J O U R N A L O F M AT E R I A L S S C I E N C E L E T T E R S 1 7 ( 1 9 9 8 ) 2 0 3 ± 2 0 5

The in¯uence of moisture on the electrical conductivity of áBi4:10 V1:90 O10:90 ceramics C. J. WATSON, D. C. SINCLAIR Chemistry Department, University of Aberdeen, Meston Walk, Aberdeen AB24 3UE, UK

There has been much interest in doped Bi4‡ y V2ÿ yO11ÿ y solid solutions known by the acronym BIMEVOX owing to their high oxide anionic conductivity [1±4]. These materials are Auriviliustype phases where Bi2 O2 sheets alternate with VO3:5 u0:5 perovskite-like layers containing oxygen vacancies which are responsible for the electrical properties. It has been established that the oxygen ion transport number in these materials is close to unity [5]; however, the conductivity is highly anisotropic [6] and very dependent on thermal history [6, 7], especially at temperatures below about 400 8C. We have recently been studying the electrical properties of Bi4:1 V1:9O10:90 ceramics below 400 8C using alternating-current (a.c.) impedance spectroscopy and became concerned that moisture may be a contributing factor to the observed time- and temperature-dependent conductivity behaviour. Since the ionic conduction mechanism in these materials occurs within the oxygen-de®cient perovskite-like layers, and the fact that many oxygen-de®cient perovskites are excellent protonic conductors, e.g., acceptor-doped BaCeO3 [8, 9] and SrZrO3 [10±13], we decided to test the in¯uence of water vapour on the conduction properties of the á polymorph of Bi4:10 V1:90O10:90 below 300 8C. The results demonstrate that the bulk conductivity is sensitive to water vapour pressure, suggesting that these materials may exhibit mixed protonic±ionic conduction. Bi2 O3 (purity, 99.99%) and V2O5 (purity, 99.6%) reagents were dried at 300 8C overnight and stored in a desiccator prior to use. Appropriate quantities of the reagents were weighed to give a Bi-to-V ratio of 4.10 to 1.90 and mixed into a paste with acetone using an agate mortar and pestle. In order to limit volatilization of the reagents, approximately two thirds of the powder was cold pressed into pellets, placed in an Au foil boat and covered with the remaining powder. A heating sequence of 500 8C for 2 h, 650 8C for 2 h, 800 8C for 12 h, regrind and repellet, and then 750 8C for 2 h and 830 8C for 12 h was found to be adequate to obtain equilibrium. Pellets for electron probe micro-analysis (EPMA) and a.c. impedance measurements were cold pressed, covered with powder of similar composition and sintered at 830 8C for a further 12 h. Phase purity was determined by X-ray diffraction (XRD) using a HaÈgg Guinier camera using Cu Ká1 radiation. The ®nal composition of the pellets was determined using a Cameca SX51 electron microp0261-8028 # 1998 Chapman & Hall

robe analyser with an incident beam energy of 20 kV and a current of 50 nA. The sample was polished to less than 1 ìm and carbon coated. The standards used were Bi2 CuO4 for Bi La and V2O5 for V Ka ; oxygen was calculated by stoichiometry. A.c. impedance measurements were conducted using a Hewlett±Packard 4192A impedance analyser over the frequency range 5Hz±5MHz with an applied voltage of 100 mV. Electrodes were fabricated from Pt organopaste; electroded pellets were ®red at 800 8C overnight to decompose the paste and to harden the Pt residue. Pellets were attached to the Pt measuring leads of a controlled-atmosphere conductivity jig and placed in a horizontal tube furnace whose temperature was controlled and measured to within 3 8C. The atmosphere was initially laboratory air and was then changed sequentially to ¯owing N2, dried over silica gel to wet ¯owing N2, bubbled through distilled water and ®nally to laboratory air. The samples prepared were phase pure by XRD and EPMA. The XRD pattern was fully indexed on an orthorhombic cell, of space group Amam with Ê b ˆ 5:593(2) A Ê lattice parameters a ˆ 5:518(2) A, Ê and c ˆ 15:239(5) A. These values are consistent with those obtained for á polymorphs of similar composition, as reported by Lee et al. [2]. EPMA established that there were no signi®cant problems associated with the volatility of the reagents as the analysed pellets had the same stoichiometry as the starting composition, within experimental errors, i.e., Bi4:10(2) V1:90(2)O10:90. It is well documented that the conductivity of BIMEVOX materials can exhibit hysteresis, especially below about 400 8C. For example, on quenching Bi4:10 V1:90O10:90 onto a brass block from 830 8C, the conductivity at about 250 8C was found to decrease from approximately 0.1 to 0.015 ìS cmÿ1 over a period of about 72 h. It should be noted that there is no signi®cant difference in XRD patterns or lattice parameters for samples prepared at 830 8C and rapidly quenched or slowly cooled over several hours. Although this effect is not well understood, it is presumably associated with oxygen reordering within the conduction planes. An in-depth analysis of impedance data associated this phenomenon will be reported elsewhere [14]; the purpose of the present communication is to establish the in¯uence of moisture on the conductivity of á-Bi4:10 V1:90O10:90 ceramics below 300 8C. Typical impedance plane plots for a Bi4:10 203

V1:90O10:90 sample slowly cooled in air from 830 8C and treated in various atmospheres at 275 8C are shown in Fig. 1. The slow cooling treatment included air equilibration at 275 8C for about 24 h before any atmospheric changes. This ensured full equilibration of the sample and therefore no signi®cant time-dependent decrease in conductivity associated with oxygen reordering, as is observed with rapidly quenched samples. The response at 275 8C in laboratory air consists of two overlapping arcs and a low-frequency inclined ``spike'' (Fig. 1a). Capacitance values associated with the arcs were calculated using the relationship ùRC ˆ 1 at the arc maxima (where ù ˆ 2ð f is the angular frequency) and were determined to be about 20 pF and 30 ìF, for the high- and low-frequency arcs, respectively. The ®rst value is a typical bulk value [15]; hence the bulk resistance, Rb , is taken as the intercept of the higher-frequency arc on the real ( Z9) axis as shown approximately in Fig. 1. The bulk conductivity is therefore the reciprocal of Rb . The second capacitance is a typical charge-transfer sample±electrode interfacial value [15]. The associated resistance of the charge-transfer process is small compared with the total resistance, which is given by the low-frequency intercept of the combined arcs. At the lowest frequencies a small additional effect is observed, with an associated capacitance in the microfarad range and is also associated with the sample±electrode interface. This low-frequency spike is attributable to ionic polarization and diffusion-limited phenomena at the electrodes and supports the idea that conduction is mainly by means of (oxygen) ions. Thus, this inclined spike is similar to that expected for a Warburg impedance with an ideal slope of 458. On changing the atmosphere from laboratory air to dry ¯owing N2 for about 24 h, a small decrease in bulk conductivity from 2.62 to 2.22 ìF cmÿ1 was observed. On changing to a moist N2 atmosphere, the bulk conductivity increased substantially, as shown by the decrease in size of the high-frequency

``bulk arc'' in the complex impedance plane plot (Fig. 1b). The capacitance of the high-frequency arc remained unchanged at a value of about 20 pF. The low-frequency data show a more complicated response; the ``spike'' is no longer apparent in the measured frequency range and only a very broad and non-ideal Debye-like response is observed. We have not attempted to model or interpret this lowfrequency response and choose only to extract information associated with the high-frequency arc, i.e., the bulk response. The bulk conductivity in the moist N2 atmosphere was 12.3 ìS cmÿ1 after about 80 h. The variation in bulk conductivity as a function of time in the moist N2 atmosphere is shown in Fig. 2 and demonstrates that full equilibration requires more than about 80 h. Changing the moist N2 atmosphere to laboratory air results in a decrease in the bulk conductivity, as shown by the increase in size of the high-frequency bulk arc in Fig. 1c. Again there was no signi®cant change in the bulk capacitance value of about 20 pF but with increasing time the low-frequency response obtained in dry ¯owing N2 (Fig. 1a) was reestablished. More than about 100 h in laboratory air was required to re-establish the original bulk conductivity value of about 2 ìS cmÿ1 . The a.c. impedance results clearly demonstrate that the bulk conductivity in Bi4:1 V1:9O10:90 increases in moist atmospheres below 300 8C. Although we have no direct evidence for protonic conduction in these materials, similar results have been observed in other defect perovskites [8±13] and in the brown millerite Ba2 In2 O5 [16] where protonic conduction has been established at temperatures around 300± 400 8C. Kruer [17] has recently reviewed the complexity of interactions in proton conduction phenomena in a wide variety of materials including oxides. Although many defect equilibrium mechanisms can be used to hypothesize how protonic defects maybe incorporated into such oxides, e.g., * 2H: ‡ OX (1) H2 O(g) ‡ V:O: ) i O : X :: * (2) H2 O(g) ‡ OO ‡ VO ) 2(OH)O in-depth structural and electromotive force studies will be required before a full explanation can be

(a) 20 kHz 20.1

12 10

(b)

20.1

σ (µS cm21)

Z 0 (MΩ cm)

Rb

100 kHz

Rb (c) 20.1

2

Rb 0.2 Z 9 (MΩ cm)

0.4

Figure 1 Complex impedance plane plots for Bi4:10 V1:90O10:90 at 275 8C: (a) in air; (b) moist N2 for 77 h; (c) after re-equilibration in air for 26 h. Selected frequencies are indicated.

204

6 4

30 kHz

0

8

0

0

20

40 Time (h)

60

80

Figure 2 Variation in bulk conductivity of Bi4:10 V1:90O10:90 at 275 8C in moist N2 as a function of time.

offered for the moisture enhancement of conductivity in BIMEVOX materials.

K L E I T Z , J. C . B O I V I N and G . M A I R E S S E , ibid. 48

(1991) 257.

6. P. K U R E K , J. R . DY G A S and M . W. B R E I T E R ,

J. Electroanal. Chem. 378 (1994) 177.

Acknowledgements We wish to thank Dr Alison Coats for EPMA measurements and the Engineering and Physical Sciences Research Council for ®nancial support of the EPMA facility and the University of Aberdeen for a studentship for C.J.W.

7. J. R . DY G A S , P. K U R E K and M . W. B R E I T E R ,

Electrochimica Acta 40 (1995) 1545.

8. H . I WA H A R A , T. E S A K A , H . U C H I D A and N . M A E D A , Solid State Ionics 3±4 (1981) 359.

9. H . U C H I D A , H . YO S H I K AWA and H . I WA H A R A , ibid.

34 (1989) 103.

10. J. F. L I U and A . S . N OW I C K , ibid. 50 (1992) 131. 11. T. H I B I N O , K . M I Z U TA N I , T. YA J I M A and H . I WA H A R A , ibid. 57 (1992) 303.

12. K . C . L I A N G , I . Y. L E E and A . S . N OW I C K , Mater. Res.

References

1. F. A B R A H A M , M . F. D E B R E U I L L E - G R E S S E , G .

M A I R E S S E and G . N OWO G R O C K I , Solid State Ionics 28±30 (1988) 529. 2. C . K . L E E , D. C . S I N C L A I R and A . R . W E S T, ibid. 62 (1993) 193. 3. O . J O U B E RT, A . J O UA N N E AU X , M . G A N N E , R . N . VA N N I E R and G . M A I R E S S E , ibid. 73 (1994) 309. 4. E . P E R N O T, M . A N N E , M . B A C M A N N , P. S T R O B E L , J. F O U L E T I E R , R . N . VA N N I E R , G . M A I R E S S E , F. A B R A H A M and G . N OWO G R O C K I , ibid. 70±71 (1994)

259.

5. T. I H A R A D A , A . H A M M O U C H E , J. F O U L E T I E R , M .

Soc. Symp. Proc. 293 (1993) 355.

13. Y. D U and A . S . N OW I C K , J. Amer. Ceram. Soc. 78 (1995)

3033.

14. C . J. WAT S O N , MSc thesis, University of Aberdeen (1997). 15. J. T. S . I RV I N E , D. C . S I N C L A I R and A . R . W E S T ,

Adv. Mater. 2 (1990) 132.

16. G . B . Z H A N G and D. M . S M Y T H , Solid State Ionics 82

(1995) 153.

17. K . D. K R U E R , Chem. Mater. 8 (1996) 610.

Received 16 June and accepted 11 September 1997

205