The Effects of Dopants and A:B Site Nonstoichiometry onProperties of Perovskite-Type Proton Conductors

I Guan,b S. E. Dorris,° U. Balachandran,° and M. Uu*tb

aArgonne National Laboratory, Energy Technology Division, Argonne, Illinois 60439, USA1School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332, USA

ABSTRACT

Investigations of perovskite-type BaCeO3 and SrCeO2 with various dopants (Y, Gd, Nd, and Ni) indicate that theirmicrostructures and electrical properties are strongly influenced by the type and amount of dopants. Grain growth anddensification of sintered samples are influenced by dopant level and A:B site nonstoichiometry. The conductivity ofBaCe1_Y039 increases with the yttrium content in hydrogen and wet Ar; and exhibits a maximum in oxygen at an yttri-um content of 10 to 20%. BaCe08Y0203_1 has the highest conductivity in a hydrogen atmosphere: —4.54 X 10-2 11' cm' at600°C, and —-4.16 >< 10 fl1 cm1 at 800°C. The effect of BaO excess depends on the concentration of dopant. Comparedwith BaCe091Y01503_0, doped BaCeO3 with BaO excess (Ba00.90Ce020.025Y203) has a higher total conductivity in allatmospheres studied (02, H2, and wet Ar), whereas the conductivity of BaCeO3 with excess BaO (Ba00.85Ce020.05Y203)is lower than that of BaCe09Y61O26. BaCeO3 based materials have higher conductivities than those of SrCeO2 based mate-rials, whereas SrCeO3 based materials show higher proton transference numbers.

IntroductionMixed ionic electronic conductors (MIECs) are used in

many solid-state electrochemical systems such as solidoxide fuel cells (SOFCs), solid-state gas sensors, and mem-branes for gas separation.12 Two well-known MIECs arethe partially substituted perovskite-type oxides BaCeO2and SrCeO3, in which substitution for Ce by trivalentcations causes the formation of oxygen vacancies andother charged defects and gives rise to mixed conductionin atmospheres containing 02, H2, and H20 vapor. Becausemuch of the charge transport in BaCeO2 and SrCeO2 is byprotons, these materials are being investigated as possiblehydrogen separation membranes. To be suitable for hydro-gen separation, a material must have a high selectivity forhydrogen, so its proton transference number must be muchhigher than its transference number for oxygen ion con-duction. To be useful in a nongalvanic mode, the transfer-ence number for electronic conduction should be compa-rable to that for protonic conduction, and the protonic andelectronic conductivities should be sufficiently high (>

X 10 fl' cm'). In addition, the materials must ex-hibit high catalytic activity for the dissociation and re-combination of hydrogen at the gas/solid interfaces.

Transport properties of perovskites are strongly influ-enced by the ionic radii of dopants. Kilner and Brook3 usedlattice simulation techniques to model ionic conduction inperovskites and concluded that the overall activation ener-gy for conduction should be minimal when the host anddopant cations have similar ionic radii. Based on this crite-rion, Bonanos et al. suggested that Gd would maximize theconductivity of BaCeO2. In fact, BaCe61Gd22023 has beenwidely studied for SOFCs because of its high conductivityin 02-containing atmospheres and under fuel cell condi-tions.4-6 Iwahara et al.7 studied the mixed conduction ofYb-, Y-, Dy-, Gd-, Sm-, and Nd-doped BaCeO2, and report-ed that the proton transference number decreased whilethe oxygen ion transference number increased withincreasing dopant ionic radius. They reasoned that dopantswith large ionic radii made oxygen ion conduction morefavorable by enlarging the spacing along the a axis.

In atmospheres containing water vapor, protons can beformed through the reaction

H20+V+0#20H [1]

Liu and Nowick found that Nd-doped BaCeO3 was a verygood proton conductor when exposed to a water vapor-containing atmosphere8 but that Eu-, Yb-, and Gd-dopedBaCeO3 were not.9 Proton conduction in moist atmos-pheres was also studied by Slade and Singh, who showedthat Gd-doped BaCeO2 had the highest conductivity

Electrochemical Society Student Member.Electrochemical Society Active Member.

among compounds doped with Y, Gd, Nd, and La. 10Stevenson et al. also found that 15% Gd-doped BaCeO3had the highest total conductivity among Yb-, Nd-, andGd-doped samples in wet Ar.

Formation of protons in alkaline earth cerates occursreadily in hydrogen-containing atmospheres according to

H2+2OP2OH+2e' [2]

Nagamoto and Yamada have studied the effect of dopantsfor A and B sites on conductivity in hydrogen.12 AmongBaCeO3 samples doped with La, Nd, Yb, Er, Tb, and Nb,Nb-doped BaCeO3 showed the highest conductivity inhydrogen. High conductivity of Nb-doped BaCeO3 in H2was also observed by Rauch and Liu.'3 However, electro-motive force measurements indicated an absence of protonconduction in Nb-doped BaCeO3. 12 At the same time, Iwa-hara et al. observed that Y-, Sm-, and Nd-doped BaCeO3had high conductivities in hydrogen,14"1 and that 20% Sm-doped BaCeO3 exhibited very high total conductivity inhydrogen, (—4 >< 10_2 fi2 cm at 800°C). However; densesamples were difficult to prepare. According to a morerecent report,2 BaCe01Y22031 may have the highest con-ductivity reported to date, —5.3 >< 10 fl cm at 800°Cin hydrogen.

SrCeO3 doped with Yb 26 and Y 12 have also been studiedas proton conductors. Compared to BaCeO3 based materi-als, SrCeO2 based materials have lower total conductivitiesbut higher proton transference numbers.16 Their higher pro-ton transference numbers may make these materials desir-able for hydrogen separation due to increased hydrogenselectivity. However; the electronic transference number ofY-doped SrCeO3 (and BaCeO3) was <0.2 at 800°C, based onopen-cell voltage measurements on hydrogen concentrationcells?0'20 Also, the materials exhibited high interfacialpolarization during hydrogen permeation tests.2° In order toimprove catalytic properties and electronic conductivity,appropriate dopants or a second phase must be introduced.Typically, elements in group VIII, such as Fe, Co, and Ni, aregood catalysts, and Ni-oxide cermet has already been usedas anode material in SOFCs because of its good catalyticproperties and high electronic conductivity.21

In addition to stoichiometric alkaline earth cerates,BaCeO3 containing excess BaD 22 and complex perovskiteswith the formula of A3B+Nb2091 (A = Ba, Sr, Ca, andB' = Sr, Ca, Mg)23 have been studied. After water uptakemeasurements and conductivity analysis, K.reuer et al.22concluded that excess BaO formed a grain boundaryphase. Recently, Shima and Haile24 reported that excessBaO aids in densification and enhances the conductivity of15% Gd-doped BaCeO2 in wet Ar at 500°C, but sampleswith 4% BaO excess eventually lost mechanical integritydue to reaction with atmospheric CO2.

1780 J. Electrochem. Soc., Vol. 145, No. 5, May 1998 The Electrochemical Society, Inc.

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP

J. Electrochem. Soc., Vol. 145, No. 5, May 1998 The Electrochemica? Society, Inc. 1781

In this study, stoichiometric and nonstoichiometricBaCeO3 and SrCeO3 were doped with various materials (Gd,Nd, Y, and Ni) at different dopant levels (5, 10, and 20%), andthe effects of these composition variations on sinterability,microstructure, and conductivity were characterized.

ExperimentalSample preparation.—BaCeO3 and SrCeO3 based mate-

rials were prepared by using solid-state reactions.'92° Allpowders in this study had a perovskite structure exceptSrCe0 8(Ni0 1Y01)O3, in which CeO2 was detected by X-raydiffraction (Scintag, XDS 2000). Pellets of doped BaCeO3were sintered in air at 15 50°C for 10 h. SrCeO3 based mate-rials without Ni dopant were sintered in air at 1500°C for10 h, whereas, materials with Ni dopant were sintered inair at 1300°C for 5 h. Both sides of each pellet were thenpolished with 600 grit SiC polishing paper. Platinum(Heraeus CL11-5 100) or silver (Heraeus C1000) paste waspainted on the polished surfaces and dried in air at 150°Cfor 5 h and subsequently fired in air for 12 mm at 12 00°C(for Pt paste) or at 850°C (for Ag paste) to form porouselectrodes for electrochemical measurements.

Microstructure.—Microstructures of materials werestudied by scanning electron microscopy (SEM; JEOLJSM-5400) on as-sintered surfaces, fracture surfaces, andetched surfaces. To prepare an etched surface, the pelletwas first polished with 600 grit SiC polishing paper, thenimmersed in 1 N HNO3 solution for 2 mm, and finallycleaned in isopropyl achohol.

Impedance spectroscopy.—An impedance analyzer(HP4192A LF) was used to acquire impedance spectra inthe frequency range from 5 Hz to 13 MHz at temperaturesbetween 550 and 800°C. Before the impedance measure-ments, each cell was first equilibrated at 800°C for 5 h andthen cooled to 550°C in the desired atmosphere (02, H2, orwet Ar). Water vapor was obtained by bubbling Arthrough deionized water at room temperature (—22°C).

H102 transference number—The ratios of proton andoxygen ion transference numbers to the total ionic trans-ference numbers were obtained by discharging a fuel cell

02, Pt I doped SrCeO3 I Pt, 4% H1 + Ar

with the external circuit shorted. The ratios (tH./tIOfl andt02 /t) were calculated from discharging currents andwater vapor evolution rates of the cathode and anode com-partments472° using

= - [3a]

tfln = t}l + t0 [3c]

where We and w are the rates of water vapor evolutionfrom the cathode side and anode side, respectively, and wis the total rate of water vapor evolution from both sidesor calculated from the short-circuit currents (using Fara-day's law). The rates of water vapor evolution due to elec-trochemical process were calculated from the flow rates ofcarrier gas and humidity measurements by a hygrometer(Fisher Scientific) in both sides of the cell under bothopen-circuit and short-circuit conditions. This procedureprovides relative transference numbers only. To measurethe absolute transference numbers, a separate experimentwould be needed to determine the electronic contribution.

Results and DiscussionMaterial sintering.—Initially, green pellets were placed

directly on an A1203 setter for sintering. After sintering,the surface areas in contact with alumina were yellowish,while the uncontacted surface areas were brown or darkgreen, depending on the initial material composition. The

large difference in color suggested a reaction with the set-ter. As expected, the X-ray diffraction analysis of the con-tacted surface areas revealed the presence of CeO2 as themajor phase and BaCeO3 as a minor phase (Fig. la). An X-ray diffraction analysis of the uncontacted surface areas(the so-called free surface) showed the perovskite struc-ture of BaCeO3, and possibly a trace amount of CeO2, dueto BaO loss at high temperature (Fig. ib). Cross-sectionalviews (Fig. 2a) showed clearly the formation of a reacted

(a) Contacted surface

__ - __L.(b) Free surface__ -- 4-

I I I

20 30 40 50 60 7028 (degree)

(c) CeO2, JCPDS 34-394Ii H(d) BaCeO3, JCPDS 22-74

____ - . p

Fig. 1. X-ray diffraction patterns of 20% Gd-doped BaCeO3,showing a composition change when contacted by alumina setterduring sintering.

Fig. 2. SEM photomicrographs of (a) cross section of contactedsurface and (b) cross section of free surface of 20% Gd-dopedBaCeO3.

I

IU

and

tflfl W

= [3b]tmn Wt

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP

1782 J. Electrochem. Soc., Vol. 145, No. 5, May 1998 The Electrochemical Society, Inc.

Fig. 3. SEM photomicro-graphs of as-sintered surfacesof (a) SrCeO3, (b) 5% Y-SrCeO3,Cc) 10% Y-SrCeO3, and (d)SrO0.90CeO20.025Y2O3. Allsamples were sintered in air at1500°C.

layer 10 to 15 m thick, whereas no such layer was evidentin materials that did not contact A1003 (Fig. 2b). Energydispersive X-ray (EDX) analysis showed that the molarratio of Ba over Ce in the reacted layer was —1:9. Similarreactions were evident after sintering other doped BaCeO3and SrCeO7 on A1203 setters, suggesting a reaction such as

Ba(Sr)Ce03 + A1903 CeO2 + Ba(Sr)A1104

Such a reaction between Ba(Sr)Ce03 and Al90.3 must beconsidered in fabricating thin films of Ba(Sr)Ce03 on alu-mina substrates.2526 To prevent the reaction in the presentstudy, powders with the same composition as the greenpellets were used as a buffer layer between the green pel-lets and the setter.

Microstructure.—Photomicrographs of as-sintered sur-faces of SrCeO3 based materials (Fig. 3) show that both sto-ichiometric and nonstoichiometric (i.e., with excess SrO)SrCeO3 can be densified at 1500°C. Increasing the temper-

Fig.4. SEM photomicrographsof (a) fracture surface of 5%Y-BaCeO3, (b) fracture surfaceof 8a00.90CeO20.025Y203,(c) as-sintered surface of10% Y-BaCeO3, and (d) as-sintered surface of8a00.85CeO20.05Y203. Allsamples were sintered in air at1550°C.

ature to 1550°C caused extra grain growth. UndopedSrCeO showed a rough surface with honeycomblike struc-ture (Fig. 3a). A similar structure was observed by Du andNowick on a thermally etched Ba3Ca1 Nb surface.27Under identical sintering conditions (temperature anddwell time), grain growth increased as dopant level in-creased (Fig. 3a, b, and c).

[4] Photomicrographs of BaCeO3 based materials sinteredat 1550°C are shown in Fig. 4. In this study, stoichiometric5% Y-doped BaCeO did not densify well (—89 % dense;Fig. 4a), whereas a sample with 5% excess BaO(BaO0.90CeO20.025Y2O3) had a very dense fracture sur-face (—94 %; Fig. 4b). This suggests that the properamount of excess BaO promotes sintering for this dopantlevel (5% Y). At higher dopant concentration (10% Y),however, the stoichiometric material was very dense(—98 %; Fig. 4c), whereas, the sample with 5% excess BaO(BaO0.85CeO70.05Y9O7) was not very dense (—91 %dense) and had finer grains (Fig. 4d). The observation of

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP

J. Elect rochem. Soc., Vol. 145, No. 5, May 1998 The Electrochemical Society, Inc. 1783

Kulscar28 that 2 mol % excess BaO in BaTiO3 acted as agrain size refiner suggests that the excess BaO may inhib-it grain growth in the 10% Y-doped sample.

Excess BaO may incorporate into the perovskite up tothe solubility limit according to defect reactions2429

BaO Ba + O + V + 2V

2BaO Ba. + Ba + 2 O + VAt low dopant concentrations, the forward reactions may befavored because the concentration of oxygen vacanciesintroduced by the dopant is relatively small. As a result, ex-cess BaO can be easily incorporated into the structure tocreate additional oxygen vacancies, which improve vacancydiffusion and thereby promote sintering. Higher dopantconcentrations, however, may saturate the material withoxygen vacancies, in which case the forward reactions in 5aand 5b became less favorable. As a result, Ba-rich phasesmay segregate at grain boundaries, thus impeding sintering.

Effects of excess BaO on microstructure were alsostudied by examining 5% Y-doped BaCeO3 with and with-out excess BaO after acid etching. After acid etching for2 mm in HNO3, the sample with 5% excess BaO(BaO. 0.9OCeO,• 0.025Y903) showed clearly distinguishedgrains (Fig. 5a), but the stoichiometric doped BaCeO3(BaCe095Y005O3 ,) was randomly corroded (Fig. 5b) withfiner grains. Although excess BaO acted as a grain sizerefiner in BaO0.85CeO,0.05Y2O3 (Fig. 4d), the observa-tion of coarser grains in BaO0.90CeO20.025Y2O3 ratherthan in BaCeO 95Y0 9503 is exactly opposite. It is possiblethat most of excess BaO in the sample with low dopantconcentration (BaO0.90CeO20.025Y2O3) had already be-come incorporated into the matrix according to Eq. 5. Thisincorporation could create more vacancies and thus pro-mote grain growth and densification.3° In the sample withthe high dopant concentration (BaO.0.85CeO20.05Y203),however, most of the excess BaO may exist as grain bound-

ary phases due to oxygen vacancy saturation in thematrix. The existence of excess BaO at grain boundaries inBaO•0.85CeO20.05Y2O3, however, could act as a grain-sizerefiner and impede sintering.28

As for stability, samples with compositions ofSr00.90CeO20.025Y203, SrO0.85CeO20.05Y03, and1.05SrOCeO, appeared to be acceptable immediatelyafter sintering, but crumbled into pieces after exposureto air for several days. After exposure to air formany weeks, as-sintered BaO0.90CeO2 0.025Y203 andBa O .85CeO2 0.05Y203 pellets, however, retained goodmechanical integrity. However, if the as-sintered pelletswere broken into two pieces and then exposed to air,BaO0.90CeO20.025Y2O3 maintained good integritywhile Ba0.85CeO70.05Y203 crumbled. The lower stabilityin air of BaO0.85CeO20.05Y2O3 than that ofBaO0.90CeO20.025Y,O3 suggested that the former mayhave more Ba-rich phases at grain boundaries than thelatter. Scanning electron microscopy (SEM) examinationon fracture surfaces of BaO0.90CeO20.025Y2O3 showedthat the entire fracture surface is dense, as shown inFig. 4b, and that there was no observable change after thefracture surface was exposed to air for four weeks. How-ever, after exposure of sintered Ba00.85CeO20.05Y203 toair for two weeks, examination of the center portion of thefracture surface revealed a Ba-rich phase in the grainboundaries (Fig. 6). EDX analysis showed that the molarratio of Ba over Ce is —91:9 in the Ba-rich phase, whileneighboring grains had a molar ratio of Ba:Ce:Y —49:44:7,close to the mixture ratio of Ba:Ce:Y = 100:85:10 when thematerial was prepared. The observation of a Ba-rich phasein BaO0.85CeO20.05Y2O3 confirmed our assumption thatthe excess BaO likely exists in the grain boundary area ofthe sample with high dopant concentration due to the pos-sible saturation of oxygen vacancies. Such a Ba-rich phasewas not observed in the surface layer of the sintered pelletbecause BaO was probably vaporized at high temperatureduring sintering. As such, the stable surface layer prevent-ed the Ba-rich phase in the center of the sintered bodyfrom further reacting with atmospheric C02, therebymaintaining the good integrity of the as-sintered pellets.However, after the pellet was broken, the Ba-rich phaseinside the sintered body was readily accessed by atmos-pheric CO,, forming BaCO3 and consequently causing thecrumbling of the pellet.

Conductivity of BaCeO3 based materials—Temperaturedependence of stoichiometric doped BaCeO3—Shown inFig. 7a are the total conductivities of doped BaCeO3 in1 atm oxygen as determined by impedance spectroscopy.The total conductivity of Y-doped BaCeO increased withtemperature. The conductivity of BaCe08Gd02O3 6 wasslightly higher than that of BaCe09Y01O39 andBaCe08Y0,036 in this work at temperatures <700°C, butwas almost the same at >700°C.

and[5a]

[5b]

Fig. 5. SEM photomicrographs of (a) acid-etched surface ofBa00.9OCeO2O.025Y203 and (b) acid-etched surface of 5% Y-BaCeO3

Fig. 6. SEM photomicrograph of fracture surface ofBaOO.85CeO2-O.05Y203 after exposure to air for 2 weeks.

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP

1784

6

I 0

t00

J. Electrochem. Soc., Vol. 145, No. 5, May 1998 The Electrochemical Society, Inc.

I 0" -. I

Fig. 7. Temperature dependence of conductivity of BaCeO3 based materials in (a) 02, (b] 4% H2, and (c) wet Ar.

Shown in Fig. 7b is the temperature dependence of con-ductivity in 4% H2. The conductivities of samples with 5,10, and 20% Y increased with dopant concentration. ForBaCeO3 doped with 20% Y, Nd, and Gd, the conductivityincreased in the order of Nd, Gd, and Y at a given temper-ature. Contrary to expectation, the conductivity of 20%Nd-doped BaCeO3 was not very high. Moreover, 20% Y-doped BaCeO2 had the highest conductivity in this work,—1.54 X 10-2 f'l' cm' at 600°C and —4.16 X 10' 11-' cmat 800°C. Compared to the results of Iwahara et al.2 for20% Y-doped BaCeO2 in H2, the conductivities in this workare lower, partially because measurements in the presentstudy were made in 4% H2 instead of 100% pure hydrogen.

Shown in Fig. 7c is the temperature dependence of con-ductivity in wet Ar. The conductivity of Y-doped BaCeO3increased with dopant concentration. Among the sampleswith 20% dopants, conductivity increased in the order ofNd-, Y-, and Gd-doped BaCeO, in the temperature rangeof 550 to 800°C. The 20% Gd-doped BaCeO3 had the high-est conductivity with the lowest activation energy(—0.31 eV) in wet Ar, 6 whereas the activation energies forother materials were very close to one another, —0.4 eV.

Composition dependence of stoichiometric dopedBaCeO3.—For Y-doped BaCeO3, at a given temperature,conductivity measured in wet Ar (Fig. 7c) and 4% H2(Fig. 8a) which is dominated by ionic conduction, increasedmonotonically with dopant concentration from 5 to 20%. In°2, however, there is a significant contribution from elec-tron hole conduction; conductivity increased with dopantconcentration from 5 to 10%, and then decreased slightlywith dopant concentration up to 20%. The smooth curve inFig. 8b implies a possible maximum between 10 and 20%.The increase in conductivity with dopant concentrationcould be related to an increase in charge carrier concentra-tion through the reaction

CeO2Y202 2Ye + 3 O + Vc°

However, sintered density and defect structure may alsoaffect total conductivities.

Effects of A:B site nonstoichiometry.—As shown inFig. 9, samples with BaO excess had higher conductivitythan the stoichiometric doped samples with 5% Y as thedopant. However, at a higher dopant concentration (10%Y), excess BaO reduced the total conductivity, except at alower temperature (<600°C), in H2. This variation of con-ductivity with BaO excess could be correlated to the den-sity change with BaO excess. At a 5% dopant level,BaO0.90CeO20.025Y2O2 had a greater density (—94 %dense) than BaCe0,,,Y,,,,O,_, (—89 % dense), whereas, atthe 10% level, BaO0.85CeO20.05Y2O2 had a lesser density(—91 %) than BaCe,,Y,102, (—98 %; Fig. 4). Howevei;Shima et al. found that 4% excess BaO reduced the con-ductivity of 15% Gd-doped BaCeO, in wet Ar at lower

temperatures (250 to 400°C) and enhanced it at highertemperatures (500°C), even though the sample with excessBaO had increased density. They attributed the reducedconductivity at low temperature to defect association.24This variation of conductivity with excess BaO may also

0.035

U0.030

0.025UV

0.0200

0.000 0.05 0.1 0.15 0.2 0.25

V Dopant Level, x

Fig. 8. Effects of dopant concentration on conductivity of Y-doped BaCeO> in (a) 4% H2 and (b) 02.

10'

(a) in oxygen

—200d, Liu at al. [610

2< 20951'o 10951'o 5y

to" i.. I.,..0.9 9.95 1 lOS 1.1 1.15 1.2 1.25

1000/T (1/K)

0

C

0

IS'

in 4% hydrogen>6

2<

8 2<

o ><o

2 O%__ 2<tO 0 0 00 0 o

0 00

20% 5. Iwahara at at. [2) 02< 20955o 10955o 20% Ndo s%y

— 20% Cd, LW at sI. [01

0,9 0.95 I 1.05 1.1 1.15 1.2 1.25

1000ff (1/K)

6

gI 0'

0

(c) in wet argon

- 06aV [61

x._. 0.41 aV'0.. 0.43 eV '

0.39 av—w''Ø:...0. ,

—20%Gd,Liuatal.[6[ o..4< I%y10955 0.40 aV

--O--20%Nd-0- 5955

0.9 0.95 1 1.05 1.1 1.15 1.2 1.25

1000ff (I/K)

0.045

0.040 (a) in 4% hydrogen—0-- 600°C-Cioo'c--o---eoo'c

,0

2°

if0

____________ I •

0 0.05 0.1 0.15 0.2 0.25

V Dopent Level, x

0.015

0.010

0.005

0.10

0.08(b) in oxygen

—0—- 800°C

_J 700°C--0--600°C[6]

E

0.06

0.040Ca0

'4/0.02

0

/ .00/a

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP

J. Electrochem. Soc., Vol. 145, No. 5, May 1998 The Electrochemical Society, Inc. 1785

5.9 0.95 1,05 1.1 115 1.2 125

100011 (I/K)

10°

(b) in 4% hydrogen

0• 0• . 00•¶0' 0 •

0 • U0

0010%Y• BaO0.85CeO,0.05Y,O,• BaO'0.90C80,0.025Y,O,o 5%Y

10W'

100011 (1/K)

Fig. 9. Effects of nonstoichiometry on conductivity of BaCeO3 based materials in (a) 02, (b) 4% H2, and (c) wet Ar.

be related to defect structure. At a low dopant concentra-tion (5% Y), incorporation of excess BaO into the per-ovskite matrix induces high vacancy concentration andincreases conductivity, whereas at a high dopant concen-tration (100/. Y), excess BaO tends to reside in grainboundary phases that may inhibit conduction in the tem-perature range studied.

Conductivity of doped SrCeO3—Shown in Fig. 10 are thetotal conductivities of doped SrCeO3 in oxygen and hydro-gen. The conductivities for BaCe0 9Y01O3. under compara-ble conditions are reproduced from Fig. 7 for reference.Obviously, doped SrCeO3 had lower conductivities in 02and H2 thar those of BaCeO3 based materials. As withBaCeO3, conductivity increased as dopant concentrationincreased from 5 to 10% in both 02 and H2. The activationenergy for conduction in H2 was lower than that in 02.Conductivity of SrCe08(Ni01Y01)O3. was surprisingly low(--1o fl- cm'1 at 650°C). The Ni was probably not a gooddopant in terms of conductivity,3132 although Ni-oxide cer-mets were reported to have good catalytic properties inSOFC applications.2'

H 702- conduction—Figure 11 shows the relative ionictransference, numbers of SrCe095Y70502 in this work, aswell as those of the 10% Y- and 10% Gd-doped BaCeO3that were reported by Iwahara et al.7 Compared to theresults of Ba CeO2 based materials, SrCeO3 based materialshave higher proton transference numbers and lower oxy-

E(2

0a,a(2'0C00

gen ion transference numbers. In SrCe092Y0,0503_,, morethan 85% of the ionic conductivity was carried by protonsin the testing temperature range (600-800°C) and oxygenions contributed to only '-15% of the ionic conductivity.Ionic conduction was completely protonic below 600°C.

ConclusionsDopant species and concentration strongly affected the

microstructures and electrical properties of BaCeO3 andSrCeO3. Grain growth and densification of sintered sam-ples were influenced by dopant level and nonstoichiometryon A and B sites. The conductivity of BaCe1_,Y,03 (x =0.05, 0.1, and 0.2) increased in hydrogen and wet Ar withYcontent. Of the doped BaCeO3, BaCe03Y0203 showed thehighest conductivity in hydrogen ('-1.54 >< i0 fi' cm1 at600°C and '-4.16 X l0 fl cm'1 at 800°C). AlthoughSrCeO2 based materials had lower conductivities thanthose of BaCeO3 based materials, they exhibited higherproton and lower oxygen ion transference numbers thanBaCeO3, which may be useful for improving their selectiv-ity for hydrogen separation.

AcknowledgmentsThis work was supported, and this manuscript created,

under Contract W-31-109-Eng-38, of the U.S. Departmentof Energy, Federal Energy Technology Center.

C41.0E 0.822zC2a101CCI—

15a

S

0

0C)

• (a) in oxygen• U 00 • U(1 • U

• 010.' 0

010%Y 0 •• Ba00.85C60,'0.05Y0

• eao'0.900e0,'o.025Y,O,o 5%f

10W'

S

V

C)

E

0

(.2

(c) in wet argon

V

—0— l0%Y 0.40 ect°BaO'0.B5CeO'0.OSYO

•-BaO'O.gOCeO '0.025Y 0,'O'5%V

09 0.95 1 1.05 1 15 1.2 ¶25 5.9 0.95 I 1.05 1,1 1.15 1.2 1.25

100011 (1/K)

1.2

10.'

I02

—.4— BaC4, 9O 1 • nO, ''':U BCO, ,; ,O, , on 4% H,

'— C. s,ce 1 0 in 0 lJ.2,,:- -C.-. : '

rCe, ,Y, 0, , ho 4%H, 'u..- :1-::OrGa 0 on 4% .4

(w+w)/w7

proton, l0%Y-BaCeO3 [7[ proton

proton. 10%Gd.BaCeO (71 .

0.6

0

0.4

A

oxi0e ion, lO%Gd-BaCeO3 [71 .c

oside ion, 10%YBaceO (7] . Jó on

I .

0.2

0

0.9 0.95 1 1.05 1.1 1.15 1.2 1.25

1000/T (1/K)

Fig. 10. Conductivities of doped SrCeO3 and 10% Y-dopedBaCeO3 in °2 and 4% H2.

550 600 650 700 750 800 850

Temperature ('C)

Fig. I I Relative ionic transference numbers under fuel cell con-dition of 5% YSrCeO3 in this work, and in 10% Y-doped BaCeO3,and 10% Gd-doped BaCeO3 by Iwahara et aI.

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP

1786 J. Electrochern. Soc., Vol. 145, No. 5, May 1998 The Electrochemical Society, Inc.

Manuscript submitted November 5, 1997; revised manu-script received January 27, 1998.

Argonne National Laboratory assisted in meeting thepublication costs of this article.

REFERENCES1. K. Kreuer, Chem. Mater., 8, 610 (1996).2. H. Iwahara, Solid State lonics, 77, 289 (1995).3. J. A. Kilner and B. J. Brook, Solid State lonics, 6, 237

(1982).4. N. Bonanos, B. Ellis, K. S. Knight, and M. N. Mah-

mood, Solid State lonics, 35, 179 (1989).5. N. Taniguchi, K. Hatoh, J. Niikura, and T. Gamo, Solid

State lonics, 53-56, 998 (1992).6. M. Liu, H. Hu, and W. Rauch, in Proceedings of the 1st

International Symposium on Ceramic Membranes,H. U. Anderson, A. C. Khandkar, and M. Liu, Edi-tors, PV 95-24, p. 191, The Electrochemical Society,Pennington, NJ (1996).

7. H. Iwahara, T Yajima, and H. Ushida, Solid State Ion-ics, 70/71, 267 (1994).

8. J. F Liu and A. S. Nowick, Solid State lonics, 50, 131(1992).

9. J. F Liu and A. S. Nowick, Mater. Res. Soc. Symp.Proc.: Solid State lonics II, 210, 675 (1991).

10. R. C. T Slade and N. Singh, Solid State lonics, 46, 111(1991).

11. P. A. Stevenson, N. Jiang, H. M. Buchanan, andF E.G.Henn, Solid State lonics, 62, 279 (1993).

12. H. Nagamoto and H. Yamada, in Proceedings of the2nd International Symposium on Ionic and MixedConducting Ceramics, T A. Ramanarayanan, W. L.Worrell, and H. L. Thller, Editors, PV 94-12, p. 39,The Electrochemical Society Proceedings Series,Pennington, NJ (1994).

13. W. Rauch and M. Liu, in Proceedings of the 1st Inter-national Symposium on Ceramic Membranes, H. U.Anderson, A. C. Khandkar, and M. Liu, Editors,

PV 95-24, p. 146, The Electrochemical Society Pro-ceedings Series, Pennington, NJ (1996).

14. H. Iwahara, H. Uchida, K. Ono, and K. Ogaki, J. Elec-trochem., Soc., 135, 529 (1988).

15. H. Iwahara, T. Yajima, T. Hibino, and H. Ushida, .J.Electrochem., Soc., 140, 1687 (1993).

16. H. Iwahara, T Hibino, and T Sunano, J. Appl. Elec-trochem., 26, 829 (1996).

17. U. Reichel, H. R. Arons, and W. Schilling, Solid Statelonics, 86-88, 639 (1996).

18. K. S. Knight and N. Bonanos, Mater. Res. Bull., 30, 347(1995).

19. J. Guan, S. E. Porris, U. Balachandran, and M. Liu,Solid State lonics, 100, 45 (1997).

20. J. Guan, S. E. Dorris, U. Balachandran, and M. Liu,Solid State lonics, Submitted.

21. T. Setoguchi, K. Okamoto, K. Eguchi, and H. Arai, J.Electrochem. Soc., 139, 2875 (1992).

22. K. Kreuer, E. Schbnherr, and J. Maier, in Proceedings ofthe 14th Rim International Symposium on MaterialsScience, p. 193, Rise National Laboratory, Roskildc,Denmark (1993).

23. A. S. Nowick and Y. Pu, Solid State lonics, 77, 137(1995).

24. D. Shima and S. M. Haile, Solid State lonics, 97, 443(1997).

25. I. Kosacki and H. Anderson, Appl. Phys. Lett., 69, 4171(1996).

26. M. Liu and P. Wang, J. Mater Res., 10, 3210 (1995).27. Y. Pu and A. S. Nowick, J. Am. Ceram. Soc., 78, 3033

(1995).28. F Kulcsar, J. Am. Ceram. Soc., 39, 13 (1956).29. 5. Witek and P. M. Smith, J. Am. Ceram. Soc., 67, 372

(1984).30. C. Lu and L. C. Dc Jonghe, J. Am. Ceram. Soc., 77,

2523 (1994).31. V. Agarwal and M. Liu, J. Electrochem. Soc., 144, 1035

(1997).32. F S. Galasso, Structure, Properties, and Preparation of

Perovskite-type Compounds, Pergamon Press, Ox-ford, England (1969).

Determination of the Trap Densily in Amorphous Silicon byQuasi-Static Capacitance-Voltage Measurements

W. R. Fahrnert and S. iofflera

Department of Electrical Engineering, University of Hagen, D-58084 Hagen, Germany

Y. Chan, S. Kwong, and K. Man

City University of Hong Kong, Kowloon, Hong Kong

ABSTRACT

A special metal-oxide-semiconductor structure based on hydrogenated amorphous silicon has been fabricated. Thequasi-static capacitance-voltage (CV) curves of this device are calculated for various trap densities of the amorphous sil-icon. Due to the occurrence of punch-through, Poisson's equation cannot be solved analytically. Thus, a finite elementsapproach has been used to compute the potential distribution and the charge density in the semiconductor. Differentiationof the charge with respect to the applied voltage delivers the low-frequency CV curve. In the last step, this CV curve isIitted to a measured one in order to determine the trap density. We find a trap density of Nbs =9.1016V' cm3 at midgap.

IntroductionAmorphous silicon (a-Si) is a material widely used in thin

film transistors (TFTs) and solar cells. In a large scale man-ufacturing process routine control is desirable. With respectto the above applications, the control should be based onelectrical measurements. The a-Si characterisation, howev-er, involves optical methods to a large extent. This is espe-cially true for a characteristic material parameter, namely,the bulk trap density, Nb. It is the purpose of this work todeliver an electrical measurement method for N.

* Eleetrochemical Society Active Member.Present address: Siemens AG, Burlington, VT, USA.

In this method a specially designed metal-oxide-a-Sistructure is used1 and its CV curve is recorded. CV tech-niques are known from monocrystalline silicon MOS capac-itance measurements where they are among the major diag-nostic tools used to investigate trap properties. There is abig difference, however, between monocrystalline andamorphous metal-oxide-semiconductor (MOS) structures:in the monocrystalline case, the CV characteristics are con-trolled by surface states (the bulk states being negligible),while for amorphous silicon, the CV characteristics arecontrolled by the bulk states (the surface states being neg-ligible). Among the first to report on this observation were

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 128.61.138.76Downloaded on 2013-05-29 to IP