M. T. EL-GEMAL et al. : Ionic Conductivity of Ag714P04 Solid Electrolyte 499

phys. stat. sol. (a) 57, 499 (1980)

Subject classification: 14; 22.8.1

Solid State Phgsics Laboratories, A1 Faleh University, Tripoli1.)

Ionic Conductivity of Ag&P04 Solid Electrolyte BY M. T. EL-GEMAL, M. SALEEM, and M. N. AVASTHI

The conductivity of Agl14P04 material in the form of pellets prepared a t 4000 kp/cme pressure is measured in the temperature range 4 to 79 "C. At room temperature (25 "C) the conductivity u is 0.015 Q-l cm-l. The activation energy is found t o be 0.81 eV through the Ig (d') versus ( 103/T) plot. The behaviour of conductivity with temperature can be expressed by the equation Ig (ST) = = -0.21 (103/kT) + 4.19. The contact resistance is taken into consideration for determining conductivity and activation energy throughout the temperature range 4 to 79 O C . Indications are found showing that this material has also average structure, i.e., the number of available lattice sites exceeds by far that of cations. A method of pressing the electrodes, made of silver powder and powdered Ag,I,PO,, is given, which seems to drastically reduce the contact resistance.

Es wird die Leitfahigkeit von Ag, J4P04. das bei 4000 kp/cm2 als Pellets hergestellt wurde, im Tem- peraturbereirh von 4 bis 79 "C gemessen. Bei Zimrnertemperatur (25 "C) betragt die Leitfiihigkeit u = 0,015 cm-l. Die Aktivierungsenergie wird zu 0,21 eV aus der 18 (uT)-(lO3/T)-Kurve bestimmt. Das Verhalten der Leitfahigkeit mit der Temperatur la5t sich durch die Gleichung Ig (0) = -0,21 (103/kT) + 4,19 darstellen. Zur Bestimmung der Leitfahigkeit und der Aktivie- rungsenergie wird im gesamten Temperaturbereich zwischen 4 und 79 "C der Kontaktwiderstand beriicksichtigt. Anzeichen deuten darauf hin, da5 das Material cine Mittelstruktur hat, d. h., die Zahl der vorhandenen Gitterpliitze iibersteigt bei weitem die Zahl der Rationen. Es wird e k e Methode zum Aufpressen der Elektroden, die aus Silberpulver und gepulvertem Ag,J4P0, bestehen, beschrieben, die den Kontaktwiderstmd drastisch zu reduzieren scheint.

1. Introduction Solid state cells are known to present a number of convincing advantages over tradi- tional ones, widely used a t present. This advantage increases with the increase of the conductivity of the electrolyte; efforts were therefore made to approach the conduct- ivity of electrolytes in solutions. With this end in view, quite a few solid electrolytes were developed - their number, a t present, exceeds fifty, which have conductivities higher than 10-4 cm-l a t room temperature. The discovery of highly conducting compounds with a general formula AaMI, (where M stands for I<, Rb, or NH,) 11, 21 led to intensive work on them [3 to 10,381. These compounds are well known to be unsta- ble in moisture and iodine atmosphere. Takahashi [ l l ] and Raleigh [20] have reviewed compounds discovered u p to 1972. I n view of the technical applicability of such mate- rials [12 to 151, a good deal of research work has been done to produce new high-conduct- ivity ionic materials which may be more stable while in contact with moisture and iodine. Recently Takahashi et al. [16] established a highly conducting compound, while studying the AgI-Ag,P04 system. The conducting compound Ag,I,PO,resulting from the system has been found to be more stable while in contact with iodine. They studied the variation of its conductivity with temperature for the determination of activation energy in the range 20 to 79 "C, transport number, and cell performance using this new compound Ag7T4P04 as electrolyte. The study confirms that this electro- lyte is a purely cationic (Agf) conductor. Though Ag71,P04 has a lower 0-value than Ag4RbI,, the cell perforniance using the former as electrolyte is better than the latter,

l) P. 0. Box 13619, Tripoli, Libya.

500 M. T. EL-GEMAL, M. SALEEM, and M. N. AVASTHI

as far as the internal resist,ance is concerned. The internal resistance of the Ag,I,PO, cell gradually increases for a month and then shows a constant value, while the Ag,RbI, cell internal resistance increases 100 times during the same period [l6].

Since (i) the conductivity of these solids depends on the method of preparation of the material [17] and (ii) their resistance is comparable with the contact resistance (electrode-electrolyte contact), different research workers have reported different con- ductivities; e.g., theconductivity ofAg,RbI, hasbeenreported to beaslowas0.12 Q-l X x cm-l and as high as 0.30 0-l cm-1 a t room temperature (see Table 1).

T a b l e 1 The specific conductivity D of Ag,RbI, a t room temperature reported by different authors

a temp. reference (a-1 cm-l) ("C)

0.12 20 Bradley and Greene (1967) c51

0.26 - Oxley and Humphrey (1968) ~ 7 1

0.30 30 Owens and Argue( 1970) [41

0.21 20 Owens and Argue (1967) P I

0.27 27 Raleigh (1970, 1972) [19, 201

0.25 25 Scrosati e t al. (1971) r211

0.17 25 Takahashi and Yamamoto (1970) [18]

I n view of the above, it was thought worthwhile (i) to investigate different super- ionic conductors preparing them under identical conditions, (ii) to devise a method of reducing the contact resistance to minimum and determine it (contact resistance) a t every temperature within the temperature range of interst, and (iii) to use the above t o estimate the room temperature conductivity free of contact resistance and also (iv) to determine the activation energy. The present paper on the conductivity of the Ag,I,PO, solid electrolyte forms a part of the above programme.

2. Experimental

2.1 Muterial. preparation

Silver phosphate (Ag,PO,) was obtained from Eastman Kodak Co., Rochester (N.Y.), silver iodide (99.9% pure) from Riedel Co., FRG, and silver powder (200 mesh) from Hopkins & Williams, Ltd., England. The compound Ag,I,PO, was prepared by mixjng stoichiometric amounts of reactant,s (4: 1 mole ratio of silver iodide and silver phos- phate). The mixture was ground and mixed and was then vacuum-sealed in a pyrex glass bulb. The mixture was finally heated at 400 "C for 30 h in a furnace to ensure complete mixing and compound formation. The melt, was then cooled in open atmos- phere to room temperature (25 "C). The material was ground for 2 h to fine powder. The electrode material was prepared by mixing intimately the above fine- grain material with 200 mesh silver powder in the ratio of 3 : 1 by weight, using a stain- less steel knife.

2.2 Preparaiion of tablets (pellets)

It has been pointed out by a number of workers [4, 19 to 21,28, 39,401 that the elec- trode-electrolyte contact resistance problem is the main difficulty in making accurate electrical conductivity measurements. I n general, low but irreproducible conductivity values are observed (see Table 1). Out of all the electrodes used by various workers it

I ~ ~ n i c Conductivity of Ag7t,P0, Solid Electrolyte 50 I

seems that the pressed electrodes made of a mixture of electrolyte and 200 mesh silver powder are most widely used. These electrodes themselves show quite high contact resistance, but much less than for other electrodes. It was therefore thought worthwhile to prepare such electrodes in different ways. The method which resulted in the least contact resistance is described below.

First the weighed amount of the electrode material was dropped in the steel die. It was then lightly, though repeatedly, pressed to ensure the flatness of the surface of the electrode material. Later, again the weighed amount of the material Ag,I,PO, (finely powdered) was gently dropped in the die; it in its own turn was lightly pressed by the steel die cylinder to make the surface of the material also flat and smooth. Lastly, the weighed amount of the contact material was gently dropped over the flat and smooth surface. This three-layered stack of powders was then gently pressed with increasing pressure up to 4000 kp/cm2, using a hand press (Perkin Elmer). It was observed that this method of pressing the electrodes onto the electrolyte material under investigation showed a very small contact resistance - always less than 0.2 R, generally in the region of 0.1 to 0.2 R (see Fig. 2). The tablets were prepared in a number of other ways, but all of them resulted in electrode-electrolyte contact resistance from 0.5 to a few ohms; thus the above method proved to be the best. It seems to the authors that in this way the silver powder comes in good contact with the material under investi- gation.

2.3 Sample holder

The typical cell designed to hold pellets during the conductivity measurements is shown in Fig. 1. The figure is self-explanat,ory. Here, the contact material of the pellet was put in intimate contact with thin silver foils (each 0.01 cm thick) which were spot- welded to thick silver wire leads leading to the measuring instruments. This was done to avoid contact potential.

2.4 Conductivity measurements

Conductivity measurements were made using a 4800 A vector impedance-meter (Hew- lett Packard) a t 1 kHz. Later the measurements were also made by measuring the current and potential difference (ac) across the specimen, using digital multimeters P M 2421 (Philips). The ac power a t 1 kHz was taken from a function generator (3311 A,

Hewlett Packard). Throughout ali the experiments, all the instruments were fed by the stabilized power supply PE 1Gll (220 V/2 kVA, Philips). Both types of measurements gave the same results. The tempera- tures of the top and the bottom of the pellets were measured by two separate copper-constantan thermocouples. These were constructed from the

Fig. 1. Cell for the measurement of electrical conductivity 0.

A asbestos sheet, E silver electrodes, HIHl ambient heater, H2H2 upper heater, R.1 metnl block, P circular metal plate, R metal rod, T thermocouples, RG rod with springs

502 &I. T. EL-GEMAL, M. SALEEM, and M. N. AVASTHI

wires of respective materials of 0.005 in. diameter. The junctions were electri- cally fused. The thermo-e.m.f.'s were measured by digital multimeters PM 2421 (Philips). Thus a temperature difference of 0.25 K between the two faces of the pellets could be detected. Observations were taken after the thermocouples showed constant temperature of the pellets for a t least 20 min each time and the difference between the temperatures of the two faces was not more than 0.25 I<. Conductivity measurements were made on four different tablets of varying thickness (0.156,0.210,0.268,0.322 cm). The results obtained in every case (tablet/pellet), within experimental error, were the same. The graph lg (0) against (103/T) shown in Fig. 4 depicts the mean measurements on all the four tab1et.s.

3. Results and Discussion 3.1 Contact resistance

The determination of electrical resistance a t various temperatures of the system Ag/4AgI-Ag3P0,/Ag is shown in Fig. 2, where the electrical resistance of this system as a whole is plotted against the varying thicknesses of the pellets - all were of the same cross-sectional area (3.14 cmz) and were prepared a t the same pressure (4000 kp/ cmz). The electrode-electrolyte interfacial resistance, usually termed as contact resi- stance, which in the cells has solid state configuration, has been evaluated through extrapolation to zero thickness. The intercept on the resistance axis yielded contact resistance [4]. It was estimated a t 10, 20, 30,40, 50, 60, and 70 "C. Another graph was plotted between the estimatedresistance from Fig. 2 and the temperature. This is shown in Fig. 3. It was found that with the method used for the preparation of the electrodes, the contact resistance ranged between 0.2 and 0.1 SZ. The latter graph (Fig. 3) was used to correct the resistance measured a t different temperatures between 4 and 79 "C. To the authors' knowledge, this is the first time the contact resistance has been taken into account for the accurate measurement of resistance, throughout the temperature range of interest, and thus for the estimation of 0 and the activation energy of any solid electrolyte. After applying these corrections (even though they are very small) lg (aT) as a function of (103/T) is shown in Fig. 4.

3.2 Electrical conductivity

Defects are known to be thermally generated in conventional ionic solids, e.g. alkali halides. Lidiard [22] has shown that within a certain temperature range, the conduc-

1 "c

thickness fcml -

t i

I

Fig. 2 Fig. 3

Fig. 2. Resistance a t different temperature vs. thickness of the pellet of Ag,I,PO,. 0 T = 10, o 20, + 30, A 40, ~ 5 0 , x 60, A 70 "C Fig. 3. Plot of contact resist anc evs. temperature

Ionic Conductivity of Ag,I,PO, Solid Electrolyte 503

Fig. 4. Logarithm of specific conductivity multiplied by the absolute temperature vs. inverse absolute temperature of Ag,14P0, in the temperature range 4 to 79 “C

tivity of ionic solids varies as

o--exp(--&) 1 (-h), T 2kT

where T is the absolute temperature, H , the enthalpy of formation of a defect pair of either Frenkel- or Schottky-type, h , the activation enthalpy for migration of mobile species, and k the Boltzmann constant.

The silver ion electrolyte Ag,14P0, like some others - e.g., Ag.,RbI, [1, 21, a-Ag,SI

- I,, [28, 291, etc. - has unusually high ionic conductivity even a t room temperature. The large conductivity of silver ion based conductors is attributed to their unusual crystal structure. X-ray crystal structure analysis of a number of such compounds, e.g. AgI(a), Ag,SI(a), Ag,RbI,, had been reported by Wiedersich and Geller [30], and of Ag13(Me4N),I,, by Geller and Lind [31]. Their analysis of structures of these com- pounds shows that mobile Ag+ ions are statistically distributed in a disordered way among the lattice sites available in the nonconducting lattice frame. Far more lattice sites than silver ions per unit cell are available. This type of structure is called average structure. The conductivity of these solids is not only high even a t room temperature, but varies very slowly with change of temperature. I n other words, this means that the number of ions available for the conduction process are not only large, but are fixed (almost) as well. I n its own turn, this means that Ag+ ions exist in a “liquid-like” [20, 321 or “free-ion-like” [33] state and all ions may be considered to be available for conduction.

Though the crystal structure of t,he electrolyte Ag,I,PO, is not yet known, i t seems reasonable to assume a similar average structure, which is compatible with its high ionic conductivity even a t room temperature. As a consequence, the concept of thermally generated defects in such solids (and so in Ag,I,PO,) vanishes, and in equation (1) we take H , = 0. Thus, the conductivity expression of these compounds may be written as

~ 3 , 2 4 1 , A~,I,WO, [8,25,261, A~,(c,H,NH)I, ~271, A ~ ~ ~ [ ( c H ~ ) ~ N I ~ I ~ ~ , A~,,[(C~H~),NI,.

or

This means that a lg (oT) against (1 /T) plot should be a straight line with a slope equal to ( -h , /k ) instead of being dependent on ($ H , + h,) as is in the case of so-called con- ventional ionic conductors (e.g. alkali halides, silver halides, etc.). We have plotted lg (oT) against (103/T), as shown in Fig. 4. The plot is linear within the temperature

504 M. T. EL-GEMAL, M. SALEEM, and M. N. AVASTHI

,

range of interest (4 to 79 "C). The result can be expressed by the following Nernst- Einstein-type equation :

ig ( U T ) = -0.21 (103/ i i~) + 4.19. (4) From the above, the activation energy h, of the Ag,I,PO, electrolyte was estimated

to be 0.21 eV, and the pre-exponential term a, = 4.6 cm-l. The direct Arrhenius plot yields h& = 0.18 eV and a,, = 4.5 l2-l cm-l and can be expressed by the equa- tion

lg cs = -0.18(103/kT) + 1.2 . ( 5 ) The value can fairly well be compared with the value reported by Takahashi's [16] Arrhenius activation energy, 0.165 eV, the only other work available for comparison.

I n this solid electrolyte, as in case of others [34], u is high even a t room temperature, which is interpreted as due to the peculiar (average) crystal structure [8, 28, 35 to 371, which contains a large number of equivalent vacant lattice sites as compared to the smaller number of Ag+ ions available for charge transport. Another important aspect (often ignored) is that a depends only weakly on temperature for all such ionic con- ductors. This may be due to the cation mobility being temperature-dependent.

As described earlier, the excess of the number of lattice sites over the number of Ag+ ions available for conduction (i.e. lattice disorder) is one of the essential reasons for the high conductivity of superionic conductors, including Ag,I,PO,. Direct evidence of lattice disorder comes from X-ray analysis. I n absence of X-ray analysis of Ag,I,PO,, we have tried to estimate the disorder qualitatively as follows.

I n Fig. 5, for several superionic conductors (for which n/n& is available in litera- ture) we have plotted their activation energy against (nlna;), where n is the number of lattice sites available for nA2 silver ions per unit cell. The ratio n/nA; may be taken to be the quantity which gives the number of lattice sites available per conducting silver ion per unit cell, and to be a measure of structural disorder in the lattice (see Table 2). The activation energy is found to vary linearly with n/nA;, and the equation of the straight line [34] may be written as

h, = 0.63 - 0.14 n/nA;: eV . (6) The above relation does not hold good for a-AgI. The reason seems to be that k T a t

room temperature is 0.026 eV and t,he activation energy cannot be less than this. More- over it is meaningless to plot the a-T dependence if the activation energy is of the same order as k T [22]. We have used this plot to obtain the value of n/nA+, a measure

of disorder of the lattice, in absence of any such data from X-ray analysis. For the activation energy of Ag,T4P0, = 0.21 eV, n/nA+g N 3, which is compatible with the high ionic conductivity of this material even at room temperature.

03

order n/n& (n and n& being, respectively, the number of

Ag4Rb15 0 1 aAgr1 a-Ag3SI Fig. 5. Activation energy h, as function of structural dis-

506 M. T. EL-GEMAL, M. SALEEM, and M. N. AVASTRI

In Table 2 we have compared electrical conductivities and activation energies of some of the silver ionic conductors. Their activation energies estimated from radio- active tracer techniques have been included. It is noteworthy that the activation ener- gies obtained from this technique are very close to those obtained from the lg (aT) versus ( l / T ) plots, i.e. Nernst-Einstein plots. We conclude that our results can be con- firmed through diffusion experiments. Work in this direction is in progress in this laboratory.

Aclnrowledgement

The authors acknowledge with thanks the laboratory facilities provided by Dr. S. Swe- dan, Chairman of the Department of Physics, Al-Fateh University, Tripoli.

References

[I] J. N. BRADLEY and P. D. GREENE, Trans. Faraday SOC. 62, 2069 (1966). [2] B. B. OWENS and G. R. ARGUE, Science (Washington) 167, 308 (1967). [3] K. SHAHI and 8. CHANDRA, J. Phys. C 9, 3105 (1976). [4] B. B. OWENS and G . R. ARGUE, J. Electrochem. SOC. 117, 898 (1970). [5] J. N. BRADLEY and P. D. GREENE, Trans. Faraday SOC. 63,424 (1967). [6] T. TAKAHASHI, 0. YAMAMOTO, and S. IKEDA, Denki Kagaku 37, 843 (1969). [7] F. P. BUNDY, M. J. MOORE, and J. S. KASPER, General Electric Res. Developm. Rep. 70-C-325

[8] K. SHAHI and 8. CHANDRA, phys. stat. sol. (a) 28, 653 (1975). [9] T. TAKAHASHI, 8. IKEDA, and 0. YAMAMOTO, J. Electrochem. SOC. 120, 647 (1973).

(1970).

[lo] S. CHANDRA and V. K. MOHABEY, J. Phys. D 8, 576 (1975). [ll] T. TAKAHASHI, J. appl. Electrochem. 3, 79 (1973). [I21 W. VAN GOOL, in: Fast Transport in Solids, Ed. W. VAN GOOL, North-Holland Publ. Co.

[13] R. T. FOLEY, cited in T. TAKAHASHI, in: Physics of Electrolytes, Vol. 2, Ed. J. HLADIE,

[14] R. T. FOLEY, J. Electrochem. SOC. 116, 13C (1969). [15] J. H. KENNEDY, see [13] (p. 931). [16] T. TARAHASHT, 8. IKEDA, and 0. YAMAMOTO, J. Electrochem. SOC. 119.477 (1972). [I71 J. E. OXLEY and J. R. HUMPHREY, Atom Internat. Rep. Conf., Evanston, July 22 t o October

[18] T. TAKAHASHI and 0. YAMAMOTO, J. Electrochem. SOC. 117, 1 (1970). [I91 D. 0. RALEIQH, J. appl. Phys. 41, 1876 (1970). [20] D. 0. RALEIGH in: Electrode Processes in Solid Electrolyte Systems in Electroanalytical

[21] B. SCROSATI, G. GERMANO, and G. PISLOIA, J. Electrochem. SOC. 118, 86 (1971). [22] A. B. LIDIARD, in: Hdb. Phys., Vol. XX, Springer-Verlag, Berlin 1957 (p. 246). [23] B. REUTER and K. HARDEL, Ber. Bunsenges. phys. Chem. 70, 82 (1966). [24] B. REUTER and K. HARDEL, Naturwissenschaften 48, 161 (1961). [25] G. CHIODELLI, A. MAOISTRIS, and A. SCHIRALDI, Electrochim. Acta (London) 19, 655 (1971). [26] T. TAKAHASHI, S. IREDA, and 0. YAMAMOTO, J. Electrochem. SOC. 120, 647 (1973). [27] S. GELLER and B. B. OWENS, J. Phys. Chem. Solids 33,1241 (1972). [28] K. SHAHI and S. CHANDRA, J. Phys. C 8, 2255 (1975). [29] B. B. OWENS, J. Electrochem. SOC. 117, 1536 (1970). [30] H. WIEDERSICH and S. GELLER, in: The Chemistry of Extended Defects in Non-Metallic

[31] S. GELLER and M. D. LIND, J. chem. Phys. 62, 5854 (1970). [32] L. W. STROCK, Z. phys. Chem. BEG, 441 (1934). [33] M. J. RICE and W. L. RQTH, J. Solid State Chem. 4, 294 (1972).

Amsterdam 1973.

Academic Press, London 1972 (p. 960).

22, 1968.

Chemistry: a Series of Advances, Ed. Q. BARD, Springer-Verlag, Berlin 1972.

Solids, Ed. L. ERYING and M. O'KEEFFE, North-Holland Publ. Co., Amsterdam 1970.

[34] K. SHAHI, Transport Studies on Superionic Conductors (Rev.. Art.), phys. stat. sol. (a) 41, 11 (1977).

[35] K. SHAG and N. K. SANYAL, J. Phys. Chem. Solids 36, 1349 (1975).

Ionic Conductivity of Ag71,P0, Solid Electrolyte 507

[36] D. 0. RALEIGH, J. Electrochem. SOC. 124, 1157 (1977). [37] D. 0. RALEIGH, in: Topics in Pure and Applied Electrochemistry, Ed. S. K. RANGARAJAN,

[38] K. E. JOHNSON, S. J. SIME. and J. DUDLEY. J. Chem. SOC. (Faraday Trans. I) 68,2015 (1972). [39] D. 0. RALEIGH, in: Progress in Solid State Chemistry, Vol. 3, Ed. H. REISS, Pergamon Press,

[40] M. LAZZARI, B. SCROSATI, and C. A. VINCENT, Electrochim. Acta (London) 22, 51 (1977). [41] C. TUBANDT and E. LORENZ, Z. phys. Chem. 8 i , 513 (1914). [42] L. SUCHOW and G. R. POND, J. Amer. Chem. SOC. i 6 , 5243 (1953). [43] R. WEIL and A. W. LAWSON. J. chem. Phys. 41, 832 (1964). [44] H. OKAZARI, J. Phys. SOC. Japan 23, 355 (1967). [45] S. GELLER, P. M. SKARSTAD, and S. A. WILBER, J. Electrochem. SOC. 122, 333 (1975). [46] P. JORDAN and M. POCHON, Helv. phys. Acta 30, 33 (1957). [47] K. E. ZIMEN, G. JOHNSON, and M. HILLART, J. Chem. SOC. (Suppl.) 2, S392 (1949). [48] R. L. ALLEN and W. J. MOORE, 5. phys. Chem. 63, 223 (1959). [49] G. G. BENTLE, J. appl. Phys. 39, 4036 (1968). [50] J. A. A. KETELAAR, Z. Krist. ABi, 434 (1934). [51] S. GELLER, see [I21 (p. 607).

SAEST Commemor. Publication, Karaikudi (India) 1975 (p. 105).

Oxford 1966 (p. 85).

(Received November 6 , 1979)