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

Подождите немного. Документ загружается.

of doped zirconia show a maximum at a specific concentration of dopant, which

wasreportedbyArachietal.[2]fortheZrO

–Ln

(Ln = lanthanide) system. It

is obvious that the electrical conductivity in ZrO

is strongly dependent on the

dopant element and its concentration. For a low concentration of dopant, the

conductivity monotonically increases with dopant amount, as is expected from

theory, and evidently the defects behave as a point defect. Therefore, conductivity

is mainly determined by the amount of oxygen vacancy, namely, the amount of

dopant. On the other hand, conductivity as well as activation energy for conduc-

tion are strongly affected by dopant ionic size. It is evident that the conductivity

increases with decreasing ionic size of doped cations. Explanations for this

conductivity behavior have been tried based on structural effects. The content

of dopant with the highest conductivity in the ZrO

–Ln

system decreases with

increasing dopant ionic radius. The dopants Dy

3þ

and Gd

þ3

with larger ionic

radii show a limiting value of 8 mol%. The dopant Sc

3þ

, which has the closest

ionic radius to the host ion, Zr

4þ

, shows the highest conductivity and the highest

dopant content at which cluster formation starts. Similar conductivity depen-

dence on the dopant level was observed in the CeO

system. Figure 4.2 shows the

conductivity of CeO

as a function of ionic size of Ln

. The highest conduc-

tivity was found at 10 mol% for the Sm

dopant and at 4 mol% for Y

.The

diffusion of oxide ion vacancies is affected by the local strain energy, which is

related to the mismatch between the host and dopant cation size [3]. Therefore,

not only dopant concentration but also ionic size is an important factor for

achieving high oxide ion conductivity. Recent studies on oxide ion conductivity

reveal that the clusters form at a dopant concentration much lower than that

considered previously [4]. Therefore, to achieve high oxide ion conductivity,

design of the dopant and its concentration is highly important.

Dopant ionic radius /nm

0.10

0.11 0.12

2.0

1.8

1.6

0.5

0.4

0.3

0.2

0.1

log(σT/Scm

–1

Binding energy/eV

Fig. 4.2 Conductivity and

binding energies of rare

earth dopants in CeO

as a

function of ionic size

4 Oxide Ion Conductivity in Perovskite Oxide for SOFC Electrolyte 67

Although at present YSZ is most commonly used for the electrolyte in a

SOFC, higher oxide ion conductivity is a major requirement for SOFC operation

at lower temperatures. However, in the case of tetravalent oxides with fluorite

structure, up to now a higher oxide ion conductor is only achieved with lower

chemical stability in reducing atmosphere and so the alternative candidates to

YSZ are quite limited: only Sc

–ZrO

doped with 1 mol% CeO

,orSm

doped with CeO

. Similar to the fluorite oxides, the perovskite structure

also has a large free volume, and so oxide ion conductivity in the perovskite oxide

presents an alternative to the fluorites for use as an SOFC electrolyte.

4.3 Oxide Ion Conductivity in Perovskite Oxides

Although oxides with perovskite structure are predicted to be good oxide ion

conductors, typical perovskite oxides such as LaCoO

and LaFeO

are widely

known as mixed electronic an d oxide ionic conductors, which can be used as

cathodes in SOFCs. Therefore, these mixed conducting perovskite oxides pre-

sent a promising material group for cathode catalysts for SOFC or oxygen-

permeating membranes. The large majority of perovskite oxides exhibiting

oxide ion conduction are classified as mixed conductors, which show both

electronic and oxide ionic conduction and thus cannot be used as the electrolyte

of an SOFC.

Takahashi and Iwahara have done pioneering work on perovskite oxide ion

conductors [5]. They reported fast oxide ion conductivity in Ti- and Al-based

perovskite oxides, and it is evident that Al- or Mg-doped CaTiO

exhibits high

conductivity, but it is still lower than that of YSZ. Iwahara and Takahashi

investigated the oxide ion conductivity in CaTiO

in detail [5]. On the other

hand, although the oxide ion exhibits a high transport number in CaTi

0.95

0.05

at intermediate temperatures, Ca-doped LaAlO

is another attractive candidate as

an oxide ion conductor, since no electronic conduction appears in a reducing

atmosphere and transport number is higher than 0.9 over the entire temperature

range.

After the report of Takahashi and Iwahara [5], many researchers investigated

the oxide ion conductivity in LaAlO

-based oxide. However, the reported oxide

ion conductors with the perovskite structure exhibited lower ionic conductivity

than that of Y

–ZrO

. In the conventional study of perovskite oxides, ABO

,it

was believed that the electric or dielectric properties were strongly dependent on

B-site cations. However, a migrating oxide ion has to pass through the triangular

orifice consisting of two large A- and one small B-site cations in the crystal lattice.

Therefore, the ionic size of the A-site cation seems to influence greatly the oxide

ion conductivity. Effects of A-site cations on oxide ion conductivity in LnAlO

based perovskite oxide were reported. Figure 4.3 shows the electrical conductivity

in LnAlO

-based oxides [6]. Electrical conductivity of Al-based perovskite oxides

increased with increasing the ionic size of A site cations, which suggests that the

68 T. Ishihara

larger unit cell volume is important for higher oxide ion conductivity because of

the larger free volume. Therefore, doping larger cations for the B site is also

important for achieving high oxide ion conductivity. The oxide ion conductivity

in NdAlO

doped with Ga at Al sites was studied [6]. The addition of Ga

3þ

larger ion than Al

3þ

, to B sites of NdAlO

is effective for improving oxide ion

conductivity [7]. Figure 4.4 shows the oxide ion conductivity at 1123 K as a

function of Ga content. In accordance with the prediction, oxide ion conductivity

increased with an increasing amount of Ga, and it attained the maximum (log s/

Scm

–1

) ¼ –1.5 at 1223 K when 50 mol% Ga was doped at Al sites in LaAlO

Since no oxygen vacancy is formed by doping Ga

3þ

because of the same valence

Fig. 4.3 Arrhenius plot of the electrical conductivity of Ca-doped LnAlO

[Ln = La (h), Pr

(

), Nd (*), Sm (

),Gd (

), Y (

), Yb (^)] (a) and electrical conductivity at 1223 K as a

function of A-site ion size (b)

X in Na

0.9

0.1

1-X

0 0.5 1.0

log(σ/Scm

–1

)

–2

–1

Fig. 4.4 Oxide ion

conductivity in

0.9

0.1

1–x

1123 K as a function of Ga

amount

4 Oxide Ion Conductivity in Perovskite Oxide for SOFC Electrolyte 69

as Al, higher oxide ion conductivity is brought about by the improved mobility of

oxide ion by increasing the unit cell volume, or free volume in the unit cell.

Although high oxide ion conductivity was obtained on Nd

0.9

0.1

0.5

the oxide ion conductivity of this compound is still lower than that of YSZ.

However, it is of great interest that higher oxide ion conductivity is reported for

LaGaO

-based perovskite oxide, which is the end composition of that in Fig. 4.4.

Oxide ion conductivity in LaGaO

-based oxide is overviewed in detail in the next

section.

Another type of perovskite oxide ion conductor is LaScO

, which is also

reported as a high-t emperature proton conductor [8]. Figure 4.5 shows the

temperature dependence of four different perovskite oxides with similar com-

position [9]. In spite of the similar compositions, the oxide ion conductivity is

very different, i.e., higher for LaGaO

. However, LaAlO

, LaScO

, and LaInO

show lower oxide ion conduction with hole conduction in the higher P

range.

A similar study was performed by Nomura et al., and the order of oxide ion

conductivity in these perovskite oxides is not simply explained by free volume

size, but by size matching of dopant, in particular, Mg to B-site cations. Mg is

too large as a dopant on the B site of these four perovskite oxides; however, it is

the one closest to Ga in size [10]. Comparison of the defect association energy is

further required, as discussed later; howeve r, it is evident that LaGaO

is a

promising oxide ion conductor over a wide P

range [11]. Oxide ion conduc-

tivity in LaGaO

is discussed in the next section.

1000/T/K

–1

–2

–3

–4

–5

–6

0.7 1.2 1.7 2.2

log( σT/Scm

–1

Fig. 4.5 Arrhenius plot of four different perovskite oxide in air. La

0.9

0.1

0.9

0.1

(^);

0.9

0.1

0.9

0.1

(h); La

0.9

0.1

0.9

0.1

(*); La

0.9

0.1

0.9

0.1

(

) (Cited

from Ref. (9))

70 T. Ishihara

4.4 LaGaO

-Based Oxide Doped with Sr and Mg (LSGM)

as a New Oxide Ion Conductor

4.4.1 Effects of Dopant for La and Ga Site

The high oxide ion conductivity in LaGaO

-based perovskites, which is a pure

oxide ion conductor, was first reported in 1994 [12]. The high oxide ion con-

ductivity in this oxide is achieved by double doping of a lower valence cation

into both the A- and the B sites of perovskite oxide, ABO

. It is obvious that the

oxide ion conductivity strongly depends on the cations on the A site, which is

similar to the Al-based oxide, and the highest conductivity is achieved for

LaGaO

, which has also the largest unit cell volume of the Ga-based perovs-

kites. The electrical conductivity of all the Ga-based perovskite oxides is almost

independent of the oxygen partial pressure, and, theref ore, it is expected that

the oxide ion conduction will dominate in all Ga-based perovskite oxides.

Doping with a lower valence cation generally form s oxygen vacancies due to

the requirement for electric neutrality; the oxide ion conductivity increases with

increasing the amount of oxygen vacancies. Therefore, doping alkaline earth

cations onto La sites was investigated, and the oxide ion conductivity is shown

in Fig. 4.6 [12]. The electrical conductivity of LaGaO

depends strongly on the

alkaline earth cations and also increase in the following order Sr > Ba > Ca.

1000/T /K

1000

900 800 700 600 500

Temperature /°C

log(

/Scm

–1

)

Fig. 4.6 Effects of the nature

of alkaline earth cations on

the La site on oxide ion

conductivity in LaGaO

4 Oxide Ion Conductivity in Perovskite Oxide for SOFC Electrolyte 71

Therefore, strontium, of which the ionic size is almost the same as that of La

3þ

is the most suitable dopant for the La sites in LaGaO

. Theoretically, increasing

the amount of Sr will increase the amount of oxygen vacancies and hence the

oxide ion conductivity. However, solid solubility of Sr into La sites of LaGaO

is poor, and the secondary phases, SrGaO

or La

SrO

, form when the amount

of Sr becomes higher than 10 mol%. Thus, the concentration of oxygen vacan-

cies introduced by La site doping is not large.

The effects of dopant on Ga sites of La

0.9

0.1

GaO

were also studied for

further improvement of electrical conductivity. It is found that doping with Mg

is very effective at increasing the conductivity because additional oxide ion

vacancies are formed. The oxide ion conductivity is further increased by

increasing the amount of Mg added; the maximum conductivity is attained at

20 mol% Mg doped for Ga sites. The lattice parameter also increases by doping

Mg for Ga sites as the ionic radius of Mg is larger than that of Ga. The solid

solubility of Sr into the LaGaO

lattice seems to reach a limit around 10 mol%

without Mg; however, it increases up to 20 mol% by doping Mg for Ga. This

enlargement in the limit of Sr solid solution was also reported by Majewski et al.

[13], which seems to be a result of the enlarged crystal lattice. In any case, the

highest oxide ion conductivity in the LaGaO

-based oxide is reported at the

composition of La

0.8

0.2

0.8

0.2

[14].

Because this oxide consists of four elements, the optimum composition varies

slightly from group to group. Oxide ion conductivity in LaGaO

-based oxide

was investigated by several groups [15, 16], and various cations were examined

as a dopant for LaGaO

-based oxides. Huang and Petric investigated the oxide

ion conductivity of various compositions [16] and expressed the oxide ion

conductivity in contour maps [17] (Fig. 4.7), in which the optimum composition

reported by two other groups is shown. Huang et al. reported that the highest

oxide ion conductivity was obtained at the composi tion of La

0.8

0.2-

0.85

0.15

[17] On the other hand, Huang et al. and Huang and Good-

enough reported the optimized composition in La

1–X

1–Y

at X =

0.2, Y = 0.17 [17, 18]. However, the optimized composition among the three

groups is close to each other, and the optimized composition in Sr- and Mg-

doped LaGaO

exists between Y = 0.15 to 0.2 in La

0.8

0.2

1–Y

. The

difference may come from the uniformity of composition and also grain size.

Figure 4.8 shows the comparison of oxide ion conductivity of doubly doped

LaGaO

with the conventional fluorite oxide ion conductors. It is obvious that

the oxide ion conductivity in La

0.8

0.2

0.8

0.2

is higher than the typical

conductivity of ZrO

- or CeO

-based oxides and somewhat lower than those of

-based oxides. It is well known that electronic conduction is dominant in

CeO

-orBi

-based oxides under a reducing atmosphere; furthermore, thermal

stability is not satisfactory in Bi

-based oxides. In contrast, La

0.8

0.2

0.8

0.2

exhibits wholly ionic conduction from P

¼10

–20

to 1 atm. Therefore, doubly

doped LaGaO

perovskite oxide shows great promise as the solid electrolyte

for fuel cell and oxygen sensor.

72 T. Ishihara

Optimized composition

IshiharaGoodenough

Petric

Fig. 4.7 Contour plot of

conductivity at 1073 K as a

function of x and y in

1–X

1–Y

[

]

1000/T /K

-1

Temperature/°C

1000 900 800 700 600

Fig. 4.8 Comparison of

oxide ion conductivity in

doubly doped LaGaO

with

conventional oxide ion

conductors

4 Oxide Ion Conductivity in Perovskite Oxide for SOFC Electrolyte 73

4.4.2 Transition Metal Doping Effects on Oxide Ion Conductivity

in LSGM

To improve oxide ion conductivity, several groups have already investigated the

effects of the various cation dopants on oxide ion conductivity in LaGaO

-based

oxides [19]. Doping transition metal cations such as Co, Ni, or Fe is not preferable

for a solid electrolyte due to the formation of partial electronic conduction. How-

ever, it is reported that oxide ion conductivity is increased by doping Co without

decreasing the transport number of oxide ions [20], if the amount of Co is smaller

than 10 mol%. In this section, the effect of transition metal doping on oxide ion

conductivity is briefly summarized. As shown in Fig. 4.8, the Arrhenius plot for

LSGM is slightly bent around 1000 K, suggesting a change in conduction mechan-

isms. Detailed crystal structure analysis by neutron diffraction suggests that the

crystal phase changes from triclinic to pseudo-cubic lattice. This phase change might

be related to the mismatch of the Mg

2þ

ionic size with that of Ga

3þ

. Doping with a

trivalent cation similar in size to that of Mg

2þ

might be effective in stabilizing the

high-temperature cubic phase. Considering the ionic size of trivalent cation with a 6-

coordinated number, it seems that Fe, Co, and Ni are candidate ions.

Baker et al. investigated the effects of doping with transition metals Cr and

Fe on the oxide ion conductivity of LaGaO

-based oxide [21]. It was reported

that doping with Cr or Fe on Ga sites induces hole conduction in LaGaO

based oxides, resulting in decreased stability against reduction. Figure 4.9

shows an Arrhenius plot of electrical conductivities of the LaGaO

-based

oxides doped with some transition metal cations on Ga sites [20]. It was observed

1000/T /K

–1

in N

Fig. 4.9 Arrhenius plot of

electrical conductivities of

the LaGaO

-based oxides

doped with some transition

metal cations for Ga sites

74 T. Ishihara

that the conductivity is improved by doping with Co and Fe and is lowered by

doping with Cu and Mn. In the case of Ni, the conductivity decreased with

increasing temperature above 1073 K, whereas below 973 K, it increased with

increasing temperature. Such a decrease in conductivity in spite of increasing

temperature may result from the significant electronic conductivity caused by the

thermal reduction of Ni. From the P

dependence of the electrical conductivity,

n-type conductivity is greatly enhanced by doping with Mn and Ni, and p-type

conduction increases by doping with Cu. Kharton et al. also investigated the

effects of transition metal dopant on oxide ion conductivity in LaGa

0.8

0.2

[22]. Although the amount of doped transition metal is much larger, i.e., 40 mol%

for Ga sites, they also reported that doping with Mn and Cr decreases oxide ion

conductivity. However, Thangaduari et al. reported that the La

0.9

0.1-

0.8

0.2

exhibits an oxide ionic conductivity that is comparable with that

of La

0.9

0.1

0.8

0.2

[23]. In addition, the activation energy for ion con-

ductivity in the Mn-doped sample is much smaller than that of Mg-doped ones.

However, the small activation energy may suggest dominant electronic conduc-

tion in this oxide. In contrast, the total conductivity of Fe- or Co-doped speci-

mens is almost independent of the oxygen partial pressure, which suggests that

oxide ion conductivity increases by doping Co or Fe [24]. Therefore, oxide ion

conductivity in Co-doped LaGaO

-based oxide will be important. On the other

hand, it is reported that the higher Fe-doped La(Sr)GaO

exhibits mixed con-

ductivity with both hole and oxide ions contributions. This material shows large

oxygen permeation flux when used as an oxygen separation membrane.

Figure 4.10 shows the electrical conductivity of Co-doped LSGM at 1273 K,

¼ 10

–5

atm and the transport number of the oxide ion at 1273 K as a

total

0.00 0.05 0.10 0.15 0.20

–0.5

–0.4

–0.3

–0.2

–0.1

0.0

0.2

0.4

0.6

0.8

1.0

Transport Number, T

X in La

0.8

0.2

0.8

0.2–x

log(

/Scm

–1

)

Fig. 4.10 Electrical

conductivity of Co-doped

LSGM at 1273 K, P

5

atm and the transport

number of oxide ion at

1273 K as a function of Co

content

4 Oxide Ion Conductivity in Perovskite Oxide for SOFC Electrolyte 75

function of Co content [20]. The electrical conductivity increases, whereas the

transport number of the oxide ion decreases with an increasing amount of Co.

The oxide ion conductivity values estimated from the transport number and

the total conductivity are shown in Fig. 4.11. T he electrical conductivity

becomes higher with increasing the amount of Co and attains a maximum

value at around 10 mol%. The apparent activation energy for the electronic

conduction monotonically decreased with increasing the Co concentration

and reaches a value of 0.45 eV for 10 mol% C o, which is almost half of the

value reported for YSZ. Although the highest oxide ion c onductivity is

obtained at X ¼ 0.1, the transport number of oxide ion becomes smaller

than 0.9. Since the decreased transport number of the oxide ion leads to a

decrease in the energy conversion efficiency of a SOFC, it is considered that

the desirable composition as the electrolyte for SOFC is La

0.8

0.2-

0.8

0.115

0.085

(denoted as LSGMC-8.5) or one with even lower Co

content.

In Fig. 4.8, the temperature dependence of oxide ion conductivity in

Co- doped LaGaO

-based oxide is also shown. It is seen that Co-doped

LaGaO

-based oxides exhibit even higher conductivity than that of LSGM

and Gd-doped CeO

. Another interesting point is the disappearance of the

slope change around 1000 K, suggesting that the high-temperature cubic

phase is stabilized. The conductivity value of this Co-doped LSGM is close

to that of Bi

-based oxide, which exhibits pure oxide ion conduction in a

limited P

range. Therefore, Co-doped LaGaO

, in particular, 8.5 mol% Co-

doped LSGM, is a good choice to use as the electrolyte of a solid oxide fuel cell

operable at intermediate temperatures.

Estimatedσ

ion

0.00 0.05 0.10 0.15 0.20

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

–0.5

–0.4

–0.3

–0.2

–0.1

0.0

X in La

0.8

0.2

0.8

0.2–X

log(σ/Scm

–1

)

Fig. 4.11 Oxide ion

conductivity estimated from

the transport number and

the total conductivity in

LSGM

76 T. Ishihara