounds

Ionic species that contribute to high conductivity of perovskite-type compounds

are rather limited. They are listed in Table 3.1 with representative compounds,

their conductivities, and distinctive features. The mobile ionic species, except

hydrogen, are host components of the compounds, whereas protons are unique

in that they are incorporated from water vapor or hydrogen gas in the ambient

atmosphere at elevated temperature. Of these, oxide ion conductors are best

known, and oxide ionic conduction in various kinds of perovskite and perovskite-

related oxides has been studied.

Confirmation of ionic conduction in the electrically conductive oxides can be

made in different ways. One of the most convenient methods is to examine the

electromotive force (emf ) of an electrochemical oxygen concentration cell:

Pt; O

ðP

ð1ÞÞ=oxide specimen=O

ðP

ð2ÞÞ; Pt cell ½1

using the oxide sample as an electrolyte diaphragm at elevated temperature.

The concept of the oxygen concentration cell is schematically shown in Fig. 3.1.

Table 3.1 Examples of ionic conduction in perovskite-type compounds

Mobile

ions

Examples of ionic

conductor

Conductivity /

Scm

1

(at8C) Remarks

2

0.9

0.1

0.8

0.2

2.85

1.5x10

1

(8008C) Doped single perovskite oxide

SrCe

0.95

0.05

3a

1x10

2

(9008C) Doped Single perovskite oxide

under hydrogen-containing

atmosphere

0.51

0.34

TiO

2.94

1.4x10

3

(278C) Host oxide: La

2/3

TiO

A-site

deficient perovskite



CsPbCl

1.2x10

3

(5008C) Non-oxide perovskite



CsPbBr

8x10

4

(5008C) Non-oxide perovskite

SI 1 x 10

2

(258C) Anti-perovskite-type

structure. Non-oxide

anti-perovskite Averaged

structure for Ag

46 H. Iwahara

If the observed emf is close to the theoretical value E

calculated from Nernst’s

equation given by Eq. (3.1), the oxide can be regarded as an ionic conductor.

1ðÞ

2ðÞ

(3:1)

If no emf is observed, the charge carriers in the oxide would be electrons or

electron holes. When emf is not zero but smaller than E

in Eq. (3.1), conduc-

tion in the oxide would be partially ionic and partially electronic. However,

readers sho uld note that this method does not give any information about

which ions in the oxide are mobile and that the rati o of observed to theoretical

emf, E/E

, does not always give the correct value of ionic transp ort number. To

determine which ions are mobile, additional experiments are necessary, such as

electrochemical mass transport measurement or tracer technique.

Transport behavior of mobile ions in perovskite-type compounds has been

investigated by many researchers using various experimental methods and

computer modeling. It is acceptable that the transport mechanism of mobile

host ions in the perovskite is based on hopping via vacant sites of the ionic

species. In that case, tolerance factor t and free vo lume v of the crystal are

thought to be important factors affecting the mobility of the ions. The toler ance

factor is a measure of symmetry of the crystal structure of the perovskite and is

written as follows:

t ¼

þ r

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

2 r

þ r

ðÞ

(3:2)

where r

, r

, and r

are ionic radii of A, B, and X ions, respectively, in ABX3.

Empirically and qualitatively, good ionic conductors of perovskite compounds

Ion conductor

(1) O

(2)

(1) > P

(2)

ದ

Electromotive force

Porous electrodePorous electrode

Fig. 3.1 Concept of an

electrochemical oxygen

concentration cell

3 Ionic Conduction in Perovskite-Type Compounds 47

have a relatively large value of t. The free volume is a measure of degree of

packing in a unit cell, which is given by the difference between measured unit

cell volume and the sum of ionic volume in the unit cell calculated from known

ionic radii. In many cases, the free volume of the good ionic conductors is large,

although quantitative studies have not yet been completed. Of course, other

factors such as the concentration of vacancy, ionic radius, and polarizability are

also important for good ionic conduction. Detailed discussions are presented in

the following chapters.

As described in the Introduction section, many perovskite-type compounds

can deviate from their stoichiometric composition to a considerable extent

because of the strong stability of the perovskite-type structure. This deviation

results in the formation of electronic defects such as excess electrons or electron

holes, which cause n-type or p-type electronic conduction. Thus, it should be

noted that, in general, perovskite-type compounds are have facile electronic

conduction, and that the ionic conduction is often accompanied by electronic

conduction. As the concentration of the electronic defects depends on the

deviation of A/B and/or (A + B)/X from their stoichiometric ratio, the content

of impur ity, atmosphere and temperature, and contribution of electronic con-

duction varies with tho se conditions.

In general, electronic conductivity of an oxide electrolyte at elevated

temperature is influenced by partial pressure of oxygen, P

, in the atmosphe re;

n-type electronic conductivity s

increases with decreasing P

, whereas p-type

electronic conductiv ity s

increases with increasing P

. It is known that P

dependence of s

or s

is given by the following:

¼ s

exp P

1=n

(3:3)

and

¼ s

exp P

1=n

(3:4)

where n is some natural number, and s

and s

are the constants that are

independent of the partial pressure of oxygen [1]. It is accepted that the ionic

conductivity s

itself is independent of P

for many oxide electrolytes. Accord-

ingly, the total conductivity s is expressed as follows:

s ¼ s

þ s

exp P

1=n

þ s

exp P

1=n

(3:5)

The relationship between each logarithm of conductivity and logarithm of

is illustrated in Fig. 3.2. The hatched regions indicate the mixed conduction

domains, and between them there exists an ionic conduction domain where

electronic conductivity is negligibly low. The outer sides of the mixed conduction

domains are electronic conduction regions. An example of the experimental

48 H. Iwahara

results on s P

plots is shown in the following section for oxide ion–electron

mixed conductors.

3.3 Early Studies on Ionic Conduction in Perovskite-Type Oxides

Most of the possible combinations of large A cations and smaller B ions, which

is needed to form perovskite-type oxides ABO

, had been tried by 1955, as

described by F.S. Galasso in his famous book [2] entitled Structure, Properties

and Preparation of Perovskite-Type Compounds, published in 1969. This book

compiled almost all available data at that time concerning structure, properties,

and preparation of perovskite-type compounds. In this book, although lattice

defects in the perovskite-type crystal were described, the author did not touch

on ionic conduction in the perovskite except for a very brief description of

BaTiO

. However, in the 1960s, several pioneering studies on ionic conduction

in perovskite-type oxides were performed.

Ionic conduction in perovskite-type oxides was first a source of interest in

ferroelectric materials. S. Swanson showed that DC conductivity of BaTiO

ceramics was significantly influenced by their fabrication history, which suggests

that there would be an intimate relationship between the solid-state reactions of

raw materials and ionic conduction [3]. In the 1960s, when research and devel-

opment of perovskite-type oxides as a dielectric or ferroelectric material such as

BaTiO

and PbTi

1x

had become active, some of the researchers paid

attention to the conduction behavior of these perovskite-type oxides. They

log σ

log P

Ionic domain

Mixed

cond.

domain

Mixed

cond.

domain

Mixed

cond.

domain

Fig. 3.2 Dependence of

conductivity on partial

pressure of oxygen

3 Ionic Conduction in Perovskite-Type Compounds 49

considered that ionic conduction in the ferroelectric materials would be affected

by their manufacturing and the characteristics of pyroelectric properties [4, 5].

In 1961, Stephenson and Flanagan thought that the anomalous pyroelectric

behavior revealed in lead zirconate titanate (PZT) would probably be caused by

ionic conduction in the oxides [5]. To test the existence of ionic conduction in

this oxide, they adopted the electrochemical oxygen concentration cell method

that had been reported by Kuikkola and Wagner [6] 4 years previously for

stabilized zirconias and which subsequently became a very familiar method to

solid-state electrochemists.

Stephenson and Flanagan constructed the concentration cell using PZT as a

diaphragm of two oxygen electrode chambers.

Pt; Pb; PbO=PbZ

r0:53

0:47

=Cu; Cu

O=Pt Cell ½2

The electromotive force (emf) of this cell in purified nitrogen gas has been

measured. The observed emf was 0.205 0.005 V at 2508C, which is close to the

theoretical value calculated from Nernst’s equation given in Eq. (3.1), where

(1) and P

(2) were thermodynamic oxygen partial pressures in each gas

chamber. This result means that conduction was purely ionic, as described in

the previous section. The authors reported that there was no conclusive evi-

dence as to whether the ionic conduction is caused by a cation or an anion

migration. However, from the observed cell behavior, the authors considered

that a t least some of the conduction is contributed by oxygen ion migration.

Heckman et al. [7] studied the conduction properties of polycrystalline lead

zirconate titanates Pb(Zr

1x

+ 1 w% Nb

, constructing electroche-

mical oxygen concentration cells

Pt ð1:0 atm O

Þ = PTZ = Pt ð0:01 atm O

ÞCell ½3

in the temperature range from room temperature (r.t.) to 6008C, and they

published the results 2 years after Stephenson’s report. The cells sh owed a

stable emf, suggesting that the specimens exhibit ionic conduction. However,

in contrast to Stephenson’s report that the conduction was entirely ionic, the

results indicated that the conduction was partly assigned to an ionic state and

partly an electronic one, which is dependent on temperature and the composi-

tion of specimens. The difference in the ionic contribution in the conduction

between the two experiments would be caused by the differences in the sample

composition and the condition of oxygen concentration cells. Stephenson’s cell

[2] had a low er average partial pressure of oxygen than that of Heckman’s cell

[3]. In any case, the absolute values of ionic conductivity are low; the specific

resistivity of 1 w% Nb

-contained Pb(Zr

1x

is reported to be 5  10

to 1  10

ohm cm at 3008C.

Subsequently, Ezis et al. studied the dependence of the transport number of

ions on composition x in Pb(Zr

1x

in detail using an electrochemical

oxygen concentration cell [8]. Their experiment showed that the transport

50 H. Iwahara

number decreased with increasing value of x over the range 0.05 < x < 0.35

and that electronic conduction became dominant above x ¼0.40. Ezis et al.

considered that the reduction of T

to the trivalent state might be a mechanism

for defect formation, which tended to counter the effects of Nb

included in all

specimens investigated.

In addition to the studies on PZT, Glower et al. examined the ionic conduc-

tion in a ferroelectric material ,Ca

0.1

0.9

TiO

, by means of the same method as

that used for PTZ [9]. They showed that the conduction in this oxide is ionic

below 3008C, but electronic conduction appears sharply near 3008Cand

becomes predominant above 5008C. They identified the mobile ions as calcium

ions by using so-called activation analysis, which was a kind of irradiation

analysis of the surface of the solid after passing DC current across the solid

conductor.

An important study of charge carrier in a typical ferroelectric perovskite-type

oxide BaTiO

was published in 1964 by Glower and Heckman [10]. The title

of the paper ‘‘Conduction – Ionic or Electronic in BaTiO

’’ was attractive for

materials researchers in those days. In their paper, they reported about conduc-

tion species in barium titanate as follows: ‘‘Usual practice of authors has been to

make the tacit assumption that conduction is exclusively electronic. There is no

priori assurance that this is true.’’ Also, they wrote ‘‘ In fact, to our knowledge,

no one has to date addressed himself to the primary problem of transport,

i.e., does conduction occur via the motion of electrons or of ions?’’ The purpose

of their study was to answer this essential question from experimental results.

They applied Cell [1] to BaTiO

specimens for different oxygen partial pressures

and p

Pt; O

ðp

Þ = BaTiO

specimen = O

ðP

Þ; Pt Cell ½4

The cell using a pure single crystal showed no significant emf in an y cell

examined, indicating that its conduction was not ionic but electronic. On the

other hand, the single crystal containing 0.1 at% of iron showed strong emf

dependency on temperature and oxygen partial pressures at two electrodes. The

values were less than the theoretical one calculated from Nersnt’s equation,

suggesting that the conduction was partially ionic. BaTiO

showed theoret ical

emf below 2508 C, indicating that the conduction was purely ionic but at 5408C,

emf of the concentration cell was far lower than the theoretical one, suggesting

that contribution of ionic conduction to total one is rather small. Thus, it was

clarified that the contribution of ionic conduction in BaTiO

depends on its

purity and morphology, single crystal or polycrystal ceramic. The conduction

in a pure single crystal is electronic and that in impure single crystal and

polycrystalline ceramic is partially ionic, but the ionic contribution decreases

with increasing temperature.

However, gas concentration cell experiments give essentially no information

about which kind of ion among the constituent ions is mobile in the solids. To

investigate this, Glower and Heckman have also applied the activation analysis

3 Ionic Conduction in Perovskite-Type Compounds 51

method to the surface of an iron-containing single crystal of BaTiO

before and

after applying DC voltage of 22 V/mil for 52 h. The result showed that the iron

ions were mobile and contributed to ionic conduction but that titanium was not

mobile.

These studies were not intended to seek a good ionic conductor but to

confirm the conduction species to clarify the phenomena characteristic to the

ferroelectric or pyroelectric materials. It should be noted that the conductivity

of ferroelectric materials mentioned above is very low and most of the research-

ers in those days took no notice of the value of conductivity itself. Studies on

highly conductive ionic conductors of perovskite-type compounds were started

in the second half of the 1960s to search for a good oxide ion-conducting

electrolyte for fuel cells and oxygen sensors. These are described in the following

sections.

3.4 Oxide Ion Conduction

When a cation A or B in a perovskite-type oxide ABO

is partially replaced by

a cation M of lower valence, it sometimes gives rise to a relatively large

number of oxygen ion vacancies in the lattice so as to maintain the electrical

neutrality of the crystal. Chemical composition of the oxide can be expressed as

1x

3a

or AB

1x

3a

, where a is an average number of oxygen

deficiencies per unit formula. In such crystals, appreciable oxide ion conduction

may be expected at elevated temperatures if the energy needed for the oxygen

ions to jump from their original sites to adjacent vacant sites is not so high.

In this case, similarly to the case of stabilized zirconia, oxide ions can migrate

through the crystal lattice with the assistance of oxide ion vacancies.

There are several methods to confirm oxide ion conduction in the oxide

specimens. One of the most convenient methods is to examine the discharge

performance of the oxygen concentration cell. If the conduction in the sample is

oxygen ionic, a steady and stable current with a reasonable value can be taken

from the oxygen concentration cell, which is composed of the specimen oxide as

a solid electrolyte. Therefore, for example, one can regard the conduction as the

oxide ion conduction, if the oxygen concentration cell shown in Cell [1] gives

rise to stable emf and a steady and reasonable current can be taken from the cell.

Historically, oxygen ion conduction in the perovskite-type oxide was reported

for the first time by Stephenson and Flanagan [4], as described in the previous

section (3.2). They tried to construct a fuel cell using modification of oxygen

concentration cell shown in Cell [3]. In the experiment, PZT was used as an

electrolyte, and hydrogen gas and oxygen gas were supplied to each side of the

electrolyte separately. They observed a steady terminal voltage of 0.81 V at 3258C

and 0.41 V at 7008C. The current output was estimated to be a few mA/cm

.The

authors wrote as follows: ‘‘These results would indicate that at least some of the

conduction is by oxygen ion migration.’’ This study would historically be the first

52 H. Iwahara

experiments to confirm oxygen ion conduction in perovskite-type oxide and be

the first experiment of SOFC using a perovskite-type oxide as a solid electrolyte,

although their intention was not to develop a fuel cell and its solid electrolyte.

The research for developing the good ionic conductors with perovskite-type

structure was first considered by van Gool, who was known as the first to

propose a one-chamber solid oxide fuel cell. In 1965, he published a paper about

one-chamber fuel cells entitled ‘‘The possible use of surface migration in fuel

cells and heterogeneous catalysis’’ [11]. In this paper, he touched on the oxygen-

deficient perovskite-type oxides as a candidate for an oxide ion conductor

applicable to a fuel cell electrolyte. However, he thought that the perovskite

structure seemed to be less favorable because A and O in ABO

would make a

closed packing structure in which the ion migration might be difficult.

Studies on highly conductive ionic conductors of perovskite-type compounds

were started in the second half of the 1960s to find a superior electrolyte for fuel

cells and sensors. From the analogy of oxygen ion conduction in fluorite-type

oxides such as stabilized zirconias, it was thought that considerable concentration

of oxygen vacancy would be essential for high oxygen ion conductivity. The

present author and coworker have paid attention to the solid solution based on

LaAlO

that is composed of large-sized trivalent cation La and a small-sized

trivalent cation Al. In this oxide, calcium ions are partially substituted for

lanthanum ions and, as a result, oxygen ion vacancies are formed to compensate

charge neutrality in the crystal [12]; i.e., the composition is expressed as

1x

AlO

3a

. Having studied the behaviors of oxygen concentration cells

and fuel cells with La

1x

AlO

3a

(x = 0.1, 0.2, and 0.3) ceramics as a solid

electrolyte, they confirmed that the conduction was partly oxygen ionic and partly

electronic (due to electron holes) in air at elevated temperatures and that, under

the fuel cell condition, the conduction is predominantly oxide ionic [13]. The

CaTiO

can take aluminum to form a solid solution CaTi

1x

3a

(x  0.5) in

which almost stoichiometric amounts of oxygen vacancies are generated [14].

It was confirmed that this solid solution exhibits conduction behavior similar to

that of La

1x

AlO

3a

[15], and that the oxide ion conductivity is rather higher

than that of the latter. These studies were reported in a Japanese journal in 1967 and

1979, and these results were summarized in English and published in 1971 [16].

Steele et al. also reported oxygen ionic conduction in CaTiO

-based ceramic [17].

CaTiO

is a typical 2:4-type perovskite composed of large-sized divalent

cation Ca and small-sized tetravalent cation Ti, whereas the aforementioned

LaAlO

is a typical 3:3 type composed of large trivalent cation La and small

trivalent cation Al. The excellent high oxide ion conductor that was discovered

by Ishihara and reported in 1994 [1 8] is also based on 3:3-type perovskite

LaGaO

, as described in detail in Chapter 4.

Oxide ion conductors with the perovskite structure mentioned above belong

to so-called single perovskites, which can be expressed as the simple form,

ABO

. Besides these, there are different types of perovskite-related oxides and

some of them are known to show oxide ion conduction. One of them is

Brownmillerite, Ba

. This composition can be written as BaInO

2.5

, and

3 Ionic Conduction in Perovskite-Type Compounds 53

it is regarded that the ½O per unit formula of a perovskite is deficient; i.e., it can

be written as BaInO

2.5

0.5

, where hexpresses oxygen vacancy. At temperature

lower than 9308C, the arrangement of oxygen vacant sites is ordered [19].

Goodenough et al. found that the oxide ion conductivity in this Brownmillerite

oxide jumps up by more than one order of magnitude above 9308C, as shown in

Fig. 3.3 [20]. Above this temperature, the arrangement of oxide ion vacanci es

becomes disordered, and oxide ions can move easily in assistance of disordered

vacancies.

A characteristic feature of perovskite-type oxide ion conductors is that they

are often accompanied with p-type electronic conduction under an oxidizing

atmosphere such as air at elevated temperatures. As described in Section 3.2,

the contribution of electronic conduction depends on P

in the atmosphere

and temperature. As a typical example, Fig. 3.4 shows the P

dependence

of conductivities of CaTiO

- and SrTiO

-based solid solutions at 8008C [16].

P-type electronic conduction appears in the region of high P

and n-type one

under low oxygen partial pressure, i.e., a reducing atmosphere. The shape of the

curve lns lnP

is essentially the same as that shown schematically in Fig. 3.1.

In many fluorite-type oxide ion conductors such as stabilized zirconias and

Fig. 3.3 Conductivity of

as a function of

temperature [20]

54 H. Iwahara

doped cerias, p-type electronic conduction is rarely observed under oxidizing

atmosphere even under p(O

) = 1 atm, whereas many perovskite-type oxide

ion conductors become the mixed conductors (O

2

þh

) under oxidizing

atmosphere at elevated temperature. Hole conduction arises from the defect

equilibrium with oxygen in gas phase:

þ 1=2O

!



þ 2h

(3:6)

In many perovskite-type oxide ion conductors, the equilibrium constant K

so large that hole conduction appears even at a relatively weak oxidizing atmo-

sphere such as air, whereas in the case of fluorite-type oxides such as stabilized

zirconias, the equilibrium constant K

is too small to cause p-type electronic

conduction in air at elevated temperatures.

Some perovskite-type oxides having transition elements at B sites exhibit

mixed conduction at elevated temperatures. A typical example is doped lantha-

num cobaltite, in which oxide ions and holes are charge carriers. The electronic

conductivity is a few orders of magnitudes higher than that of the oxide-ionic

although the ionic conductiv ity itself is sufficiently high (>10

1

Scm

1

several 1008C). This kind of mixed conductor is a promising candidate for the

electrode materials of SOFCs and ceramic membrane reactors and is described

in Chapters 7 and 8 in detail.

3.5 Proton Conduction

Some perovskite-type oxides exhibit proton conduction under hydrogen-

containing atmosphere at elevated temperatures. Cerates or zirconates of

alkaline earth elements in which some trivalent cations are partially substituted

Fig. 3.4 Dependence of

conductivities of CaTiO

based and SrTiO

-based

solid solutions on oxygen

partial pressure at 8008C [16]

3 Ionic Conduction in Perovskite-Type Compounds 55