Solid State Ionics 122 (1999) 127–136

Conductivity measurements on nasicon and nasicon-modified

materials
a, b b
O. Bohnke *, S. Ronchetti , D. Mazza
a
Laboratoire des Fluorures, UPRESA 6010 CNRS, Universite´ du Maine, Avenue O. Messiaen, 72085 Le Mans Cedex 9, France
b
Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, Corso Duca Abruzzi, 24 10129 Torino, Italy
Received 4 November 1998; accepted 18 February 1999

Abstract

Conductivity measurements have been performed on the solid solution series, Na 3 Zr 22x / 4 Si 22x P11x O 12 , for 0 , x , 2.
These materials are Zr-deficient nasicon. A monoclinic to rhombohedral phase transition occurs around x 5 0.333. The
variation of the ionic conductivity as a function of the composition x is investigated in the temperature range 150–600 K. It
is shown that the bulk ionic conductivity decreases slightly from 6 3 10 24 S cm 21 for the Zr non-deficient nasicon (x 5 0) to
6 3 10 25 S cm 21 for the Zr highest-deficient material (x 5 1.667). The presence of the above mentioned structural change
does not affect the ionic conductivity. The mobility of the Na 1 ions in these materials seems to be mostly influenced by the
size of the bottlenecks through which the ions have to pass. The variations of both the ionic conductivity and the activation
energy of the ionic motion process in the bulk of the material can be explained by structural considerations.  1999
Elsevier Science B.V. All rights reserved.

Keywords: Nasicon; Zr-deficient nasicon; DC conductivity; Xerogels

1. Introduction crystallize in the nasicon structure. These structures

The discovery of nasicon structure,
Na 11x Zr 2 Si x P32x O 12 , represented an important im-
provement in solid electrolyte developments because
their three-dimensional framework structure presents
a high ionic conductivity comparable to that of
two-dimensional networks, such as b-alumina. One
of the first investigations was performed by Hong
[1], who demonstrated that solids in the above
compositional range (0 , x , 3) heated to 12008C
*Corresponding author. Tel.: 133-2-4383-3354; fax: 133-2-
4383-3506.
E-mail address: odile.bohnke@univ-lemans.fr (O. Bohnke)
have rhombohedral symmetry except in the interval
1.8 , x , 2.2 where a small distortion to monoclinic
symmetry occurs. Even though the nasicon type
structures are stable in a large triangular zone of the
quaternary phase diagram Na 2 O–ZrO 2 –P2 O 5 –SiO 2 ,
delimited by the compositions NaZr 2 P3 O 12 ,
Na 4 Zr 2 Si 3 O 12 , and Na 4 Zr 1.25 P2 O 12 [2]. Most of the
studies [3–7] have been concentrated on the above
reported formula, all around the composition x 5 2.
In fact the highest conductivity values were observed
in the range 1.8 , x , 2.4. A commercial product
(NKG-Co. Ltd Japan) having nominal composition
Na 3 Zr 2 Si 2 PO 12 , was already produced and investi-
gated as a carbon dioxide sensor [8–10].

0167-2738 / 99 / $ – see front matter  1999 Elsevier Science B.V. All rights reserved.
PII: S0167-2738( 99 )00062-4
128 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136
Only few studies were dedicated to other com-
positions: Kohler et al. [2] and later Rudolf et al.
[11,12] succeeded in preparing Zr-deficient mono-
clinic solids. Von Alpen et al. [13] proposed a
different general formula, Na 11x Zr 22x / 3 Si x P32x O 12 ,
in order to obtain a zirconia free material. However,
even though zirconia-free, the Von Alpen materials
consist of two phases, when generally the Hong
compositions contain two (when zirconia-free) or
even three different phases. In fact, many authors
demonstrated that a glassy phase (which is basically
a sodium silicophosphate with some dissolved zir-
conia) forms in many compositions, mostly in the
silica rich ones. This phase has a high ionic con-
ductivity, but unfortunately, it rapidly degrades upon
contact with humidity. In addition, the appearance of
new phases after high temperature treatments, princi-
pally liquid phases, was observed and correlated to
an increase in powder sinterability.
In a previous paper [14] we investigated further
the ternary field extending among NaZr 2 P3 O 12 ,
Na 4 Zr 2 Si 3 O 12 and Na 4 Zr 1.25 P2 O 12 . In particular we
selected those compositions which have a constant
number of three Na 1 ions per structural building unit
(SBU) in order to maintain the constant amount of
ionic carriers already present in Na 3 Zr 2 Si 2 PO 12 . The
aim of this work is to assess the ionic conductivity of
the above compositions expressed by the relation,
Na 3 Zr 22x / 4 Si 22x P11x O 12 , and to correlate it to the x
value.
2. Experimental
2.1. Sample preparation
Owing to the difficulty in obtaining high-purity
nasicon ceramics through conventional solid state
reactions, we have elaborated a sol-gel preparative
technique in order to obtain the preliminary forma-
tion of an amorphous solid (xerogel). The higher
reactivity of this precursor leads more easily to
monophasic compositions, without impurities at the
grain boundary, and at the same time originates very
small grains which are sintered more easily.
We have obtained these xerogels through the
following preparative steps.
2.1.1. Solution A
A measured volume of tetraethylorthosilicate
(TEOS) is added to an equal volume of ethanol and
half a volume of distilled water. The pH of the
aqueous alcoholic solution is adjusted to 0.5–1 with
a few drops of 1 M HNO 3 . In this way TEOS is
firstly partially hydrolyzed to silanol monomers.
2.1.2. Solution B
Stoichiometric quantities of NaNO 3 (BDH Chemi-
cals, purity . 99.5%) and of zirconium dinitrate
oxide ZrO(NO 3 ) 2 .xH 2 O (Ventron, ZrO 2 5 45.68%)
are dissolved in a little amount of a mixture of
water / nitric acid (1:1) and added to the solution A
while stirring carefully until a clear phase is ob-
tained.
2.1.3. Solution C
A suitable portion of (NH 4 ) 2 HPO 4 (Merck,
purity . 99.5%) is dissolved in a mixture of distilled
water and nitric acid (1:1) and then quickly added to
the (A 1 B) solution prepared above, always stirring
vigorously. In this way a dense gel forms, which is
firstly dried at 1058C (under stirring) and then
cautiously and gradually heated in air up to 4008C,
thus removing excess water, alcohol and gaseous
products formed by decomposition of nitrates, so
obtaining a white amorphous crispy powder.
The following samples x 5 0, 0.333, 0.667, 1.000,
1.333 and 1.667 of the solid solution series
Na 3 Zr 22x / 4 Si 22x P11x O 12 have been prepared. Each
amorphous sample was submitted to simultaneous
DTA and TG analysis in order to determine the
crystallization temperature of the xerogels from the
position of the relative exotherms. These tempera-
tures were comprised for all the samples between
650 and 7008C. At these low temperatures the XRD
patterns showed very broad peaks of the nasicon
structure. The samples, after crystallization at the
above temperatures, were ground in an agate mortar,
then axially cold pressed at 0.1 GPa and finally
sintered directly at 12008C for 30 min. Some sam-
ples were also sintered by slow heating (58C min 21 )
from room temperature up to 12008C. Sintering at
12008C confirmed the poor sinterability of these
powders. Even though starting from high density
grain (55–65% of theoretical value) only some
 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136
compositions were able to overcome 90% of the
theoretical value [15]. The density of the sample has
been measured picnometrically on the samples used
for conductivity. For x 5 0 it was less than 70%.
However for x 5 0.333 the density of the sample was
76% and for x 5 0.667 and x 5 1.333 it was greater
than 90%. SEM observations made on
Na 3 Zr 1.833 Si 1.333 P1.667 O 12 sintered at 12008C showed
the presence of a glassy phase, supporting the
assumption already suggested in the literature that
nasicon undergoes a liquid phase sintering [5,16,17].
The appearance of this liquid phase may induce the
segregation of zirconia crystals by a dissolution–
precipitation process, particularly if the samples are
sintered by slow heating (58C min 21 ) [15].
2.2. XRD characterization
The X-ray powder patterns were recorded at room
temperature with a Philips PW1710 diffractometer
equipped with a graphite monochromator on the
diffracted beam and a sample spinner. The relevant
powder diffraction data together with the unit cell
dimensions refined therefrom for the monophasic
studied samples are reported in Ref. [14]. The
composition x 5 0 corresponds to the classical
nasicon composition Na 3 Zr 2 Si 2 PO 12 . The unit cell
parameters and the space groups of the above
monophasic compositions are reported in Table 1.
All the samples have been sintered directly at
12008C.
2.3. Conductivity measurements
Conductivity measurements have been carried out
with a Frequency Response Analyzer 1276 (Solar-
Table 1
Unit cell parameters and space group of Na 3 Zr 22x / 4 Si 22x P11x O 12 as a function of composition x
Sample ˚
a 0 (A)
A Na 3 Zr 2 Si 2 PO 12 15.586
B Na 3 Zr 1.917 Si 1.667 P1.333 O 12 15.591
C Na 3 Zr 1.833 Si 1.333 P1.667 O 12 9.000
D Na 3 Zr 1.750 SiP2 O 12 8.966
E Na 3 Zr 1.667 Si 0.667 P2.333 O 12 8.902
F Na 3 Zr 1.583 Si 0.333 P2.667 O 12 8.878
129
tron) from 150 K to 600 K in the frequency range 4
MHz–1 Hz. An applied voltage of 100 or 200 mV
(r.m.s.) was used (the response of the electrochemi-
cal cell remains linear up to 500 mV). Either
platinum paste or sputtered gold were used as
ionically blocking electrodes. The dimensions of the
samples were typically 1 cm in diameter and 1 to 2
mm in thickness (the form factor was then around
0.2 cm 21 ). For measurements above room tempera-
ture the sample was fixed up in a two-electrode cell
in an oven under pure dehydrated N 2 atmosphere
(less than 5 ppm of O 2 ). For measurements below
room temperature the sample was fixed up in a
two-electrode cell in a cryogenerator under vacuum.
The samples were allowed to equilibrate for 1 to 2 h
before measurement until obtaining reproducible
impedance spectra. The grain boundary conductivity
was better observed in samples with sputtered gold
electrodes than in samples with platinum paste
electrodes due to a diffusion process which occurs at
low frequencies in the platinum paste, this diffusion
partly masking the grain boundary circle. On the
other hand, with gold electrode the grain boundary
circle and the electrode polarization are well sepa-
rated as shown in Fig. 3a. To model bulk and grain
boundary conduction we used a series circuit made
of two ZARC-Cole elements (ZC). Each element is a
distributed element and is a combination of a con-
stant phase element (CPE) and a resistor R [18]. The
CPE element takes into account the electrical in-
homogeneity of the sample. The impedance of such a
ZC element is given by the relationship:
ZZC 5 R /(1 1 RC(iv )f ) (1)
and is represented by a complex plane impedance
˚
b 0 (A) ˚
c 0 (A) b Space
group
9.029 9.205 123.78 C2 / c [1]
9.022 9.197 123.78 C2 / c [14]
– 22.964 – ¯ [14]
R3c
– 22.912 – ¯ [14]
R3c
– 22.805 – ¯ [14]
R3c
– 22.789 – ¯ [14]
R3c
130 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136

plot which forms an arc of a circle (hence ZARC)

with the center displaced from the real axis. The
dimension of C is equivalent to a capacitance and
0 # f # 1. The smaller f is, the more the center is
displaced. For the bulk and the grain boundary
conduction the electrical elements are determined by
fitting the experimental values (Nyquist impedance
and Nyquist modulus plots) with such a model.

3. Results and discussion

The nasicon framework structure is characterized

by structural building units (SBU) made of two
overlaying ZrO 6 octahedra linked by (Si,P)O 4 tetra-
hedra. Into these tetrahedra silicon and phosphorus
are randomly distributed on sites with fourfold
coordination. These SBU are connected by (Si,P)O 4
tetrahedra to form infinite ribbons parallel to the
c-axis. The Na 1 interstitial sites, located in the
interSBU positions, are connected by a more or less
wide channel where Na 1 ions can diffuse. These
ions have to pass through bottlenecks formed by
ZrO 6 octahedra and (Si,P)O 4 tetrahedra, as shown in
Fig. 1. Since these bottlenecks exceed 4.8 A, ˚ the
1
Na ions can be mobile in this material. In the solid
solution series Na 3 Zr 22x / 4 Si 22x P11x O 12 three so- Fig. 1. Part of the structure, showing the columns of NaO 6 , ZrO 6
dium ions balance the negative charge of the SBU. and (Si,P)O 4 polyhedra. Large dark circles represent Na(1)
In the ternary diagram delimited by the compositions positions, large open square Na(2) position, arrow represents the
conduction path of Na 1 ions.
NaZr 2 P3 O 12 –Na 4 Zr 2 Si 3 PO 12 –Na 4 Zr 1.25 P3 O 12 the
studied samples are located on the line extending
from Na 3 Zr 2 Si 2 PO 12 , and Na 3 Zr 1.5 P3 O 12 as shown
41
in Fig. 2. In these materials Si are partly (or
completely) replaced by P 51 in the tetrahedra. This
substitution leads to the removal of Zr 41 in some of
the octahedra to compensate the gain of the positive
charge in the SBU. The Zr non-deficient Nasicon
(x50) structure is monoclinic at room temperature
[1] and presents three different Na 1 sites not fully
occupied. According to Mazza et al. [14], a mono-
clinic to rhombohedral phase transition occurs in the
Zr-deficient Nasicon for x lying between 0.333 and
0.667. In the rhombohedral phase only two different
Na 1 sites exist.
The Nyquist plots of the impedance obtained on
Na 3 Zr 22x / 4 Si 22x P11x O 12 , with sputtered gold elec-
trodes, display two different shapes depending on the
composition x. For 0#x#1 the impedance diagrams Fig. 2. Composition line of the examined samples.
 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136 131

show clearly two semicircles, as illustrated in Fig.
3a, for x50.667. The radius of these circles in-
creases when temperature decreases. At frequency
below 50 Hz, the straight line observed can be
ascribed to the electrode polarization, i.e. the Na 1
blocking response at the surface of the gold elec-
trode. For x$1.333 (Fig. 4a, for x51.333) the
impedance diagrams present only one well shaped
semicircle, even at low temperature, and a straight
line at low frequency. Figs. 3b and 4b show the
Nyquist representation of the modulus (M 5 jv C0 Z,
with C0 5 ´0 s /l). For all the samples the modulus
diagrams display two semicircles and then two
relaxation processes. This means that in all materials
bulk and grain boundary conductivities are present.
The high frequency semicircle can be attributed to
the bulk conductivity and the low frequency one to
the grain boundary conductivity. The impedance
diagrams for high x values display only one well
shaped semicircle because the capacitance of the

Fig. 3. Nyquist plots of the impedance diagrams (a) and modulus diagrams (b) obtained in the frequency range 1 MHz–1 Hz on
Na 3 Zr 22x / 4 Si 22x P11x O 12 (x50.667) at 248 K (upper diagrams) and 225 K (lower diagrams).
132 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136

Fig. 4. (a) Nyquist plots of the impedance diagrams (a) and modulus diagrams (b) obtained in the frequency range 1 MHz–1 Hz on
Na 3 Zr 22x / 4 Si 22x P11x O 12 (x51.333) at 244 K (upper diagrams) and 226 K (lower diagrams).

bulk and the grain boundary are of the same order of
magnitude in these materials, may be due to the heat
treatment conditions and / or to the composition. Figs.
3a and 4a also show that the centers of the ex-
perimental semicircles are displaced below the real
axis. This feature is generally observed in most of
the ceramic materials. It indicates the presence of
distributed elements in the material–electrode system
and explains the choice of the model we used (with
CPE elements) to estimate the bulk and grain bound-
ary conductivities. The parameter f is then related to
the width of the relaxation time distribution.
In order to quantitatively estimate the bulk and
grain boundary conductivities of these samples we
used the equivalent circuit shown in Fig. 5 with two
ZC elements. The resistances represent the ionic
 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136 133

conductive paths in the bulk (R b ) and in the grain

boundary (R gb ), the CPEs represent the space charge
polarizations of the bulk (CPE b ), the grain
boundaries (CPE gb ) and at the interface electrode–
electrolyte at low frequency (CPE p ). For all the
compositions these parameters have been determined
as a function of temperature by fitting the experimen-
tal data with the above described electrical model.
From the values of R b and R gb the bulk and grain
boundary conductivities have been calculated.
As an example, Fig. 6 shows the conductivity
variations obtained as a function of inverse of
temperature for the composition x50.667. The grain
boundary conductivity is roughly one order of mag-
nitude smaller than the bulk conductivity. The
capacitance associated with the high frequency arc is
found to be around 10 211 F and the one associated to
the lower frequency arc lies between 10 29 and 10 28
F. The high frequency arc of Fig. 3a can then be
identified as a bulk property from the fact that the
capacitance is of the order of pF (Cb ¯3 pF cm 21 ,
the form factor of the sample is 0.32 cm 21 ) and the
lower frequency arc to the grain boundary property

Fig. 4. (b)

Fig. 6. Bulk (a) and grain boundary conductivities (b) as a

Fig. 5. The equivalent circuit used to describe the behavior shown function of inverse of temperature obtained on
in Figs. 3 and 4. Na 3 Zr 22x / 4 Si 22x P11x O 12 (x50.333).
134 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136

since the capacitance is much greater. The third s0 Ea

capacitance represented by CPE p is found to be of s 5 ] exp 2 ] (2)
T kT
the order of mF and can be associated to the
polarization of the electrode. where s is the conductivity, s0 the preexponential
The analysis of the impedance data for compounds factor, T the temperature and k the Boltzmann
with x.1.333 (Figs. 4 and 5) leads to the same constant. Fig. 8 shows the variations of the room
behavior for conductivity but the capacitances of
both bulk and grain boundary are of the same order
of magnitude (¯10 29 F) explaining the presence of
only one well shaped semicircle in the Nyquist
impedance plots but the presence of two semicircles
in the Nyquist modulus plots.
Fig. 7 presents the bulk conductivity of all the
studied compounds as a function of inverse of
temperature. All the samples have a conductivity of
the same order of magnitude. The conductivity
follows an Arrhenius law with different activation
energy. From the curves of Fig. 7, the room tempera-
ture conductivity and the activation energy (Ea ) can
be obtained according to the well-known Arrhenius
law:

Fig. 8. Room temperature conductivity (a), activation energy (b)

Fig. 7. Bulk conductivities of the solid solution series for the solid solution series Na 3 Zr 22x / 4 Si 22x P11x O 12 (0#x,2) as
Na 3 Zr 22x / 4 Si 22x P11x O 12 (0#x,2) as a function of the inverse a function of the composition parameter x, for the Na 1 hopping
temperature. motion in the bulk (n) and grain boundaries (s).
 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136
temperature conductivity (a) and the activation
energy (b) for both the bulk and grain boundary
conductivities as a function of composition x. The
room temperature conductivity decreases slightly
with composition and remains between 10 23 and
10 24 S cm 21 , the activation energy increases from
0.33 to 0.42 eV at x.1. For x,1 both the con-
ductivity and the activation energy remain constant
to the value of the classical nasicon Na 3 Zr 2 Si 2 PO 12
(i.e. 6310 24 S cm 21 at 258C and 0.33 eV, respec-
tively [17]).
The change of activation energy observed at x51
cannot be ascribed to the monoclinic to rhombohedr-
al structural change since this structural change does
not occur for this composition. However the dimen-
sions of the cell seem to play an important role in the
mobility of the Na 1 ions as shown in Fig. 9. The
decrease of the a and c parameters and consequently
Fig. 9. Variation of the cell parameters a and c (curves a and b,
respectively) and of the cell volume (curve c) of the solid solution
series Na 3 Zr 22x / 4 Si 22x P11x O 12 as a function of the composition
parameter x.
135
of the bottleneck size of the ion pathways decreases
the ion mobility and may increase the activation
energy of the motion process.
As x increases, phosphate tetrahedra replace sil-
icate tetrahedra leading to a tilting in such a way to
decrease the c-axis and to close the bottlenecks. This
process becomes important around x51 as shown in
Fig. 9b. This process can increase the barrier to ion
motion, as shown in Fig. 8b, and decrease the
conductivity. Moreover the substitution of Si 41 ions
by P 51 ions leads to the removal of (x / 4)Zr 41 in the
octahedra to compensate the gain of the positive
charges in the SBU. The presence of Zr-free octa-
hedra can alter the distribution of Na 1 in the
different available sites. If some of the Na 1 ions can
be located in the Zr-free octahedra they will become
immobile. Consequently the conductivity may also
decrease because of a decrease of the number of
mobile ions. More studies are necessary to clarify
these points and to locate the Na 1 ions in the
structure as the octahedra are more and more Zr-free.
4. Conclusion
The conductivity measurements performed on the
solid solution series Na 3 Zr 22x / 4 Si 22x P11x O 12 show
that the mobility of the Na 1 ions in these materials is
mostly influenced by the size of the bottlenecks
through which the ions have to pass. The presence of
a structural change from monoclinic to rhombohedral
symmetry, occurring around x50.333, does not
affect the ionic conductivity. A shrinkage of the
rhombohedral unit cell dimensions around x51
could be further evidence of the structural rearrange-
ment which leads to the contemporary conductivity
decrease. Moreover the formation of empty octahed-
ra occurring when Si 41 is replaced by P 51 may also
explain the decrease of conductivity. Some Na 1 may
fill these empty octahedra and become less mobile
leading to a decrease in the number of species
participating in the conduction process. Indeed
radioluminescence experiments performed on com-
positions Na 12x Zr 2 Si x P32x O 12 (x50, 2, 3) confirmed
that Frenkel pairs, originated by the displacement of
sodium ions into lacunar sites, are the dominant
defects in this type of structure [19]. The exact
location of the Na 1 ions would be of great interest to
136 O. Bohnke et al. / Solid State Ionics 122 (1999) 127 – 136

understand the conductivity behavior in these materi- [9] Y. Sadaoka, Y. Sakai, M. Matsumoto, T. Manabe, J. Mater.
Sci. 28 (1993) 5783–5792.
als.
[10] S. Nakayama, Y. Sadoaka, J. Mater. Chem. 4 (5) (1994)
663–668.
[11] P.R. Rudolf, M.A. Subramanian, A. Clearfield, J.D. Jorgen-
References sen, Mater. Res. Bull. 20 (1985) 643–651.
[12] P.R. Rudolf, A. Clearfield, J.D. Jorgensen, J. Solid State
Chem. 72 (1988) 100–112.
[1] H.Y.-P. Hong, Mater. Res. Bull. 11 (1976) 173–182.
[13] U. von Alpen, M.F. Bell, H.H. Hofer, Solid State Ionics 3 / 4
[2] H. Kohler, H. Schulz, O. Melnikov, Mater. Res. Bull. 18
(1981) 215–218.
(1983) 1143–1152.
´ G. Desplanches, D. Le Potier, Mater. [14] M. Lucco-Borlera, D. Mazza, L. Montanaro, A. Negro, S.
[3] J.P. Boilot, J.P. Salanie,
Ronchetti, Powder Diffraction 12 (3) (1997) 171–174.
Res. Bull. 14 (1979) 1469–1477.
[15] A. Negro, M. Lucco-Borlera, D. Mazza, L. Montanaro, E.
[4] Ph. Colomban, in: P. Barret, L.-C. Dufour (Eds.), Reactivity
Traversa, in: The Third East Asian Conference On Chemical
of Solids — Proceedings 10th International Symposium On
Sensors, Seoul (Korea), 1997.
the Reactivity of Solids, Materials Science Monographs, Vol.
[16] N. Gasmi, N. Gharbi, H. Zarrhouk, P. Barboux, R. Morineau,
28, Elsevier, 1985, pp. 487–492, Part A.
J. Livage, J. Sol-gel Sci. Tech. 4 (1995) 231.
[5] A.K. Kuriakose, T.A. Wheat, A. Ahmad, J. Dirocco, J. Am.
[17] K.D. Kreuer, H. Kohler, J. Maier, in: High Conductivity
Ceram. Soc. 67 (3) (1984) 179–183.
Solid Ionic Conductors, World Scientific Publishing, 1989,
[6] H. Perthuis, Ph. Colomban, Mater. Res. Bull. 19 (1984)
pp. 242–279.
621–631.
[18] J.R. MacDonald, Solid State Ionics 15 (1985) 159–161.
[7] S. Yao, Y. Shimizu, N. Miura, N. Yamazoe, Jpn. J. Appl.
[19] P. Brovetto, V. Maxia, M. Salis, M. Lucco-Borlera, D. Mazza,
Phys. 31 (1992) L197–L199.
On the lattice defects in nasicon compounds, Il Nuovo
[8] N. Miura, S. Yao, Y. Shimizu, N. Yamazoe, Sensors Ac-
Cimento (in press).
tuators B9 (1992) 165–170.