characterisation of the oxygen transfer in bimevox membranes under applied current conditions

8
Characterisation of the oxygen transfer in BIMEVOX membranes under applied current conditions R.N. Vannier a, * , R.J. Chater b , S.J. Skinner b , J.A. Kilner b , G. Mairesse a a Laboratoire de Cristallochimie et Physicochimie du Solide, UMR CNRS 8012, ENSCL, Universite ´ des Sciences et Technologies de Lille, B.P. 108, 59652 Villeneuve d’Ascq Cedex, France b Centre for Ion Conducting Membranes (CICM), Department of Materials, Imperial College of Science Technology and Medicine, London SW7 2BP, UK Received 28 January 2003; received in revised form 14 May 2003; accepted 15 May 2003 Abstract Oxygen isotopic exchange has been studied for a number of materials in the BIMEVOX family of compounds. The exchanges were undertaken at 620 jC with gold grid electrodes on the samples and with a constant current flowing through the samples during the exchange anneals. These conditions simulate those used when these materials are employed in oxygen separation devices where substantial oxygen fluxes can be sustained using such simple gold grid electrodes. The results showed that samples exchanged under current flow conditions exhibit substantial oxygen exchange at the cathode, in contrast to samples where no electrical bias is applied. This effect was sustained in regions remote from the sputtered gold electrode. Complementary studies of the samples using X-ray diffraction revealed subtle changes in the diffraction patterns following experiments with current flow. These changes are ascribed to a reduction of V 5+ to V 4+ at the cathode locally transforming the BIMEVOX material into a mixed conducting material, and hence enhancing the oxygen isotopic exchange process. D 2003 Elsevier B.V. All rights reserved. Keywords: BIMEVOX; 18 O/ 16 O isotopic exchange; Oxide ion conduction; Ceramic oxygen generator 1. Introduction The use of electrically driven ceramic oxygen generators could reduce the cost of oxygen produc- tion, at point of use, by a factor of 60% compared to classic methods, e.g. cryogenic extraction, compres- sion and the subsequent handling of gas cylinders. The principle of such a device is very similar to the solid oxide fuel cell (SOFC), in that it relies on the ability of oxide ions to migrate through a ceramic material under an electric field. The first step in this process is when the oxygen molecules from the air are dissociated into oxide ions at the cathode according to the reaction: O 2 þ 4e ! 2O 2 : These oxide ions migrate, under the influence of the electric field, to the anode where they recombine 0167-2738/03/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0167-2738(03)00206-6 * Corresponding author. Tel.: +33-3-20-43-65-83; fax: +33-3- 20-43-68-14. E-mail address: [email protected] (R.N. Vannier). www.elsevier.com/locate/ssi Solid State Ionics 160 (2003) 327 – 334

Upload: independent

Post on 21-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

www.elsevier.com/locate/ssi

Solid State Ionics 160 (2003) 327–334

Characterisation of the oxygen transfer in BIMEVOX membranes

under applied current conditions

R.N. Vanniera,*, R.J. Chaterb, S.J. Skinnerb, J.A. Kilnerb, G. Mairessea

aLaboratoire de Cristallochimie et Physicochimie du Solide, UMR CNRS 8012, ENSCL,

Universite des Sciences et Technologies de Lille, B.P. 108, 59652 Villeneuve d’Ascq Cedex, FrancebCentre for Ion Conducting Membranes (CICM), Department of Materials,

Imperial College of Science Technology and Medicine, London SW7 2BP, UK

Received 28 January 2003; received in revised form 14 May 2003; accepted 15 May 2003

Abstract

Oxygen isotopic exchange has been studied for a number of materials in the BIMEVOX family of compounds. The

exchanges were undertaken at 620 jC with gold grid electrodes on the samples and with a constant current flowing through the

samples during the exchange anneals. These conditions simulate those used when these materials are employed in oxygen

separation devices where substantial oxygen fluxes can be sustained using such simple gold grid electrodes.

The results showed that samples exchanged under current flow conditions exhibit substantial oxygen exchange at the

cathode, in contrast to samples where no electrical bias is applied. This effect was sustained in regions remote from the sputtered

gold electrode. Complementary studies of the samples using X-ray diffraction revealed subtle changes in the diffraction patterns

following experiments with current flow. These changes are ascribed to a reduction of V5 + to V4 + at the cathode locally

transforming the BIMEVOX material into a mixed conducting material, and hence enhancing the oxygen isotopic exchange

process.

D 2003 Elsevier B.V. All rights reserved.

Keywords: BIMEVOX; 18O/16O isotopic exchange; Oxide ion conduction; Ceramic oxygen generator

1. Introduction The principle of such a device is very similar to the

The use of electrically driven ceramic oxygen

generators could reduce the cost of oxygen produc-

tion, at point of use, by a factor of 60% compared to

classic methods, e.g. cryogenic extraction, compres-

sion and the subsequent handling of gas cylinders.

0167-2738/03/$ - see front matter D 2003 Elsevier B.V. All rights reserve

doi:10.1016/S0167-2738(03)00206-6

* Corresponding author. Tel.: +33-3-20-43-65-83; fax: +33-3-

20-43-68-14.

E-mail address: [email protected] (R.N. Vannier).

solid oxide fuel cell (SOFC), in that it relies on the

ability of oxide ions to migrate through a ceramic

material under an electric field. The first step in this

process is when the oxygen molecules from the air are

dissociated into oxide ions at the cathode according to

the reaction:

O2 þ 4e� ! 2O2�:

These oxide ions migrate, under the influence of

the electric field, to the anode where they recombine

d.

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334328

into oxygen molecules according to the reverse reac-

tion. This process allows for the production of con-

trolled amounts of very high purity (>99.99%)

oxygen, which can be delivered under pressure with-

out the use of any mechanical device.

In devices such as ceramic oxygen generators and

solid oxide fuel cells, two steps govern the oxygen

transport: (1) the oxygen exchange at the surface of

the material and (2) the oxygen diffusion through the

material. The limiting step in the whole process is

often the gas–solid transfer at the surface of the

material and, to help this transfer, electrode materials

are usually added at the surface of the membrane. For

high temperature devices the most commonly used

electrolyte for oxygen generator applications is yttria

stabilised zirconia (YSZ), with La0.9Sr0.10MnO3� x

(LSM) most commonly used as the cathode material.

For lower temperatures, a variety of materials have

been proposed. An understanding of the detailed

mechanisms involved in the transfer of oxygen at

such electrodes has not yet been achieved because

of the complex nature of the cathodes and the inter-

play of electrochemical and microstructural factors.

Some recent studies aimed at unravelling this process

are discussed below.

It is often very difficult to characterise the oxygen

transfer process using electrochemical methods alone.

One much less ambiguous technique is the use of18O/16O Isotopic Exchange Depth Profile technique

(IEDP) based on secondary ion mass spectrometry

(SIMS) [1–4]. This technique has recently been used

to visualise active sites for oxygen reduction in a

LSM/YSZ membrane [5–8]. An LSM grid was de-

posited onto a YSZ electrolyte and 18O/16O exchange

was performed under cathodic polarisation. SIMS

analysis of 18O� revealed that the active sites for

oxygen reduction were localised at the O2/YSZ/LSM

triple points and the YSZ/LSM boundary. The exper-

iment was repeated with a gold grid as an electrode

[7,8], which is known to be a poor electrode for the

dissociation of oxygen. In this case, the active sites for

oxygen reduction were only localised at the O2/YSZ/

gold triple point. This difference between the two

electrodes was explained by the low diffusion coeffi-

cient of oxygen in the gold.

In contrast to the findings with YSZ, high oxygen

fluxes have been measured when using BIMEVOX

electrolytes [9] with gold electrodes. These electrolyte

materials are well known for their attractive properties

at moderate temperatures (between 300 and 600 jC)making them suitable for lower temperature devices.

At 300 jC, they exhibit the same conductivity as

stabilised zirconia at 800 jC. They are derived from

the parent compound, Bi4V2O11, by the partial sub-

stitution for vanadium with a metal [10]. BICU-

VOX.10, for instance, is obtained by partial

substitution for vanadium with 10% of copper. Oxy-

gen fluxes were measured at temperatures between

430 and 600 jC on membranes, composed of a

BIMEVOX electrolyte (BICUVOX, BICOVOX,

BIZNVOX) that was simply co-sintered between

two gold grids. Excellent performances with close to

100% efficiency and current densities of up to 1 A

cm� 2 were observed [9]. It must be stressed that no

additional electrode material was added, except for the

gold grid, which was primarily acting as a current

collector. It was then concluded the BIMEVOX elec-

trolyte itself was acting as an electrode material. This

was confirmed by Boukamp who showed evidence

that a BICUVOX membrane was more active as an

electrode than a porous Pt electrode [11,12]. The same

activity had already been demonstrated on erbia

stabilised bismuth oxide by combining 18O Isotope

exchange and dc polarisation [13].

In order to characterise and gain an understanding

of the oxygen transfer mechanism in BIMEVOX

compounds, the IEDP technique was used under

equilibrium and simulated operational conditions.

Experiments performed, at equilibrium, under dry

oxygen atmosphere, confirmed the high oxygen dif-

fusion rates in these materials but revealed surprising-

ly slow surface exchange kinetics [14]. To complete

this study, a second set of experiments was carried out

under simulated working conditions, i.e. an electrical

bias was applied during the isotopic exchange.

2. Experimental

To perform isotopic exchanges under electrical

bias, a special cell was constructed. Three BIMEVOX

compositions were selected: BIBIVOX.02, BICO-

VOX.10 and BICUVOX.10. Cylindrical pellets (10

mm in diameter, 2.5 mm in thickness) with a relative

density higher than 95% were prepared as described

in [14]. One side of each pellet was polished flat using

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334 329

successive diamond spray polishing medium down to

1/4 Am to obtain a mirror like surface. Half of this

surface was then masked and the sample faces coated

with gold by dc sputtering. This resulted in a cylin-

drical sample with one face half-coated in gold as

shown in Fig. 1. A gold grid was then applied to this

sputtered region using gold paste. The other side of

the pellet was painted with gold paste and stuck to a

gold grid, held on a circular alumina disk. Each gold

grid was attached to gold wires for the current

collection. To sinter the gold paste, the whole cell

was annealed for 1 h at 650 jC in air with a heating

and cooling rate of 4 jC/min. It was then introduced

into a silica tube. A schematic of the cell and the 18O

exchange apparatus are given in Fig. 1. A pre-anneal-

ing was performed for two hours at 620 jC under 200

mbar of 16O (zero grade, 99.6%). The membrane was

heated and cooled at a rate of 20 jC/min to avoid any

risk of cracking of the pellet. The cell was then heated

to 620 jC at a rate of 20 jC/min in a 200 mbar18O/16O dry atmosphere. Once the anneal temperature

was steady, a constant current of 80 mA was applied

for two hours. Once this anneal time was completed,

the current was turned off and the sample immediately

quenched. In a similar experiment, a BIBIVOX.02

membrane was also studied with and without the

application of a constant current of 8 mA. A cell

potential of about 2 V was measured for membranes

under 80 mA; it was 0.5 V for the membrane under a

8 mA bias. When considering the high oxide ion

conduction of these materials (0.1–0.2 S/cm at 620

jC), these potentials would lead to rather high polar-

isation voltage but they included all the electrical

Fig. 1. Schematic of the 18O exchang

contacts, which were very poor and it was not

possible from these measurements to deduce an esti-

mation of the polarisation voltage. Moreover, the

anode, made of a thick gold layer, detached easily

after experiment.

After the experiment, the 18O/(18O + 16O) isotopic

fractions at the membrane surfaces were determined

by secondary ion mass spectroscopy (Atomika 6500).

An 8 keV Xe+ beam was used with a beam spot size

of 40 Am. 18O� and 16O� ions from the surface of the

pellet were collected by line scanning the beam along

a line from the gold electrode to the uncovered

ceramic. The 18O isotopic fraction from 200� 200

Am craters, located at given distances from the gold

border, were also measured. After the SIMS analysis,

the depths of the craters and lengths of the line scans

were obtained by optical interference microscopy

(Zygo New View 200). The 18O isotopic fraction in

the gas phase was determined by oxidation of a silicon

wafer at 1050 jC before and after each set of experi-

ments, followed by SIMS analysis of the newly grown

SiO2. This fraction was found to be of 0.600(2) for all

the experiments except those involving BICUVOX,

for which the isotopic fraction in the gas phase was

0.500(3).

Focused Ion Beam (FIB) microscopy images were

obtained by the means of a FIB-SIMS apparatus

(Philips, FEI, Oregon USA). X-ray diffraction (Sie-

mens D5000) was performed to characterise the

surface of the membranes after isotopic exchange.

The gold covered and gold free surfaces of the

samples were analysed separately by covering the

counter part with a thick absorbing gold sheet.

e cell and BIMEVOX sample.

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334330

3. Results and discussion

FIB microscopy images obtained from a BICO-

VOX membrane after polarisation with an 80 mA

current are reported in Fig. 2. A mean grain size of

about 5 Am was revealed for the ceramic. The same

microstructure was observed for all the pellets studied

under the same conditions, including those which

were not polarised. The images also show the border

of the gold sputtered region at the ceramic surface.

The gold sputtering was shown to be inhomogeneous,

consisting of isolated clusters of 1–2 Am in diameter.

No degradation at the interface of the gold coating and

the gold free surface was observed after the experi-

ment.

SIMS analysis of the pellet which has been

annealed with no electrical bias revealed an isotopic

fraction close to the natural isotopic background level

(0.002). In contrast, high isotopic fractions were

observed after polarisation. Fig. 3 shows normalised18O isotopic fractions, obtained in both the line scan

mode, and from selected areas in the depth profile

mode, as a function of the distance from the gold

border on BIBIVOX.02. As an example, after a

constant current of 80 mA for one hour, an isotopic

ratio of 0.20 was observed at 1 mm from the gold

electrode border. The same behaviour was observed

for the BICUVOX.10 and BICOVOX.10 composi-

tions (Fig. 4). However, the 18O isotopic fraction as a

Fig. 2. FIB microscopy images on a BICOV

function of the distance from the gold border de-

creased more rapidly for BICOVOX than the two

other compositions. The measured isotopic fraction

was found to be a function of the applied current

density. When a current of 8 mA was applied to a

separate sample, a normalised isotopic fraction ten

times lower was measured (Fig. 3).

After the exchange experiment, the phase compo-

sition of the surface of the pellets exchanged with a

bias current was checked by X-ray. The gold covered

and gold free surfaces of the samples were analysed

separately by covering the counter part with a thick

absorbing gold sheet. In Fig. 5, the diffractograms

obtained on the BIBIVOX.02 pellet polarised with a

80 mA current are compared to that obtained on the

unpolarised pellet with the same composition. For the

gold free areas, diffraction patterns very close to those

of the original materials were found. No extra phase

was observed in this domain, nevertheless small

modifications were observed. These modifications

are subtle but help to explain the interesting behaviour

of these materials. The following points should be

noted.

(1) A strong preferred orientation of (hkl) planes

with h < k compared to (khl) was obvious and can be

explained by the ferroelectric properties of this com-

position at room temperature. The structure of these

materials is tetragonal from 620 jC to room temper-

ature for BICOVOX.10 and BICUVOX.10. In con-

OX pellet after polarisation at 80 mA.

Fig. 3. Normalised oxygen-18 isotopic fractions obtained in line scan mode and depth profile mode (rectangles indicate the depth profile crater

positions) as a function of the distance from the gold border on BIBIVOX.02 membranes polarised at 8 and 80 mA.

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334 331

trast, BIBIVOX.02 transforms to an orthorhombic

form when cooling the sample to room temperature,

this orthorhombic form is ferroelectric. This leads to a

Fig. 4. Normalised oxygen-18 isotopic fractions obtained in line scan mod

border on BICUVOX.10 and BICOVOX.10 membranes polarised at 80 m

splitting of the (hkl/khl) Bragg peaks into two peaks of

similar intensity. The (0 2 0/2 0 0), (0 2 6/2 0 6), (1 3

3, 3 1 3) doublets indexed in the 5.5, 5.6, 15.3 unit

e and depth profile mode as a function of the distance from the gold

A.

Fig. 5. Diffractograms obtained on the BIBIVOX.02 pellet polarised with an 80 mA current (a)-obtained on the surface located under the gold

electrode, (b)-on the gold free surface, (c)-compared to that obtained on the non polarised pellet with the same composition. *Peaks of extra

phase BiVO4-see text for explanation.

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334332

cell are indicated on Fig. 5. The decrease of the

intensity of the (0 2 0), (0 2 6) and (1 3 3) Bragg

peaks is obvious. The samples were immediately

quenched after the polarisation experiment; however,

a small polarisation remained during the cooling

periods. This leads to a preferred orientation, with

the [1 0 0] direction mainly oriented along the electric

field.

(2) The space between the (hkl/khl) doublets was

smaller for the polarised sample. In contrast, the (1 1

9) and (2 2 6) peaks, which overlapped for the

unpolarised samples, were split after electrical polar-

isation. The same behaviour was observed when these

materials were studied under a reducing atmosphere

Fig. 6. Diffractograms obtained on the BICOVOX.10 pellet polarised with

electrode, (b)-on the gold free surface. *Peaks of extra phase BiVO4-see

and is characteristic of a reduction of small amounts

of V5 + to V4 + in the structure [15]. An intermediate

effect was noticed on the pellet with a constant current

of 8 mA.

Pellets of the same composition were recently

analysed under similar conditions using X-ray diffrac-

tion [16–18]. These experiments were performed on

pellets of 9 mm diameter and 2–2.5 mm thickness

with faces covered with two 50 mesh gold grids, fixed

by a porous layer of BIMEVOX. At 620 jC, for

cathodic polarisation at current densities up to 2 A/

cm2, a reduction of the material was observed. This

was maintained whilst a current was maintained but

when the current was switched off the material

an 80 mA current (a)-obtained on the surface located under the gold

text for explanation.

Fig. 7. Diffractograms obtained on the BICUVOX.10 pellet polarised with an 80 mA current (a)-obtained on the surface located under the gold

electrode, (b)-on the gold free surface. *Peaks of non identified extra phase-see text for explanation.

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334 333

reverted to its initial state. In the present study, this

reduced state was quenched to room temperature,

confirming the transformation of the surface of these

materials over a long distance to allow the oxygen

transfer at least 1 mm from the gold current collector.

In contrast, an extra phase was observed in the

domain located under the gold current collector,

indicating a degradation of the materials in this part

of the cell. This phase was easily identified as BiVO4.

The same degradation was observed under the gold

electrode for BICOVOX.10 under 80 mA (Fig. 6), an

extra phase was also observed for BICUVOX in the

same domain but was not BiVO4 and could not be

identified (Fig. 7). However, in all cases, no extra

phase was observed on the gold free domain.

4. Conclusions

Compared to classical electrolytes such as stabi-

lised zirconia, bismuth-based materials exhibit novel

behaviour. Previous experiments had shown very slow

kinetics for dry oxygen exchange at the surface of

BIMEVOX membranes under equilibrium conditions.

Under a bias the situation changes and in the case of

BIMEVOX materials, the transfer of oxygen is not

limited at the electrolyte/electrode/gas triple point

boundary but occurs directly at the surface of the

BIMEVOX electrolyte. This is explained by a partial

reduction of V5 + into V4 + at the sample surface. The

layered structure of these materials is maintained

under reduction; however, it reversibly transforms to

its original state once the electrical stimulus is re-

moved. Thus it would appear that under polarisation,

the electrolyte locally transforms into a mixed (elec-

tronic/oxide ion) conductor and behaves as an elec-

trode material. This is a surprising and unexpected

result and opens up the exciting possibility of having

very simple structures for cathodes for fuel cells and

oxygen separation membranes. This study also

emphasises the need to characterise materials under

working conditions to fully understand them.

Acknowledgements

R.N.V. is grateful to the CNRS for giving her a

delegation and to the European Community for

funding this project through a Marie-Curie Fellow-

ship. She is also grateful to Franc�oise Ratajczak for

her help in the powder preparation and to Prof. Guy

Nowogrocki for numerous advice.

References

[1] S. Carter, A. Selcuk, R.J. Chater, J. Kaida, J.A. Kilner, B.C.H.

Steele, Solid State Ionics 53–56 (1992) 597.

[2] R.J. Chater, S. Carter, J.A. Kilner, B.C.H. Steele, Solid State

Ionics 53–56 (1992) 859.

[3] P.S. Manning, J.D. Sirman, J.A. Kilner, Solid State Ionics 93

(1997) 125.

R.N. Vannier et al. / Solid State Ionics 160 (2003) 327–334334

[4] R.A. De Souza, J.A. Kilner, Solid State Ionics 106 (1998) 175.

[5] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, T. Kawada, T.

Kato, Solid State Ionics 127 (2000) 55.

[6] T. Kawada, T. Horita, N. Sakai, H. Yokokawa, M. Dokiya, J.

Mizusaki, Solid State Ionics 131 (2000) 199.

[7] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, Electrochemical

Society Proceedings, 2001, pp. 538.

[8] T. Horita, K. Yamaji, N. Sakai, Y. Xiong, H. Yokokawa, T.

Kawada, Solid State Ionics 8 (2002) 108.

[9] J.C. Boivin, C. Pirovano, G. Nowogrocki, G. Mairesse, P.

Labrune, G. Lagrange, Solid State Ionics 113–115 (1998)

639.

[10] F. Abraham, J.C. Boivin, G. Mairesse, G. Nowogrocki, Solid

State Ionics 40/41 (1990) 934.

[11] B.A. Boukamp, Solid State Ionics 136–137 (2000) 75–82.

[12] B.A. Boukamp, ACH-Models in Chemistry 137 (2–3) (2000)

225–242.

[13] B.A. Boukamp, B.A. Van Hassel, I.C. Vinke, K.J De Vries, A.J.

Burggraaf, Electrochimica Acta 38 (14) (1993) 1817–1825.

[14] R.N. Vannier, S.J. Skinner, R.J. Chater, J.A. Kilner, G. Mair-

esse, Solid State Ionics 160 (2003) 85–92.

[15] S. Patoux, R.N. Vannier, G. Mairesse, G. Nowogrocki, J.M.

Tarascon, Chemistry of Materials 13 (2001) 500–507.

[16] C. Pirovano, R.N. Vannier, E. Capoen, G. Nowogrocki, M.

Anne, J.C. Boivin, G. Mairesse, in: T.A. Ramanaryanan, W.L.

Worrell, M. Mogensen (Eds.), Proceedings of the Electro-

chemical Society, Ionic and Mixed Conducting Ceramics,

vol. IV, 2001 Fall, PV 2001-28 San Francisco.

[17] C. Pirovano, R.N. Vannier, E. Capoen, G. Nowogrocki, J.C.

Boivin, G. Mairesse, M. Anne, E. Dooryhee, P. Strobel, Solid

State Ionics 159 (2003) 167–179.

[18] C. Pirovano, R.N. Vannier, G. Nowogrocki, J.C. Boivin, G.

Mairesse, Solid State Ionics 159 (2003) 181–191.