J. Electroanal. Chem, 286 (1990) 245-251 Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
Na+ exchange at the Nasicon/
E. Siebert, A. Caneiro *, P. Fabry and M. Levy Laboratoire d’lonique et d’Electrochimie D’Heres Cedex (France)
des solides de Grenoble, ENSEEG,
BP75, 38402 Saint Martin
(Received 18 January 1990; in revised form 27 March 1990)
Polycrystalhne Nasicon, Na,Zr,Si,PO,,, has been proposed as a sensitive membrane in Na+ ion selective electrodes [1,2] since this material is a fast Na+ ionic conductor. In addition its rigid structure was expected to provide good selectivity
Since the work of Hong and Goodenough [4,5], the mechanism of Na+ transport within Nasicon type solid electrolytes has been investigated extensively. Nasicon-type materials can be formulated as a solid solution of the general formula Their structure consists of ZrO, octahedra linked by comers Na l+XZr,Si,P,_,O,,. to (P,Si)O, tetrahedra. This rigid oxide network forms a three-dimensional lattice of intersecting conduction channels in which are localized the Na+ ions. There are essentially two types of Na sites, referred to as Na(1) and Na(2) (rhombohedric structure). The conduction path is thought to involve both Na sites. For migration along this Na(l)-Na(2) path, Na+ ions must pass through “bottlenecks” formed by triangles of oxygen. The size of these bottlenecks is a determining factor in the conductivity of Nasicon. It can, for example, be adjusted as a function of the composition. For instance, as x increases, that is as Si4+ replaces P’+, the tetrahedra tilt so as to open the bottlenecks. This process reaches a maximum at about x = 2. Thus, the best conductivity is obtained for the composition Na,Zr,Si,PO,,. A detailed description of the conduction mechanism as a function of structural modifications has appeared [6,7]. Up to now the ionic transfer at the membrane aqueous solution interface has not been studied. Though some compositions have been reported to react with water , Nasicon can be regarded as being inert to water insertion [9,10] so that the Na+
* Permanent address: Centro Atom& 0022-0728/90/$03.50
de Bariloche, 8400-SC de Bariloche, Argentina.
0 1990 - Elsevier Sequoia S.A.
exchange involves the movement of Na+ from its hydration shell in the aqueous phase into the Na+ site in the membrane phase. This can be depicted by: (Na+) + (Na+)
where (Na+) represents Na+ ions surrounded by water molecules, and (Na+) represents Na+ ions surrounded by oxygen atoms in Nasicon. Moreover, since the structure is specific to Na+ ions (size of the Na+ cavities and size of the bottlenecks well suited to Na+ ions), this material is expected to show a higher chemical affinity for Na+ ions than for other cations, so that equilibrium (1) should be little disturbed by the presence of a foreign cation. Recently, we measured the potentiometric selectivity coefficient for Na+ over K+, Li+, Ca2+ and H+ and found that Nasicon performed better than existing commercial electrodes based on glass membranes (PNAV Tacussel) . In this note, we present results on the exchange of the Na+ ions between a liquid electrolyte and the Nasicon solid phase. Four electrode impedance measurements were performed to measure the rate of exchange of Na+. However, the collection of reliable kinetic data is frequently hampered by the high resistivity of the membrane. The purpose of this note is to show that impedance measurements can be applied to study the ion exchange reaction at the interface between Nasicon and aqueous solution. As a first step in this study, some preliminary data on the Na+ exchange across this interface are presented. EXPERIMENTAL
Highly reactive Nasicon powder of composition x = 2 was prepared by the sol-gel technique as described by Colomban [ll]. Dense Nasicon ceramics (density greater than 97% of the theoretical density) were obtained by isostatic pressing and sintering at high temperature. Before they were tested, the solid electrolytes were polished with abrasive paper (ESCIL, grain 1200). The samples were characterized by scanning electron microscopy and X-ray diffraction. Their conductivity was measured by impedance spectroscopy in the temperature range 20-200” C. Measurements were carried out with a Hewlett-Packard impedance meter (HP 4192A) in the frequency range 13 MHz-5 Hz. Platinum or graphite were used as the contacting electrodes. The samples were painted on both sides with a suspension of powdered graphite (DAG 154) or platinum (Degussa 308A). In the present work, the membranes were sintered at loo0 o C or 1074 o C. For each temperature, several specimens were prepared, one for the conductivity measurements and the rest for the impedance measurements in contact with ionic solutions. Microscopic investigation and X-ray diffraction analysis showed that all the compounds were single-phase. At such temperatures, rhombohedral symmetry is obtained. The composition of the samples was checked and found to be x = 2. The conductivity of the samples, reported in Table 1, was about 1O-3 Q-’ cm-’ at room temperature, in agreement with literature data . The membranes were between
241 TABLE 1 Activation energy and conductivity of Nasiwn sintering temperature T/O C
103 u(25 o C)/ !X’ cm-’
0.2 and 0.3 cm thick and 1 to 1.1 cm in diameter so that the resistance of the membrane did not exceed 400 Q. The aqueous electrolytes were obtained by dissolution of NaCl in a buffer solution at pH = 8 (Tris T8 THC, Tacussel). This solution is commonly used to test the ion selective electrodes. Some solutions were prepared from distilled water. Four electrode impedance measurements were performed with the cell depicted in fig. 1. Reference electrodes on both sides of the membrane were chloridized silver wires, selected so as to avoid any pollution of the working solution, such as the K+ ions from the calomel saturated electrode which can interfere. The two current-carrying electrodes were made of platinum. The active membrane area in contact with the solution was about 0.5 cm’. The impedance data were obtained using a Solartron 1250 Frequency Response Analyser coupled to a 1186 Electrochemical Interface (Schlumberger). The amplitude of the applied sinusoidal voltage was 20 mV rms. Impedance measurements were made on symmetrically bathed membranes after a controlled soaking time (1 h). The measurements were performed at room temperature (21 + 2” C).
Fig. 1. Diagram of the four-electrode cell.
Figure 2 shows the impedance diagram of a membrane which is in contact with a 0.1 M NaCl solution. The impedance spectrum shows only one semi-circle, with low frequency behavior which is difficult to resolve. The high frequency intercept of the arc with the real axis corresponds to the resistance of the membrane and the solution (about 890 52). This type of diagram was obtained for all the membranes. The bulk semi-circle at the highest frequencies, corresponding to Nasicon, could not be observed on the diagram since the relaxation frequency (5: 1 MHz) of the process is outside the frequency range of the instrument. The impedance spectra can therefore be related to the interface which behaves like a resistance (R) in parallel with a capacitance (C). For a two-sided membrane, there are two identical RC circuits in series. Generally, the semi-circles are reasonably well defined and the semi-circle parameters can therefore be determined accurately. The resistance R was calculated from the radius of the semi-circle and the capacitance R from the relationship RCw,,= 1, where w, is the frequency at the maximum height of the semi-circle. It must be emphasized that the semi-circular center lies below the real axis (depression angle = 20 ” ) and the impedance would therefore be more adequately described by a Cole-Cole type formula and can be explained in terms of the presence of distributed elements (constant phase element impedance) . The equivalent circuit of the electrode should then consist of a frequency-dependent capacitance in parallel with a frequency-independent resistance. The capacitance determined by (Ro,,)-'is therefore of limited significance and must be viewed only as an order of magnitude approximation of the relevant capacitive effects. The impedance data for the various membranes in contact with NaCl solution are summarized in Table 2. As expected for an interface phenomenon, the impedance is highly dependent on the Na+ concentration in the aqueous contacting
Fig. 2. Impedance
diagram for Nasicon
in contact with a 0.1 M NaCl solution made from Tris.
249 TABLE 2 Impedance data and kinetics parameters for Nasicon in contact with NaCl aqueous solutions. No.: sample number; T: sintering temperature of Nasicon membrane; R: resistance for 0.2 M NaCl; C: capacitance (mean value); 8: depression angle (mean value); j,,: exchange current density for 0.1 M NaCl No.
membrane sintering temperature/ o C
2 3 4
1074 1000 1000
solution. R decreases systematically relationship:
95 250 80 150
0.4 2 1.2 2.6
15 20 16 17
1.06 0.39 1.3 0.67
as the concentration
mA cm-* 0.45 0.46 0.51
of Na+ is increased. The
R a [Na+]-”
held closely for Na+ concentrations ranging from 10e3 to 1 M (see Fig. 3). The values of n for all the membranes used are reported in Table 2. The n coefficient was near to 0.5 for all the samples. For a single specimen, the values of R were reasonably reproducible, but they were highly dependent on the “history” of the sample. For example, the resistance always increases after the membrane has been in contact with a KC1 solution (see curve b, Fig. 3) whereas the n coefficient decreases to 0.4. In this respect, impedance measurements are likely to provide information on possible interference mechanisms. For several samples prepared with the same experimental condition, the values of R were less reproducible,
[Na+] I M 1
Fig. 3. Variation of the resistance R asa function of the Na+ concentration in solution (Tris). (a) “Fresh membrane”; (b) membrane which had previously been in contact with a 0.1 M KC1 solution.
probably because of modification of the active surface area. The depression angle and the capacitance were independent of the Na+ concentration. The mean values are reported in Table 2. The capacitance was found to be 1 PF cm-‘, and which is similar to that in distilled water or Tris solution. The semi-circle may be related to the ionic transfer across the interface, and thus corresponds to the charge transfer resistance (R = R,,) parallel to the double layer capacitance (C = C,,). The exchange current density, j,, was calculated from the charge transfer resistance at the equilibrium potential according to the classical relationship: jo =
RT/( F&, >
where the symbols have their usual meanings. j, values as high as 1 mA cme2 were obtained (cf. Table 2), suggesting that the exchange of Na+ between the aqueous solution and Nasicon is rapid. These exchange current densities are higher than those obtained under the same experimental conditions for a PVC membrane containing valinomycin . For comparison, here are some values of exchange current density from the literature: K+ exchange at the aqueous solution/Valinomycin-PVC membrane interface: j, = 0.17 mA cm-*
for 1O-2 M KC1 
j, > 0.03 mA cm-2
for 10-l M KC1 
ion transfer between water and nitrobenzene: 0.26
ceramics and molten mixtures, CO(NH,),
The n = 0.5 result could indicate that the situation is analogous to that occurring at the interface between an aqueous solution and a PVC membrane containing valinomycin  that was treated as though it were formed between two immiscible electrolytes 1191.In such a type of interface, the relationship: R,, a [ (Na+)] -1’2 is expected since the exchange current density is expressed by: j, = Fk[(Na+)] a[(Na+)]l-a
Here k is the rate constant for the reaction, a is the transfer coefficient, [(Na’)] is the concentration in the aqueous phase, and [(Na+)] is the concentration in the solid phase. The electrical double layer at the interface can be regarded in terms of series combination of the double layer in contact with the bathing electrolyte and in the membrane. Interestingly earlier theoretical studies on the description of the contact between an ionic solid and a liquid electrolyte indicate that ionic crystal can develop
a diffuse space charge [20-221. The experimental results, i.e. no variation of the capacitance as a function of Na+ concentration in the contacting solution and no influence of foreign ions in the aqueous solution (Tris instead of distilled water), indicate that C should be related to the double layer in the solid. The capacitance (= 1 PF cm-*) is indeed lower than that expected for a NaCl aqueous solution (typically 10 PF cm-* ). It can be compared with that obtained for the double layer capacitance of an Au/Na-~-alumina interface . It is likely that the double layer in Nasicon arises from finite ion size effects (inner layer capacitance) rather than a diffuse space charge. The results reported herein show that impedance measurements can be applied to study the ion exchange reaction at the interface between Nasicon and aqueous solutions and that this method is certainly well-suited to obtaining information on possible interference mechanisms. There is evidence for the rapid exchange of Na+ between Nasicon and aqueous solutions. It illustrates that Nasicon is a promising material for use as a membrane in aqueous solutions, such as an ion-selective electrode for potential measurement in biomedical applications. REFERENCES 1 2 3 4 5 6 7 8 9 10
11 12 13 14 15 16 17 18 19 20 21 22 23 24
J. Engell and S. Mortensen, Radiometer Int. Patent w o 84/01829 (1984). P. Fabry, J.P. Gros, J.F. Million-Brodaz and M. Kleitz, Sensors Actuators, 15 (1988) 33. M. Kleitz, J.F. Million-Brodaz and P. Fabry, Solid State Ionics, 22 (1987) 295. H.Y.P. Hong, Mat. Res. Bull., 11 (1976) 173. J.B. Goodenough, H.Y.P. Hong and J.A. Kafalas, Mat. Res. Bull., 11 (1976) 203. G. Collin and J.P. Boilot, in A.L. Laskar and S. Chandra (Eds.), Superionic Solids and Solid Electrolytes: Recent Trends, Academic Press, New York 1989, p. 227. Nasicon: Solid State Ionics, 9-10 (1983) 795 ff.; 18-19 (1986) 935 ff.; 28-30 (1988) 403 ff. A. Ahmad, T.A. Wheat, A.K. Kuriakose, J.D. Canodey and A.G. Mcdonald, Solid State Ionics, 24 (1987) 89. J.J. Aubom and D.W. Johnson Jr., Solid State Ionics, 5 (1981) 315. G.R. Miller, B.J. MC Entire, T.D. Hadnagy, J.R. Rasmussen, R.S. Gordon and A.V. Virkar in P. Vashista, J.N. Mundy and G.K. Shenoy (Eds.), Fast Ion Transport in Solids, Elsevier, Amsterdam, 1979, p. 83. H. Perthuis and P. Colomban, Mat. res. Bull., 19 (1984) 621. P. Colomban, Solid State Ionics, 21 (1986) 97. J.R. Macdonald, Impedance Spectroscopy. Emphasizing Solid Materials and Systems, Wiley-Interscience, New York, 1987. R.D. Armstrong, J.C. Lockhart and M. Todd, Electrochim. Acta, 31 (1986) 591. R.D. Armstrong, Electrochim. Acta, 32 (1987) 1549. S.L. Xie and K. Camman, J. Electroanal Chem., 229 (1987) 249. R.P. Buck and W.E. Bronner, J. Electroanal. Chem., 197 (1986) 179. E.A. Ukshe, solid State Ionics, 36 (1989) 143. J. Koryta, Electrochim. Acta, 33 (1988) 189. T.B. Grimley and N.F. Mott, Discuss. Faraday Sot., 1 (1947) 3. D.O. Raleigh in M. Kleitz and J. Dupuy (Eds.), Electrode Processes in Solid State Ionics, Reidel, Dordrecht, 1975, p. 119. J.R. MacDonald in ref. 23, p. 149. M.J. Spamaay, The Electrical Double Layer, Pergamon Press, Oxford, 1972. R.D. Armstrong and W.I. Archer, J. Electroanal. Chem., 87 (1978) 221.