Electrical conductivity of stabilized low and high temperature modifications of tantalum pentoxide (LTa2O5 and HTa2O5) I: Electrical conductivity of LTa2O5 with additions of TiO2, HfO2 and Cr2O3

Journal of the ~ess-~o~~o~

Metals, 98 (1984)

253

253 - 266

ELECTRICAL CONDUCTIVITY OF STABILIZED LOW AND HIGH TEMPERATURE MODIFICATIONS OF TANTALUM PENTOXIDE ( L-Ta205 AND H-Ta,O,) I: ELECTRICAL CONDUCTIVITY OF L-Ta,O, WITH ADDITIONS OF TiO,, Hf02 AND Cr,03 @WIND JOHANNESEN* Institute of &~e~~si~, (Received

August

and PER KOFSTAD U~~vers~t~of Oslo, P.O. Box 1033, 3l~n~er~, Oslo 3 ~~orwa~~

26, 1983)

Summary Foreign oxides have no doping effect on the low temperature modification of TazO, (L-Ta,O,), i.e. the ionic conductivity is not increased or extended as a result of the presence of substitutionally dissolved cations of lower valency. Pure L-Ta,O$ exhibits the highest ionic conductivity, and additions of foreign oxides often result in an overall decrease in the electrical conductivity compared with that of the pure material. In general additions of oxides appear to have no significant effect on the temperature and oxygen pressure dependences of the electrical conductivity. The conductivity measurements are related to earlier structural investigations where it was concluded that the L-Taz05-type superstructure is stabilized by a number of oxides. It is tentatively suggested, on the basis of the present electrical conductivity measurements, that the ionic conductivity is related to the presence of extended defects in the stabilized L-Ta20s -type structure.

1, Introduction The electrical conductivity and ionic transport numbers of the low temperature modification of TazO, (L-Ta,O,) have recently been measured as a function of the partial pressure po, of oxygen at temperatures from 800 to 1270 “C [l]. The conductivity is predominantly ionic at higher values of po, (1 - 10e4 atm), whereas L-Ta,Os is an n-type conductor at low oxygen pressures (10e9 - lo-l8 atm). The defect structure of L-TazOs was interpreted in terms of an oxygen vacancy model where the ionic conductivity is related to the concentration of mobile oxygen vacancies formed as a result of the presence of impurities with lower valencies. *Present address: Laboratory of Silicate Science, Norwegian The University of Trondheim, 7034 Trondheim NTH, Norway. OOZZ-5088/84/$3.00

Institute

0 Elsevier Sequoia/Printed

of Technology,

in The Netherlands

254

The results of the study reported in ref. 1 suggested that the ionic conductivity of L-Ta20s could be increased and extended by the substitutional dissolution of controlled amounts of impurities with lower valencies, and accordingly a programme was initiated to investigate the effect of foreign oxide additions to L-Ta,Oj [ 1, 21. However, it was observed that pure L-Ta20, exhibited the highest ionic conductivity, and additions of oxides of lower valency resulted in an overall decrease in the electrical conductivity. Further, the oxide additions appeared to have no effect on the temperature and oxygen pressure dependences of the electrical conductivity [ 1,2]. Owing to the similarities in their electrical behaviour, the data for L-Ta,OS with foreign oxide additions were interpreted using the same defect model as that proposed for L-Ta20s, and the results of this analysis yielded enthalpy terms which were closely related to those of pure L-Ta20, [ 1,2]. It is of great interest to relate the electrical conductivity data to earlier X-ray and transmission electron microscopy (TEM) studies [3 - 71 as well as to more recent X-ray and TEM studies performed at this university. The conductivities of L-Ta,OS and of L-Ta20s with additions of foreign oxides can be considered in terms of a modulated structure. This means that addition of foreign oxides results in the formation of a series of L-Ta,O,type superstruct~es where the basic difference between the L-Ta,Os structure and the L-Ta,O,-type structure is the length of the b axis. Therefore it is of interest to study the electrical conductivity of L-TazOs with additions of TiOz, HfO, and Cr203. 2. Materials, specimen

prep~ation

and electrical

me~uremen~

Ceramic samples were prepared from Ta,OS and foreign oxide powders. The impurity contents were determined using mass spectrographic analysis. Cold-pressed tablets of mechanically mixed oxide powders were pressure sintered in a graphite die at a low partial pressure of oxygen (about 10Pz4 atm). Detailed descriptions of the results of the mass spectro~aphic analysis and the specimen preparation are given elsewhere [ 21. The electrical conductivity was measured using a standard two-probe a.c. technique with a conductance bridge operating at 1592 Hz. Ionic transport numbers were obtained by measuring the e.m.f. across the specimen while maintaining different oxygen pressures on opposite sides of the specimen. Various partial pressures of oxygen were obtained by using Ar-O2 mixtures for the range 1 - 10m4 atm O2 and CO-CO2 mixtures for the range lop4 - 10F2’ atm OZ. The total pressure in the apparatus was always 1 atm. The partial pressure po, of oxygen was determined directly from the APO, mixtures in the high pressure region and was calculated from the CO-CO2 reaction in the low pressure region. The relative amounts of gases present were controlled by flowmeters. The po, values were cross checked using a calcia-stabilized zirconia cell situated near the specimen. A more detailed description of both the apparatus and the procedure is given elsewhere [ 2).

255

3. TazOs with additions

of TiO,! at temperatures

below 1200 “C

The electrical conductivity and the ionic transport numbers of Ta,O, containing 10 mol.% TiO, ((Taz0,),.y(Ti02)0.,) were measured as a function of the partial pressure of oxygen from 990 to 1300 “C (Figs. 1 and 2). The conductivity and the ionic transport number measurements at temperatures above 1100 “C will be discussed in a subsequent paper.

OXYGEN

PRESSURE, atm

Fig. 1. The electrical conductivity of (TazO&,~(TiO&i as a function of the partial pressure of oxygen at temperatures from 990 to 1300 “C: A, results obtained at 1200 “C before heating to 1300 “C; m, results obtained at 1200 “C after heating to 1300 “C. Measurements above 1100 “C will be discussed in a subsequent paper.

OXVGEN PRESSURE,

Fig. 2. Ionic transport number of oxygen at 1100 “C.

atm of (Ta,05)o,a(TiOz)a.i

as a function

of the partial

pressure

t I, 10-15

I

,,I

I

1O“O

I

I

I

I,

1

10‘5

1

I

OXYGEN PRESSURE, atm

Fig. 3. The electrical conductivity of Ta205 (+, A) and (TazOS)o_9(TiOz)o.l (*, 0, 0) as a function of ~0,. The total conductivity is subdivided into (i) ionic conductivity (the horizontal line at the minimum in the electrical conductivity), (ii) n-type conductivity a,, and (iii) p-type conductivity a,,.

At temperatures between 990 and 1100 “C the conductivity dependence on po, was found to be closely related to that of pure L-TazO, (Fig. 3). Thus at higher values of po, (1 - 10e4 atm) the conductivity exhibits a shallow minimum where the total electrical conductivity is lower than that inof L-Ta205. In the low po, region (10S8 - lo-l4 atm) the conductivity creases with decreasing po,, and at oxygen pressures below lo-” atm it is of approximately proportional to po, -1’4 . Further, the specific conductivity (Ta,Os)o~g(TiOz)o,l approaches that of L-Ta,O, with decreasing po,, and in the lowest pressure range (i.e. po, < lo-lo atm) it exceeds that of pure L-Ta*Os (Fig. 3 ). is essentially a pure electron At low values of po, (Taz0,)o,s(Ti02)o,i conductor, whereas at near-atmospheric pressures of oxygen it exhibits significant ionic conductivity (Fig. 2). The ionic transport number ti is approximately 0.7 in this region, which is lower than the reported value of about 0.9 for pure L-Ta20s [l]. Waring and Roth [ 41 have studied the phase equilibrium relations in the Ta*Os-Ti02 system. The X-ray powder patterns of the system were difficult to interpret. However, the complex nature of the patterns can be considered in relation to the Ta#-WO, system which has been investigated in greater detail [5, 8 - II] and where the examination of the powder patterns has been directly related to the indexing of single-crystal patterns [5]. On this basis Waring and Roth [3,4] concluded that TiOz does not enter into solid

solutions with L-Ta,05 but apparently forms new discrete phases with structures which are directly related to the structure of LLTazOS. The true superlattice multiplicity is that of a large unit cell built up by the ordered repetition of a set of subunits (i.e. UO,-type subcells). A series of structures can then be produced by the ordered stacking of these subunits along the b direction. The basic difference between the original structure and the superstructure is only the length of the true b axis. Thus Waring and Roth [4] report the existence of at least two discrete phases formed by adding 2 mol.% TiO, and 12.5 mol.% TiO, to Ta20,. Further, their X-ray investigation revealed the existence of TiTazO,. Preliminary X-ray powder and TEM investigations at this university [12] on (Ta,?O,),_g(TiO,)O,I quenched from 990 - 1100 “C revealed, in agreement with the work of Waring and Roth [4], a predominance of phase(s) directly related to L-Ta,O,. The TEM study showed the presence of a small amount of a second phase. X-ray examination suggested that no TiO, was present. However, the TEM study indicated small amounts of a rutile structure which may have been unreacted TiOz or TiTa,O,. Phases directly related to L-Ta,O, will for convenience be called R phase(s) in this paper. Consideration of the electrical conductivity data at 990 and 1100 “C on the basis of the structural investigations reported above suggests that the results reflect conductivity in a two-phase or multiphase mixture of R phase(s) and an additional rutile phase. Accordingly, it is difficult to present an unequivocal interpretation of the results. Although it is beyond the scope of this work to give an adequate interpretation of the conductivity data for between 990 and 1100 “C, a number of (Ta~O~)o.P(T~O~)o.~ at temperatures conditions which may affect the conductivity data will be discussed below. As has already been emphasized, the structure of the R phase(s) is directly related to the structure of pure L-Ta20s . The main difference is the number of ordered UO, subunits stacked along the b direction [13]. Provided that the electrical properties of the R phase(s) are effectively the same as those of L-Ta,O, , it can be assumed that the minima of the electrical conductivity of (Taz05)o,s( TiO,),. , , which are almost pressure independent, occur when the conductivity of the R matrix is almost completely ionic. Therefore the conductivity data can be analysed in the same way as for pure L-Ta205 by separating the total electrical conductivity into ionic and electronic contributions ] 11. According to the classical oxygen vacancy model for the R phase(s) the ionic conductivity oion represents the transport of doubly charged oxygen vacancies and is given by uion

=2e[Vo2']uvoz-

= constant

1

X F exp

where LiLH, is the activation energy for the mobility of oxygen vacancies, of Uvo2’ is the mobility of oxygen vacancies and [Vo2+] is the concentration doubly charged oxygen vacancies (KrBger-Vink notation). At high partial

258

pressures of oxygen (1 - lop4 atm) the variation in the electrical conductivity with po, is due mainly to the extrinsic region of the R phase, and [Vo2’] is determined by the presence of impurities of lower valency substitutionally dissolved in the R phase. Further, in accordance with the oxygen vacancy model the n-type conductivity un and the p-type conductivity (Th in the extrinsic region are proportional to P,~-“~ and po,1’4 respectively [ 11. The total electrical conductivity ut of (Ta205),,,s(Ti02)0.i at 990 and 1100 “C was separated into ionic and electronic conductivities by subtracting the ionic conductivity, which is equal to Utti, from the total conductivity. The results of such an analysis performed at 1100 “C are shown in Fig. 3 which also includes a similar analysis of pure L-Ta,O, performed at the same temperature. According to eqn. (1) a plot of lOg(Ui,,T) uersus l/T should which in this case is 164 kJ mol-‘, i.e. yield wH, for (Ta205)0.9(TiO&.i almost equal to the value of 163 kJ mol-’ obtained for L-Ta,Os (Fig. 4). This appears to be reasonable in view of the similarity between the structural features of the R phase and L-Ta20s. If the effect of the possible existence of small amounts of a second phase is neglected, it can be tentatively concluded that the increased specific conductivity of (Ta20s)0.s(Ti02)0.i compared with that of L-Ta20s at low pressures is due to the fact that at these pressures u,, for (Ta20s,)0.s(Ti02)o.l is higher than u, for pure L-Ta20, (Fig. 3). Further, in agreement with the oxygen vacancy model, the electrical conductivity of (Ta205)0.9(Ti02)o.1 at low oxygen partial pressures (PO, < lOPi atm) depends on po, according to u ape -1’4. If the possible existence of a multiphase mixture of small amour& of a second phase embedded in the R matrix is disregarded, it can be argued that the decrease in the ionic transport number ti from about 0.95 at high oxygen partial pressures (i.e. 10e4 < po, < 1 atm) to about 0.7 at low

Fig. 4. U,T

uersus l/T for Taz05 (01, (Taz05)o.~(TiOz)o.l (m) and (Ta20&,.9(HfO2)o.l

(*I.

259

pressures is due to a lower concentration of mobile oxygen vacancies in the extrinsic region of the R phase. Although X-ray studies of (Ta20s)0,9(Ti0,),+1 at this university resulted in powder patterns which could be related only to R phase(s), care should be taken not to assume that there is no second phase present. This is primarily due to the inability of ordinary X-ray powder techniques to detect small amounts of coexisting phases. Thus in view of the results of the TEM studies, it is possible that the (Ta,05)0.9(Ti02)o., sample is a multiphase mixture of R and small amounts of a rutile phase 1141 (see above). Further, the results of lattice parameter calculations suggest that some TiTa04 or TiO, is embedded in the R matrix. (It was difficult to separate these phases because of the similarity in their lattice parameters 1141. However, both TiTaO,, [15] and TiO, are assumed to be n-type semiconductors in the pressure region under investigation.) TiO, is an n-type semiconductor, and its electrical conductivity has been reported by a number of workers to depend on the oxygen partial pressure according to CJ0~pozP1’” where n is between 4 and 6. The defect structure is still a subject of controversy, and we have chosen to use the structure proposed by Kofstad [16] who suggested that oxygen vacancies predominate at high values of po,, whereas Ti3+ and Ti4+ are the dominant defects at low values of po,. Near atmospheric pressures o = ~~,i’~, while at lower values of po, (r 0: po, -1’4 . If the oxide is in the extrinsic region, the conductivity will be proportions to poz-1’4 even at high values of po, [ 1’71. None of the theories proposed in the literature appears to be adequate to describe the ionic conductivity at high values of po, of (TazOS)0.9(TiOz),+1 with a dispersed n-type semiconducting phase embedded in the R matrix. However, consideration of the observed decrease in the ionic transport number of (Taz05)0,9(Ti0,),_1 at high values of po, compared with that of L-Ta*O, suggests that the electron concentration increases as a result of donor action by the dispersed particles. On this basis an increase in the total n-type conductivity of the R matrix would be expected. At low values of po,, i.e. lop9 atm or less, the R phase is an n-type semiconductor 11, Z], and if small amounts of TiO, are present as a second dispersed phase in the R matrix the conductivity data for (Ta~O~)~.~(TiO~)~.~ should be characteristic of a mixture where both phases exhibit essentially pure n-type semiconductor behaviour. Then, according to classical dispersion theory, the ratio of the electrical conductivity u, of a mixture to the electrical conductivity u,, of a continuous matrix at each poz< 10P9 atm can be obtained from the isothermal expression

where (z+ is the conductivity of each volume fraction fi of the dispersed phase i (fi is not a function of po,)_ Equation (2) refers to a random distribution of spherical dispersoids of uniform size. However, a similar expression

260

can be given for a large range of dispersoid sizes [ 18 1. It follows easily that eqn. (2) reduces to the well-known Maxwell expression unl -=-

1-f

00

1++t‘

at the limiting condition of a non-conducting dispersoid, i.e. adJo = 0 for i= 1. For a two-phase mixture only, with the ratio of the conductivity oti of the dispersed phase to the conductivity o. of the R matrix defined as n, eqn. (2) becomes (n + Z)/(n - 1) +

0x33 -= 00

2f

@+W(n-11-f

which can be rearranged IJ, -00

to

3f (n+W@--1)-f

=I+

where the 3f/{(n -t Z)/(n - 1) -f) in eqn, (4) represents the increase (or decrease) in the electrical conductivity of the two-phase mixture compared with that of the matrix phase. Since the X-ray investigation indicated only the presence of the R phase, the amount of TiOz present (based on the TEM study [14]) is assumed to be low. Thus if the amount of TiO, is about 1 mol.%, the volume fraction f of TiO, is about 0.04 (the density of the R phase was set equal to the density of L-Ta*Os, i.e. 8.2 g cme3, and the density of TiOz is 4.26 g cmP3). At low values of p o, (less than lo-* atm) the conductivity of TiOz as measured by Blumenthal et al. [ 191 shows a po, dependence u cxp. -1’4. The specific conductivity vaIues at 1100 “C vary between about 10-i ZG?-’cm-’ at po, = 10-s atm and about 10 a-* cm-’ at po, = lo-l4 atm. If it is assumed that the specific conductivity of the dispersed phase of TiOz is infinitely higher than that of the R matrix, i.e. u~/uo= 00, the limiting vafue of u,/cte (f is const~t) is

n+w

1

.!i =1+

lim

i

f constant

00

_

3f

1-f

(5)

which for f = 0.04 amounts to only o,/uo = 1.125 or log u, - log u. = 0.05 S2-l cm-‘. Accordingly, the increase in the conductivity of (Taz0,)0.9containing TiOz as a second phase compared with that of the R (TiWo.i phase (assuming that the conductivity of TiOz is infinitely higher than that of the R phase) is only 0.05 S’ cm-’ on a logarithmic scale. This is of course an extreme situation (even when Ta,Os starts to dissolve interstiti~y or substitution~ly in TiO,), in view of the basic assumption that as a conse-

261

quence of the structural similarities the specific conductivity of the R phase is close to that of pure L-Ta20s. This result is consistent with the interpretation given above where the existence of small amounts of a second phase has been disregarded. Accordingly, the conductivity data for (Ta20s),.9(Ti02),,, are fundamentally related to those of the R phase. The electrical properties of oxides have quite successfully been described by the classical defect model introduced by Wagner and Schottky [ 201. However, as stressed more recently [21], it is now apparent that the electrical properties of the materials, and particularly the ionic conductivity, should be related to the real crystal structure. Therefore, in addition to studying the dependence of the electrical properties on the defect structure, it is of great interest to relate the results to the real crystal structure of the L-TazO,-type modifications. This is not a simple task since the crystal structure of the L-Ta,O,-type modifications is, despite numerous investigations, still a subject for further discussions. We have chosen to relate the conductivity data to the structural interpretations of the L-Ta20s-type phases proposed by Roth and Stephenson this means that the R phase(s) should be ]5,131. For U'GW~.dTiW~.~ interpreted in terms of an “infinitely adaptive” L-TazOs-type structure. The L-Ta,Os-type structures can generally be visualized as ordered stacking of U03 subunits along the b direction where the basic structure is built up by folding chains of edge-sharing regular pentagonal bipyramids. Owing to anionic packing, distortions occur in the structure. These are minimized by rearranging coordination polyhedra along recurrent planes parallel to (OlO), where each rearrangement eliminates one anion site (i.e. forms one oxygen vacancy) per unit cell [5]. A single distortion or reduction in the coordination number gives rise to a distortion plane (DP plane) running parallel to (010). Thus the oxygen deficiency is related to the rearrangement of coordination polyhedra along DP planes parallel to (OlO), and according to a model proposed by Anderson [6] each threefold group of edge-linked pentagonal bipyramids may result in the formation of one or two oxygen vacancies. For the ionic conductivity of (Ta,0,),.9(TiOz),,l at 990 and 1100 “C it can be assumed that in the high pressure range (1 - 10M4atm 0,) the oxygen vacancy concentration is controlled by vacancies due to DP planes in the R phase. A simple approach to a defect model is that the total oxygen vacancy concentration [ Voltot can be expressed as the sum

Woltot = [VoW’)I + Wol

(6)

where [V,(DP)] refers to the oxygen vacancy concentration due to the concentration of DP planes in R and [V,] is the additional concentration of randomly distributed oxygen vacancies. The formation of a doubly charged oxygen vacancy is expressed as O. = Vo2’ + 2e’ + +02 (in Kroger-Vink

notation)

(7) and the corresponding

mass action expression

is

262

[v,2-]n2po,“2 = KV02.

(8)

of oxygen vawhere Kvoz- is the equilibrium constant for the formation cancies. At this point it should be noted that great care should be taken when relating the classical term “point defect” to a real crystal structure. However, if it is assumed that there is thermodynamic equilibrium between point defects and extended defects such as DP planes, we can write [Vo(DP)I

s

[VoIpoint

(9)

defect

For the ionic conductivity of the R phase at 990 and 1100 ‘C, it is tentatively suggested that the ionic transport is directly related to the migration of randomly distributed DP planes. If it is assumed that the R phase is stabilized by small amounts of substitutionally dissolved impurities (i.e. the impurities stabilize the R phase), the electrons in the extrinsic region may be localized at impurity cations (acting as acceptors) and not trapped at DP planes. We can then write V,“(DP)

+ 2Mfr,” = V,“(DP)

+ 2Mf,’

(10)

where Mf,” denotes a substitutionally dissolved cation which stabilizes the R phase by acting as an acceptor for the DP plane. In the extrinsic region where R shows predominating ionic behaviour we can make the assumption [V,(DP) J = constant

(11)

and for

(12)

WoW’)I > Wol where [V,] is as given in eqn. (6) the electroneutrality 2[Vo2-(DP)] According

= [Mf’] S n

condition

is given by (13)

to eqn. (8) it follows that

~2~Vo2*(DP)]~o~-i’2 = KV02-

(14)

and n = ( [vo~;p,,)

1’2po;l/4

= (7;;;;P

)112p*;l/4

(15)

which gives us qualitatively the same a, dependence as assumed earlier. In view of this result the observed decrease in the ionic conductivity of compared with that of L-Ta205 may reflect a possible (Ta2C5)0.9tTi02)0,1 decrease in the [Vo2’(DP)] as a result of small structural changes in the R phase. (From eqn. (9), [V,*‘(DP)] = K[Vo2*]point defect and from eqn. (15) Kvoz*/K = Kvoz* DP where K,,z- Dp includes the enthalpy and entropy of for,

1

263

mation of doubly charged oxygen vacancies as a result of the formation of a DP plane.) More recently it has been suggested that the L-Ta,O,-type structures can best be described as vernier structures [22]. Thus, it is assumed that, rather than forming discrete phases directly related to L-Ta,O, , additions of foreign oxides yield true solid solutions in L-Ta,Os. However, as has been emphasized elsewhere [2], the effects of solid solutions must be related to the observed electrical properties of the L-TazOs-type structures. In fact the electrical properties may prove enlightening when we consider the complex nature of the stabilizing effects of the L-TazOS-type structures.

4. Ta,Os with additions

of HfOz and Cr,03

In view of our interest in an increased and extended ionic conductivity caused by the substitutional dissolution of oxides of lower valence in Ta205, the effects of additions of HfO? and Cr,O, were investigated [2]. Figures 5 and 6 show the electrical conductivity and the corresponding ionic transport numbers of (Ta205)0.9fHf02)0_1. At temperatures from 1000 to 1200 “C!the data are closely related to those of L-Ta*O, [I], except that the totdl conductivity is lower. X-ray studies of samples quenched from 1000 and 1200 “C to room temperat~e show only the existence of an L-Ta,O,-related phase (R phase). HfO, was not observed. (More advanced structural studies including high resolution TEM will be performed later.) The data can be analysed in terms of an oxygen vacancy model (Fig. 7) and the results of such an analysis give enthalpy terms very similar to those of L-Ta20, [ 11.

OXYGEN PRESSURE, atm

Fig. 5. The electrical conductivity of (Ta~O~)c.~(~fO~)~.~ as a function of the partial pressure of oxygen.

264

10-15

lO-'Q

10-S

OXYGEN PRESSURE,

1

atm

Fig. 6. The ionic transport number of (Ta~Os)~_~~f~~)o,~ sure of oxygen: 0,lOOO %;A, 1100 “C; a,1200 “C.

as a function

of the partial pres-

OXYGEN PRESSURE, atm

Fig. 7. The electrical conductivity of (Ta~~s)~.~~fO~)~.~ as a function of no,. The total conductivity is subdivided into (i) ionic conductivity (the horizontal line at the minimum in the electrical conductivity, (ii) n-type conductivity un and (iii) p-type conductivity crP Fig. 8. The electrical conductivity of (Ta~O~)*.~~Cr s 0 s ) 0.e~ as a function of the partial pressure of oxygen at 1100 - 1300 “C. measurements above 1200 “C will be considered in a later publication.

Thus the band gap is estimated to be 396 kJ mol-’ (4.1 eV) and the activation energy at constant po, in the assumed extrinsic range (a, a -1’4) is 237 kJ mol-i, which is essentially equal to 236 kJ mol-“ for pure po, L-Ta20s [l]. The activation energy AM, = 167 kJ mol-’ (Fig. 4) for the mobility of oxygen vacancies is equal to that of L-Ta205 [I J within the limits of experimental error. It is concluded that HfOz has no doping effect on L-Ta20S, i.e. there is no increase in the ionic conductivity as a result of an increased concentration

265

of mobile oxygen vacancies produced by substitutionally dissolved hafnium cations. As has already been emphasized in a previous paper [23], it is unlikely, in view of the pronounced ionic conductivity of (TazO,),~,(HfO,),.I in the L-Ta,O,-type at high values of po,, that Hf4” dissolves interstitially matrix. When Hf 4+ dissolves interstitially the electron concentration will increase (i.e. the n-type conductivity will increase), whereas [V,*‘] will decrease [ 231. Furthermore, the electrical properties of (TaaOs)0.9(Hf02)0.1 at temperatures below 1200 “C reflect the structural similarity between L-Ta205 and the R phase(s). The electrical properties of (Ta Z0 s )*.~,(~r*O~)~,~~ have been subjected to a less detailed investigation than those of ~Ta~O~}~.~(~fO~)~.~. However, apart from a generally lower conductivity, the variation in the electrical conductivity was similar to that of pure L-Ta205 (Fig. 8). A substantial increase in the electrical conductivity of (Ta 2 0 s )o.97f~2°3)0.03 was ObseI-ved at 1300 “C; this is due to the fact that Cr,Os stabilizes the high temperature form of Ta20s. The electrical properties of (Ta20s)o~97(Cr,03)o~03 will be considered further in a later publication.

Acknowledgments The authors would like to express their thanks to Professor J. Bruce Wagner, Jr., for his continued interest in the problem and for helpful discussions. We also gratefully acknowledge a grant from the Norwegian Oil and Energy Department through the Norwegian Council for General Sciences and Humanities.

References 1 @. Johannesen and P. Kofstad, in J. Nowotny (ed.), Proc. 1st Int. Conf. on Transport in Non-stoichiomefric Compounds, Cracow-Mogilany, Elsevier, Amsterdam, 1982, p. 100. 2 @. Johannesen and P. Kofstad, J. Phys. Chem. Solids, 44 (1984), in the press. 3 R. S. Roth and J. L. Waring, J. Res. Natl. Bur. Stand., Sect. A, 74 (4) (1970) 485. 4 J. L. Waring and R. S. Roth, J. Res. Natl. Bur. Stand., Sect. A, 72 (2) (1968) 175. 5 R. S. Roth and N. C. Stephenson, in L. Eyring and M. O’Keeffe (eds.), The Chemistry of Extended Defects in Non-metallic Sotids, North-Holland, Amsterdam, 1970, p. 167. 6 J. S. Anderson, J. Chem. Sot., Dalton Trans., (1973) 1107. 7 S. Iijima, Crystal structures of L-Nb20s and L-Ta20s, in G. W. Bailey (ed.), Proc. 35th Anna. Electron Microscopy Con/‘., Electron Microscopy Society of America, Boston, MA, 1977, p. 188. 8 N. C. Stephenson and R. S. Roth, Acta Crystallogr., Sect. B, 27 (1971) 1010. 9 N. C. Stephenson and R. S. Roth, Acta Crystutlogr., Sect. B, 27 (1971) 1018. 10 N. C. Stephenson and R. S. Roth, Acta Crystallogr., Sect. B, 27 (1971) 1031. 11 R. S. Roth, J. L. Waring and H. S. Parker, J. Solid State Chem., 2 (1970) 445. 12 S. Furuseth, K. Selte and J. Gj#nnes, personal communication, 1981. 13 N. C. Stephenson and R. S. Roth, Acta CrystaZlogr., Sect. B, 27 (1971 f 1037.

266 14 J. Gjdnnes, personal communication, 1981. 15 F. A. Rozhdestvenskii, G. G. Kasimov and E. I. Krylov, Sou. Phys. - Solid State, 11 (1969) 1366. 16 P. Kofstad, J. Less-Common Vet., 13 (1967) 635. 17 P. Kofstad, Nonstoichiometry, Diffusion and Electrical Conductivity in Binary Oxides, Wiley-Interscience, New York, 1972, p. 98. 18 R. E. Meredith and C. W. Tobias, Adu. Electrochem. Eng., 2 (1962) 26. 19 R. N. Blumenthal, J. Baukus and W. M. Hirthe, J. Electrochem. Sot., 114 (1967) 172. 20 C. Wagner and W. Schottky, 2. Phys. Chem., Abt. B, 11 (1931) 163. 21 R. A. Huggins, in P. Hagenmulier and W. van Go01 (eds.), Solid Electrolytes, Academic Press, New York, 1978, p. 27. 22 R. S. Roth, Prog. Solid State Chem., 13 (1980) 159. 23 0. Johannesen and P. Kofstad, in G. Petot-Ervas (ed.), Solid State Ionics, Proc. 2nd Int. Conf. on Transport in Non-stoichiometric Compounds, Perpignan-Alenya, NorthHolland, Amsterdam, 1983.